Merge branch 'tb/commit-graph-no-check-oids' into master
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "config.h"
6 #include "tempfile.h"
7 #include "quote.h"
8 #include "diff.h"
9 #include "diffcore.h"
10 #include "delta.h"
11 #include "xdiff-interface.h"
12 #include "color.h"
13 #include "attr.h"
14 #include "run-command.h"
15 #include "utf8.h"
16 #include "object-store.h"
17 #include "userdiff.h"
18 #include "submodule-config.h"
19 #include "submodule.h"
20 #include "hashmap.h"
21 #include "ll-merge.h"
22 #include "string-list.h"
23 #include "argv-array.h"
24 #include "graph.h"
25 #include "packfile.h"
26 #include "parse-options.h"
27 #include "help.h"
28 #include "promisor-remote.h"
29
30 #ifdef NO_FAST_WORKING_DIRECTORY
31 #define FAST_WORKING_DIRECTORY 0
32 #else
33 #define FAST_WORKING_DIRECTORY 1
34 #endif
35
36 static int diff_detect_rename_default;
37 static int diff_indent_heuristic = 1;
38 static int diff_rename_limit_default = 400;
39 static int diff_suppress_blank_empty;
40 static int diff_use_color_default = -1;
41 static int diff_color_moved_default;
42 static int diff_color_moved_ws_default;
43 static int diff_context_default = 3;
44 static int diff_interhunk_context_default;
45 static const char *diff_word_regex_cfg;
46 static const char *external_diff_cmd_cfg;
47 static const char *diff_order_file_cfg;
48 int diff_auto_refresh_index = 1;
49 static int diff_mnemonic_prefix;
50 static int diff_no_prefix;
51 static int diff_relative;
52 static int diff_stat_graph_width;
53 static int diff_dirstat_permille_default = 30;
54 static struct diff_options default_diff_options;
55 static long diff_algorithm;
56 static unsigned ws_error_highlight_default = WSEH_NEW;
57
58 static char diff_colors[][COLOR_MAXLEN] = {
59         GIT_COLOR_RESET,
60         GIT_COLOR_NORMAL,       /* CONTEXT */
61         GIT_COLOR_BOLD,         /* METAINFO */
62         GIT_COLOR_CYAN,         /* FRAGINFO */
63         GIT_COLOR_RED,          /* OLD */
64         GIT_COLOR_GREEN,        /* NEW */
65         GIT_COLOR_YELLOW,       /* COMMIT */
66         GIT_COLOR_BG_RED,       /* WHITESPACE */
67         GIT_COLOR_NORMAL,       /* FUNCINFO */
68         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
69         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
70         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
71         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
72         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
73         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
74         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
75         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
76         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
77         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
78         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
79         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
80         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
81         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
82 };
83
84 static const char *color_diff_slots[] = {
85         [DIFF_CONTEXT]                = "context",
86         [DIFF_METAINFO]               = "meta",
87         [DIFF_FRAGINFO]               = "frag",
88         [DIFF_FILE_OLD]               = "old",
89         [DIFF_FILE_NEW]               = "new",
90         [DIFF_COMMIT]                 = "commit",
91         [DIFF_WHITESPACE]             = "whitespace",
92         [DIFF_FUNCINFO]               = "func",
93         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
94         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
95         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
96         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
97         [DIFF_FILE_NEW_MOVED]         = "newMoved",
98         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
99         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
100         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
101         [DIFF_CONTEXT_DIM]            = "contextDimmed",
102         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
103         [DIFF_FILE_NEW_DIM]           = "newDimmed",
104         [DIFF_CONTEXT_BOLD]           = "contextBold",
105         [DIFF_FILE_OLD_BOLD]          = "oldBold",
106         [DIFF_FILE_NEW_BOLD]          = "newBold",
107 };
108
109 define_list_config_array_extra(color_diff_slots, {"plain"});
110
111 static int parse_diff_color_slot(const char *var)
112 {
113         if (!strcasecmp(var, "plain"))
114                 return DIFF_CONTEXT;
115         return LOOKUP_CONFIG(color_diff_slots, var);
116 }
117
118 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
119                                 struct strbuf *errmsg)
120 {
121         char *params_copy = xstrdup(params_string);
122         struct string_list params = STRING_LIST_INIT_NODUP;
123         int ret = 0;
124         int i;
125
126         if (*params_copy)
127                 string_list_split_in_place(&params, params_copy, ',', -1);
128         for (i = 0; i < params.nr; i++) {
129                 const char *p = params.items[i].string;
130                 if (!strcmp(p, "changes")) {
131                         options->flags.dirstat_by_line = 0;
132                         options->flags.dirstat_by_file = 0;
133                 } else if (!strcmp(p, "lines")) {
134                         options->flags.dirstat_by_line = 1;
135                         options->flags.dirstat_by_file = 0;
136                 } else if (!strcmp(p, "files")) {
137                         options->flags.dirstat_by_line = 0;
138                         options->flags.dirstat_by_file = 1;
139                 } else if (!strcmp(p, "noncumulative")) {
140                         options->flags.dirstat_cumulative = 0;
141                 } else if (!strcmp(p, "cumulative")) {
142                         options->flags.dirstat_cumulative = 1;
143                 } else if (isdigit(*p)) {
144                         char *end;
145                         int permille = strtoul(p, &end, 10) * 10;
146                         if (*end == '.' && isdigit(*++end)) {
147                                 /* only use first digit */
148                                 permille += *end - '0';
149                                 /* .. and ignore any further digits */
150                                 while (isdigit(*++end))
151                                         ; /* nothing */
152                         }
153                         if (!*end)
154                                 options->dirstat_permille = permille;
155                         else {
156                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
157                                             p);
158                                 ret++;
159                         }
160                 } else {
161                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
162                         ret++;
163                 }
164
165         }
166         string_list_clear(&params, 0);
167         free(params_copy);
168         return ret;
169 }
170
171 static int parse_submodule_params(struct diff_options *options, const char *value)
172 {
173         if (!strcmp(value, "log"))
174                 options->submodule_format = DIFF_SUBMODULE_LOG;
175         else if (!strcmp(value, "short"))
176                 options->submodule_format = DIFF_SUBMODULE_SHORT;
177         else if (!strcmp(value, "diff"))
178                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
179         /*
180          * Please update $__git_diff_submodule_formats in
181          * git-completion.bash when you add new formats.
182          */
183         else
184                 return -1;
185         return 0;
186 }
187
188 int git_config_rename(const char *var, const char *value)
189 {
190         if (!value)
191                 return DIFF_DETECT_RENAME;
192         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
193                 return  DIFF_DETECT_COPY;
194         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
195 }
196
197 long parse_algorithm_value(const char *value)
198 {
199         if (!value)
200                 return -1;
201         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
202                 return 0;
203         else if (!strcasecmp(value, "minimal"))
204                 return XDF_NEED_MINIMAL;
205         else if (!strcasecmp(value, "patience"))
206                 return XDF_PATIENCE_DIFF;
207         else if (!strcasecmp(value, "histogram"))
208                 return XDF_HISTOGRAM_DIFF;
209         /*
210          * Please update $__git_diff_algorithms in git-completion.bash
211          * when you add new algorithms.
212          */
213         return -1;
214 }
215
216 static int parse_one_token(const char **arg, const char *token)
217 {
218         const char *rest;
219         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
220                 *arg = rest;
221                 return 1;
222         }
223         return 0;
224 }
225
226 static int parse_ws_error_highlight(const char *arg)
227 {
228         const char *orig_arg = arg;
229         unsigned val = 0;
230
231         while (*arg) {
232                 if (parse_one_token(&arg, "none"))
233                         val = 0;
234                 else if (parse_one_token(&arg, "default"))
235                         val = WSEH_NEW;
236                 else if (parse_one_token(&arg, "all"))
237                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
238                 else if (parse_one_token(&arg, "new"))
239                         val |= WSEH_NEW;
240                 else if (parse_one_token(&arg, "old"))
241                         val |= WSEH_OLD;
242                 else if (parse_one_token(&arg, "context"))
243                         val |= WSEH_CONTEXT;
244                 else {
245                         return -1 - (int)(arg - orig_arg);
246                 }
247                 if (*arg)
248                         arg++;
249         }
250         return val;
251 }
252
253 /*
254  * These are to give UI layer defaults.
255  * The core-level commands such as git-diff-files should
256  * never be affected by the setting of diff.renames
257  * the user happens to have in the configuration file.
258  */
259 void init_diff_ui_defaults(void)
260 {
261         diff_detect_rename_default = DIFF_DETECT_RENAME;
262 }
263
264 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
265 {
266         if (!strcmp(var, "diff.indentheuristic"))
267                 diff_indent_heuristic = git_config_bool(var, value);
268         return 0;
269 }
270
271 static int parse_color_moved(const char *arg)
272 {
273         switch (git_parse_maybe_bool(arg)) {
274         case 0:
275                 return COLOR_MOVED_NO;
276         case 1:
277                 return COLOR_MOVED_DEFAULT;
278         default:
279                 break;
280         }
281
282         if (!strcmp(arg, "no"))
283                 return COLOR_MOVED_NO;
284         else if (!strcmp(arg, "plain"))
285                 return COLOR_MOVED_PLAIN;
286         else if (!strcmp(arg, "blocks"))
287                 return COLOR_MOVED_BLOCKS;
288         else if (!strcmp(arg, "zebra"))
289                 return COLOR_MOVED_ZEBRA;
290         else if (!strcmp(arg, "default"))
291                 return COLOR_MOVED_DEFAULT;
292         else if (!strcmp(arg, "dimmed-zebra"))
293                 return COLOR_MOVED_ZEBRA_DIM;
294         else if (!strcmp(arg, "dimmed_zebra"))
295                 return COLOR_MOVED_ZEBRA_DIM;
296         else
297                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
298 }
299
300 static unsigned parse_color_moved_ws(const char *arg)
301 {
302         int ret = 0;
303         struct string_list l = STRING_LIST_INIT_DUP;
304         struct string_list_item *i;
305
306         string_list_split(&l, arg, ',', -1);
307
308         for_each_string_list_item(i, &l) {
309                 struct strbuf sb = STRBUF_INIT;
310                 strbuf_addstr(&sb, i->string);
311                 strbuf_trim(&sb);
312
313                 if (!strcmp(sb.buf, "no"))
314                         ret = 0;
315                 else if (!strcmp(sb.buf, "ignore-space-change"))
316                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
317                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
318                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
319                 else if (!strcmp(sb.buf, "ignore-all-space"))
320                         ret |= XDF_IGNORE_WHITESPACE;
321                 else if (!strcmp(sb.buf, "allow-indentation-change"))
322                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
323                 else {
324                         ret |= COLOR_MOVED_WS_ERROR;
325                         error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
326                 }
327
328                 strbuf_release(&sb);
329         }
330
331         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
332             (ret & XDF_WHITESPACE_FLAGS)) {
333                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
334                 ret |= COLOR_MOVED_WS_ERROR;
335         }
336
337         string_list_clear(&l, 0);
338
339         return ret;
340 }
341
342 int git_diff_ui_config(const char *var, const char *value, void *cb)
343 {
344         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
345                 diff_use_color_default = git_config_colorbool(var, value);
346                 return 0;
347         }
348         if (!strcmp(var, "diff.colormoved")) {
349                 int cm = parse_color_moved(value);
350                 if (cm < 0)
351                         return -1;
352                 diff_color_moved_default = cm;
353                 return 0;
354         }
355         if (!strcmp(var, "diff.colormovedws")) {
356                 unsigned cm = parse_color_moved_ws(value);
357                 if (cm & COLOR_MOVED_WS_ERROR)
358                         return -1;
359                 diff_color_moved_ws_default = cm;
360                 return 0;
361         }
362         if (!strcmp(var, "diff.context")) {
363                 diff_context_default = git_config_int(var, value);
364                 if (diff_context_default < 0)
365                         return -1;
366                 return 0;
367         }
368         if (!strcmp(var, "diff.interhunkcontext")) {
369                 diff_interhunk_context_default = git_config_int(var, value);
370                 if (diff_interhunk_context_default < 0)
371                         return -1;
372                 return 0;
373         }
374         if (!strcmp(var, "diff.renames")) {
375                 diff_detect_rename_default = git_config_rename(var, value);
376                 return 0;
377         }
378         if (!strcmp(var, "diff.autorefreshindex")) {
379                 diff_auto_refresh_index = git_config_bool(var, value);
380                 return 0;
381         }
382         if (!strcmp(var, "diff.mnemonicprefix")) {
383                 diff_mnemonic_prefix = git_config_bool(var, value);
384                 return 0;
385         }
386         if (!strcmp(var, "diff.noprefix")) {
387                 diff_no_prefix = git_config_bool(var, value);
388                 return 0;
389         }
390         if (!strcmp(var, "diff.relative")) {
391                 diff_relative = git_config_bool(var, value);
392                 return 0;
393         }
394         if (!strcmp(var, "diff.statgraphwidth")) {
395                 diff_stat_graph_width = git_config_int(var, value);
396                 return 0;
397         }
398         if (!strcmp(var, "diff.external"))
399                 return git_config_string(&external_diff_cmd_cfg, var, value);
400         if (!strcmp(var, "diff.wordregex"))
401                 return git_config_string(&diff_word_regex_cfg, var, value);
402         if (!strcmp(var, "diff.orderfile"))
403                 return git_config_pathname(&diff_order_file_cfg, var, value);
404
405         if (!strcmp(var, "diff.ignoresubmodules"))
406                 handle_ignore_submodules_arg(&default_diff_options, value);
407
408         if (!strcmp(var, "diff.submodule")) {
409                 if (parse_submodule_params(&default_diff_options, value))
410                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
411                                 value);
412                 return 0;
413         }
414
415         if (!strcmp(var, "diff.algorithm")) {
416                 diff_algorithm = parse_algorithm_value(value);
417                 if (diff_algorithm < 0)
418                         return -1;
419                 return 0;
420         }
421
422         if (git_color_config(var, value, cb) < 0)
423                 return -1;
424
425         return git_diff_basic_config(var, value, cb);
426 }
427
428 int git_diff_basic_config(const char *var, const char *value, void *cb)
429 {
430         const char *name;
431
432         if (!strcmp(var, "diff.renamelimit")) {
433                 diff_rename_limit_default = git_config_int(var, value);
434                 return 0;
435         }
436
437         if (userdiff_config(var, value) < 0)
438                 return -1;
439
440         if (skip_prefix(var, "diff.color.", &name) ||
441             skip_prefix(var, "color.diff.", &name)) {
442                 int slot = parse_diff_color_slot(name);
443                 if (slot < 0)
444                         return 0;
445                 if (!value)
446                         return config_error_nonbool(var);
447                 return color_parse(value, diff_colors[slot]);
448         }
449
450         if (!strcmp(var, "diff.wserrorhighlight")) {
451                 int val = parse_ws_error_highlight(value);
452                 if (val < 0)
453                         return -1;
454                 ws_error_highlight_default = val;
455                 return 0;
456         }
457
458         /* like GNU diff's --suppress-blank-empty option  */
459         if (!strcmp(var, "diff.suppressblankempty") ||
460                         /* for backwards compatibility */
461                         !strcmp(var, "diff.suppress-blank-empty")) {
462                 diff_suppress_blank_empty = git_config_bool(var, value);
463                 return 0;
464         }
465
466         if (!strcmp(var, "diff.dirstat")) {
467                 struct strbuf errmsg = STRBUF_INIT;
468                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
469                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
470                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
471                                 errmsg.buf);
472                 strbuf_release(&errmsg);
473                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
474                 return 0;
475         }
476
477         if (git_diff_heuristic_config(var, value, cb) < 0)
478                 return -1;
479
480         return git_default_config(var, value, cb);
481 }
482
483 static char *quote_two(const char *one, const char *two)
484 {
485         int need_one = quote_c_style(one, NULL, NULL, 1);
486         int need_two = quote_c_style(two, NULL, NULL, 1);
487         struct strbuf res = STRBUF_INIT;
488
489         if (need_one + need_two) {
490                 strbuf_addch(&res, '"');
491                 quote_c_style(one, &res, NULL, 1);
492                 quote_c_style(two, &res, NULL, 1);
493                 strbuf_addch(&res, '"');
494         } else {
495                 strbuf_addstr(&res, one);
496                 strbuf_addstr(&res, two);
497         }
498         return strbuf_detach(&res, NULL);
499 }
500
501 static const char *external_diff(void)
502 {
503         static const char *external_diff_cmd = NULL;
504         static int done_preparing = 0;
505
506         if (done_preparing)
507                 return external_diff_cmd;
508         external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
509         if (!external_diff_cmd)
510                 external_diff_cmd = external_diff_cmd_cfg;
511         done_preparing = 1;
512         return external_diff_cmd;
513 }
514
515 /*
516  * Keep track of files used for diffing. Sometimes such an entry
517  * refers to a temporary file, sometimes to an existing file, and
518  * sometimes to "/dev/null".
519  */
520 static struct diff_tempfile {
521         /*
522          * filename external diff should read from, or NULL if this
523          * entry is currently not in use:
524          */
525         const char *name;
526
527         char hex[GIT_MAX_HEXSZ + 1];
528         char mode[10];
529
530         /*
531          * If this diff_tempfile instance refers to a temporary file,
532          * this tempfile object is used to manage its lifetime.
533          */
534         struct tempfile *tempfile;
535 } diff_temp[2];
536
537 struct emit_callback {
538         int color_diff;
539         unsigned ws_rule;
540         int blank_at_eof_in_preimage;
541         int blank_at_eof_in_postimage;
542         int lno_in_preimage;
543         int lno_in_postimage;
544         const char **label_path;
545         struct diff_words_data *diff_words;
546         struct diff_options *opt;
547         struct strbuf *header;
548 };
549
550 static int count_lines(const char *data, int size)
551 {
552         int count, ch, completely_empty = 1, nl_just_seen = 0;
553         count = 0;
554         while (0 < size--) {
555                 ch = *data++;
556                 if (ch == '\n') {
557                         count++;
558                         nl_just_seen = 1;
559                         completely_empty = 0;
560                 }
561                 else {
562                         nl_just_seen = 0;
563                         completely_empty = 0;
564                 }
565         }
566         if (completely_empty)
567                 return 0;
568         if (!nl_just_seen)
569                 count++; /* no trailing newline */
570         return count;
571 }
572
573 static int fill_mmfile(struct repository *r, mmfile_t *mf,
574                        struct diff_filespec *one)
575 {
576         if (!DIFF_FILE_VALID(one)) {
577                 mf->ptr = (char *)""; /* does not matter */
578                 mf->size = 0;
579                 return 0;
580         }
581         else if (diff_populate_filespec(r, one, NULL))
582                 return -1;
583
584         mf->ptr = one->data;
585         mf->size = one->size;
586         return 0;
587 }
588
589 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
590 static unsigned long diff_filespec_size(struct repository *r,
591                                         struct diff_filespec *one)
592 {
593         struct diff_populate_filespec_options dpf_options = {
594                 .check_size_only = 1,
595         };
596
597         if (!DIFF_FILE_VALID(one))
598                 return 0;
599         diff_populate_filespec(r, one, &dpf_options);
600         return one->size;
601 }
602
603 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
604 {
605         char *ptr = mf->ptr;
606         long size = mf->size;
607         int cnt = 0;
608
609         if (!size)
610                 return cnt;
611         ptr += size - 1; /* pointing at the very end */
612         if (*ptr != '\n')
613                 ; /* incomplete line */
614         else
615                 ptr--; /* skip the last LF */
616         while (mf->ptr < ptr) {
617                 char *prev_eol;
618                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
619                         if (*prev_eol == '\n')
620                                 break;
621                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
622                         break;
623                 cnt++;
624                 ptr = prev_eol - 1;
625         }
626         return cnt;
627 }
628
629 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
630                                struct emit_callback *ecbdata)
631 {
632         int l1, l2, at;
633         unsigned ws_rule = ecbdata->ws_rule;
634         l1 = count_trailing_blank(mf1, ws_rule);
635         l2 = count_trailing_blank(mf2, ws_rule);
636         if (l2 <= l1) {
637                 ecbdata->blank_at_eof_in_preimage = 0;
638                 ecbdata->blank_at_eof_in_postimage = 0;
639                 return;
640         }
641         at = count_lines(mf1->ptr, mf1->size);
642         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
643
644         at = count_lines(mf2->ptr, mf2->size);
645         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
646 }
647
648 static void emit_line_0(struct diff_options *o,
649                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
650                         int first, const char *line, int len)
651 {
652         int has_trailing_newline, has_trailing_carriage_return;
653         int needs_reset = 0; /* at the end of the line */
654         FILE *file = o->file;
655
656         fputs(diff_line_prefix(o), file);
657
658         has_trailing_newline = (len > 0 && line[len-1] == '\n');
659         if (has_trailing_newline)
660                 len--;
661
662         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
663         if (has_trailing_carriage_return)
664                 len--;
665
666         if (!len && !first)
667                 goto end_of_line;
668
669         if (reverse && want_color(o->use_color)) {
670                 fputs(GIT_COLOR_REVERSE, file);
671                 needs_reset = 1;
672         }
673
674         if (set_sign) {
675                 fputs(set_sign, file);
676                 needs_reset = 1;
677         }
678
679         if (first)
680                 fputc(first, file);
681
682         if (!len)
683                 goto end_of_line;
684
685         if (set) {
686                 if (set_sign && set != set_sign)
687                         fputs(reset, file);
688                 fputs(set, file);
689                 needs_reset = 1;
690         }
691         fwrite(line, len, 1, file);
692         needs_reset = 1; /* 'line' may contain color codes. */
693
694 end_of_line:
695         if (needs_reset)
696                 fputs(reset, file);
697         if (has_trailing_carriage_return)
698                 fputc('\r', file);
699         if (has_trailing_newline)
700                 fputc('\n', file);
701 }
702
703 static void emit_line(struct diff_options *o, const char *set, const char *reset,
704                       const char *line, int len)
705 {
706         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
707 }
708
709 enum diff_symbol {
710         DIFF_SYMBOL_BINARY_DIFF_HEADER,
711         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
712         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
713         DIFF_SYMBOL_BINARY_DIFF_BODY,
714         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
715         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
716         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
717         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
718         DIFF_SYMBOL_STATS_LINE,
719         DIFF_SYMBOL_WORD_DIFF,
720         DIFF_SYMBOL_STAT_SEP,
721         DIFF_SYMBOL_SUMMARY,
722         DIFF_SYMBOL_SUBMODULE_ADD,
723         DIFF_SYMBOL_SUBMODULE_DEL,
724         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
725         DIFF_SYMBOL_SUBMODULE_MODIFIED,
726         DIFF_SYMBOL_SUBMODULE_HEADER,
727         DIFF_SYMBOL_SUBMODULE_ERROR,
728         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
729         DIFF_SYMBOL_REWRITE_DIFF,
730         DIFF_SYMBOL_BINARY_FILES,
731         DIFF_SYMBOL_HEADER,
732         DIFF_SYMBOL_FILEPAIR_PLUS,
733         DIFF_SYMBOL_FILEPAIR_MINUS,
734         DIFF_SYMBOL_WORDS_PORCELAIN,
735         DIFF_SYMBOL_WORDS,
736         DIFF_SYMBOL_CONTEXT,
737         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
738         DIFF_SYMBOL_PLUS,
739         DIFF_SYMBOL_MINUS,
740         DIFF_SYMBOL_NO_LF_EOF,
741         DIFF_SYMBOL_CONTEXT_FRAGINFO,
742         DIFF_SYMBOL_CONTEXT_MARKER,
743         DIFF_SYMBOL_SEPARATOR
744 };
745 /*
746  * Flags for content lines:
747  * 0..12 are whitespace rules
748  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
749  * 16 is marking if the line is blank at EOF
750  */
751 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
752 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
753 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
754 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
755 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
756
757 /*
758  * This struct is used when we need to buffer the output of the diff output.
759  *
760  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
761  * into the pre/post image file. This pointer could be a union with the
762  * line pointer. By storing an offset into the file instead of the literal line,
763  * we can decrease the memory footprint for the buffered output. At first we
764  * may want to only have indirection for the content lines, but we could also
765  * enhance the state for emitting prefabricated lines, e.g. the similarity
766  * score line or hunk/file headers would only need to store a number or path
767  * and then the output can be constructed later on depending on state.
768  */
769 struct emitted_diff_symbol {
770         const char *line;
771         int len;
772         int flags;
773         int indent_off;   /* Offset to first non-whitespace character */
774         int indent_width; /* The visual width of the indentation */
775         enum diff_symbol s;
776 };
777 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
778
779 struct emitted_diff_symbols {
780         struct emitted_diff_symbol *buf;
781         int nr, alloc;
782 };
783 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
784
785 static void append_emitted_diff_symbol(struct diff_options *o,
786                                        struct emitted_diff_symbol *e)
787 {
788         struct emitted_diff_symbol *f;
789
790         ALLOC_GROW(o->emitted_symbols->buf,
791                    o->emitted_symbols->nr + 1,
792                    o->emitted_symbols->alloc);
793         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
794
795         memcpy(f, e, sizeof(struct emitted_diff_symbol));
796         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
797 }
798
799 struct moved_entry {
800         struct hashmap_entry ent;
801         const struct emitted_diff_symbol *es;
802         struct moved_entry *next_line;
803 };
804
805 struct moved_block {
806         struct moved_entry *match;
807         int wsd; /* The whitespace delta of this block */
808 };
809
810 static void moved_block_clear(struct moved_block *b)
811 {
812         memset(b, 0, sizeof(*b));
813 }
814
815 #define INDENT_BLANKLINE INT_MIN
816
817 static void fill_es_indent_data(struct emitted_diff_symbol *es)
818 {
819         unsigned int off = 0, i;
820         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
821         const char *s = es->line;
822         const int len = es->len;
823
824         /* skip any \v \f \r at start of indentation */
825         while (s[off] == '\f' || s[off] == '\v' ||
826                (s[off] == '\r' && off < len - 1))
827                 off++;
828
829         /* calculate the visual width of indentation */
830         while(1) {
831                 if (s[off] == ' ') {
832                         width++;
833                         off++;
834                 } else if (s[off] == '\t') {
835                         width += tab_width - (width % tab_width);
836                         while (s[++off] == '\t')
837                                 width += tab_width;
838                 } else {
839                         break;
840                 }
841         }
842
843         /* check if this line is blank */
844         for (i = off; i < len; i++)
845                 if (!isspace(s[i]))
846                     break;
847
848         if (i == len) {
849                 es->indent_width = INDENT_BLANKLINE;
850                 es->indent_off = len;
851         } else {
852                 es->indent_off = off;
853                 es->indent_width = width;
854         }
855 }
856
857 static int compute_ws_delta(const struct emitted_diff_symbol *a,
858                             const struct emitted_diff_symbol *b,
859                             int *out)
860 {
861         int a_len = a->len,
862             b_len = b->len,
863             a_off = a->indent_off,
864             a_width = a->indent_width,
865             b_off = b->indent_off,
866             b_width = b->indent_width;
867         int delta;
868
869         if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
870                 *out = INDENT_BLANKLINE;
871                 return 1;
872         }
873
874         if (a->s == DIFF_SYMBOL_PLUS)
875                 delta = a_width - b_width;
876         else
877                 delta = b_width - a_width;
878
879         if (a_len - a_off != b_len - b_off ||
880             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
881                 return 0;
882
883         *out = delta;
884
885         return 1;
886 }
887
888 static int cmp_in_block_with_wsd(const struct diff_options *o,
889                                  const struct moved_entry *cur,
890                                  const struct moved_entry *match,
891                                  struct moved_block *pmb,
892                                  int n)
893 {
894         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
895         int al = cur->es->len, bl = match->es->len, cl = l->len;
896         const char *a = cur->es->line,
897                    *b = match->es->line,
898                    *c = l->line;
899         int a_off = cur->es->indent_off,
900             a_width = cur->es->indent_width,
901             c_off = l->indent_off,
902             c_width = l->indent_width;
903         int delta;
904
905         /*
906          * We need to check if 'cur' is equal to 'match'.  As those
907          * are from the same (+/-) side, we do not need to adjust for
908          * indent changes. However these were found using fuzzy
909          * matching so we do have to check if they are equal. Here we
910          * just check the lengths. We delay calling memcmp() to check
911          * the contents until later as if the length comparison for a
912          * and c fails we can avoid the call all together.
913          */
914         if (al != bl)
915                 return 1;
916
917         /* If 'l' and 'cur' are both blank then they match. */
918         if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
919                 return 0;
920
921         /*
922          * The indent changes of the block are known and stored in pmb->wsd;
923          * however we need to check if the indent changes of the current line
924          * match those of the current block and that the text of 'l' and 'cur'
925          * after the indentation match.
926          */
927         if (cur->es->s == DIFF_SYMBOL_PLUS)
928                 delta = a_width - c_width;
929         else
930                 delta = c_width - a_width;
931
932         /*
933          * If the previous lines of this block were all blank then set its
934          * whitespace delta.
935          */
936         if (pmb->wsd == INDENT_BLANKLINE)
937                 pmb->wsd = delta;
938
939         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
940                  !memcmp(a, b, al) && !
941                  memcmp(a + a_off, c + c_off, al - a_off));
942 }
943
944 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
945                            const struct hashmap_entry *eptr,
946                            const struct hashmap_entry *entry_or_key,
947                            const void *keydata)
948 {
949         const struct diff_options *diffopt = hashmap_cmp_fn_data;
950         const struct moved_entry *a, *b;
951         unsigned flags = diffopt->color_moved_ws_handling
952                          & XDF_WHITESPACE_FLAGS;
953
954         a = container_of(eptr, const struct moved_entry, ent);
955         b = container_of(entry_or_key, const struct moved_entry, ent);
956
957         if (diffopt->color_moved_ws_handling &
958             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
959                 /*
960                  * As there is not specific white space config given,
961                  * we'd need to check for a new block, so ignore all
962                  * white space. The setup of the white space
963                  * configuration for the next block is done else where
964                  */
965                 flags |= XDF_IGNORE_WHITESPACE;
966
967         return !xdiff_compare_lines(a->es->line, a->es->len,
968                                     b->es->line, b->es->len,
969                                     flags);
970 }
971
972 static struct moved_entry *prepare_entry(struct diff_options *o,
973                                          int line_no)
974 {
975         struct moved_entry *ret = xmalloc(sizeof(*ret));
976         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
977         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
978         unsigned int hash = xdiff_hash_string(l->line, l->len, flags);
979
980         hashmap_entry_init(&ret->ent, hash);
981         ret->es = l;
982         ret->next_line = NULL;
983
984         return ret;
985 }
986
987 static void add_lines_to_move_detection(struct diff_options *o,
988                                         struct hashmap *add_lines,
989                                         struct hashmap *del_lines)
990 {
991         struct moved_entry *prev_line = NULL;
992
993         int n;
994         for (n = 0; n < o->emitted_symbols->nr; n++) {
995                 struct hashmap *hm;
996                 struct moved_entry *key;
997
998                 switch (o->emitted_symbols->buf[n].s) {
999                 case DIFF_SYMBOL_PLUS:
1000                         hm = add_lines;
1001                         break;
1002                 case DIFF_SYMBOL_MINUS:
1003                         hm = del_lines;
1004                         break;
1005                 default:
1006                         prev_line = NULL;
1007                         continue;
1008                 }
1009
1010                 if (o->color_moved_ws_handling &
1011                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1012                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
1013                 key = prepare_entry(o, n);
1014                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
1015                         prev_line->next_line = key;
1016
1017                 hashmap_add(hm, &key->ent);
1018                 prev_line = key;
1019         }
1020 }
1021
1022 static void pmb_advance_or_null(struct diff_options *o,
1023                                 struct moved_entry *match,
1024                                 struct hashmap *hm,
1025                                 struct moved_block *pmb,
1026                                 int pmb_nr)
1027 {
1028         int i;
1029         for (i = 0; i < pmb_nr; i++) {
1030                 struct moved_entry *prev = pmb[i].match;
1031                 struct moved_entry *cur = (prev && prev->next_line) ?
1032                                 prev->next_line : NULL;
1033                 if (cur && !hm->cmpfn(o, &cur->ent, &match->ent, NULL)) {
1034                         pmb[i].match = cur;
1035                 } else {
1036                         pmb[i].match = NULL;
1037                 }
1038         }
1039 }
1040
1041 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1042                                             struct moved_entry *match,
1043                                             struct hashmap *hm,
1044                                             struct moved_block *pmb,
1045                                             int pmb_nr, int n)
1046 {
1047         int i;
1048         char *got_match = xcalloc(1, pmb_nr);
1049
1050         hashmap_for_each_entry_from(hm, match, ent) {
1051                 for (i = 0; i < pmb_nr; i++) {
1052                         struct moved_entry *prev = pmb[i].match;
1053                         struct moved_entry *cur = (prev && prev->next_line) ?
1054                                         prev->next_line : NULL;
1055                         if (!cur)
1056                                 continue;
1057                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1058                                 got_match[i] |= 1;
1059                 }
1060         }
1061
1062         for (i = 0; i < pmb_nr; i++) {
1063                 if (got_match[i]) {
1064                         /* Advance to the next line */
1065                         pmb[i].match = pmb[i].match->next_line;
1066                 } else {
1067                         moved_block_clear(&pmb[i]);
1068                 }
1069         }
1070
1071         free(got_match);
1072 }
1073
1074 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1075                                          int pmb_nr)
1076 {
1077         int lp, rp;
1078
1079         /* Shrink the set of potential block to the remaining running */
1080         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1081                 while (lp < pmb_nr && pmb[lp].match)
1082                         lp++;
1083                 /* lp points at the first NULL now */
1084
1085                 while (rp > -1 && !pmb[rp].match)
1086                         rp--;
1087                 /* rp points at the last non-NULL */
1088
1089                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1090                         pmb[lp] = pmb[rp];
1091                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1092                         rp--;
1093                         lp++;
1094                 }
1095         }
1096
1097         /* Remember the number of running sets */
1098         return rp + 1;
1099 }
1100
1101 /*
1102  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1103  *
1104  * Otherwise, if the last block has fewer alphanumeric characters than
1105  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1106  * that block.
1107  *
1108  * The last block consists of the (n - block_length)'th line up to but not
1109  * including the nth line.
1110  *
1111  * Returns 0 if the last block is empty or is unset by this function, non zero
1112  * otherwise.
1113  *
1114  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1115  * Think of a way to unify them.
1116  */
1117 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1118 {
1119         int i, alnum_count = 0;
1120         if (o->color_moved == COLOR_MOVED_PLAIN)
1121                 return block_length;
1122         for (i = 1; i < block_length + 1; i++) {
1123                 const char *c = o->emitted_symbols->buf[n - i].line;
1124                 for (; *c; c++) {
1125                         if (!isalnum(*c))
1126                                 continue;
1127                         alnum_count++;
1128                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1129                                 return 1;
1130                 }
1131         }
1132         for (i = 1; i < block_length + 1; i++)
1133                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1134         return 0;
1135 }
1136
1137 /* Find blocks of moved code, delegate actual coloring decision to helper */
1138 static void mark_color_as_moved(struct diff_options *o,
1139                                 struct hashmap *add_lines,
1140                                 struct hashmap *del_lines)
1141 {
1142         struct moved_block *pmb = NULL; /* potentially moved blocks */
1143         int pmb_nr = 0, pmb_alloc = 0;
1144         int n, flipped_block = 0, block_length = 0;
1145
1146
1147         for (n = 0; n < o->emitted_symbols->nr; n++) {
1148                 struct hashmap *hm = NULL;
1149                 struct moved_entry *key;
1150                 struct moved_entry *match = NULL;
1151                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1152                 enum diff_symbol last_symbol = 0;
1153
1154                 switch (l->s) {
1155                 case DIFF_SYMBOL_PLUS:
1156                         hm = del_lines;
1157                         key = prepare_entry(o, n);
1158                         match = hashmap_get_entry(hm, key, ent, NULL);
1159                         free(key);
1160                         break;
1161                 case DIFF_SYMBOL_MINUS:
1162                         hm = add_lines;
1163                         key = prepare_entry(o, n);
1164                         match = hashmap_get_entry(hm, key, ent, NULL);
1165                         free(key);
1166                         break;
1167                 default:
1168                         flipped_block = 0;
1169                 }
1170
1171                 if (!match) {
1172                         int i;
1173
1174                         adjust_last_block(o, n, block_length);
1175                         for(i = 0; i < pmb_nr; i++)
1176                                 moved_block_clear(&pmb[i]);
1177                         pmb_nr = 0;
1178                         block_length = 0;
1179                         flipped_block = 0;
1180                         last_symbol = l->s;
1181                         continue;
1182                 }
1183
1184                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1185                         last_symbol = l->s;
1186                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1187                         continue;
1188                 }
1189
1190                 if (o->color_moved_ws_handling &
1191                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1192                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1193                 else
1194                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1195
1196                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1197
1198                 if (pmb_nr == 0) {
1199                         /*
1200                          * The current line is the start of a new block.
1201                          * Setup the set of potential blocks.
1202                          */
1203                         hashmap_for_each_entry_from(hm, match, ent) {
1204                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1205                                 if (o->color_moved_ws_handling &
1206                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1207                                         if (compute_ws_delta(l, match->es,
1208                                                              &pmb[pmb_nr].wsd))
1209                                                 pmb[pmb_nr++].match = match;
1210                                 } else {
1211                                         pmb[pmb_nr].wsd = 0;
1212                                         pmb[pmb_nr++].match = match;
1213                                 }
1214                         }
1215
1216                         if (adjust_last_block(o, n, block_length) &&
1217                             pmb_nr && last_symbol != l->s)
1218                                 flipped_block = (flipped_block + 1) % 2;
1219                         else
1220                                 flipped_block = 0;
1221
1222                         block_length = 0;
1223                 }
1224
1225                 if (pmb_nr) {
1226                         block_length++;
1227                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1228                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1229                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1230                 }
1231                 last_symbol = l->s;
1232         }
1233         adjust_last_block(o, n, block_length);
1234
1235         for(n = 0; n < pmb_nr; n++)
1236                 moved_block_clear(&pmb[n]);
1237         free(pmb);
1238 }
1239
1240 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1241   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1242 static void dim_moved_lines(struct diff_options *o)
1243 {
1244         int n;
1245         for (n = 0; n < o->emitted_symbols->nr; n++) {
1246                 struct emitted_diff_symbol *prev = (n != 0) ?
1247                                 &o->emitted_symbols->buf[n - 1] : NULL;
1248                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1249                 struct emitted_diff_symbol *next =
1250                                 (n < o->emitted_symbols->nr - 1) ?
1251                                 &o->emitted_symbols->buf[n + 1] : NULL;
1252
1253                 /* Not a plus or minus line? */
1254                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1255                         continue;
1256
1257                 /* Not a moved line? */
1258                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1259                         continue;
1260
1261                 /*
1262                  * If prev or next are not a plus or minus line,
1263                  * pretend they don't exist
1264                  */
1265                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1266                             prev->s != DIFF_SYMBOL_MINUS)
1267                         prev = NULL;
1268                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1269                             next->s != DIFF_SYMBOL_MINUS)
1270                         next = NULL;
1271
1272                 /* Inside a block? */
1273                 if ((prev &&
1274                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1275                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1276                     (next &&
1277                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1278                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1279                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1280                         continue;
1281                 }
1282
1283                 /* Check if we are at an interesting bound: */
1284                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1285                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1286                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1287                         continue;
1288                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1289                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1290                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1291                         continue;
1292
1293                 /*
1294                  * The boundary to prev and next are not interesting,
1295                  * so this line is not interesting as a whole
1296                  */
1297                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1298         }
1299 }
1300
1301 static void emit_line_ws_markup(struct diff_options *o,
1302                                 const char *set_sign, const char *set,
1303                                 const char *reset,
1304                                 int sign_index, const char *line, int len,
1305                                 unsigned ws_rule, int blank_at_eof)
1306 {
1307         const char *ws = NULL;
1308         int sign = o->output_indicators[sign_index];
1309
1310         if (o->ws_error_highlight & ws_rule) {
1311                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1312                 if (!*ws)
1313                         ws = NULL;
1314         }
1315
1316         if (!ws && !set_sign)
1317                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1318         else if (!ws) {
1319                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1320         } else if (blank_at_eof)
1321                 /* Blank line at EOF - paint '+' as well */
1322                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1323         else {
1324                 /* Emit just the prefix, then the rest. */
1325                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1326                             sign, "", 0);
1327                 ws_check_emit(line, len, ws_rule,
1328                               o->file, set, reset, ws);
1329         }
1330 }
1331
1332 static void emit_diff_symbol_from_struct(struct diff_options *o,
1333                                          struct emitted_diff_symbol *eds)
1334 {
1335         static const char *nneof = " No newline at end of file\n";
1336         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1337         struct strbuf sb = STRBUF_INIT;
1338
1339         enum diff_symbol s = eds->s;
1340         const char *line = eds->line;
1341         int len = eds->len;
1342         unsigned flags = eds->flags;
1343
1344         switch (s) {
1345         case DIFF_SYMBOL_NO_LF_EOF:
1346                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1347                 reset = diff_get_color_opt(o, DIFF_RESET);
1348                 putc('\n', o->file);
1349                 emit_line_0(o, context, NULL, 0, reset, '\\',
1350                             nneof, strlen(nneof));
1351                 break;
1352         case DIFF_SYMBOL_SUBMODULE_HEADER:
1353         case DIFF_SYMBOL_SUBMODULE_ERROR:
1354         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1355         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1356         case DIFF_SYMBOL_SUMMARY:
1357         case DIFF_SYMBOL_STATS_LINE:
1358         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1359         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1360                 emit_line(o, "", "", line, len);
1361                 break;
1362         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1363         case DIFF_SYMBOL_CONTEXT_MARKER:
1364                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1365                 reset = diff_get_color_opt(o, DIFF_RESET);
1366                 emit_line(o, context, reset, line, len);
1367                 break;
1368         case DIFF_SYMBOL_SEPARATOR:
1369                 fprintf(o->file, "%s%c",
1370                         diff_line_prefix(o),
1371                         o->line_termination);
1372                 break;
1373         case DIFF_SYMBOL_CONTEXT:
1374                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1375                 reset = diff_get_color_opt(o, DIFF_RESET);
1376                 set_sign = NULL;
1377                 if (o->flags.dual_color_diffed_diffs) {
1378                         char c = !len ? 0 : line[0];
1379
1380                         if (c == '+')
1381                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1382                         else if (c == '@')
1383                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1384                         else if (c == '-')
1385                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1386                 }
1387                 emit_line_ws_markup(o, set_sign, set, reset,
1388                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1389                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1390                 break;
1391         case DIFF_SYMBOL_PLUS:
1392                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1393                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1394                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1395                 case DIFF_SYMBOL_MOVED_LINE |
1396                      DIFF_SYMBOL_MOVED_LINE_ALT |
1397                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1398                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1399                         break;
1400                 case DIFF_SYMBOL_MOVED_LINE |
1401                      DIFF_SYMBOL_MOVED_LINE_ALT:
1402                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1403                         break;
1404                 case DIFF_SYMBOL_MOVED_LINE |
1405                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1406                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1407                         break;
1408                 case DIFF_SYMBOL_MOVED_LINE:
1409                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1410                         break;
1411                 default:
1412                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1413                 }
1414                 reset = diff_get_color_opt(o, DIFF_RESET);
1415                 if (!o->flags.dual_color_diffed_diffs)
1416                         set_sign = NULL;
1417                 else {
1418                         char c = !len ? 0 : line[0];
1419
1420                         set_sign = set;
1421                         if (c == '-')
1422                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1423                         else if (c == '@')
1424                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1425                         else if (c == '+')
1426                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1427                         else
1428                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1429                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1430                 }
1431                 emit_line_ws_markup(o, set_sign, set, reset,
1432                                     OUTPUT_INDICATOR_NEW, line, len,
1433                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1434                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1435                 break;
1436         case DIFF_SYMBOL_MINUS:
1437                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1438                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1439                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1440                 case DIFF_SYMBOL_MOVED_LINE |
1441                      DIFF_SYMBOL_MOVED_LINE_ALT |
1442                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1443                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1444                         break;
1445                 case DIFF_SYMBOL_MOVED_LINE |
1446                      DIFF_SYMBOL_MOVED_LINE_ALT:
1447                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1448                         break;
1449                 case DIFF_SYMBOL_MOVED_LINE |
1450                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1451                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1452                         break;
1453                 case DIFF_SYMBOL_MOVED_LINE:
1454                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1455                         break;
1456                 default:
1457                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1458                 }
1459                 reset = diff_get_color_opt(o, DIFF_RESET);
1460                 if (!o->flags.dual_color_diffed_diffs)
1461                         set_sign = NULL;
1462                 else {
1463                         char c = !len ? 0 : line[0];
1464
1465                         set_sign = set;
1466                         if (c == '+')
1467                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1468                         else if (c == '@')
1469                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1470                         else if (c == '-')
1471                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1472                         else
1473                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1474                 }
1475                 emit_line_ws_markup(o, set_sign, set, reset,
1476                                     OUTPUT_INDICATOR_OLD, line, len,
1477                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1478                 break;
1479         case DIFF_SYMBOL_WORDS_PORCELAIN:
1480                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1481                 reset = diff_get_color_opt(o, DIFF_RESET);
1482                 emit_line(o, context, reset, line, len);
1483                 fputs("~\n", o->file);
1484                 break;
1485         case DIFF_SYMBOL_WORDS:
1486                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1487                 reset = diff_get_color_opt(o, DIFF_RESET);
1488                 /*
1489                  * Skip the prefix character, if any.  With
1490                  * diff_suppress_blank_empty, there may be
1491                  * none.
1492                  */
1493                 if (line[0] != '\n') {
1494                         line++;
1495                         len--;
1496                 }
1497                 emit_line(o, context, reset, line, len);
1498                 break;
1499         case DIFF_SYMBOL_FILEPAIR_PLUS:
1500                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1501                 reset = diff_get_color_opt(o, DIFF_RESET);
1502                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1503                         line, reset,
1504                         strchr(line, ' ') ? "\t" : "");
1505                 break;
1506         case DIFF_SYMBOL_FILEPAIR_MINUS:
1507                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1508                 reset = diff_get_color_opt(o, DIFF_RESET);
1509                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1510                         line, reset,
1511                         strchr(line, ' ') ? "\t" : "");
1512                 break;
1513         case DIFF_SYMBOL_BINARY_FILES:
1514         case DIFF_SYMBOL_HEADER:
1515                 fprintf(o->file, "%s", line);
1516                 break;
1517         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1518                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1519                 break;
1520         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1521                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1522                 break;
1523         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1524                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1525                 break;
1526         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1527                 fputs(diff_line_prefix(o), o->file);
1528                 fputc('\n', o->file);
1529                 break;
1530         case DIFF_SYMBOL_REWRITE_DIFF:
1531                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1532                 reset = diff_get_color_opt(o, DIFF_RESET);
1533                 emit_line(o, fraginfo, reset, line, len);
1534                 break;
1535         case DIFF_SYMBOL_SUBMODULE_ADD:
1536                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1537                 reset = diff_get_color_opt(o, DIFF_RESET);
1538                 emit_line(o, set, reset, line, len);
1539                 break;
1540         case DIFF_SYMBOL_SUBMODULE_DEL:
1541                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1542                 reset = diff_get_color_opt(o, DIFF_RESET);
1543                 emit_line(o, set, reset, line, len);
1544                 break;
1545         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1546                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1547                         diff_line_prefix(o), line);
1548                 break;
1549         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1550                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1551                         diff_line_prefix(o), line);
1552                 break;
1553         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1554                 emit_line(o, "", "", " 0 files changed\n",
1555                           strlen(" 0 files changed\n"));
1556                 break;
1557         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1558                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1559                 break;
1560         case DIFF_SYMBOL_WORD_DIFF:
1561                 fprintf(o->file, "%.*s", len, line);
1562                 break;
1563         case DIFF_SYMBOL_STAT_SEP:
1564                 fputs(o->stat_sep, o->file);
1565                 break;
1566         default:
1567                 BUG("unknown diff symbol");
1568         }
1569         strbuf_release(&sb);
1570 }
1571
1572 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1573                              const char *line, int len, unsigned flags)
1574 {
1575         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1576
1577         if (o->emitted_symbols)
1578                 append_emitted_diff_symbol(o, &e);
1579         else
1580                 emit_diff_symbol_from_struct(o, &e);
1581 }
1582
1583 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1584 {
1585         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1586 }
1587
1588 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1589 {
1590         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1591 }
1592
1593 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1594 {
1595         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1596                          path, strlen(path), 0);
1597 }
1598
1599 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1600 {
1601         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1602                          path, strlen(path), 0);
1603 }
1604
1605 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1606 {
1607         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1608                          header, strlen(header), 0);
1609 }
1610
1611 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1612 {
1613         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1614 }
1615
1616 void diff_emit_submodule_pipethrough(struct diff_options *o,
1617                                      const char *line, int len)
1618 {
1619         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1620 }
1621
1622 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1623 {
1624         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1625               ecbdata->blank_at_eof_in_preimage &&
1626               ecbdata->blank_at_eof_in_postimage &&
1627               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1628               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1629                 return 0;
1630         return ws_blank_line(line, len, ecbdata->ws_rule);
1631 }
1632
1633 static void emit_add_line(struct emit_callback *ecbdata,
1634                           const char *line, int len)
1635 {
1636         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1637         if (new_blank_line_at_eof(ecbdata, line, len))
1638                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1639
1640         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1641 }
1642
1643 static void emit_del_line(struct emit_callback *ecbdata,
1644                           const char *line, int len)
1645 {
1646         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1647         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1648 }
1649
1650 static void emit_context_line(struct emit_callback *ecbdata,
1651                               const char *line, int len)
1652 {
1653         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1654         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1655 }
1656
1657 static void emit_hunk_header(struct emit_callback *ecbdata,
1658                              const char *line, int len)
1659 {
1660         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1661         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1662         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1663         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1664         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1665         static const char atat[2] = { '@', '@' };
1666         const char *cp, *ep;
1667         struct strbuf msgbuf = STRBUF_INIT;
1668         int org_len = len;
1669         int i = 1;
1670
1671         /*
1672          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1673          * it always is at least 10 bytes long.
1674          */
1675         if (len < 10 ||
1676             memcmp(line, atat, 2) ||
1677             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1678                 emit_diff_symbol(ecbdata->opt,
1679                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1680                 return;
1681         }
1682         ep += 2; /* skip over @@ */
1683
1684         /* The hunk header in fraginfo color */
1685         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1686                 strbuf_addstr(&msgbuf, reverse);
1687         strbuf_addstr(&msgbuf, frag);
1688         if (ecbdata->opt->flags.suppress_hunk_header_line_count)
1689                 strbuf_add(&msgbuf, atat, sizeof(atat));
1690         else
1691                 strbuf_add(&msgbuf, line, ep - line);
1692         strbuf_addstr(&msgbuf, reset);
1693
1694         /*
1695          * trailing "\r\n"
1696          */
1697         for ( ; i < 3; i++)
1698                 if (line[len - i] == '\r' || line[len - i] == '\n')
1699                         len--;
1700
1701         /* blank before the func header */
1702         for (cp = ep; ep - line < len; ep++)
1703                 if (*ep != ' ' && *ep != '\t')
1704                         break;
1705         if (ep != cp) {
1706                 strbuf_addstr(&msgbuf, context);
1707                 strbuf_add(&msgbuf, cp, ep - cp);
1708                 strbuf_addstr(&msgbuf, reset);
1709         }
1710
1711         if (ep < line + len) {
1712                 strbuf_addstr(&msgbuf, func);
1713                 strbuf_add(&msgbuf, ep, line + len - ep);
1714                 strbuf_addstr(&msgbuf, reset);
1715         }
1716
1717         strbuf_add(&msgbuf, line + len, org_len - len);
1718         strbuf_complete_line(&msgbuf);
1719         emit_diff_symbol(ecbdata->opt,
1720                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1721         strbuf_release(&msgbuf);
1722 }
1723
1724 static struct diff_tempfile *claim_diff_tempfile(void)
1725 {
1726         int i;
1727         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1728                 if (!diff_temp[i].name)
1729                         return diff_temp + i;
1730         BUG("diff is failing to clean up its tempfiles");
1731 }
1732
1733 static void remove_tempfile(void)
1734 {
1735         int i;
1736         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1737                 if (is_tempfile_active(diff_temp[i].tempfile))
1738                         delete_tempfile(&diff_temp[i].tempfile);
1739                 diff_temp[i].name = NULL;
1740         }
1741 }
1742
1743 static void add_line_count(struct strbuf *out, int count)
1744 {
1745         switch (count) {
1746         case 0:
1747                 strbuf_addstr(out, "0,0");
1748                 break;
1749         case 1:
1750                 strbuf_addstr(out, "1");
1751                 break;
1752         default:
1753                 strbuf_addf(out, "1,%d", count);
1754                 break;
1755         }
1756 }
1757
1758 static void emit_rewrite_lines(struct emit_callback *ecb,
1759                                int prefix, const char *data, int size)
1760 {
1761         const char *endp = NULL;
1762
1763         while (0 < size) {
1764                 int len;
1765
1766                 endp = memchr(data, '\n', size);
1767                 len = endp ? (endp - data + 1) : size;
1768                 if (prefix != '+') {
1769                         ecb->lno_in_preimage++;
1770                         emit_del_line(ecb, data, len);
1771                 } else {
1772                         ecb->lno_in_postimage++;
1773                         emit_add_line(ecb, data, len);
1774                 }
1775                 size -= len;
1776                 data += len;
1777         }
1778         if (!endp)
1779                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1780 }
1781
1782 static void emit_rewrite_diff(const char *name_a,
1783                               const char *name_b,
1784                               struct diff_filespec *one,
1785                               struct diff_filespec *two,
1786                               struct userdiff_driver *textconv_one,
1787                               struct userdiff_driver *textconv_two,
1788                               struct diff_options *o)
1789 {
1790         int lc_a, lc_b;
1791         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1792         const char *a_prefix, *b_prefix;
1793         char *data_one, *data_two;
1794         size_t size_one, size_two;
1795         struct emit_callback ecbdata;
1796         struct strbuf out = STRBUF_INIT;
1797
1798         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1799                 a_prefix = o->b_prefix;
1800                 b_prefix = o->a_prefix;
1801         } else {
1802                 a_prefix = o->a_prefix;
1803                 b_prefix = o->b_prefix;
1804         }
1805
1806         name_a += (*name_a == '/');
1807         name_b += (*name_b == '/');
1808
1809         strbuf_reset(&a_name);
1810         strbuf_reset(&b_name);
1811         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1812         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1813
1814         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1815         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1816
1817         memset(&ecbdata, 0, sizeof(ecbdata));
1818         ecbdata.color_diff = want_color(o->use_color);
1819         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1820         ecbdata.opt = o;
1821         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1822                 mmfile_t mf1, mf2;
1823                 mf1.ptr = (char *)data_one;
1824                 mf2.ptr = (char *)data_two;
1825                 mf1.size = size_one;
1826                 mf2.size = size_two;
1827                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1828         }
1829         ecbdata.lno_in_preimage = 1;
1830         ecbdata.lno_in_postimage = 1;
1831
1832         lc_a = count_lines(data_one, size_one);
1833         lc_b = count_lines(data_two, size_two);
1834
1835         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1836                          a_name.buf, a_name.len, 0);
1837         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1838                          b_name.buf, b_name.len, 0);
1839
1840         strbuf_addstr(&out, "@@ -");
1841         if (!o->irreversible_delete)
1842                 add_line_count(&out, lc_a);
1843         else
1844                 strbuf_addstr(&out, "?,?");
1845         strbuf_addstr(&out, " +");
1846         add_line_count(&out, lc_b);
1847         strbuf_addstr(&out, " @@\n");
1848         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1849         strbuf_release(&out);
1850
1851         if (lc_a && !o->irreversible_delete)
1852                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1853         if (lc_b)
1854                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1855         if (textconv_one)
1856                 free((char *)data_one);
1857         if (textconv_two)
1858                 free((char *)data_two);
1859 }
1860
1861 struct diff_words_buffer {
1862         mmfile_t text;
1863         unsigned long alloc;
1864         struct diff_words_orig {
1865                 const char *begin, *end;
1866         } *orig;
1867         int orig_nr, orig_alloc;
1868 };
1869
1870 static void diff_words_append(char *line, unsigned long len,
1871                 struct diff_words_buffer *buffer)
1872 {
1873         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1874         line++;
1875         len--;
1876         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1877         buffer->text.size += len;
1878         buffer->text.ptr[buffer->text.size] = '\0';
1879 }
1880
1881 struct diff_words_style_elem {
1882         const char *prefix;
1883         const char *suffix;
1884         const char *color; /* NULL; filled in by the setup code if
1885                             * color is enabled */
1886 };
1887
1888 struct diff_words_style {
1889         enum diff_words_type type;
1890         struct diff_words_style_elem new_word, old_word, ctx;
1891         const char *newline;
1892 };
1893
1894 static struct diff_words_style diff_words_styles[] = {
1895         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1896         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1897         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1898 };
1899
1900 struct diff_words_data {
1901         struct diff_words_buffer minus, plus;
1902         const char *current_plus;
1903         int last_minus;
1904         struct diff_options *opt;
1905         regex_t *word_regex;
1906         enum diff_words_type type;
1907         struct diff_words_style *style;
1908 };
1909
1910 static int fn_out_diff_words_write_helper(struct diff_options *o,
1911                                           struct diff_words_style_elem *st_el,
1912                                           const char *newline,
1913                                           size_t count, const char *buf)
1914 {
1915         int print = 0;
1916         struct strbuf sb = STRBUF_INIT;
1917
1918         while (count) {
1919                 char *p = memchr(buf, '\n', count);
1920                 if (print)
1921                         strbuf_addstr(&sb, diff_line_prefix(o));
1922
1923                 if (p != buf) {
1924                         const char *reset = st_el->color && *st_el->color ?
1925                                             GIT_COLOR_RESET : NULL;
1926                         if (st_el->color && *st_el->color)
1927                                 strbuf_addstr(&sb, st_el->color);
1928                         strbuf_addstr(&sb, st_el->prefix);
1929                         strbuf_add(&sb, buf, p ? p - buf : count);
1930                         strbuf_addstr(&sb, st_el->suffix);
1931                         if (reset)
1932                                 strbuf_addstr(&sb, reset);
1933                 }
1934                 if (!p)
1935                         goto out;
1936
1937                 strbuf_addstr(&sb, newline);
1938                 count -= p + 1 - buf;
1939                 buf = p + 1;
1940                 print = 1;
1941                 if (count) {
1942                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1943                                          sb.buf, sb.len, 0);
1944                         strbuf_reset(&sb);
1945                 }
1946         }
1947
1948 out:
1949         if (sb.len)
1950                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1951                                  sb.buf, sb.len, 0);
1952         strbuf_release(&sb);
1953         return 0;
1954 }
1955
1956 /*
1957  * '--color-words' algorithm can be described as:
1958  *
1959  *   1. collect the minus/plus lines of a diff hunk, divided into
1960  *      minus-lines and plus-lines;
1961  *
1962  *   2. break both minus-lines and plus-lines into words and
1963  *      place them into two mmfile_t with one word for each line;
1964  *
1965  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1966  *
1967  * And for the common parts of the both file, we output the plus side text.
1968  * diff_words->current_plus is used to trace the current position of the plus file
1969  * which printed. diff_words->last_minus is used to trace the last minus word
1970  * printed.
1971  *
1972  * For '--graph' to work with '--color-words', we need to output the graph prefix
1973  * on each line of color words output. Generally, there are two conditions on
1974  * which we should output the prefix.
1975  *
1976  *   1. diff_words->last_minus == 0 &&
1977  *      diff_words->current_plus == diff_words->plus.text.ptr
1978  *
1979  *      that is: the plus text must start as a new line, and if there is no minus
1980  *      word printed, a graph prefix must be printed.
1981  *
1982  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1983  *      *(diff_words->current_plus - 1) == '\n'
1984  *
1985  *      that is: a graph prefix must be printed following a '\n'
1986  */
1987 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1988 {
1989         if ((diff_words->last_minus == 0 &&
1990                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1991                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1992                 *(diff_words->current_plus - 1) == '\n')) {
1993                 return 1;
1994         } else {
1995                 return 0;
1996         }
1997 }
1998
1999 static void fn_out_diff_words_aux(void *priv,
2000                                   long minus_first, long minus_len,
2001                                   long plus_first, long plus_len,
2002                                   const char *func, long funclen)
2003 {
2004         struct diff_words_data *diff_words = priv;
2005         struct diff_words_style *style = diff_words->style;
2006         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
2007         struct diff_options *opt = diff_words->opt;
2008         const char *line_prefix;
2009
2010         assert(opt);
2011         line_prefix = diff_line_prefix(opt);
2012
2013         /* POSIX requires that first be decremented by one if len == 0... */
2014         if (minus_len) {
2015                 minus_begin = diff_words->minus.orig[minus_first].begin;
2016                 minus_end =
2017                         diff_words->minus.orig[minus_first + minus_len - 1].end;
2018         } else
2019                 minus_begin = minus_end =
2020                         diff_words->minus.orig[minus_first].end;
2021
2022         if (plus_len) {
2023                 plus_begin = diff_words->plus.orig[plus_first].begin;
2024                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2025         } else
2026                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2027
2028         if (color_words_output_graph_prefix(diff_words)) {
2029                 fputs(line_prefix, diff_words->opt->file);
2030         }
2031         if (diff_words->current_plus != plus_begin) {
2032                 fn_out_diff_words_write_helper(diff_words->opt,
2033                                 &style->ctx, style->newline,
2034                                 plus_begin - diff_words->current_plus,
2035                                 diff_words->current_plus);
2036         }
2037         if (minus_begin != minus_end) {
2038                 fn_out_diff_words_write_helper(diff_words->opt,
2039                                 &style->old_word, style->newline,
2040                                 minus_end - minus_begin, minus_begin);
2041         }
2042         if (plus_begin != plus_end) {
2043                 fn_out_diff_words_write_helper(diff_words->opt,
2044                                 &style->new_word, style->newline,
2045                                 plus_end - plus_begin, plus_begin);
2046         }
2047
2048         diff_words->current_plus = plus_end;
2049         diff_words->last_minus = minus_first;
2050 }
2051
2052 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2053 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2054                 int *begin, int *end)
2055 {
2056         if (word_regex && *begin < buffer->size) {
2057                 regmatch_t match[1];
2058                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2059                                  buffer->size - *begin, 1, match, 0)) {
2060                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2061                                         '\n', match[0].rm_eo - match[0].rm_so);
2062                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2063                         *begin += match[0].rm_so;
2064                         return *begin >= *end;
2065                 }
2066                 return -1;
2067         }
2068
2069         /* find the next word */
2070         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2071                 (*begin)++;
2072         if (*begin >= buffer->size)
2073                 return -1;
2074
2075         /* find the end of the word */
2076         *end = *begin + 1;
2077         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2078                 (*end)++;
2079
2080         return 0;
2081 }
2082
2083 /*
2084  * This function splits the words in buffer->text, stores the list with
2085  * newline separator into out, and saves the offsets of the original words
2086  * in buffer->orig.
2087  */
2088 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2089                 regex_t *word_regex)
2090 {
2091         int i, j;
2092         long alloc = 0;
2093
2094         out->size = 0;
2095         out->ptr = NULL;
2096
2097         /* fake an empty "0th" word */
2098         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2099         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2100         buffer->orig_nr = 1;
2101
2102         for (i = 0; i < buffer->text.size; i++) {
2103                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2104                         return;
2105
2106                 /* store original boundaries */
2107                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2108                                 buffer->orig_alloc);
2109                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2110                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2111                 buffer->orig_nr++;
2112
2113                 /* store one word */
2114                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2115                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2116                 out->ptr[out->size + j - i] = '\n';
2117                 out->size += j - i + 1;
2118
2119                 i = j - 1;
2120         }
2121 }
2122
2123 /* this executes the word diff on the accumulated buffers */
2124 static void diff_words_show(struct diff_words_data *diff_words)
2125 {
2126         xpparam_t xpp;
2127         xdemitconf_t xecfg;
2128         mmfile_t minus, plus;
2129         struct diff_words_style *style = diff_words->style;
2130
2131         struct diff_options *opt = diff_words->opt;
2132         const char *line_prefix;
2133
2134         assert(opt);
2135         line_prefix = diff_line_prefix(opt);
2136
2137         /* special case: only removal */
2138         if (!diff_words->plus.text.size) {
2139                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2140                                  line_prefix, strlen(line_prefix), 0);
2141                 fn_out_diff_words_write_helper(diff_words->opt,
2142                         &style->old_word, style->newline,
2143                         diff_words->minus.text.size,
2144                         diff_words->minus.text.ptr);
2145                 diff_words->minus.text.size = 0;
2146                 return;
2147         }
2148
2149         diff_words->current_plus = diff_words->plus.text.ptr;
2150         diff_words->last_minus = 0;
2151
2152         memset(&xpp, 0, sizeof(xpp));
2153         memset(&xecfg, 0, sizeof(xecfg));
2154         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2155         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2156         xpp.flags = 0;
2157         /* as only the hunk header will be parsed, we need a 0-context */
2158         xecfg.ctxlen = 0;
2159         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2160                           diff_words, &xpp, &xecfg))
2161                 die("unable to generate word diff");
2162         free(minus.ptr);
2163         free(plus.ptr);
2164         if (diff_words->current_plus != diff_words->plus.text.ptr +
2165                         diff_words->plus.text.size) {
2166                 if (color_words_output_graph_prefix(diff_words))
2167                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2168                                          line_prefix, strlen(line_prefix), 0);
2169                 fn_out_diff_words_write_helper(diff_words->opt,
2170                         &style->ctx, style->newline,
2171                         diff_words->plus.text.ptr + diff_words->plus.text.size
2172                         - diff_words->current_plus, diff_words->current_plus);
2173         }
2174         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2175 }
2176
2177 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2178 static void diff_words_flush(struct emit_callback *ecbdata)
2179 {
2180         struct diff_options *wo = ecbdata->diff_words->opt;
2181
2182         if (ecbdata->diff_words->minus.text.size ||
2183             ecbdata->diff_words->plus.text.size)
2184                 diff_words_show(ecbdata->diff_words);
2185
2186         if (wo->emitted_symbols) {
2187                 struct diff_options *o = ecbdata->opt;
2188                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2189                 int i;
2190
2191                 /*
2192                  * NEEDSWORK:
2193                  * Instead of appending each, concat all words to a line?
2194                  */
2195                 for (i = 0; i < wol->nr; i++)
2196                         append_emitted_diff_symbol(o, &wol->buf[i]);
2197
2198                 for (i = 0; i < wol->nr; i++)
2199                         free((void *)wol->buf[i].line);
2200
2201                 wol->nr = 0;
2202         }
2203 }
2204
2205 static void diff_filespec_load_driver(struct diff_filespec *one,
2206                                       struct index_state *istate)
2207 {
2208         /* Use already-loaded driver */
2209         if (one->driver)
2210                 return;
2211
2212         if (S_ISREG(one->mode))
2213                 one->driver = userdiff_find_by_path(istate, one->path);
2214
2215         /* Fallback to default settings */
2216         if (!one->driver)
2217                 one->driver = userdiff_find_by_name("default");
2218 }
2219
2220 static const char *userdiff_word_regex(struct diff_filespec *one,
2221                                        struct index_state *istate)
2222 {
2223         diff_filespec_load_driver(one, istate);
2224         return one->driver->word_regex;
2225 }
2226
2227 static void init_diff_words_data(struct emit_callback *ecbdata,
2228                                  struct diff_options *orig_opts,
2229                                  struct diff_filespec *one,
2230                                  struct diff_filespec *two)
2231 {
2232         int i;
2233         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2234         memcpy(o, orig_opts, sizeof(struct diff_options));
2235
2236         ecbdata->diff_words =
2237                 xcalloc(1, sizeof(struct diff_words_data));
2238         ecbdata->diff_words->type = o->word_diff;
2239         ecbdata->diff_words->opt = o;
2240
2241         if (orig_opts->emitted_symbols)
2242                 o->emitted_symbols =
2243                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2244
2245         if (!o->word_regex)
2246                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2247         if (!o->word_regex)
2248                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2249         if (!o->word_regex)
2250                 o->word_regex = diff_word_regex_cfg;
2251         if (o->word_regex) {
2252                 ecbdata->diff_words->word_regex = (regex_t *)
2253                         xmalloc(sizeof(regex_t));
2254                 if (regcomp(ecbdata->diff_words->word_regex,
2255                             o->word_regex,
2256                             REG_EXTENDED | REG_NEWLINE))
2257                         die("invalid regular expression: %s",
2258                             o->word_regex);
2259         }
2260         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2261                 if (o->word_diff == diff_words_styles[i].type) {
2262                         ecbdata->diff_words->style =
2263                                 &diff_words_styles[i];
2264                         break;
2265                 }
2266         }
2267         if (want_color(o->use_color)) {
2268                 struct diff_words_style *st = ecbdata->diff_words->style;
2269                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2270                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2271                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2272         }
2273 }
2274
2275 static void free_diff_words_data(struct emit_callback *ecbdata)
2276 {
2277         if (ecbdata->diff_words) {
2278                 diff_words_flush(ecbdata);
2279                 free (ecbdata->diff_words->opt->emitted_symbols);
2280                 free (ecbdata->diff_words->opt);
2281                 free (ecbdata->diff_words->minus.text.ptr);
2282                 free (ecbdata->diff_words->minus.orig);
2283                 free (ecbdata->diff_words->plus.text.ptr);
2284                 free (ecbdata->diff_words->plus.orig);
2285                 if (ecbdata->diff_words->word_regex) {
2286                         regfree(ecbdata->diff_words->word_regex);
2287                         free(ecbdata->diff_words->word_regex);
2288                 }
2289                 FREE_AND_NULL(ecbdata->diff_words);
2290         }
2291 }
2292
2293 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2294 {
2295         if (want_color(diff_use_color))
2296                 return diff_colors[ix];
2297         return "";
2298 }
2299
2300 const char *diff_line_prefix(struct diff_options *opt)
2301 {
2302         struct strbuf *msgbuf;
2303         if (!opt->output_prefix)
2304                 return "";
2305
2306         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2307         return msgbuf->buf;
2308 }
2309
2310 static unsigned long sane_truncate_line(char *line, unsigned long len)
2311 {
2312         const char *cp;
2313         unsigned long allot;
2314         size_t l = len;
2315
2316         cp = line;
2317         allot = l;
2318         while (0 < l) {
2319                 (void) utf8_width(&cp, &l);
2320                 if (!cp)
2321                         break; /* truncated in the middle? */
2322         }
2323         return allot - l;
2324 }
2325
2326 static void find_lno(const char *line, struct emit_callback *ecbdata)
2327 {
2328         const char *p;
2329         ecbdata->lno_in_preimage = 0;
2330         ecbdata->lno_in_postimage = 0;
2331         p = strchr(line, '-');
2332         if (!p)
2333                 return; /* cannot happen */
2334         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2335         p = strchr(p, '+');
2336         if (!p)
2337                 return; /* cannot happen */
2338         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2339 }
2340
2341 static void fn_out_consume(void *priv, char *line, unsigned long len)
2342 {
2343         struct emit_callback *ecbdata = priv;
2344         struct diff_options *o = ecbdata->opt;
2345
2346         o->found_changes = 1;
2347
2348         if (ecbdata->header) {
2349                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2350                                  ecbdata->header->buf, ecbdata->header->len, 0);
2351                 strbuf_reset(ecbdata->header);
2352                 ecbdata->header = NULL;
2353         }
2354
2355         if (ecbdata->label_path[0]) {
2356                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2357                                  ecbdata->label_path[0],
2358                                  strlen(ecbdata->label_path[0]), 0);
2359                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2360                                  ecbdata->label_path[1],
2361                                  strlen(ecbdata->label_path[1]), 0);
2362                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2363         }
2364
2365         if (diff_suppress_blank_empty
2366             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2367                 line[0] = '\n';
2368                 len = 1;
2369         }
2370
2371         if (line[0] == '@') {
2372                 if (ecbdata->diff_words)
2373                         diff_words_flush(ecbdata);
2374                 len = sane_truncate_line(line, len);
2375                 find_lno(line, ecbdata);
2376                 emit_hunk_header(ecbdata, line, len);
2377                 return;
2378         }
2379
2380         if (ecbdata->diff_words) {
2381                 enum diff_symbol s =
2382                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2383                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2384                 if (line[0] == '-') {
2385                         diff_words_append(line, len,
2386                                           &ecbdata->diff_words->minus);
2387                         return;
2388                 } else if (line[0] == '+') {
2389                         diff_words_append(line, len,
2390                                           &ecbdata->diff_words->plus);
2391                         return;
2392                 } else if (starts_with(line, "\\ ")) {
2393                         /*
2394                          * Eat the "no newline at eof" marker as if we
2395                          * saw a "+" or "-" line with nothing on it,
2396                          * and return without diff_words_flush() to
2397                          * defer processing. If this is the end of
2398                          * preimage, more "+" lines may come after it.
2399                          */
2400                         return;
2401                 }
2402                 diff_words_flush(ecbdata);
2403                 emit_diff_symbol(o, s, line, len, 0);
2404                 return;
2405         }
2406
2407         switch (line[0]) {
2408         case '+':
2409                 ecbdata->lno_in_postimage++;
2410                 emit_add_line(ecbdata, line + 1, len - 1);
2411                 break;
2412         case '-':
2413                 ecbdata->lno_in_preimage++;
2414                 emit_del_line(ecbdata, line + 1, len - 1);
2415                 break;
2416         case ' ':
2417                 ecbdata->lno_in_postimage++;
2418                 ecbdata->lno_in_preimage++;
2419                 emit_context_line(ecbdata, line + 1, len - 1);
2420                 break;
2421         default:
2422                 /* incomplete line at the end */
2423                 ecbdata->lno_in_preimage++;
2424                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2425                                  line, len, 0);
2426                 break;
2427         }
2428 }
2429
2430 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2431 {
2432         const char *old_name = a;
2433         const char *new_name = b;
2434         int pfx_length, sfx_length;
2435         int pfx_adjust_for_slash;
2436         int len_a = strlen(a);
2437         int len_b = strlen(b);
2438         int a_midlen, b_midlen;
2439         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2440         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2441
2442         if (qlen_a || qlen_b) {
2443                 quote_c_style(a, name, NULL, 0);
2444                 strbuf_addstr(name, " => ");
2445                 quote_c_style(b, name, NULL, 0);
2446                 return;
2447         }
2448
2449         /* Find common prefix */
2450         pfx_length = 0;
2451         while (*old_name && *new_name && *old_name == *new_name) {
2452                 if (*old_name == '/')
2453                         pfx_length = old_name - a + 1;
2454                 old_name++;
2455                 new_name++;
2456         }
2457
2458         /* Find common suffix */
2459         old_name = a + len_a;
2460         new_name = b + len_b;
2461         sfx_length = 0;
2462         /*
2463          * If there is a common prefix, it must end in a slash.  In
2464          * that case we let this loop run 1 into the prefix to see the
2465          * same slash.
2466          *
2467          * If there is no common prefix, we cannot do this as it would
2468          * underrun the input strings.
2469          */
2470         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2471         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2472                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2473                *old_name == *new_name) {
2474                 if (*old_name == '/')
2475                         sfx_length = len_a - (old_name - a);
2476                 old_name--;
2477                 new_name--;
2478         }
2479
2480         /*
2481          * pfx{mid-a => mid-b}sfx
2482          * {pfx-a => pfx-b}sfx
2483          * pfx{sfx-a => sfx-b}
2484          * name-a => name-b
2485          */
2486         a_midlen = len_a - pfx_length - sfx_length;
2487         b_midlen = len_b - pfx_length - sfx_length;
2488         if (a_midlen < 0)
2489                 a_midlen = 0;
2490         if (b_midlen < 0)
2491                 b_midlen = 0;
2492
2493         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2494         if (pfx_length + sfx_length) {
2495                 strbuf_add(name, a, pfx_length);
2496                 strbuf_addch(name, '{');
2497         }
2498         strbuf_add(name, a + pfx_length, a_midlen);
2499         strbuf_addstr(name, " => ");
2500         strbuf_add(name, b + pfx_length, b_midlen);
2501         if (pfx_length + sfx_length) {
2502                 strbuf_addch(name, '}');
2503                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2504         }
2505 }
2506
2507 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2508                                           const char *name_a,
2509                                           const char *name_b)
2510 {
2511         struct diffstat_file *x;
2512         x = xcalloc(1, sizeof(*x));
2513         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2514         diffstat->files[diffstat->nr++] = x;
2515         if (name_b) {
2516                 x->from_name = xstrdup(name_a);
2517                 x->name = xstrdup(name_b);
2518                 x->is_renamed = 1;
2519         }
2520         else {
2521                 x->from_name = NULL;
2522                 x->name = xstrdup(name_a);
2523         }
2524         return x;
2525 }
2526
2527 static void diffstat_consume(void *priv, char *line, unsigned long len)
2528 {
2529         struct diffstat_t *diffstat = priv;
2530         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2531
2532         if (line[0] == '+')
2533                 x->added++;
2534         else if (line[0] == '-')
2535                 x->deleted++;
2536 }
2537
2538 const char mime_boundary_leader[] = "------------";
2539
2540 static int scale_linear(int it, int width, int max_change)
2541 {
2542         if (!it)
2543                 return 0;
2544         /*
2545          * make sure that at least one '-' or '+' is printed if
2546          * there is any change to this path. The easiest way is to
2547          * scale linearly as if the allotted width is one column shorter
2548          * than it is, and then add 1 to the result.
2549          */
2550         return 1 + (it * (width - 1) / max_change);
2551 }
2552
2553 static void show_graph(struct strbuf *out, char ch, int cnt,
2554                        const char *set, const char *reset)
2555 {
2556         if (cnt <= 0)
2557                 return;
2558         strbuf_addstr(out, set);
2559         strbuf_addchars(out, ch, cnt);
2560         strbuf_addstr(out, reset);
2561 }
2562
2563 static void fill_print_name(struct diffstat_file *file)
2564 {
2565         struct strbuf pname = STRBUF_INIT;
2566
2567         if (file->print_name)
2568                 return;
2569
2570         if (file->is_renamed)
2571                 pprint_rename(&pname, file->from_name, file->name);
2572         else
2573                 quote_c_style(file->name, &pname, NULL, 0);
2574
2575         if (file->comments)
2576                 strbuf_addf(&pname, " (%s)", file->comments);
2577
2578         file->print_name = strbuf_detach(&pname, NULL);
2579 }
2580
2581 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2582                 int files, int insertions, int deletions)
2583 {
2584         struct strbuf sb = STRBUF_INIT;
2585
2586         if (!files) {
2587                 assert(insertions == 0 && deletions == 0);
2588                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2589                                  NULL, 0, 0);
2590                 return;
2591         }
2592
2593         strbuf_addf(&sb,
2594                     (files == 1) ? " %d file changed" : " %d files changed",
2595                     files);
2596
2597         /*
2598          * For binary diff, the caller may want to print "x files
2599          * changed" with insertions == 0 && deletions == 0.
2600          *
2601          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2602          * is probably less confusing (i.e skip over "2 files changed
2603          * but nothing about added/removed lines? Is this a bug in Git?").
2604          */
2605         if (insertions || deletions == 0) {
2606                 strbuf_addf(&sb,
2607                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2608                             insertions);
2609         }
2610
2611         if (deletions || insertions == 0) {
2612                 strbuf_addf(&sb,
2613                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2614                             deletions);
2615         }
2616         strbuf_addch(&sb, '\n');
2617         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2618                          sb.buf, sb.len, 0);
2619         strbuf_release(&sb);
2620 }
2621
2622 void print_stat_summary(FILE *fp, int files,
2623                         int insertions, int deletions)
2624 {
2625         struct diff_options o;
2626         memset(&o, 0, sizeof(o));
2627         o.file = fp;
2628
2629         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2630 }
2631
2632 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2633 {
2634         int i, len, add, del, adds = 0, dels = 0;
2635         uintmax_t max_change = 0, max_len = 0;
2636         int total_files = data->nr, count;
2637         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2638         const char *reset, *add_c, *del_c;
2639         int extra_shown = 0;
2640         const char *line_prefix = diff_line_prefix(options);
2641         struct strbuf out = STRBUF_INIT;
2642
2643         if (data->nr == 0)
2644                 return;
2645
2646         count = options->stat_count ? options->stat_count : data->nr;
2647
2648         reset = diff_get_color_opt(options, DIFF_RESET);
2649         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2650         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2651
2652         /*
2653          * Find the longest filename and max number of changes
2654          */
2655         for (i = 0; (i < count) && (i < data->nr); i++) {
2656                 struct diffstat_file *file = data->files[i];
2657                 uintmax_t change = file->added + file->deleted;
2658
2659                 if (!file->is_interesting && (change == 0)) {
2660                         count++; /* not shown == room for one more */
2661                         continue;
2662                 }
2663                 fill_print_name(file);
2664                 len = strlen(file->print_name);
2665                 if (max_len < len)
2666                         max_len = len;
2667
2668                 if (file->is_unmerged) {
2669                         /* "Unmerged" is 8 characters */
2670                         bin_width = bin_width < 8 ? 8 : bin_width;
2671                         continue;
2672                 }
2673                 if (file->is_binary) {
2674                         /* "Bin XXX -> YYY bytes" */
2675                         int w = 14 + decimal_width(file->added)
2676                                 + decimal_width(file->deleted);
2677                         bin_width = bin_width < w ? w : bin_width;
2678                         /* Display change counts aligned with "Bin" */
2679                         number_width = 3;
2680                         continue;
2681                 }
2682
2683                 if (max_change < change)
2684                         max_change = change;
2685         }
2686         count = i; /* where we can stop scanning in data->files[] */
2687
2688         /*
2689          * We have width = stat_width or term_columns() columns total.
2690          * We want a maximum of min(max_len, stat_name_width) for the name part.
2691          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2692          * We also need 1 for " " and 4 + decimal_width(max_change)
2693          * for " | NNNN " and one the empty column at the end, altogether
2694          * 6 + decimal_width(max_change).
2695          *
2696          * If there's not enough space, we will use the smaller of
2697          * stat_name_width (if set) and 5/8*width for the filename,
2698          * and the rest for constant elements + graph part, but no more
2699          * than stat_graph_width for the graph part.
2700          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2701          * for the standard terminal size).
2702          *
2703          * In other words: stat_width limits the maximum width, and
2704          * stat_name_width fixes the maximum width of the filename,
2705          * and is also used to divide available columns if there
2706          * aren't enough.
2707          *
2708          * Binary files are displayed with "Bin XXX -> YYY bytes"
2709          * instead of the change count and graph. This part is treated
2710          * similarly to the graph part, except that it is not
2711          * "scaled". If total width is too small to accommodate the
2712          * guaranteed minimum width of the filename part and the
2713          * separators and this message, this message will "overflow"
2714          * making the line longer than the maximum width.
2715          */
2716
2717         if (options->stat_width == -1)
2718                 width = term_columns() - strlen(line_prefix);
2719         else
2720                 width = options->stat_width ? options->stat_width : 80;
2721         number_width = decimal_width(max_change) > number_width ?
2722                 decimal_width(max_change) : number_width;
2723
2724         if (options->stat_graph_width == -1)
2725                 options->stat_graph_width = diff_stat_graph_width;
2726
2727         /*
2728          * Guarantee 3/8*16==6 for the graph part
2729          * and 5/8*16==10 for the filename part
2730          */
2731         if (width < 16 + 6 + number_width)
2732                 width = 16 + 6 + number_width;
2733
2734         /*
2735          * First assign sizes that are wanted, ignoring available width.
2736          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2737          * starting from "XXX" should fit in graph_width.
2738          */
2739         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2740         if (options->stat_graph_width &&
2741             options->stat_graph_width < graph_width)
2742                 graph_width = options->stat_graph_width;
2743
2744         name_width = (options->stat_name_width > 0 &&
2745                       options->stat_name_width < max_len) ?
2746                 options->stat_name_width : max_len;
2747
2748         /*
2749          * Adjust adjustable widths not to exceed maximum width
2750          */
2751         if (name_width + number_width + 6 + graph_width > width) {
2752                 if (graph_width > width * 3/8 - number_width - 6) {
2753                         graph_width = width * 3/8 - number_width - 6;
2754                         if (graph_width < 6)
2755                                 graph_width = 6;
2756                 }
2757
2758                 if (options->stat_graph_width &&
2759                     graph_width > options->stat_graph_width)
2760                         graph_width = options->stat_graph_width;
2761                 if (name_width > width - number_width - 6 - graph_width)
2762                         name_width = width - number_width - 6 - graph_width;
2763                 else
2764                         graph_width = width - number_width - 6 - name_width;
2765         }
2766
2767         /*
2768          * From here name_width is the width of the name area,
2769          * and graph_width is the width of the graph area.
2770          * max_change is used to scale graph properly.
2771          */
2772         for (i = 0; i < count; i++) {
2773                 const char *prefix = "";
2774                 struct diffstat_file *file = data->files[i];
2775                 char *name = file->print_name;
2776                 uintmax_t added = file->added;
2777                 uintmax_t deleted = file->deleted;
2778                 int name_len;
2779
2780                 if (!file->is_interesting && (added + deleted == 0))
2781                         continue;
2782
2783                 /*
2784                  * "scale" the filename
2785                  */
2786                 len = name_width;
2787                 name_len = strlen(name);
2788                 if (name_width < name_len) {
2789                         char *slash;
2790                         prefix = "...";
2791                         len -= 3;
2792                         name += name_len - len;
2793                         slash = strchr(name, '/');
2794                         if (slash)
2795                                 name = slash;
2796                 }
2797
2798                 if (file->is_binary) {
2799                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2800                         strbuf_addf(&out, " %*s", number_width, "Bin");
2801                         if (!added && !deleted) {
2802                                 strbuf_addch(&out, '\n');
2803                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2804                                                  out.buf, out.len, 0);
2805                                 strbuf_reset(&out);
2806                                 continue;
2807                         }
2808                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2809                                 del_c, deleted, reset);
2810                         strbuf_addstr(&out, " -> ");
2811                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2812                                 add_c, added, reset);
2813                         strbuf_addstr(&out, " bytes\n");
2814                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2815                                          out.buf, out.len, 0);
2816                         strbuf_reset(&out);
2817                         continue;
2818                 }
2819                 else if (file->is_unmerged) {
2820                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2821                         strbuf_addstr(&out, " Unmerged\n");
2822                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2823                                          out.buf, out.len, 0);
2824                         strbuf_reset(&out);
2825                         continue;
2826                 }
2827
2828                 /*
2829                  * scale the add/delete
2830                  */
2831                 add = added;
2832                 del = deleted;
2833
2834                 if (graph_width <= max_change) {
2835                         int total = scale_linear(add + del, graph_width, max_change);
2836                         if (total < 2 && add && del)
2837                                 /* width >= 2 due to the sanity check */
2838                                 total = 2;
2839                         if (add < del) {
2840                                 add = scale_linear(add, graph_width, max_change);
2841                                 del = total - add;
2842                         } else {
2843                                 del = scale_linear(del, graph_width, max_change);
2844                                 add = total - del;
2845                         }
2846                 }
2847                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2848                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2849                         number_width, added + deleted,
2850                         added + deleted ? " " : "");
2851                 show_graph(&out, '+', add, add_c, reset);
2852                 show_graph(&out, '-', del, del_c, reset);
2853                 strbuf_addch(&out, '\n');
2854                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2855                                  out.buf, out.len, 0);
2856                 strbuf_reset(&out);
2857         }
2858
2859         for (i = 0; i < data->nr; i++) {
2860                 struct diffstat_file *file = data->files[i];
2861                 uintmax_t added = file->added;
2862                 uintmax_t deleted = file->deleted;
2863
2864                 if (file->is_unmerged ||
2865                     (!file->is_interesting && (added + deleted == 0))) {
2866                         total_files--;
2867                         continue;
2868                 }
2869
2870                 if (!file->is_binary) {
2871                         adds += added;
2872                         dels += deleted;
2873                 }
2874                 if (i < count)
2875                         continue;
2876                 if (!extra_shown)
2877                         emit_diff_symbol(options,
2878                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2879                                          NULL, 0, 0);
2880                 extra_shown = 1;
2881         }
2882
2883         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2884         strbuf_release(&out);
2885 }
2886
2887 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2888 {
2889         int i, adds = 0, dels = 0, total_files = data->nr;
2890
2891         if (data->nr == 0)
2892                 return;
2893
2894         for (i = 0; i < data->nr; i++) {
2895                 int added = data->files[i]->added;
2896                 int deleted = data->files[i]->deleted;
2897
2898                 if (data->files[i]->is_unmerged ||
2899                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2900                         total_files--;
2901                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2902                         adds += added;
2903                         dels += deleted;
2904                 }
2905         }
2906         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2907 }
2908
2909 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2910 {
2911         int i;
2912
2913         if (data->nr == 0)
2914                 return;
2915
2916         for (i = 0; i < data->nr; i++) {
2917                 struct diffstat_file *file = data->files[i];
2918
2919                 fprintf(options->file, "%s", diff_line_prefix(options));
2920
2921                 if (file->is_binary)
2922                         fprintf(options->file, "-\t-\t");
2923                 else
2924                         fprintf(options->file,
2925                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2926                                 file->added, file->deleted);
2927                 if (options->line_termination) {
2928                         fill_print_name(file);
2929                         if (!file->is_renamed)
2930                                 write_name_quoted(file->name, options->file,
2931                                                   options->line_termination);
2932                         else {
2933                                 fputs(file->print_name, options->file);
2934                                 putc(options->line_termination, options->file);
2935                         }
2936                 } else {
2937                         if (file->is_renamed) {
2938                                 putc('\0', options->file);
2939                                 write_name_quoted(file->from_name, options->file, '\0');
2940                         }
2941                         write_name_quoted(file->name, options->file, '\0');
2942                 }
2943         }
2944 }
2945
2946 struct dirstat_file {
2947         const char *name;
2948         unsigned long changed;
2949 };
2950
2951 struct dirstat_dir {
2952         struct dirstat_file *files;
2953         int alloc, nr, permille, cumulative;
2954 };
2955
2956 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2957                 unsigned long changed, const char *base, int baselen)
2958 {
2959         unsigned long sum_changes = 0;
2960         unsigned int sources = 0;
2961         const char *line_prefix = diff_line_prefix(opt);
2962
2963         while (dir->nr) {
2964                 struct dirstat_file *f = dir->files;
2965                 int namelen = strlen(f->name);
2966                 unsigned long changes;
2967                 char *slash;
2968
2969                 if (namelen < baselen)
2970                         break;
2971                 if (memcmp(f->name, base, baselen))
2972                         break;
2973                 slash = strchr(f->name + baselen, '/');
2974                 if (slash) {
2975                         int newbaselen = slash + 1 - f->name;
2976                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2977                         sources++;
2978                 } else {
2979                         changes = f->changed;
2980                         dir->files++;
2981                         dir->nr--;
2982                         sources += 2;
2983                 }
2984                 sum_changes += changes;
2985         }
2986
2987         /*
2988          * We don't report dirstat's for
2989          *  - the top level
2990          *  - or cases where everything came from a single directory
2991          *    under this directory (sources == 1).
2992          */
2993         if (baselen && sources != 1) {
2994                 if (sum_changes) {
2995                         int permille = sum_changes * 1000 / changed;
2996                         if (permille >= dir->permille) {
2997                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2998                                         permille / 10, permille % 10, baselen, base);
2999                                 if (!dir->cumulative)
3000                                         return 0;
3001                         }
3002                 }
3003         }
3004         return sum_changes;
3005 }
3006
3007 static int dirstat_compare(const void *_a, const void *_b)
3008 {
3009         const struct dirstat_file *a = _a;
3010         const struct dirstat_file *b = _b;
3011         return strcmp(a->name, b->name);
3012 }
3013
3014 static void show_dirstat(struct diff_options *options)
3015 {
3016         int i;
3017         unsigned long changed;
3018         struct dirstat_dir dir;
3019         struct diff_queue_struct *q = &diff_queued_diff;
3020
3021         dir.files = NULL;
3022         dir.alloc = 0;
3023         dir.nr = 0;
3024         dir.permille = options->dirstat_permille;
3025         dir.cumulative = options->flags.dirstat_cumulative;
3026
3027         changed = 0;
3028         for (i = 0; i < q->nr; i++) {
3029                 struct diff_filepair *p = q->queue[i];
3030                 const char *name;
3031                 unsigned long copied, added, damage;
3032                 struct diff_populate_filespec_options dpf_options = {
3033                         .check_size_only = 1,
3034                 };
3035
3036                 name = p->two->path ? p->two->path : p->one->path;
3037
3038                 if (p->one->oid_valid && p->two->oid_valid &&
3039                     oideq(&p->one->oid, &p->two->oid)) {
3040                         /*
3041                          * The SHA1 has not changed, so pre-/post-content is
3042                          * identical. We can therefore skip looking at the
3043                          * file contents altogether.
3044                          */
3045                         damage = 0;
3046                         goto found_damage;
3047                 }
3048
3049                 if (options->flags.dirstat_by_file) {
3050                         /*
3051                          * In --dirstat-by-file mode, we don't really need to
3052                          * look at the actual file contents at all.
3053                          * The fact that the SHA1 changed is enough for us to
3054                          * add this file to the list of results
3055                          * (with each file contributing equal damage).
3056                          */
3057                         damage = 1;
3058                         goto found_damage;
3059                 }
3060
3061                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3062                         diff_populate_filespec(options->repo, p->one, NULL);
3063                         diff_populate_filespec(options->repo, p->two, NULL);
3064                         diffcore_count_changes(options->repo,
3065                                                p->one, p->two, NULL, NULL,
3066                                                &copied, &added);
3067                         diff_free_filespec_data(p->one);
3068                         diff_free_filespec_data(p->two);
3069                 } else if (DIFF_FILE_VALID(p->one)) {
3070                         diff_populate_filespec(options->repo, p->one, &dpf_options);
3071                         copied = added = 0;
3072                         diff_free_filespec_data(p->one);
3073                 } else if (DIFF_FILE_VALID(p->two)) {
3074                         diff_populate_filespec(options->repo, p->two, &dpf_options);
3075                         copied = 0;
3076                         added = p->two->size;
3077                         diff_free_filespec_data(p->two);
3078                 } else
3079                         continue;
3080
3081                 /*
3082                  * Original minus copied is the removed material,
3083                  * added is the new material.  They are both damages
3084                  * made to the preimage.
3085                  * If the resulting damage is zero, we know that
3086                  * diffcore_count_changes() considers the two entries to
3087                  * be identical, but since the oid changed, we
3088                  * know that there must have been _some_ kind of change,
3089                  * so we force all entries to have damage > 0.
3090                  */
3091                 damage = (p->one->size - copied) + added;
3092                 if (!damage)
3093                         damage = 1;
3094
3095 found_damage:
3096                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3097                 dir.files[dir.nr].name = name;
3098                 dir.files[dir.nr].changed = damage;
3099                 changed += damage;
3100                 dir.nr++;
3101         }
3102
3103         /* This can happen even with many files, if everything was renames */
3104         if (!changed)
3105                 return;
3106
3107         /* Show all directories with more than x% of the changes */
3108         QSORT(dir.files, dir.nr, dirstat_compare);
3109         gather_dirstat(options, &dir, changed, "", 0);
3110 }
3111
3112 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3113 {
3114         int i;
3115         unsigned long changed;
3116         struct dirstat_dir dir;
3117
3118         if (data->nr == 0)
3119                 return;
3120
3121         dir.files = NULL;
3122         dir.alloc = 0;
3123         dir.nr = 0;
3124         dir.permille = options->dirstat_permille;
3125         dir.cumulative = options->flags.dirstat_cumulative;
3126
3127         changed = 0;
3128         for (i = 0; i < data->nr; i++) {
3129                 struct diffstat_file *file = data->files[i];
3130                 unsigned long damage = file->added + file->deleted;
3131                 if (file->is_binary)
3132                         /*
3133                          * binary files counts bytes, not lines. Must find some
3134                          * way to normalize binary bytes vs. textual lines.
3135                          * The following heuristic assumes that there are 64
3136                          * bytes per "line".
3137                          * This is stupid and ugly, but very cheap...
3138                          */
3139                         damage = DIV_ROUND_UP(damage, 64);
3140                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3141                 dir.files[dir.nr].name = file->name;
3142                 dir.files[dir.nr].changed = damage;
3143                 changed += damage;
3144                 dir.nr++;
3145         }
3146
3147         /* This can happen even with many files, if everything was renames */
3148         if (!changed)
3149                 return;
3150
3151         /* Show all directories with more than x% of the changes */
3152         QSORT(dir.files, dir.nr, dirstat_compare);
3153         gather_dirstat(options, &dir, changed, "", 0);
3154 }
3155
3156 void free_diffstat_info(struct diffstat_t *diffstat)
3157 {
3158         int i;
3159         for (i = 0; i < diffstat->nr; i++) {
3160                 struct diffstat_file *f = diffstat->files[i];
3161                 free(f->print_name);
3162                 free(f->name);
3163                 free(f->from_name);
3164                 free(f);
3165         }
3166         free(diffstat->files);
3167 }
3168
3169 struct checkdiff_t {
3170         const char *filename;
3171         int lineno;
3172         int conflict_marker_size;
3173         struct diff_options *o;
3174         unsigned ws_rule;
3175         unsigned status;
3176 };
3177
3178 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3179 {
3180         char firstchar;
3181         int cnt;
3182
3183         if (len < marker_size + 1)
3184                 return 0;
3185         firstchar = line[0];
3186         switch (firstchar) {
3187         case '=': case '>': case '<': case '|':
3188                 break;
3189         default:
3190                 return 0;
3191         }
3192         for (cnt = 1; cnt < marker_size; cnt++)
3193                 if (line[cnt] != firstchar)
3194                         return 0;
3195         /* line[1] through line[marker_size-1] are same as firstchar */
3196         if (len < marker_size + 1 || !isspace(line[marker_size]))
3197                 return 0;
3198         return 1;
3199 }
3200
3201 static void checkdiff_consume_hunk(void *priv,
3202                                    long ob, long on, long nb, long nn,
3203                                    const char *func, long funclen)
3204
3205 {
3206         struct checkdiff_t *data = priv;
3207         data->lineno = nb - 1;
3208 }
3209
3210 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3211 {
3212         struct checkdiff_t *data = priv;
3213         int marker_size = data->conflict_marker_size;
3214         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3215         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3216         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3217         char *err;
3218         const char *line_prefix;
3219
3220         assert(data->o);
3221         line_prefix = diff_line_prefix(data->o);
3222
3223         if (line[0] == '+') {
3224                 unsigned bad;
3225                 data->lineno++;
3226                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3227                         data->status |= 1;
3228                         fprintf(data->o->file,
3229                                 "%s%s:%d: leftover conflict marker\n",
3230                                 line_prefix, data->filename, data->lineno);
3231                 }
3232                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3233                 if (!bad)
3234                         return;
3235                 data->status |= bad;
3236                 err = whitespace_error_string(bad);
3237                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3238                         line_prefix, data->filename, data->lineno, err);
3239                 free(err);
3240                 emit_line(data->o, set, reset, line, 1);
3241                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3242                               data->o->file, set, reset, ws);
3243         } else if (line[0] == ' ') {
3244                 data->lineno++;
3245         }
3246 }
3247
3248 static unsigned char *deflate_it(char *data,
3249                                  unsigned long size,
3250                                  unsigned long *result_size)
3251 {
3252         int bound;
3253         unsigned char *deflated;
3254         git_zstream stream;
3255
3256         git_deflate_init(&stream, zlib_compression_level);
3257         bound = git_deflate_bound(&stream, size);
3258         deflated = xmalloc(bound);
3259         stream.next_out = deflated;
3260         stream.avail_out = bound;
3261
3262         stream.next_in = (unsigned char *)data;
3263         stream.avail_in = size;
3264         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3265                 ; /* nothing */
3266         git_deflate_end(&stream);
3267         *result_size = stream.total_out;
3268         return deflated;
3269 }
3270
3271 static void emit_binary_diff_body(struct diff_options *o,
3272                                   mmfile_t *one, mmfile_t *two)
3273 {
3274         void *cp;
3275         void *delta;
3276         void *deflated;
3277         void *data;
3278         unsigned long orig_size;
3279         unsigned long delta_size;
3280         unsigned long deflate_size;
3281         unsigned long data_size;
3282
3283         /* We could do deflated delta, or we could do just deflated two,
3284          * whichever is smaller.
3285          */
3286         delta = NULL;
3287         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3288         if (one->size && two->size) {
3289                 delta = diff_delta(one->ptr, one->size,
3290                                    two->ptr, two->size,
3291                                    &delta_size, deflate_size);
3292                 if (delta) {
3293                         void *to_free = delta;
3294                         orig_size = delta_size;
3295                         delta = deflate_it(delta, delta_size, &delta_size);
3296                         free(to_free);
3297                 }
3298         }
3299
3300         if (delta && delta_size < deflate_size) {
3301                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3302                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3303                                  s, strlen(s), 0);
3304                 free(s);
3305                 free(deflated);
3306                 data = delta;
3307                 data_size = delta_size;
3308         } else {
3309                 char *s = xstrfmt("%lu", two->size);
3310                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3311                                  s, strlen(s), 0);
3312                 free(s);
3313                 free(delta);
3314                 data = deflated;
3315                 data_size = deflate_size;
3316         }
3317
3318         /* emit data encoded in base85 */
3319         cp = data;
3320         while (data_size) {
3321                 int len;
3322                 int bytes = (52 < data_size) ? 52 : data_size;
3323                 char line[71];
3324                 data_size -= bytes;
3325                 if (bytes <= 26)
3326                         line[0] = bytes + 'A' - 1;
3327                 else
3328                         line[0] = bytes - 26 + 'a' - 1;
3329                 encode_85(line + 1, cp, bytes);
3330                 cp = (char *) cp + bytes;
3331
3332                 len = strlen(line);
3333                 line[len++] = '\n';
3334                 line[len] = '\0';
3335
3336                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3337                                  line, len, 0);
3338         }
3339         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3340         free(data);
3341 }
3342
3343 static void emit_binary_diff(struct diff_options *o,
3344                              mmfile_t *one, mmfile_t *two)
3345 {
3346         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3347         emit_binary_diff_body(o, one, two);
3348         emit_binary_diff_body(o, two, one);
3349 }
3350
3351 int diff_filespec_is_binary(struct repository *r,
3352                             struct diff_filespec *one)
3353 {
3354         struct diff_populate_filespec_options dpf_options = {
3355                 .check_binary = 1,
3356         };
3357
3358         if (one->is_binary == -1) {
3359                 diff_filespec_load_driver(one, r->index);
3360                 if (one->driver->binary != -1)
3361                         one->is_binary = one->driver->binary;
3362                 else {
3363                         if (!one->data && DIFF_FILE_VALID(one))
3364                                 diff_populate_filespec(r, one, &dpf_options);
3365                         if (one->is_binary == -1 && one->data)
3366                                 one->is_binary = buffer_is_binary(one->data,
3367                                                 one->size);
3368                         if (one->is_binary == -1)
3369                                 one->is_binary = 0;
3370                 }
3371         }
3372         return one->is_binary;
3373 }
3374
3375 static const struct userdiff_funcname *
3376 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3377 {
3378         diff_filespec_load_driver(one, o->repo->index);
3379         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3380 }
3381
3382 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3383 {
3384         if (!options->a_prefix)
3385                 options->a_prefix = a;
3386         if (!options->b_prefix)
3387                 options->b_prefix = b;
3388 }
3389
3390 struct userdiff_driver *get_textconv(struct repository *r,
3391                                      struct diff_filespec *one)
3392 {
3393         if (!DIFF_FILE_VALID(one))
3394                 return NULL;
3395
3396         diff_filespec_load_driver(one, r->index);
3397         return userdiff_get_textconv(r, one->driver);
3398 }
3399
3400 static void builtin_diff(const char *name_a,
3401                          const char *name_b,
3402                          struct diff_filespec *one,
3403                          struct diff_filespec *two,
3404                          const char *xfrm_msg,
3405                          int must_show_header,
3406                          struct diff_options *o,
3407                          int complete_rewrite)
3408 {
3409         mmfile_t mf1, mf2;
3410         const char *lbl[2];
3411         char *a_one, *b_two;
3412         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3413         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3414         const char *a_prefix, *b_prefix;
3415         struct userdiff_driver *textconv_one = NULL;
3416         struct userdiff_driver *textconv_two = NULL;
3417         struct strbuf header = STRBUF_INIT;
3418         const char *line_prefix = diff_line_prefix(o);
3419
3420         diff_set_mnemonic_prefix(o, "a/", "b/");
3421         if (o->flags.reverse_diff) {
3422                 a_prefix = o->b_prefix;
3423                 b_prefix = o->a_prefix;
3424         } else {
3425                 a_prefix = o->a_prefix;
3426                 b_prefix = o->b_prefix;
3427         }
3428
3429         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3430             (!one->mode || S_ISGITLINK(one->mode)) &&
3431             (!two->mode || S_ISGITLINK(two->mode))) {
3432                 show_submodule_summary(o, one->path ? one->path : two->path,
3433                                 &one->oid, &two->oid,
3434                                 two->dirty_submodule);
3435                 return;
3436         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3437                    (!one->mode || S_ISGITLINK(one->mode)) &&
3438                    (!two->mode || S_ISGITLINK(two->mode))) {
3439                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3440                                 &one->oid, &two->oid,
3441                                 two->dirty_submodule);
3442                 return;
3443         }
3444
3445         if (o->flags.allow_textconv) {
3446                 textconv_one = get_textconv(o->repo, one);
3447                 textconv_two = get_textconv(o->repo, two);
3448         }
3449
3450         /* Never use a non-valid filename anywhere if at all possible */
3451         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3452         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3453
3454         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3455         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3456         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3457         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3458         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3459         if (lbl[0][0] == '/') {
3460                 /* /dev/null */
3461                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3462                 if (xfrm_msg)
3463                         strbuf_addstr(&header, xfrm_msg);
3464                 must_show_header = 1;
3465         }
3466         else if (lbl[1][0] == '/') {
3467                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3468                 if (xfrm_msg)
3469                         strbuf_addstr(&header, xfrm_msg);
3470                 must_show_header = 1;
3471         }
3472         else {
3473                 if (one->mode != two->mode) {
3474                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3475                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3476                         must_show_header = 1;
3477                 }
3478                 if (xfrm_msg)
3479                         strbuf_addstr(&header, xfrm_msg);
3480
3481                 /*
3482                  * we do not run diff between different kind
3483                  * of objects.
3484                  */
3485                 if ((one->mode ^ two->mode) & S_IFMT)
3486                         goto free_ab_and_return;
3487                 if (complete_rewrite &&
3488                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3489                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3490                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3491                                          header.buf, header.len, 0);
3492                         strbuf_reset(&header);
3493                         emit_rewrite_diff(name_a, name_b, one, two,
3494                                           textconv_one, textconv_two, o);
3495                         o->found_changes = 1;
3496                         goto free_ab_and_return;
3497                 }
3498         }
3499
3500         if (o->irreversible_delete && lbl[1][0] == '/') {
3501                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3502                                  header.len, 0);
3503                 strbuf_reset(&header);
3504                 goto free_ab_and_return;
3505         } else if (!o->flags.text &&
3506                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3507                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3508                 struct strbuf sb = STRBUF_INIT;
3509                 if (!one->data && !two->data &&
3510                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3511                     !o->flags.binary) {
3512                         if (oideq(&one->oid, &two->oid)) {
3513                                 if (must_show_header)
3514                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3515                                                          header.buf, header.len,
3516                                                          0);
3517                                 goto free_ab_and_return;
3518                         }
3519                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3520                                          header.buf, header.len, 0);
3521                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3522                                     diff_line_prefix(o), lbl[0], lbl[1]);
3523                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3524                                          sb.buf, sb.len, 0);
3525                         strbuf_release(&sb);
3526                         goto free_ab_and_return;
3527                 }
3528                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3529                     fill_mmfile(o->repo, &mf2, two) < 0)
3530                         die("unable to read files to diff");
3531                 /* Quite common confusing case */
3532                 if (mf1.size == mf2.size &&
3533                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3534                         if (must_show_header)
3535                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3536                                                  header.buf, header.len, 0);
3537                         goto free_ab_and_return;
3538                 }
3539                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3540                 strbuf_reset(&header);
3541                 if (o->flags.binary)
3542                         emit_binary_diff(o, &mf1, &mf2);
3543                 else {
3544                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3545                                     diff_line_prefix(o), lbl[0], lbl[1]);
3546                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3547                                          sb.buf, sb.len, 0);
3548                         strbuf_release(&sb);
3549                 }
3550                 o->found_changes = 1;
3551         } else {
3552                 /* Crazy xdl interfaces.. */
3553                 const char *diffopts;
3554                 const char *v;
3555                 xpparam_t xpp;
3556                 xdemitconf_t xecfg;
3557                 struct emit_callback ecbdata;
3558                 const struct userdiff_funcname *pe;
3559
3560                 if (must_show_header) {
3561                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3562                                          header.buf, header.len, 0);
3563                         strbuf_reset(&header);
3564                 }
3565
3566                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3567                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3568
3569                 pe = diff_funcname_pattern(o, one);
3570                 if (!pe)
3571                         pe = diff_funcname_pattern(o, two);
3572
3573                 memset(&xpp, 0, sizeof(xpp));
3574                 memset(&xecfg, 0, sizeof(xecfg));
3575                 memset(&ecbdata, 0, sizeof(ecbdata));
3576                 if (o->flags.suppress_diff_headers)
3577                         lbl[0] = NULL;
3578                 ecbdata.label_path = lbl;
3579                 ecbdata.color_diff = want_color(o->use_color);
3580                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3581                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3582                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3583                 ecbdata.opt = o;
3584                 if (header.len && !o->flags.suppress_diff_headers)
3585                         ecbdata.header = &header;
3586                 xpp.flags = o->xdl_opts;
3587                 xpp.anchors = o->anchors;
3588                 xpp.anchors_nr = o->anchors_nr;
3589                 xecfg.ctxlen = o->context;
3590                 xecfg.interhunkctxlen = o->interhunkcontext;
3591                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3592                 if (o->flags.funccontext)
3593                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3594                 if (pe)
3595                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3596
3597                 diffopts = getenv("GIT_DIFF_OPTS");
3598                 if (!diffopts)
3599                         ;
3600                 else if (skip_prefix(diffopts, "--unified=", &v))
3601                         xecfg.ctxlen = strtoul(v, NULL, 10);
3602                 else if (skip_prefix(diffopts, "-u", &v))
3603                         xecfg.ctxlen = strtoul(v, NULL, 10);
3604
3605                 if (o->word_diff)
3606                         init_diff_words_data(&ecbdata, o, one, two);
3607                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3608                                   &ecbdata, &xpp, &xecfg))
3609                         die("unable to generate diff for %s", one->path);
3610                 if (o->word_diff)
3611                         free_diff_words_data(&ecbdata);
3612                 if (textconv_one)
3613                         free(mf1.ptr);
3614                 if (textconv_two)
3615                         free(mf2.ptr);
3616                 xdiff_clear_find_func(&xecfg);
3617         }
3618
3619  free_ab_and_return:
3620         strbuf_release(&header);
3621         diff_free_filespec_data(one);
3622         diff_free_filespec_data(two);
3623         free(a_one);
3624         free(b_two);
3625         return;
3626 }
3627
3628 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3629 {
3630         if (!is_renamed) {
3631                 if (p->status == DIFF_STATUS_ADDED) {
3632                         if (S_ISLNK(p->two->mode))
3633                                 return "new +l";
3634                         else if ((p->two->mode & 0777) == 0755)
3635                                 return "new +x";
3636                         else
3637                                 return "new";
3638                 } else if (p->status == DIFF_STATUS_DELETED)
3639                         return "gone";
3640         }
3641         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3642                 return "mode -l";
3643         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3644                 return "mode +l";
3645         else if ((p->one->mode & 0777) == 0644 &&
3646                  (p->two->mode & 0777) == 0755)
3647                 return "mode +x";
3648         else if ((p->one->mode & 0777) == 0755 &&
3649                  (p->two->mode & 0777) == 0644)
3650                 return "mode -x";
3651         return NULL;
3652 }
3653
3654 static void builtin_diffstat(const char *name_a, const char *name_b,
3655                              struct diff_filespec *one,
3656                              struct diff_filespec *two,
3657                              struct diffstat_t *diffstat,
3658                              struct diff_options *o,
3659                              struct diff_filepair *p)
3660 {
3661         mmfile_t mf1, mf2;
3662         struct diffstat_file *data;
3663         int same_contents;
3664         int complete_rewrite = 0;
3665
3666         if (!DIFF_PAIR_UNMERGED(p)) {
3667                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3668                         complete_rewrite = 1;
3669         }
3670
3671         data = diffstat_add(diffstat, name_a, name_b);
3672         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3673         if (o->flags.stat_with_summary)
3674                 data->comments = get_compact_summary(p, data->is_renamed);
3675
3676         if (!one || !two) {
3677                 data->is_unmerged = 1;
3678                 return;
3679         }
3680
3681         same_contents = oideq(&one->oid, &two->oid);
3682
3683         if (diff_filespec_is_binary(o->repo, one) ||
3684             diff_filespec_is_binary(o->repo, two)) {
3685                 data->is_binary = 1;
3686                 if (same_contents) {
3687                         data->added = 0;
3688                         data->deleted = 0;
3689                 } else {
3690                         data->added = diff_filespec_size(o->repo, two);
3691                         data->deleted = diff_filespec_size(o->repo, one);
3692                 }
3693         }
3694
3695         else if (complete_rewrite) {
3696                 diff_populate_filespec(o->repo, one, NULL);
3697                 diff_populate_filespec(o->repo, two, NULL);
3698                 data->deleted = count_lines(one->data, one->size);
3699                 data->added = count_lines(two->data, two->size);
3700         }
3701
3702         else if (!same_contents) {
3703                 /* Crazy xdl interfaces.. */
3704                 xpparam_t xpp;
3705                 xdemitconf_t xecfg;
3706
3707                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3708                     fill_mmfile(o->repo, &mf2, two) < 0)
3709                         die("unable to read files to diff");
3710
3711                 memset(&xpp, 0, sizeof(xpp));
3712                 memset(&xecfg, 0, sizeof(xecfg));
3713                 xpp.flags = o->xdl_opts;
3714                 xpp.anchors = o->anchors;
3715                 xpp.anchors_nr = o->anchors_nr;
3716                 xecfg.ctxlen = o->context;
3717                 xecfg.interhunkctxlen = o->interhunkcontext;
3718                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3719                                   diffstat_consume, diffstat, &xpp, &xecfg))
3720                         die("unable to generate diffstat for %s", one->path);
3721         }
3722
3723         diff_free_filespec_data(one);
3724         diff_free_filespec_data(two);
3725 }
3726
3727 static void builtin_checkdiff(const char *name_a, const char *name_b,
3728                               const char *attr_path,
3729                               struct diff_filespec *one,
3730                               struct diff_filespec *two,
3731                               struct diff_options *o)
3732 {
3733         mmfile_t mf1, mf2;
3734         struct checkdiff_t data;
3735
3736         if (!two)
3737                 return;
3738
3739         memset(&data, 0, sizeof(data));
3740         data.filename = name_b ? name_b : name_a;
3741         data.lineno = 0;
3742         data.o = o;
3743         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3744         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3745
3746         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3747             fill_mmfile(o->repo, &mf2, two) < 0)
3748                 die("unable to read files to diff");
3749
3750         /*
3751          * All the other codepaths check both sides, but not checking
3752          * the "old" side here is deliberate.  We are checking the newly
3753          * introduced changes, and as long as the "new" side is text, we
3754          * can and should check what it introduces.
3755          */
3756         if (diff_filespec_is_binary(o->repo, two))
3757                 goto free_and_return;
3758         else {
3759                 /* Crazy xdl interfaces.. */
3760                 xpparam_t xpp;
3761                 xdemitconf_t xecfg;
3762
3763                 memset(&xpp, 0, sizeof(xpp));
3764                 memset(&xecfg, 0, sizeof(xecfg));
3765                 xecfg.ctxlen = 1; /* at least one context line */
3766                 xpp.flags = 0;
3767                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3768                                   checkdiff_consume, &data,
3769                                   &xpp, &xecfg))
3770                         die("unable to generate checkdiff for %s", one->path);
3771
3772                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3773                         struct emit_callback ecbdata;
3774                         int blank_at_eof;
3775
3776                         ecbdata.ws_rule = data.ws_rule;
3777                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3778                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3779
3780                         if (blank_at_eof) {
3781                                 static char *err;
3782                                 if (!err)
3783                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3784                                 fprintf(o->file, "%s:%d: %s.\n",
3785                                         data.filename, blank_at_eof, err);
3786                                 data.status = 1; /* report errors */
3787                         }
3788                 }
3789         }
3790  free_and_return:
3791         diff_free_filespec_data(one);
3792         diff_free_filespec_data(two);
3793         if (data.status)
3794                 o->flags.check_failed = 1;
3795 }
3796
3797 struct diff_filespec *alloc_filespec(const char *path)
3798 {
3799         struct diff_filespec *spec;
3800
3801         FLEXPTR_ALLOC_STR(spec, path, path);
3802         spec->count = 1;
3803         spec->is_binary = -1;
3804         return spec;
3805 }
3806
3807 void free_filespec(struct diff_filespec *spec)
3808 {
3809         if (!--spec->count) {
3810                 diff_free_filespec_data(spec);
3811                 free(spec);
3812         }
3813 }
3814
3815 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3816                    int oid_valid, unsigned short mode)
3817 {
3818         if (mode) {
3819                 spec->mode = canon_mode(mode);
3820                 oidcpy(&spec->oid, oid);
3821                 spec->oid_valid = oid_valid;
3822         }
3823 }
3824
3825 /*
3826  * Given a name and sha1 pair, if the index tells us the file in
3827  * the work tree has that object contents, return true, so that
3828  * prepare_temp_file() does not have to inflate and extract.
3829  */
3830 static int reuse_worktree_file(struct index_state *istate,
3831                                const char *name,
3832                                const struct object_id *oid,
3833                                int want_file)
3834 {
3835         const struct cache_entry *ce;
3836         struct stat st;
3837         int pos, len;
3838
3839         /*
3840          * We do not read the cache ourselves here, because the
3841          * benchmark with my previous version that always reads cache
3842          * shows that it makes things worse for diff-tree comparing
3843          * two linux-2.6 kernel trees in an already checked out work
3844          * tree.  This is because most diff-tree comparisons deal with
3845          * only a small number of files, while reading the cache is
3846          * expensive for a large project, and its cost outweighs the
3847          * savings we get by not inflating the object to a temporary
3848          * file.  Practically, this code only helps when we are used
3849          * by diff-cache --cached, which does read the cache before
3850          * calling us.
3851          */
3852         if (!istate->cache)
3853                 return 0;
3854
3855         /* We want to avoid the working directory if our caller
3856          * doesn't need the data in a normal file, this system
3857          * is rather slow with its stat/open/mmap/close syscalls,
3858          * and the object is contained in a pack file.  The pack
3859          * is probably already open and will be faster to obtain
3860          * the data through than the working directory.  Loose
3861          * objects however would tend to be slower as they need
3862          * to be individually opened and inflated.
3863          */
3864         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3865                 return 0;
3866
3867         /*
3868          * Similarly, if we'd have to convert the file contents anyway, that
3869          * makes the optimization not worthwhile.
3870          */
3871         if (!want_file && would_convert_to_git(istate, name))
3872                 return 0;
3873
3874         len = strlen(name);
3875         pos = index_name_pos(istate, name, len);
3876         if (pos < 0)
3877                 return 0;
3878         ce = istate->cache[pos];
3879
3880         /*
3881          * This is not the sha1 we are looking for, or
3882          * unreusable because it is not a regular file.
3883          */
3884         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3885                 return 0;
3886
3887         /*
3888          * If ce is marked as "assume unchanged", there is no
3889          * guarantee that work tree matches what we are looking for.
3890          */
3891         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3892                 return 0;
3893
3894         /*
3895          * If ce matches the file in the work tree, we can reuse it.
3896          */
3897         if (ce_uptodate(ce) ||
3898             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3899                 return 1;
3900
3901         return 0;
3902 }
3903
3904 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3905 {
3906         struct strbuf buf = STRBUF_INIT;
3907         char *dirty = "";
3908
3909         /* Are we looking at the work tree? */
3910         if (s->dirty_submodule)
3911                 dirty = "-dirty";
3912
3913         strbuf_addf(&buf, "Subproject commit %s%s\n",
3914                     oid_to_hex(&s->oid), dirty);
3915         s->size = buf.len;
3916         if (size_only) {
3917                 s->data = NULL;
3918                 strbuf_release(&buf);
3919         } else {
3920                 s->data = strbuf_detach(&buf, NULL);
3921                 s->should_free = 1;
3922         }
3923         return 0;
3924 }
3925
3926 /*
3927  * While doing rename detection and pickaxe operation, we may need to
3928  * grab the data for the blob (or file) for our own in-core comparison.
3929  * diff_filespec has data and size fields for this purpose.
3930  */
3931 int diff_populate_filespec(struct repository *r,
3932                            struct diff_filespec *s,
3933                            const struct diff_populate_filespec_options *options)
3934 {
3935         int size_only = options ? options->check_size_only : 0;
3936         int check_binary = options ? options->check_binary : 0;
3937         int err = 0;
3938         int conv_flags = global_conv_flags_eol;
3939         /*
3940          * demote FAIL to WARN to allow inspecting the situation
3941          * instead of refusing.
3942          */
3943         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3944                 conv_flags = CONV_EOL_RNDTRP_WARN;
3945
3946         if (!DIFF_FILE_VALID(s))
3947                 die("internal error: asking to populate invalid file.");
3948         if (S_ISDIR(s->mode))
3949                 return -1;
3950
3951         if (s->data)
3952                 return 0;
3953
3954         if (size_only && 0 < s->size)
3955                 return 0;
3956
3957         if (S_ISGITLINK(s->mode))
3958                 return diff_populate_gitlink(s, size_only);
3959
3960         if (!s->oid_valid ||
3961             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3962                 struct strbuf buf = STRBUF_INIT;
3963                 struct stat st;
3964                 int fd;
3965
3966                 if (lstat(s->path, &st) < 0) {
3967                 err_empty:
3968                         err = -1;
3969                 empty:
3970                         s->data = (char *)"";
3971                         s->size = 0;
3972                         return err;
3973                 }
3974                 s->size = xsize_t(st.st_size);
3975                 if (!s->size)
3976                         goto empty;
3977                 if (S_ISLNK(st.st_mode)) {
3978                         struct strbuf sb = STRBUF_INIT;
3979
3980                         if (strbuf_readlink(&sb, s->path, s->size))
3981                                 goto err_empty;
3982                         s->size = sb.len;
3983                         s->data = strbuf_detach(&sb, NULL);
3984                         s->should_free = 1;
3985                         return 0;
3986                 }
3987
3988                 /*
3989                  * Even if the caller would be happy with getting
3990                  * only the size, we cannot return early at this
3991                  * point if the path requires us to run the content
3992                  * conversion.
3993                  */
3994                 if (size_only && !would_convert_to_git(r->index, s->path))
3995                         return 0;
3996
3997                 /*
3998                  * Note: this check uses xsize_t(st.st_size) that may
3999                  * not be the true size of the blob after it goes
4000                  * through convert_to_git().  This may not strictly be
4001                  * correct, but the whole point of big_file_threshold
4002                  * and is_binary check being that we want to avoid
4003                  * opening the file and inspecting the contents, this
4004                  * is probably fine.
4005                  */
4006                 if (check_binary &&
4007                     s->size > big_file_threshold && s->is_binary == -1) {
4008                         s->is_binary = 1;
4009                         return 0;
4010                 }
4011                 fd = open(s->path, O_RDONLY);
4012                 if (fd < 0)
4013                         goto err_empty;
4014                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4015                 close(fd);
4016                 s->should_munmap = 1;
4017
4018                 /*
4019                  * Convert from working tree format to canonical git format
4020                  */
4021                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4022                         size_t size = 0;
4023                         munmap(s->data, s->size);
4024                         s->should_munmap = 0;
4025                         s->data = strbuf_detach(&buf, &size);
4026                         s->size = size;
4027                         s->should_free = 1;
4028                 }
4029         }
4030         else {
4031                 struct object_info info = {
4032                         .sizep = &s->size
4033                 };
4034
4035                 if (!(size_only || check_binary))
4036                         /*
4037                          * Set contentp, since there is no chance that merely
4038                          * the size is sufficient.
4039                          */
4040                         info.contentp = &s->data;
4041
4042                 if (options && options->missing_object_cb) {
4043                         if (!oid_object_info_extended(r, &s->oid, &info,
4044                                                       OBJECT_INFO_LOOKUP_REPLACE |
4045                                                       OBJECT_INFO_SKIP_FETCH_OBJECT))
4046                                 goto object_read;
4047                         options->missing_object_cb(options->missing_object_data);
4048                 }
4049                 if (oid_object_info_extended(r, &s->oid, &info,
4050                                              OBJECT_INFO_LOOKUP_REPLACE))
4051                         die("unable to read %s", oid_to_hex(&s->oid));
4052
4053 object_read:
4054                 if (size_only || check_binary) {
4055                         if (size_only)
4056                                 return 0;
4057                         if (s->size > big_file_threshold && s->is_binary == -1) {
4058                                 s->is_binary = 1;
4059                                 return 0;
4060                         }
4061                 }
4062                 if (!info.contentp) {
4063                         info.contentp = &s->data;
4064                         if (oid_object_info_extended(r, &s->oid, &info,
4065                                                      OBJECT_INFO_LOOKUP_REPLACE))
4066                                 die("unable to read %s", oid_to_hex(&s->oid));
4067                 }
4068                 s->should_free = 1;
4069         }
4070         return 0;
4071 }
4072
4073 void diff_free_filespec_blob(struct diff_filespec *s)
4074 {
4075         if (s->should_free)
4076                 free(s->data);
4077         else if (s->should_munmap)
4078                 munmap(s->data, s->size);
4079
4080         if (s->should_free || s->should_munmap) {
4081                 s->should_free = s->should_munmap = 0;
4082                 s->data = NULL;
4083         }
4084 }
4085
4086 void diff_free_filespec_data(struct diff_filespec *s)
4087 {
4088         diff_free_filespec_blob(s);
4089         FREE_AND_NULL(s->cnt_data);
4090 }
4091
4092 static void prep_temp_blob(struct index_state *istate,
4093                            const char *path, struct diff_tempfile *temp,
4094                            void *blob,
4095                            unsigned long size,
4096                            const struct object_id *oid,
4097                            int mode)
4098 {
4099         struct strbuf buf = STRBUF_INIT;
4100         struct strbuf tempfile = STRBUF_INIT;
4101         char *path_dup = xstrdup(path);
4102         const char *base = basename(path_dup);
4103         struct checkout_metadata meta;
4104
4105         init_checkout_metadata(&meta, NULL, NULL, oid);
4106
4107         /* Generate "XXXXXX_basename.ext" */
4108         strbuf_addstr(&tempfile, "XXXXXX_");
4109         strbuf_addstr(&tempfile, base);
4110
4111         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4112         if (!temp->tempfile)
4113                 die_errno("unable to create temp-file");
4114         if (convert_to_working_tree(istate, path,
4115                         (const char *)blob, (size_t)size, &buf, &meta)) {
4116                 blob = buf.buf;
4117                 size = buf.len;
4118         }
4119         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4120             close_tempfile_gently(temp->tempfile))
4121                 die_errno("unable to write temp-file");
4122         temp->name = get_tempfile_path(temp->tempfile);
4123         oid_to_hex_r(temp->hex, oid);
4124         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4125         strbuf_release(&buf);
4126         strbuf_release(&tempfile);
4127         free(path_dup);
4128 }
4129
4130 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4131                                                const char *name,
4132                                                struct diff_filespec *one)
4133 {
4134         struct diff_tempfile *temp = claim_diff_tempfile();
4135
4136         if (!DIFF_FILE_VALID(one)) {
4137         not_a_valid_file:
4138                 /* A '-' entry produces this for file-2, and
4139                  * a '+' entry produces this for file-1.
4140                  */
4141                 temp->name = "/dev/null";
4142                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4143                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4144                 return temp;
4145         }
4146
4147         if (!S_ISGITLINK(one->mode) &&
4148             (!one->oid_valid ||
4149              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4150                 struct stat st;
4151                 if (lstat(name, &st) < 0) {
4152                         if (errno == ENOENT)
4153                                 goto not_a_valid_file;
4154                         die_errno("stat(%s)", name);
4155                 }
4156                 if (S_ISLNK(st.st_mode)) {
4157                         struct strbuf sb = STRBUF_INIT;
4158                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4159                                 die_errno("readlink(%s)", name);
4160                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4161                                        (one->oid_valid ?
4162                                         &one->oid : &null_oid),
4163                                        (one->oid_valid ?
4164                                         one->mode : S_IFLNK));
4165                         strbuf_release(&sb);
4166                 }
4167                 else {
4168                         /* we can borrow from the file in the work tree */
4169                         temp->name = name;
4170                         if (!one->oid_valid)
4171                                 oid_to_hex_r(temp->hex, &null_oid);
4172                         else
4173                                 oid_to_hex_r(temp->hex, &one->oid);
4174                         /* Even though we may sometimes borrow the
4175                          * contents from the work tree, we always want
4176                          * one->mode.  mode is trustworthy even when
4177                          * !(one->oid_valid), as long as
4178                          * DIFF_FILE_VALID(one).
4179                          */
4180                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4181                 }
4182                 return temp;
4183         }
4184         else {
4185                 if (diff_populate_filespec(r, one, NULL))
4186                         die("cannot read data blob for %s", one->path);
4187                 prep_temp_blob(r->index, name, temp,
4188                                one->data, one->size,
4189                                &one->oid, one->mode);
4190         }
4191         return temp;
4192 }
4193
4194 static void add_external_diff_name(struct repository *r,
4195                                    struct argv_array *argv,
4196                                    const char *name,
4197                                    struct diff_filespec *df)
4198 {
4199         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4200         argv_array_push(argv, temp->name);
4201         argv_array_push(argv, temp->hex);
4202         argv_array_push(argv, temp->mode);
4203 }
4204
4205 /* An external diff command takes:
4206  *
4207  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4208  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4209  *
4210  */
4211 static void run_external_diff(const char *pgm,
4212                               const char *name,
4213                               const char *other,
4214                               struct diff_filespec *one,
4215                               struct diff_filespec *two,
4216                               const char *xfrm_msg,
4217                               struct diff_options *o)
4218 {
4219         struct argv_array argv = ARGV_ARRAY_INIT;
4220         struct argv_array env = ARGV_ARRAY_INIT;
4221         struct diff_queue_struct *q = &diff_queued_diff;
4222
4223         argv_array_push(&argv, pgm);
4224         argv_array_push(&argv, name);
4225
4226         if (one && two) {
4227                 add_external_diff_name(o->repo, &argv, name, one);
4228                 if (!other)
4229                         add_external_diff_name(o->repo, &argv, name, two);
4230                 else {
4231                         add_external_diff_name(o->repo, &argv, other, two);
4232                         argv_array_push(&argv, other);
4233                         argv_array_push(&argv, xfrm_msg);
4234                 }
4235         }
4236
4237         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4238         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4239
4240         diff_free_filespec_data(one);
4241         diff_free_filespec_data(two);
4242         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4243                 die(_("external diff died, stopping at %s"), name);
4244
4245         remove_tempfile();
4246         argv_array_clear(&argv);
4247         argv_array_clear(&env);
4248 }
4249
4250 static int similarity_index(struct diff_filepair *p)
4251 {
4252         return p->score * 100 / MAX_SCORE;
4253 }
4254
4255 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4256 {
4257         if (startup_info->have_repository)
4258                 return find_unique_abbrev(oid, abbrev);
4259         else {
4260                 char *hex = oid_to_hex(oid);
4261                 if (abbrev < 0)
4262                         abbrev = FALLBACK_DEFAULT_ABBREV;
4263                 if (abbrev > the_hash_algo->hexsz)
4264                         BUG("oid abbreviation out of range: %d", abbrev);
4265                 if (abbrev)
4266                         hex[abbrev] = '\0';
4267                 return hex;
4268         }
4269 }
4270
4271 static void fill_metainfo(struct strbuf *msg,
4272                           const char *name,
4273                           const char *other,
4274                           struct diff_filespec *one,
4275                           struct diff_filespec *two,
4276                           struct diff_options *o,
4277                           struct diff_filepair *p,
4278                           int *must_show_header,
4279                           int use_color)
4280 {
4281         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4282         const char *reset = diff_get_color(use_color, DIFF_RESET);
4283         const char *line_prefix = diff_line_prefix(o);
4284
4285         *must_show_header = 1;
4286         strbuf_init(msg, PATH_MAX * 2 + 300);
4287         switch (p->status) {
4288         case DIFF_STATUS_COPIED:
4289                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4290                             line_prefix, set, similarity_index(p));
4291                 strbuf_addf(msg, "%s\n%s%scopy from ",
4292                             reset,  line_prefix, set);
4293                 quote_c_style(name, msg, NULL, 0);
4294                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4295                 quote_c_style(other, msg, NULL, 0);
4296                 strbuf_addf(msg, "%s\n", reset);
4297                 break;
4298         case DIFF_STATUS_RENAMED:
4299                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4300                             line_prefix, set, similarity_index(p));
4301                 strbuf_addf(msg, "%s\n%s%srename from ",
4302                             reset, line_prefix, set);
4303                 quote_c_style(name, msg, NULL, 0);
4304                 strbuf_addf(msg, "%s\n%s%srename to ",
4305                             reset, line_prefix, set);
4306                 quote_c_style(other, msg, NULL, 0);
4307                 strbuf_addf(msg, "%s\n", reset);
4308                 break;
4309         case DIFF_STATUS_MODIFIED:
4310                 if (p->score) {
4311                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4312                                     line_prefix,
4313                                     set, similarity_index(p), reset);
4314                         break;
4315                 }
4316                 /* fallthru */
4317         default:
4318                 *must_show_header = 0;
4319         }
4320         if (one && two && !oideq(&one->oid, &two->oid)) {
4321                 const unsigned hexsz = the_hash_algo->hexsz;
4322                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4323
4324                 if (o->flags.binary) {
4325                         mmfile_t mf;
4326                         if ((!fill_mmfile(o->repo, &mf, one) &&
4327                              diff_filespec_is_binary(o->repo, one)) ||
4328                             (!fill_mmfile(o->repo, &mf, two) &&
4329                              diff_filespec_is_binary(o->repo, two)))
4330                                 abbrev = hexsz;
4331                 }
4332                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4333                             diff_abbrev_oid(&one->oid, abbrev),
4334                             diff_abbrev_oid(&two->oid, abbrev));
4335                 if (one->mode == two->mode)
4336                         strbuf_addf(msg, " %06o", one->mode);
4337                 strbuf_addf(msg, "%s\n", reset);
4338         }
4339 }
4340
4341 static void run_diff_cmd(const char *pgm,
4342                          const char *name,
4343                          const char *other,
4344                          const char *attr_path,
4345                          struct diff_filespec *one,
4346                          struct diff_filespec *two,
4347                          struct strbuf *msg,
4348                          struct diff_options *o,
4349                          struct diff_filepair *p)
4350 {
4351         const char *xfrm_msg = NULL;
4352         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4353         int must_show_header = 0;
4354
4355
4356         if (o->flags.allow_external) {
4357                 struct userdiff_driver *drv;
4358
4359                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4360                 if (drv && drv->external)
4361                         pgm = drv->external;
4362         }
4363
4364         if (msg) {
4365                 /*
4366                  * don't use colors when the header is intended for an
4367                  * external diff driver
4368                  */
4369                 fill_metainfo(msg, name, other, one, two, o, p,
4370                               &must_show_header,
4371                               want_color(o->use_color) && !pgm);
4372                 xfrm_msg = msg->len ? msg->buf : NULL;
4373         }
4374
4375         if (pgm) {
4376                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4377                 return;
4378         }
4379         if (one && two)
4380                 builtin_diff(name, other ? other : name,
4381                              one, two, xfrm_msg, must_show_header,
4382                              o, complete_rewrite);
4383         else
4384                 fprintf(o->file, "* Unmerged path %s\n", name);
4385 }
4386
4387 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4388 {
4389         if (DIFF_FILE_VALID(one)) {
4390                 if (!one->oid_valid) {
4391                         struct stat st;
4392                         if (one->is_stdin) {
4393                                 oidclr(&one->oid);
4394                                 return;
4395                         }
4396                         if (lstat(one->path, &st) < 0)
4397                                 die_errno("stat '%s'", one->path);
4398                         if (index_path(istate, &one->oid, one->path, &st, 0))
4399                                 die("cannot hash %s", one->path);
4400                 }
4401         }
4402         else
4403                 oidclr(&one->oid);
4404 }
4405
4406 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4407 {
4408         /* Strip the prefix but do not molest /dev/null and absolute paths */
4409         if (*namep && !is_absolute_path(*namep)) {
4410                 *namep += prefix_length;
4411                 if (**namep == '/')
4412                         ++*namep;
4413         }
4414         if (*otherp && !is_absolute_path(*otherp)) {
4415                 *otherp += prefix_length;
4416                 if (**otherp == '/')
4417                         ++*otherp;
4418         }
4419 }
4420
4421 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4422 {
4423         const char *pgm = external_diff();
4424         struct strbuf msg;
4425         struct diff_filespec *one = p->one;
4426         struct diff_filespec *two = p->two;
4427         const char *name;
4428         const char *other;
4429         const char *attr_path;
4430
4431         name  = one->path;
4432         other = (strcmp(name, two->path) ? two->path : NULL);
4433         attr_path = name;
4434         if (o->prefix_length)
4435                 strip_prefix(o->prefix_length, &name, &other);
4436
4437         if (!o->flags.allow_external)
4438                 pgm = NULL;
4439
4440         if (DIFF_PAIR_UNMERGED(p)) {
4441                 run_diff_cmd(pgm, name, NULL, attr_path,
4442                              NULL, NULL, NULL, o, p);
4443                 return;
4444         }
4445
4446         diff_fill_oid_info(one, o->repo->index);
4447         diff_fill_oid_info(two, o->repo->index);
4448
4449         if (!pgm &&
4450             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4451             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4452                 /*
4453                  * a filepair that changes between file and symlink
4454                  * needs to be split into deletion and creation.
4455                  */
4456                 struct diff_filespec *null = alloc_filespec(two->path);
4457                 run_diff_cmd(NULL, name, other, attr_path,
4458                              one, null, &msg,
4459                              o, p);
4460                 free(null);
4461                 strbuf_release(&msg);
4462
4463                 null = alloc_filespec(one->path);
4464                 run_diff_cmd(NULL, name, other, attr_path,
4465                              null, two, &msg, o, p);
4466                 free(null);
4467         }
4468         else
4469                 run_diff_cmd(pgm, name, other, attr_path,
4470                              one, two, &msg, o, p);
4471
4472         strbuf_release(&msg);
4473 }
4474
4475 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4476                          struct diffstat_t *diffstat)
4477 {
4478         const char *name;
4479         const char *other;
4480
4481         if (DIFF_PAIR_UNMERGED(p)) {
4482                 /* unmerged */
4483                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4484                                  diffstat, o, p);
4485                 return;
4486         }
4487
4488         name = p->one->path;
4489         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4490
4491         if (o->prefix_length)
4492                 strip_prefix(o->prefix_length, &name, &other);
4493
4494         diff_fill_oid_info(p->one, o->repo->index);
4495         diff_fill_oid_info(p->two, o->repo->index);
4496
4497         builtin_diffstat(name, other, p->one, p->two,
4498                          diffstat, o, p);
4499 }
4500
4501 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4502 {
4503         const char *name;
4504         const char *other;
4505         const char *attr_path;
4506
4507         if (DIFF_PAIR_UNMERGED(p)) {
4508                 /* unmerged */
4509                 return;
4510         }
4511
4512         name = p->one->path;
4513         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4514         attr_path = other ? other : name;
4515
4516         if (o->prefix_length)
4517                 strip_prefix(o->prefix_length, &name, &other);
4518
4519         diff_fill_oid_info(p->one, o->repo->index);
4520         diff_fill_oid_info(p->two, o->repo->index);
4521
4522         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4523 }
4524
4525 static void prep_parse_options(struct diff_options *options);
4526
4527 void repo_diff_setup(struct repository *r, struct diff_options *options)
4528 {
4529         memcpy(options, &default_diff_options, sizeof(*options));
4530
4531         options->file = stdout;
4532         options->repo = r;
4533
4534         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4535         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4536         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4537         options->abbrev = DEFAULT_ABBREV;
4538         options->line_termination = '\n';
4539         options->break_opt = -1;
4540         options->rename_limit = -1;
4541         options->dirstat_permille = diff_dirstat_permille_default;
4542         options->context = diff_context_default;
4543         options->interhunkcontext = diff_interhunk_context_default;
4544         options->ws_error_highlight = ws_error_highlight_default;
4545         options->flags.rename_empty = 1;
4546         options->flags.relative_name = diff_relative;
4547         options->objfind = NULL;
4548
4549         /* pathchange left =NULL by default */
4550         options->change = diff_change;
4551         options->add_remove = diff_addremove;
4552         options->use_color = diff_use_color_default;
4553         options->detect_rename = diff_detect_rename_default;
4554         options->xdl_opts |= diff_algorithm;
4555         if (diff_indent_heuristic)
4556                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4557
4558         options->orderfile = diff_order_file_cfg;
4559
4560         if (diff_no_prefix) {
4561                 options->a_prefix = options->b_prefix = "";
4562         } else if (!diff_mnemonic_prefix) {
4563                 options->a_prefix = "a/";
4564                 options->b_prefix = "b/";
4565         }
4566
4567         options->color_moved = diff_color_moved_default;
4568         options->color_moved_ws_handling = diff_color_moved_ws_default;
4569
4570         prep_parse_options(options);
4571 }
4572
4573 void diff_setup_done(struct diff_options *options)
4574 {
4575         unsigned check_mask = DIFF_FORMAT_NAME |
4576                               DIFF_FORMAT_NAME_STATUS |
4577                               DIFF_FORMAT_CHECKDIFF |
4578                               DIFF_FORMAT_NO_OUTPUT;
4579         /*
4580          * This must be signed because we're comparing against a potentially
4581          * negative value.
4582          */
4583         const int hexsz = the_hash_algo->hexsz;
4584
4585         if (options->set_default)
4586                 options->set_default(options);
4587
4588         if (HAS_MULTI_BITS(options->output_format & check_mask))
4589                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4590
4591         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4592                 die(_("-G, -S and --find-object are mutually exclusive"));
4593
4594         /*
4595          * Most of the time we can say "there are changes"
4596          * only by checking if there are changed paths, but
4597          * --ignore-whitespace* options force us to look
4598          * inside contents.
4599          */
4600
4601         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4602                 options->flags.diff_from_contents = 1;
4603         else
4604                 options->flags.diff_from_contents = 0;
4605
4606         if (options->flags.find_copies_harder)
4607                 options->detect_rename = DIFF_DETECT_COPY;
4608
4609         if (!options->flags.relative_name)
4610                 options->prefix = NULL;
4611         if (options->prefix)
4612                 options->prefix_length = strlen(options->prefix);
4613         else
4614                 options->prefix_length = 0;
4615
4616         if (options->output_format & (DIFF_FORMAT_NAME |
4617                                       DIFF_FORMAT_NAME_STATUS |
4618                                       DIFF_FORMAT_CHECKDIFF |
4619                                       DIFF_FORMAT_NO_OUTPUT))
4620                 options->output_format &= ~(DIFF_FORMAT_RAW |
4621                                             DIFF_FORMAT_NUMSTAT |
4622                                             DIFF_FORMAT_DIFFSTAT |
4623                                             DIFF_FORMAT_SHORTSTAT |
4624                                             DIFF_FORMAT_DIRSTAT |
4625                                             DIFF_FORMAT_SUMMARY |
4626                                             DIFF_FORMAT_PATCH);
4627
4628         /*
4629          * These cases always need recursive; we do not drop caller-supplied
4630          * recursive bits for other formats here.
4631          */
4632         if (options->output_format & (DIFF_FORMAT_PATCH |
4633                                       DIFF_FORMAT_NUMSTAT |
4634                                       DIFF_FORMAT_DIFFSTAT |
4635                                       DIFF_FORMAT_SHORTSTAT |
4636                                       DIFF_FORMAT_DIRSTAT |
4637                                       DIFF_FORMAT_SUMMARY |
4638                                       DIFF_FORMAT_CHECKDIFF))
4639                 options->flags.recursive = 1;
4640         /*
4641          * Also pickaxe would not work very well if you do not say recursive
4642          */
4643         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4644                 options->flags.recursive = 1;
4645         /*
4646          * When patches are generated, submodules diffed against the work tree
4647          * must be checked for dirtiness too so it can be shown in the output
4648          */
4649         if (options->output_format & DIFF_FORMAT_PATCH)
4650                 options->flags.dirty_submodules = 1;
4651
4652         if (options->detect_rename && options->rename_limit < 0)
4653                 options->rename_limit = diff_rename_limit_default;
4654         if (hexsz < options->abbrev)
4655                 options->abbrev = hexsz; /* full */
4656
4657         /*
4658          * It does not make sense to show the first hit we happened
4659          * to have found.  It does not make sense not to return with
4660          * exit code in such a case either.
4661          */
4662         if (options->flags.quick) {
4663                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4664                 options->flags.exit_with_status = 1;
4665         }
4666
4667         options->diff_path_counter = 0;
4668
4669         if (options->flags.follow_renames && options->pathspec.nr != 1)
4670                 die(_("--follow requires exactly one pathspec"));
4671
4672         if (!options->use_color || external_diff())
4673                 options->color_moved = 0;
4674
4675         FREE_AND_NULL(options->parseopts);
4676 }
4677
4678 int parse_long_opt(const char *opt, const char **argv,
4679                    const char **optarg)
4680 {
4681         const char *arg = argv[0];
4682         if (!skip_prefix(arg, "--", &arg))
4683                 return 0;
4684         if (!skip_prefix(arg, opt, &arg))
4685                 return 0;
4686         if (*arg == '=') { /* stuck form: --option=value */
4687                 *optarg = arg + 1;
4688                 return 1;
4689         }
4690         if (*arg != '\0')
4691                 return 0;
4692         /* separate form: --option value */
4693         if (!argv[1])
4694                 die("Option '--%s' requires a value", opt);
4695         *optarg = argv[1];
4696         return 2;
4697 }
4698
4699 static int diff_opt_stat(const struct option *opt, const char *value, int unset)
4700 {
4701         struct diff_options *options = opt->value;
4702         int width = options->stat_width;
4703         int name_width = options->stat_name_width;
4704         int graph_width = options->stat_graph_width;
4705         int count = options->stat_count;
4706         char *end;
4707
4708         BUG_ON_OPT_NEG(unset);
4709
4710         if (!strcmp(opt->long_name, "stat")) {
4711                 if (value) {
4712                         width = strtoul(value, &end, 10);
4713                         if (*end == ',')
4714                                 name_width = strtoul(end+1, &end, 10);
4715                         if (*end == ',')
4716                                 count = strtoul(end+1, &end, 10);
4717                         if (*end)
4718                                 return error(_("invalid --stat value: %s"), value);
4719                 }
4720         } else if (!strcmp(opt->long_name, "stat-width")) {
4721                 width = strtoul(value, &end, 10);
4722                 if (*end)
4723                         return error(_("%s expects a numerical value"),
4724                                      opt->long_name);
4725         } else if (!strcmp(opt->long_name, "stat-name-width")) {
4726                 name_width = strtoul(value, &end, 10);
4727                 if (*end)
4728                         return error(_("%s expects a numerical value"),
4729                                      opt->long_name);
4730         } else if (!strcmp(opt->long_name, "stat-graph-width")) {
4731                 graph_width = strtoul(value, &end, 10);
4732                 if (*end)
4733                         return error(_("%s expects a numerical value"),
4734                                      opt->long_name);
4735         } else if (!strcmp(opt->long_name, "stat-count")) {
4736                 count = strtoul(value, &end, 10);
4737                 if (*end)
4738                         return error(_("%s expects a numerical value"),
4739                                      opt->long_name);
4740         } else
4741                 BUG("%s should not get here", opt->long_name);
4742
4743         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4744         options->stat_name_width = name_width;
4745         options->stat_graph_width = graph_width;
4746         options->stat_width = width;
4747         options->stat_count = count;
4748         return 0;
4749 }
4750
4751 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4752 {
4753         struct strbuf errmsg = STRBUF_INIT;
4754         if (parse_dirstat_params(options, params, &errmsg))
4755                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4756                     errmsg.buf);
4757         strbuf_release(&errmsg);
4758         /*
4759          * The caller knows a dirstat-related option is given from the command
4760          * line; allow it to say "return this_function();"
4761          */
4762         options->output_format |= DIFF_FORMAT_DIRSTAT;
4763         return 1;
4764 }
4765
4766 static const char diff_status_letters[] = {
4767         DIFF_STATUS_ADDED,
4768         DIFF_STATUS_COPIED,
4769         DIFF_STATUS_DELETED,
4770         DIFF_STATUS_MODIFIED,
4771         DIFF_STATUS_RENAMED,
4772         DIFF_STATUS_TYPE_CHANGED,
4773         DIFF_STATUS_UNKNOWN,
4774         DIFF_STATUS_UNMERGED,
4775         DIFF_STATUS_FILTER_AON,
4776         DIFF_STATUS_FILTER_BROKEN,
4777         '\0',
4778 };
4779
4780 static unsigned int filter_bit['Z' + 1];
4781
4782 static void prepare_filter_bits(void)
4783 {
4784         int i;
4785
4786         if (!filter_bit[DIFF_STATUS_ADDED]) {
4787                 for (i = 0; diff_status_letters[i]; i++)
4788                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4789         }
4790 }
4791
4792 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4793 {
4794         return opt->filter & filter_bit[(int) status];
4795 }
4796
4797 unsigned diff_filter_bit(char status)
4798 {
4799         prepare_filter_bits();
4800         return filter_bit[(int) status];
4801 }
4802
4803 static int diff_opt_diff_filter(const struct option *option,
4804                                 const char *optarg, int unset)
4805 {
4806         struct diff_options *opt = option->value;
4807         int i, optch;
4808
4809         BUG_ON_OPT_NEG(unset);
4810         prepare_filter_bits();
4811
4812         /*
4813          * If there is a negation e.g. 'd' in the input, and we haven't
4814          * initialized the filter field with another --diff-filter, start
4815          * from full set of bits, except for AON.
4816          */
4817         if (!opt->filter) {
4818                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4819                         if (optch < 'a' || 'z' < optch)
4820                                 continue;
4821                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4822                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4823                         break;
4824                 }
4825         }
4826
4827         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4828                 unsigned int bit;
4829                 int negate;
4830
4831                 if ('a' <= optch && optch <= 'z') {
4832                         negate = 1;
4833                         optch = toupper(optch);
4834                 } else {
4835                         negate = 0;
4836                 }
4837
4838                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4839                 if (!bit)
4840                         return error(_("unknown change class '%c' in --diff-filter=%s"),
4841                                      optarg[i], optarg);
4842                 if (negate)
4843                         opt->filter &= ~bit;
4844                 else
4845                         opt->filter |= bit;
4846         }
4847         return 0;
4848 }
4849
4850 static void enable_patch_output(int *fmt)
4851 {
4852         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4853         *fmt |= DIFF_FORMAT_PATCH;
4854 }
4855
4856 static int diff_opt_ws_error_highlight(const struct option *option,
4857                                        const char *arg, int unset)
4858 {
4859         struct diff_options *opt = option->value;
4860         int val = parse_ws_error_highlight(arg);
4861
4862         BUG_ON_OPT_NEG(unset);
4863         if (val < 0)
4864                 return error(_("unknown value after ws-error-highlight=%.*s"),
4865                              -1 - val, arg);
4866         opt->ws_error_highlight = val;
4867         return 0;
4868 }
4869
4870 static int diff_opt_find_object(const struct option *option,
4871                                 const char *arg, int unset)
4872 {
4873         struct diff_options *opt = option->value;
4874         struct object_id oid;
4875
4876         BUG_ON_OPT_NEG(unset);
4877         if (get_oid(arg, &oid))
4878                 return error(_("unable to resolve '%s'"), arg);
4879
4880         if (!opt->objfind)
4881                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4882
4883         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4884         opt->flags.recursive = 1;
4885         opt->flags.tree_in_recursive = 1;
4886         oidset_insert(opt->objfind, &oid);
4887         return 0;
4888 }
4889
4890 static int diff_opt_anchored(const struct option *opt,
4891                              const char *arg, int unset)
4892 {
4893         struct diff_options *options = opt->value;
4894
4895         BUG_ON_OPT_NEG(unset);
4896         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4897         ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4898                    options->anchors_alloc);
4899         options->anchors[options->anchors_nr++] = xstrdup(arg);
4900         return 0;
4901 }
4902
4903 static int diff_opt_binary(const struct option *opt,
4904                            const char *arg, int unset)
4905 {
4906         struct diff_options *options = opt->value;
4907
4908         BUG_ON_OPT_NEG(unset);
4909         BUG_ON_OPT_ARG(arg);
4910         enable_patch_output(&options->output_format);
4911         options->flags.binary = 1;
4912         return 0;
4913 }
4914
4915 static int diff_opt_break_rewrites(const struct option *opt,
4916                                    const char *arg, int unset)
4917 {
4918         int *break_opt = opt->value;
4919         int opt1, opt2;
4920
4921         BUG_ON_OPT_NEG(unset);
4922         if (!arg)
4923                 arg = "";
4924         opt1 = parse_rename_score(&arg);
4925         if (*arg == 0)
4926                 opt2 = 0;
4927         else if (*arg != '/')
4928                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4929         else {
4930                 arg++;
4931                 opt2 = parse_rename_score(&arg);
4932         }
4933         if (*arg != 0)
4934                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4935         *break_opt = opt1 | (opt2 << 16);
4936         return 0;
4937 }
4938
4939 static int diff_opt_char(const struct option *opt,
4940                          const char *arg, int unset)
4941 {
4942         char *value = opt->value;
4943
4944         BUG_ON_OPT_NEG(unset);
4945         if (arg[1])
4946                 return error(_("%s expects a character, got '%s'"),
4947                              opt->long_name, arg);
4948         *value = arg[0];
4949         return 0;
4950 }
4951
4952 static int diff_opt_color_moved(const struct option *opt,
4953                                 const char *arg, int unset)
4954 {
4955         struct diff_options *options = opt->value;
4956
4957         if (unset) {
4958                 options->color_moved = COLOR_MOVED_NO;
4959         } else if (!arg) {
4960                 if (diff_color_moved_default)
4961                         options->color_moved = diff_color_moved_default;
4962                 if (options->color_moved == COLOR_MOVED_NO)
4963                         options->color_moved = COLOR_MOVED_DEFAULT;
4964         } else {
4965                 int cm = parse_color_moved(arg);
4966                 if (cm < 0)
4967                         return error(_("bad --color-moved argument: %s"), arg);
4968                 options->color_moved = cm;
4969         }
4970         return 0;
4971 }
4972
4973 static int diff_opt_color_moved_ws(const struct option *opt,
4974                                    const char *arg, int unset)
4975 {
4976         struct diff_options *options = opt->value;
4977         unsigned cm;
4978
4979         if (unset) {
4980                 options->color_moved_ws_handling = 0;
4981                 return 0;
4982         }
4983
4984         cm = parse_color_moved_ws(arg);
4985         if (cm & COLOR_MOVED_WS_ERROR)
4986                 return error(_("invalid mode '%s' in --color-moved-ws"), arg);
4987         options->color_moved_ws_handling = cm;
4988         return 0;
4989 }
4990
4991 static int diff_opt_color_words(const struct option *opt,
4992                                 const char *arg, int unset)
4993 {
4994         struct diff_options *options = opt->value;
4995
4996         BUG_ON_OPT_NEG(unset);
4997         options->use_color = 1;
4998         options->word_diff = DIFF_WORDS_COLOR;
4999         options->word_regex = arg;
5000         return 0;
5001 }
5002
5003 static int diff_opt_compact_summary(const struct option *opt,
5004                                     const char *arg, int unset)
5005 {
5006         struct diff_options *options = opt->value;
5007
5008         BUG_ON_OPT_ARG(arg);
5009         if (unset) {
5010                 options->flags.stat_with_summary = 0;
5011         } else {
5012                 options->flags.stat_with_summary = 1;
5013                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
5014         }
5015         return 0;
5016 }
5017
5018 static int diff_opt_diff_algorithm(const struct option *opt,
5019                                    const char *arg, int unset)
5020 {
5021         struct diff_options *options = opt->value;
5022         long value = parse_algorithm_value(arg);
5023
5024         BUG_ON_OPT_NEG(unset);
5025         if (value < 0)
5026                 return error(_("option diff-algorithm accepts \"myers\", "
5027                                "\"minimal\", \"patience\" and \"histogram\""));
5028
5029         /* clear out previous settings */
5030         DIFF_XDL_CLR(options, NEED_MINIMAL);
5031         options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5032         options->xdl_opts |= value;
5033         return 0;
5034 }
5035
5036 static int diff_opt_dirstat(const struct option *opt,
5037                             const char *arg, int unset)
5038 {
5039         struct diff_options *options = opt->value;
5040
5041         BUG_ON_OPT_NEG(unset);
5042         if (!strcmp(opt->long_name, "cumulative")) {
5043                 if (arg)
5044                         BUG("how come --cumulative take a value?");
5045                 arg = "cumulative";
5046         } else if (!strcmp(opt->long_name, "dirstat-by-file"))
5047                 parse_dirstat_opt(options, "files");
5048         parse_dirstat_opt(options, arg ? arg : "");
5049         return 0;
5050 }
5051
5052 static int diff_opt_find_copies(const struct option *opt,
5053                                 const char *arg, int unset)
5054 {
5055         struct diff_options *options = opt->value;
5056
5057         BUG_ON_OPT_NEG(unset);
5058         if (!arg)
5059                 arg = "";
5060         options->rename_score = parse_rename_score(&arg);
5061         if (*arg != 0)
5062                 return error(_("invalid argument to %s"), opt->long_name);
5063
5064         if (options->detect_rename == DIFF_DETECT_COPY)
5065                 options->flags.find_copies_harder = 1;
5066         else
5067                 options->detect_rename = DIFF_DETECT_COPY;
5068
5069         return 0;
5070 }
5071
5072 static int diff_opt_find_renames(const struct option *opt,
5073                                  const char *arg, int unset)
5074 {
5075         struct diff_options *options = opt->value;
5076
5077         BUG_ON_OPT_NEG(unset);
5078         if (!arg)
5079                 arg = "";
5080         options->rename_score = parse_rename_score(&arg);
5081         if (*arg != 0)
5082                 return error(_("invalid argument to %s"), opt->long_name);
5083
5084         options->detect_rename = DIFF_DETECT_RENAME;
5085         return 0;
5086 }
5087
5088 static int diff_opt_follow(const struct option *opt,
5089                            const char *arg, int unset)
5090 {
5091         struct diff_options *options = opt->value;
5092
5093         BUG_ON_OPT_ARG(arg);
5094         if (unset) {
5095                 options->flags.follow_renames = 0;
5096                 options->flags.default_follow_renames = 0;
5097         } else {
5098                 options->flags.follow_renames = 1;
5099         }
5100         return 0;
5101 }
5102
5103 static int diff_opt_ignore_submodules(const struct option *opt,
5104                                       const char *arg, int unset)
5105 {
5106         struct diff_options *options = opt->value;
5107
5108         BUG_ON_OPT_NEG(unset);
5109         if (!arg)
5110                 arg = "all";
5111         options->flags.override_submodule_config = 1;
5112         handle_ignore_submodules_arg(options, arg);
5113         return 0;
5114 }
5115
5116 static int diff_opt_line_prefix(const struct option *opt,
5117                                 const char *optarg, int unset)
5118 {
5119         struct diff_options *options = opt->value;
5120
5121         BUG_ON_OPT_NEG(unset);
5122         options->line_prefix = optarg;
5123         options->line_prefix_length = strlen(options->line_prefix);
5124         graph_setup_line_prefix(options);
5125         return 0;
5126 }
5127
5128 static int diff_opt_no_prefix(const struct option *opt,
5129                               const char *optarg, int unset)
5130 {
5131         struct diff_options *options = opt->value;
5132
5133         BUG_ON_OPT_NEG(unset);
5134         BUG_ON_OPT_ARG(optarg);
5135         options->a_prefix = "";
5136         options->b_prefix = "";
5137         return 0;
5138 }
5139
5140 static enum parse_opt_result diff_opt_output(struct parse_opt_ctx_t *ctx,
5141                                              const struct option *opt,
5142                                              const char *arg, int unset)
5143 {
5144         struct diff_options *options = opt->value;
5145         char *path;
5146
5147         BUG_ON_OPT_NEG(unset);
5148         path = prefix_filename(ctx->prefix, arg);
5149         options->file = xfopen(path, "w");
5150         options->close_file = 1;
5151         if (options->use_color != GIT_COLOR_ALWAYS)
5152                 options->use_color = GIT_COLOR_NEVER;
5153         free(path);
5154         return 0;
5155 }
5156
5157 static int diff_opt_patience(const struct option *opt,
5158                              const char *arg, int unset)
5159 {
5160         struct diff_options *options = opt->value;
5161         int i;
5162
5163         BUG_ON_OPT_NEG(unset);
5164         BUG_ON_OPT_ARG(arg);
5165         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5166         /*
5167          * Both --patience and --anchored use PATIENCE_DIFF
5168          * internally, so remove any anchors previously
5169          * specified.
5170          */
5171         for (i = 0; i < options->anchors_nr; i++)
5172                 free(options->anchors[i]);
5173         options->anchors_nr = 0;
5174         return 0;
5175 }
5176
5177 static int diff_opt_pickaxe_regex(const struct option *opt,
5178                                   const char *arg, int unset)
5179 {
5180         struct diff_options *options = opt->value;
5181
5182         BUG_ON_OPT_NEG(unset);
5183         options->pickaxe = arg;
5184         options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5185         return 0;
5186 }
5187
5188 static int diff_opt_pickaxe_string(const struct option *opt,
5189                                    const char *arg, int unset)
5190 {
5191         struct diff_options *options = opt->value;
5192
5193         BUG_ON_OPT_NEG(unset);
5194         options->pickaxe = arg;
5195         options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5196         return 0;
5197 }
5198
5199 static int diff_opt_relative(const struct option *opt,
5200                              const char *arg, int unset)
5201 {
5202         struct diff_options *options = opt->value;
5203
5204         options->flags.relative_name = !unset;
5205         if (arg)
5206                 options->prefix = arg;
5207         return 0;
5208 }
5209
5210 static int diff_opt_submodule(const struct option *opt,
5211                               const char *arg, int unset)
5212 {
5213         struct diff_options *options = opt->value;
5214
5215         BUG_ON_OPT_NEG(unset);
5216         if (!arg)
5217                 arg = "log";
5218         if (parse_submodule_params(options, arg))
5219                 return error(_("failed to parse --submodule option parameter: '%s'"),
5220                              arg);
5221         return 0;
5222 }
5223
5224 static int diff_opt_textconv(const struct option *opt,
5225                              const char *arg, int unset)
5226 {
5227         struct diff_options *options = opt->value;
5228
5229         BUG_ON_OPT_ARG(arg);
5230         if (unset) {
5231                 options->flags.allow_textconv = 0;
5232         } else {
5233                 options->flags.allow_textconv = 1;
5234                 options->flags.textconv_set_via_cmdline = 1;
5235         }
5236         return 0;
5237 }
5238
5239 static int diff_opt_unified(const struct option *opt,
5240                             const char *arg, int unset)
5241 {
5242         struct diff_options *options = opt->value;
5243         char *s;
5244
5245         BUG_ON_OPT_NEG(unset);
5246
5247         if (arg) {
5248                 options->context = strtol(arg, &s, 10);
5249                 if (*s)
5250                         return error(_("%s expects a numerical value"), "--unified");
5251         }
5252         enable_patch_output(&options->output_format);
5253
5254         return 0;
5255 }
5256
5257 static int diff_opt_word_diff(const struct option *opt,
5258                               const char *arg, int unset)
5259 {
5260         struct diff_options *options = opt->value;
5261
5262         BUG_ON_OPT_NEG(unset);
5263         if (arg) {
5264                 if (!strcmp(arg, "plain"))
5265                         options->word_diff = DIFF_WORDS_PLAIN;
5266                 else if (!strcmp(arg, "color")) {
5267                         options->use_color = 1;
5268                         options->word_diff = DIFF_WORDS_COLOR;
5269                 }
5270                 else if (!strcmp(arg, "porcelain"))
5271                         options->word_diff = DIFF_WORDS_PORCELAIN;
5272                 else if (!strcmp(arg, "none"))
5273                         options->word_diff = DIFF_WORDS_NONE;
5274                 else
5275                         return error(_("bad --word-diff argument: %s"), arg);
5276         } else {
5277                 if (options->word_diff == DIFF_WORDS_NONE)
5278                         options->word_diff = DIFF_WORDS_PLAIN;
5279         }
5280         return 0;
5281 }
5282
5283 static int diff_opt_word_diff_regex(const struct option *opt,
5284                                     const char *arg, int unset)
5285 {
5286         struct diff_options *options = opt->value;
5287
5288         BUG_ON_OPT_NEG(unset);
5289         if (options->word_diff == DIFF_WORDS_NONE)
5290                 options->word_diff = DIFF_WORDS_PLAIN;
5291         options->word_regex = arg;
5292         return 0;
5293 }
5294
5295 static void prep_parse_options(struct diff_options *options)
5296 {
5297         struct option parseopts[] = {
5298                 OPT_GROUP(N_("Diff output format options")),
5299                 OPT_BITOP('p', "patch", &options->output_format,
5300                           N_("generate patch"),
5301                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5302                 OPT_BIT_F('s', "no-patch", &options->output_format,
5303                           N_("suppress diff output"),
5304                           DIFF_FORMAT_NO_OUTPUT, PARSE_OPT_NONEG),
5305                 OPT_BITOP('u', NULL, &options->output_format,
5306                           N_("generate patch"),
5307                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5308                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
5309                                N_("generate diffs with <n> lines context"),
5310                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_unified),
5311                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
5312                          N_("generate diffs with <n> lines context")),
5313                 OPT_BIT_F(0, "raw", &options->output_format,
5314                           N_("generate the diff in raw format"),
5315                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
5316                 OPT_BITOP(0, "patch-with-raw", &options->output_format,
5317                           N_("synonym for '-p --raw'"),
5318                           DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW,
5319                           DIFF_FORMAT_NO_OUTPUT),
5320                 OPT_BITOP(0, "patch-with-stat", &options->output_format,
5321                           N_("synonym for '-p --stat'"),
5322                           DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT,
5323                           DIFF_FORMAT_NO_OUTPUT),
5324                 OPT_BIT_F(0, "numstat", &options->output_format,
5325                           N_("machine friendly --stat"),
5326                           DIFF_FORMAT_NUMSTAT, PARSE_OPT_NONEG),
5327                 OPT_BIT_F(0, "shortstat", &options->output_format,
5328                           N_("output only the last line of --stat"),
5329                           DIFF_FORMAT_SHORTSTAT, PARSE_OPT_NONEG),
5330                 OPT_CALLBACK_F('X', "dirstat", options, N_("<param1,param2>..."),
5331                                N_("output the distribution of relative amount of changes for each sub-directory"),
5332                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5333                                diff_opt_dirstat),
5334                 OPT_CALLBACK_F(0, "cumulative", options, NULL,
5335                                N_("synonym for --dirstat=cumulative"),
5336                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5337                                diff_opt_dirstat),
5338                 OPT_CALLBACK_F(0, "dirstat-by-file", options, N_("<param1,param2>..."),
5339                                N_("synonym for --dirstat=files,param1,param2..."),
5340                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5341                                diff_opt_dirstat),
5342                 OPT_BIT_F(0, "check", &options->output_format,
5343                           N_("warn if changes introduce conflict markers or whitespace errors"),
5344                           DIFF_FORMAT_CHECKDIFF, PARSE_OPT_NONEG),
5345                 OPT_BIT_F(0, "summary", &options->output_format,
5346                           N_("condensed summary such as creations, renames and mode changes"),
5347                           DIFF_FORMAT_SUMMARY, PARSE_OPT_NONEG),
5348                 OPT_BIT_F(0, "name-only", &options->output_format,
5349                           N_("show only names of changed files"),
5350                           DIFF_FORMAT_NAME, PARSE_OPT_NONEG),
5351                 OPT_BIT_F(0, "name-status", &options->output_format,
5352                           N_("show only names and status of changed files"),
5353                           DIFF_FORMAT_NAME_STATUS, PARSE_OPT_NONEG),
5354                 OPT_CALLBACK_F(0, "stat", options, N_("<width>[,<name-width>[,<count>]]"),
5355                                N_("generate diffstat"),
5356                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_stat),
5357                 OPT_CALLBACK_F(0, "stat-width", options, N_("<width>"),
5358                                N_("generate diffstat with a given width"),
5359                                PARSE_OPT_NONEG, diff_opt_stat),
5360                 OPT_CALLBACK_F(0, "stat-name-width", options, N_("<width>"),
5361                                N_("generate diffstat with a given name width"),
5362                                PARSE_OPT_NONEG, diff_opt_stat),
5363                 OPT_CALLBACK_F(0, "stat-graph-width", options, N_("<width>"),
5364                                N_("generate diffstat with a given graph width"),
5365                                PARSE_OPT_NONEG, diff_opt_stat),
5366                 OPT_CALLBACK_F(0, "stat-count", options, N_("<count>"),
5367                                N_("generate diffstat with limited lines"),
5368                                PARSE_OPT_NONEG, diff_opt_stat),
5369                 OPT_CALLBACK_F(0, "compact-summary", options, NULL,
5370                                N_("generate compact summary in diffstat"),
5371                                PARSE_OPT_NOARG, diff_opt_compact_summary),
5372                 OPT_CALLBACK_F(0, "binary", options, NULL,
5373                                N_("output a binary diff that can be applied"),
5374                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_binary),
5375                 OPT_BOOL(0, "full-index", &options->flags.full_index,
5376                          N_("show full pre- and post-image object names on the \"index\" lines")),
5377                 OPT_COLOR_FLAG(0, "color", &options->use_color,
5378                                N_("show colored diff")),
5379                 OPT_CALLBACK_F(0, "ws-error-highlight", options, N_("<kind>"),
5380                                N_("highlight whitespace errors in the 'context', 'old' or 'new' lines in the diff"),
5381                                PARSE_OPT_NONEG, diff_opt_ws_error_highlight),
5382                 OPT_SET_INT('z', NULL, &options->line_termination,
5383                             N_("do not munge pathnames and use NULs as output field terminators in --raw or --numstat"),
5384                             0),
5385                 OPT__ABBREV(&options->abbrev),
5386                 OPT_STRING_F(0, "src-prefix", &options->a_prefix, N_("<prefix>"),
5387                              N_("show the given source prefix instead of \"a/\""),
5388                              PARSE_OPT_NONEG),
5389                 OPT_STRING_F(0, "dst-prefix", &options->b_prefix, N_("<prefix>"),
5390                              N_("show the given destination prefix instead of \"b/\""),
5391                              PARSE_OPT_NONEG),
5392                 OPT_CALLBACK_F(0, "line-prefix", options, N_("<prefix>"),
5393                                N_("prepend an additional prefix to every line of output"),
5394                                PARSE_OPT_NONEG, diff_opt_line_prefix),
5395                 OPT_CALLBACK_F(0, "no-prefix", options, NULL,
5396                                N_("do not show any source or destination prefix"),
5397                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_no_prefix),
5398                 OPT_INTEGER_F(0, "inter-hunk-context", &options->interhunkcontext,
5399                               N_("show context between diff hunks up to the specified number of lines"),
5400                               PARSE_OPT_NONEG),
5401                 OPT_CALLBACK_F(0, "output-indicator-new",
5402                                &options->output_indicators[OUTPUT_INDICATOR_NEW],
5403                                N_("<char>"),
5404                                N_("specify the character to indicate a new line instead of '+'"),
5405                                PARSE_OPT_NONEG, diff_opt_char),
5406                 OPT_CALLBACK_F(0, "output-indicator-old",
5407                                &options->output_indicators[OUTPUT_INDICATOR_OLD],
5408                                N_("<char>"),
5409                                N_("specify the character to indicate an old line instead of '-'"),
5410                                PARSE_OPT_NONEG, diff_opt_char),
5411                 OPT_CALLBACK_F(0, "output-indicator-context",
5412                                &options->output_indicators[OUTPUT_INDICATOR_CONTEXT],
5413                                N_("<char>"),
5414                                N_("specify the character to indicate a context instead of ' '"),
5415                                PARSE_OPT_NONEG, diff_opt_char),
5416
5417                 OPT_GROUP(N_("Diff rename options")),
5418                 OPT_CALLBACK_F('B', "break-rewrites", &options->break_opt, N_("<n>[/<m>]"),
5419                                N_("break complete rewrite changes into pairs of delete and create"),
5420                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5421                                diff_opt_break_rewrites),
5422                 OPT_CALLBACK_F('M', "find-renames", options, N_("<n>"),
5423                                N_("detect renames"),
5424                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5425                                diff_opt_find_renames),
5426                 OPT_SET_INT_F('D', "irreversible-delete", &options->irreversible_delete,
5427                               N_("omit the preimage for deletes"),
5428                               1, PARSE_OPT_NONEG),
5429                 OPT_CALLBACK_F('C', "find-copies", options, N_("<n>"),
5430                                N_("detect copies"),
5431                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5432                                diff_opt_find_copies),
5433                 OPT_BOOL(0, "find-copies-harder", &options->flags.find_copies_harder,
5434                          N_("use unmodified files as source to find copies")),
5435                 OPT_SET_INT_F(0, "no-renames", &options->detect_rename,
5436                               N_("disable rename detection"),
5437                               0, PARSE_OPT_NONEG),
5438                 OPT_BOOL(0, "rename-empty", &options->flags.rename_empty,
5439                          N_("use empty blobs as rename source")),
5440                 OPT_CALLBACK_F(0, "follow", options, NULL,
5441                                N_("continue listing the history of a file beyond renames"),
5442                                PARSE_OPT_NOARG, diff_opt_follow),
5443                 OPT_INTEGER('l', NULL, &options->rename_limit,
5444                             N_("prevent rename/copy detection if the number of rename/copy targets exceeds given limit")),
5445
5446                 OPT_GROUP(N_("Diff algorithm options")),
5447                 OPT_BIT(0, "minimal", &options->xdl_opts,
5448                         N_("produce the smallest possible diff"),
5449                         XDF_NEED_MINIMAL),
5450                 OPT_BIT_F('w', "ignore-all-space", &options->xdl_opts,
5451                           N_("ignore whitespace when comparing lines"),
5452                           XDF_IGNORE_WHITESPACE, PARSE_OPT_NONEG),
5453                 OPT_BIT_F('b', "ignore-space-change", &options->xdl_opts,
5454                           N_("ignore changes in amount of whitespace"),
5455                           XDF_IGNORE_WHITESPACE_CHANGE, PARSE_OPT_NONEG),
5456                 OPT_BIT_F(0, "ignore-space-at-eol", &options->xdl_opts,
5457                           N_("ignore changes in whitespace at EOL"),
5458                           XDF_IGNORE_WHITESPACE_AT_EOL, PARSE_OPT_NONEG),
5459                 OPT_BIT_F(0, "ignore-cr-at-eol", &options->xdl_opts,
5460                           N_("ignore carrier-return at the end of line"),
5461                           XDF_IGNORE_CR_AT_EOL, PARSE_OPT_NONEG),
5462                 OPT_BIT_F(0, "ignore-blank-lines", &options->xdl_opts,
5463                           N_("ignore changes whose lines are all blank"),
5464                           XDF_IGNORE_BLANK_LINES, PARSE_OPT_NONEG),
5465                 OPT_BIT(0, "indent-heuristic", &options->xdl_opts,
5466                         N_("heuristic to shift diff hunk boundaries for easy reading"),
5467                         XDF_INDENT_HEURISTIC),
5468                 OPT_CALLBACK_F(0, "patience", options, NULL,
5469                                N_("generate diff using the \"patience diff\" algorithm"),
5470                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5471                                diff_opt_patience),
5472                 OPT_BITOP(0, "histogram", &options->xdl_opts,
5473                           N_("generate diff using the \"histogram diff\" algorithm"),
5474                           XDF_HISTOGRAM_DIFF, XDF_DIFF_ALGORITHM_MASK),
5475                 OPT_CALLBACK_F(0, "diff-algorithm", options, N_("<algorithm>"),
5476                                N_("choose a diff algorithm"),
5477                                PARSE_OPT_NONEG, diff_opt_diff_algorithm),
5478                 OPT_CALLBACK_F(0, "anchored", options, N_("<text>"),
5479                                N_("generate diff using the \"anchored diff\" algorithm"),
5480                                PARSE_OPT_NONEG, diff_opt_anchored),
5481                 OPT_CALLBACK_F(0, "word-diff", options, N_("<mode>"),
5482                                N_("show word diff, using <mode> to delimit changed words"),
5483                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_word_diff),
5484                 OPT_CALLBACK_F(0, "word-diff-regex", options, N_("<regex>"),
5485                                N_("use <regex> to decide what a word is"),
5486                                PARSE_OPT_NONEG, diff_opt_word_diff_regex),
5487                 OPT_CALLBACK_F(0, "color-words", options, N_("<regex>"),
5488                                N_("equivalent to --word-diff=color --word-diff-regex=<regex>"),
5489                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_color_words),
5490                 OPT_CALLBACK_F(0, "color-moved", options, N_("<mode>"),
5491                                N_("moved lines of code are colored differently"),
5492                                PARSE_OPT_OPTARG, diff_opt_color_moved),
5493                 OPT_CALLBACK_F(0, "color-moved-ws", options, N_("<mode>"),
5494                                N_("how white spaces are ignored in --color-moved"),
5495                                0, diff_opt_color_moved_ws),
5496
5497                 OPT_GROUP(N_("Other diff options")),
5498                 OPT_CALLBACK_F(0, "relative", options, N_("<prefix>"),
5499                                N_("when run from subdir, exclude changes outside and show relative paths"),
5500                                PARSE_OPT_OPTARG,
5501                                diff_opt_relative),
5502                 OPT_BOOL('a', "text", &options->flags.text,
5503                          N_("treat all files as text")),
5504                 OPT_BOOL('R', NULL, &options->flags.reverse_diff,
5505                          N_("swap two inputs, reverse the diff")),
5506                 OPT_BOOL(0, "exit-code", &options->flags.exit_with_status,
5507                          N_("exit with 1 if there were differences, 0 otherwise")),
5508                 OPT_BOOL(0, "quiet", &options->flags.quick,
5509                          N_("disable all output of the program")),
5510                 OPT_BOOL(0, "ext-diff", &options->flags.allow_external,
5511                          N_("allow an external diff helper to be executed")),
5512                 OPT_CALLBACK_F(0, "textconv", options, NULL,
5513                                N_("run external text conversion filters when comparing binary files"),
5514                                PARSE_OPT_NOARG, diff_opt_textconv),
5515                 OPT_CALLBACK_F(0, "ignore-submodules", options, N_("<when>"),
5516                                N_("ignore changes to submodules in the diff generation"),
5517                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5518                                diff_opt_ignore_submodules),
5519                 OPT_CALLBACK_F(0, "submodule", options, N_("<format>"),
5520                                N_("specify how differences in submodules are shown"),
5521                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5522                                diff_opt_submodule),
5523                 OPT_SET_INT_F(0, "ita-invisible-in-index", &options->ita_invisible_in_index,
5524                               N_("hide 'git add -N' entries from the index"),
5525                               1, PARSE_OPT_NONEG),
5526                 OPT_SET_INT_F(0, "ita-visible-in-index", &options->ita_invisible_in_index,
5527                               N_("treat 'git add -N' entries as real in the index"),
5528                               0, PARSE_OPT_NONEG),
5529                 OPT_CALLBACK_F('S', NULL, options, N_("<string>"),
5530                                N_("look for differences that change the number of occurrences of the specified string"),
5531                                0, diff_opt_pickaxe_string),
5532                 OPT_CALLBACK_F('G', NULL, options, N_("<regex>"),
5533                                N_("look for differences that change the number of occurrences of the specified regex"),
5534                                0, diff_opt_pickaxe_regex),
5535                 OPT_BIT_F(0, "pickaxe-all", &options->pickaxe_opts,
5536                           N_("show all changes in the changeset with -S or -G"),
5537                           DIFF_PICKAXE_ALL, PARSE_OPT_NONEG),
5538                 OPT_BIT_F(0, "pickaxe-regex", &options->pickaxe_opts,
5539                           N_("treat <string> in -S as extended POSIX regular expression"),
5540                           DIFF_PICKAXE_REGEX, PARSE_OPT_NONEG),
5541                 OPT_FILENAME('O', NULL, &options->orderfile,
5542                              N_("control the order in which files appear in the output")),
5543                 OPT_CALLBACK_F(0, "find-object", options, N_("<object-id>"),
5544                                N_("look for differences that change the number of occurrences of the specified object"),
5545                                PARSE_OPT_NONEG, diff_opt_find_object),
5546                 OPT_CALLBACK_F(0, "diff-filter", options, N_("[(A|C|D|M|R|T|U|X|B)...[*]]"),
5547                                N_("select files by diff type"),
5548                                PARSE_OPT_NONEG, diff_opt_diff_filter),
5549                 { OPTION_CALLBACK, 0, "output", options, N_("<file>"),
5550                   N_("Output to a specific file"),
5551                   PARSE_OPT_NONEG, NULL, 0, diff_opt_output },
5552
5553                 OPT_END()
5554         };
5555
5556         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
5557         memcpy(options->parseopts, parseopts, sizeof(parseopts));
5558 }
5559
5560 int diff_opt_parse(struct diff_options *options,
5561                    const char **av, int ac, const char *prefix)
5562 {
5563         if (!prefix)
5564                 prefix = "";
5565
5566         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
5567                            PARSE_OPT_KEEP_DASHDASH |
5568                            PARSE_OPT_KEEP_UNKNOWN |
5569                            PARSE_OPT_NO_INTERNAL_HELP |
5570                            PARSE_OPT_ONE_SHOT |
5571                            PARSE_OPT_STOP_AT_NON_OPTION);
5572
5573         return ac;
5574 }
5575
5576 int parse_rename_score(const char **cp_p)
5577 {
5578         unsigned long num, scale;
5579         int ch, dot;
5580         const char *cp = *cp_p;
5581
5582         num = 0;
5583         scale = 1;
5584         dot = 0;
5585         for (;;) {
5586                 ch = *cp;
5587                 if ( !dot && ch == '.' ) {
5588                         scale = 1;
5589                         dot = 1;
5590                 } else if ( ch == '%' ) {
5591                         scale = dot ? scale*100 : 100;
5592                         cp++;   /* % is always at the end */
5593                         break;
5594                 } else if ( ch >= '0' && ch <= '9' ) {
5595                         if ( scale < 100000 ) {
5596                                 scale *= 10;
5597                                 num = (num*10) + (ch-'0');
5598                         }
5599                 } else {
5600                         break;
5601                 }
5602                 cp++;
5603         }
5604         *cp_p = cp;
5605
5606         /* user says num divided by scale and we say internally that
5607          * is MAX_SCORE * num / scale.
5608          */
5609         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5610 }
5611
5612 struct diff_queue_struct diff_queued_diff;
5613
5614 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5615 {
5616         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5617         queue->queue[queue->nr++] = dp;
5618 }
5619
5620 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5621                                  struct diff_filespec *one,
5622                                  struct diff_filespec *two)
5623 {
5624         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5625         dp->one = one;
5626         dp->two = two;
5627         if (queue)
5628                 diff_q(queue, dp);
5629         return dp;
5630 }
5631
5632 void diff_free_filepair(struct diff_filepair *p)
5633 {
5634         free_filespec(p->one);
5635         free_filespec(p->two);
5636         free(p);
5637 }
5638
5639 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5640 {
5641         int abblen;
5642         const char *abbrev;
5643
5644         /* Do we want all 40 hex characters? */
5645         if (len == the_hash_algo->hexsz)
5646                 return oid_to_hex(oid);
5647
5648         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5649         abbrev = diff_abbrev_oid(oid, len);
5650
5651         if (!print_sha1_ellipsis())
5652                 return abbrev;
5653
5654         abblen = strlen(abbrev);
5655
5656         /*
5657          * In well-behaved cases, where the abbreviated result is the
5658          * same as the requested length, append three dots after the
5659          * abbreviation (hence the whole logic is limited to the case
5660          * where abblen < 37); when the actual abbreviated result is a
5661          * bit longer than the requested length, we reduce the number
5662          * of dots so that they match the well-behaved ones.  However,
5663          * if the actual abbreviation is longer than the requested
5664          * length by more than three, we give up on aligning, and add
5665          * three dots anyway, to indicate that the output is not the
5666          * full object name.  Yes, this may be suboptimal, but this
5667          * appears only in "diff --raw --abbrev" output and it is not
5668          * worth the effort to change it now.  Note that this would
5669          * likely to work fine when the automatic sizing of default
5670          * abbreviation length is used--we would be fed -1 in "len" in
5671          * that case, and will end up always appending three-dots, but
5672          * the automatic sizing is supposed to give abblen that ensures
5673          * uniqueness across all objects (statistically speaking).
5674          */
5675         if (abblen < the_hash_algo->hexsz - 3) {
5676                 static char hex[GIT_MAX_HEXSZ + 1];
5677                 if (len < abblen && abblen <= len + 2)
5678                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5679                 else
5680                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5681                 return hex;
5682         }
5683
5684         return oid_to_hex(oid);
5685 }
5686
5687 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5688 {
5689         int line_termination = opt->line_termination;
5690         int inter_name_termination = line_termination ? '\t' : '\0';
5691
5692         fprintf(opt->file, "%s", diff_line_prefix(opt));
5693         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5694                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5695                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5696                 fprintf(opt->file, "%s ",
5697                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5698         }
5699         if (p->score) {
5700                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5701                         inter_name_termination);
5702         } else {
5703                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5704         }
5705
5706         if (p->status == DIFF_STATUS_COPIED ||
5707             p->status == DIFF_STATUS_RENAMED) {
5708                 const char *name_a, *name_b;
5709                 name_a = p->one->path;
5710                 name_b = p->two->path;
5711                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5712                 write_name_quoted(name_a, opt->file, inter_name_termination);
5713                 write_name_quoted(name_b, opt->file, line_termination);
5714         } else {
5715                 const char *name_a, *name_b;
5716                 name_a = p->one->mode ? p->one->path : p->two->path;
5717                 name_b = NULL;
5718                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5719                 write_name_quoted(name_a, opt->file, line_termination);
5720         }
5721 }
5722
5723 int diff_unmodified_pair(struct diff_filepair *p)
5724 {
5725         /* This function is written stricter than necessary to support
5726          * the currently implemented transformers, but the idea is to
5727          * let transformers to produce diff_filepairs any way they want,
5728          * and filter and clean them up here before producing the output.
5729          */
5730         struct diff_filespec *one = p->one, *two = p->two;
5731
5732         if (DIFF_PAIR_UNMERGED(p))
5733                 return 0; /* unmerged is interesting */
5734
5735         /* deletion, addition, mode or type change
5736          * and rename are all interesting.
5737          */
5738         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5739             DIFF_PAIR_MODE_CHANGED(p) ||
5740             strcmp(one->path, two->path))
5741                 return 0;
5742
5743         /* both are valid and point at the same path.  that is, we are
5744          * dealing with a change.
5745          */
5746         if (one->oid_valid && two->oid_valid &&
5747             oideq(&one->oid, &two->oid) &&
5748             !one->dirty_submodule && !two->dirty_submodule)
5749                 return 1; /* no change */
5750         if (!one->oid_valid && !two->oid_valid)
5751                 return 1; /* both look at the same file on the filesystem. */
5752         return 0;
5753 }
5754
5755 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5756 {
5757         if (diff_unmodified_pair(p))
5758                 return;
5759
5760         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5761             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5762                 return; /* no tree diffs in patch format */
5763
5764         run_diff(p, o);
5765 }
5766
5767 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5768                             struct diffstat_t *diffstat)
5769 {
5770         if (diff_unmodified_pair(p))
5771                 return;
5772
5773         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5774             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5775                 return; /* no useful stat for tree diffs */
5776
5777         run_diffstat(p, o, diffstat);
5778 }
5779
5780 static void diff_flush_checkdiff(struct diff_filepair *p,
5781                 struct diff_options *o)
5782 {
5783         if (diff_unmodified_pair(p))
5784                 return;
5785
5786         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5787             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5788                 return; /* nothing to check in tree diffs */
5789
5790         run_checkdiff(p, o);
5791 }
5792
5793 int diff_queue_is_empty(void)
5794 {
5795         struct diff_queue_struct *q = &diff_queued_diff;
5796         int i;
5797         for (i = 0; i < q->nr; i++)
5798                 if (!diff_unmodified_pair(q->queue[i]))
5799                         return 0;
5800         return 1;
5801 }
5802
5803 #if DIFF_DEBUG
5804 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5805 {
5806         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5807                 x, one ? one : "",
5808                 s->path,
5809                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5810                 s->mode,
5811                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5812         fprintf(stderr, "queue[%d] %s size %lu\n",
5813                 x, one ? one : "",
5814                 s->size);
5815 }
5816
5817 void diff_debug_filepair(const struct diff_filepair *p, int i)
5818 {
5819         diff_debug_filespec(p->one, i, "one");
5820         diff_debug_filespec(p->two, i, "two");
5821         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5822                 p->score, p->status ? p->status : '?',
5823                 p->one->rename_used, p->broken_pair);
5824 }
5825
5826 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5827 {
5828         int i;
5829         if (msg)
5830                 fprintf(stderr, "%s\n", msg);
5831         fprintf(stderr, "q->nr = %d\n", q->nr);
5832         for (i = 0; i < q->nr; i++) {
5833                 struct diff_filepair *p = q->queue[i];
5834                 diff_debug_filepair(p, i);
5835         }
5836 }
5837 #endif
5838
5839 static void diff_resolve_rename_copy(void)
5840 {
5841         int i;
5842         struct diff_filepair *p;
5843         struct diff_queue_struct *q = &diff_queued_diff;
5844
5845         diff_debug_queue("resolve-rename-copy", q);
5846
5847         for (i = 0; i < q->nr; i++) {
5848                 p = q->queue[i];
5849                 p->status = 0; /* undecided */
5850                 if (DIFF_PAIR_UNMERGED(p))
5851                         p->status = DIFF_STATUS_UNMERGED;
5852                 else if (!DIFF_FILE_VALID(p->one))
5853                         p->status = DIFF_STATUS_ADDED;
5854                 else if (!DIFF_FILE_VALID(p->two))
5855                         p->status = DIFF_STATUS_DELETED;
5856                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5857                         p->status = DIFF_STATUS_TYPE_CHANGED;
5858
5859                 /* from this point on, we are dealing with a pair
5860                  * whose both sides are valid and of the same type, i.e.
5861                  * either in-place edit or rename/copy edit.
5862                  */
5863                 else if (DIFF_PAIR_RENAME(p)) {
5864                         /*
5865                          * A rename might have re-connected a broken
5866                          * pair up, causing the pathnames to be the
5867                          * same again. If so, that's not a rename at
5868                          * all, just a modification..
5869                          *
5870                          * Otherwise, see if this source was used for
5871                          * multiple renames, in which case we decrement
5872                          * the count, and call it a copy.
5873                          */
5874                         if (!strcmp(p->one->path, p->two->path))
5875                                 p->status = DIFF_STATUS_MODIFIED;
5876                         else if (--p->one->rename_used > 0)
5877                                 p->status = DIFF_STATUS_COPIED;
5878                         else
5879                                 p->status = DIFF_STATUS_RENAMED;
5880                 }
5881                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5882                          p->one->mode != p->two->mode ||
5883                          p->one->dirty_submodule ||
5884                          p->two->dirty_submodule ||
5885                          is_null_oid(&p->one->oid))
5886                         p->status = DIFF_STATUS_MODIFIED;
5887                 else {
5888                         /* This is a "no-change" entry and should not
5889                          * happen anymore, but prepare for broken callers.
5890                          */
5891                         error("feeding unmodified %s to diffcore",
5892                               p->one->path);
5893                         p->status = DIFF_STATUS_UNKNOWN;
5894                 }
5895         }
5896         diff_debug_queue("resolve-rename-copy done", q);
5897 }
5898
5899 static int check_pair_status(struct diff_filepair *p)
5900 {
5901         switch (p->status) {
5902         case DIFF_STATUS_UNKNOWN:
5903                 return 0;
5904         case 0:
5905                 die("internal error in diff-resolve-rename-copy");
5906         default:
5907                 return 1;
5908         }
5909 }
5910
5911 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5912 {
5913         int fmt = opt->output_format;
5914
5915         if (fmt & DIFF_FORMAT_CHECKDIFF)
5916                 diff_flush_checkdiff(p, opt);
5917         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5918                 diff_flush_raw(p, opt);
5919         else if (fmt & DIFF_FORMAT_NAME) {
5920                 const char *name_a, *name_b;
5921                 name_a = p->two->path;
5922                 name_b = NULL;
5923                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5924                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5925                 write_name_quoted(name_a, opt->file, opt->line_termination);
5926         }
5927 }
5928
5929 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5930 {
5931         struct strbuf sb = STRBUF_INIT;
5932         if (fs->mode)
5933                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5934         else
5935                 strbuf_addf(&sb, " %s ", newdelete);
5936
5937         quote_c_style(fs->path, &sb, NULL, 0);
5938         strbuf_addch(&sb, '\n');
5939         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5940                          sb.buf, sb.len, 0);
5941         strbuf_release(&sb);
5942 }
5943
5944 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5945                 int show_name)
5946 {
5947         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5948                 struct strbuf sb = STRBUF_INIT;
5949                 strbuf_addf(&sb, " mode change %06o => %06o",
5950                             p->one->mode, p->two->mode);
5951                 if (show_name) {
5952                         strbuf_addch(&sb, ' ');
5953                         quote_c_style(p->two->path, &sb, NULL, 0);
5954                 }
5955                 strbuf_addch(&sb, '\n');
5956                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5957                                  sb.buf, sb.len, 0);
5958                 strbuf_release(&sb);
5959         }
5960 }
5961
5962 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5963                 struct diff_filepair *p)
5964 {
5965         struct strbuf sb = STRBUF_INIT;
5966         struct strbuf names = STRBUF_INIT;
5967
5968         pprint_rename(&names, p->one->path, p->two->path);
5969         strbuf_addf(&sb, " %s %s (%d%%)\n",
5970                     renamecopy, names.buf, similarity_index(p));
5971         strbuf_release(&names);
5972         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5973                                  sb.buf, sb.len, 0);
5974         show_mode_change(opt, p, 0);
5975         strbuf_release(&sb);
5976 }
5977
5978 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5979 {
5980         switch(p->status) {
5981         case DIFF_STATUS_DELETED:
5982                 show_file_mode_name(opt, "delete", p->one);
5983                 break;
5984         case DIFF_STATUS_ADDED:
5985                 show_file_mode_name(opt, "create", p->two);
5986                 break;
5987         case DIFF_STATUS_COPIED:
5988                 show_rename_copy(opt, "copy", p);
5989                 break;
5990         case DIFF_STATUS_RENAMED:
5991                 show_rename_copy(opt, "rename", p);
5992                 break;
5993         default:
5994                 if (p->score) {
5995                         struct strbuf sb = STRBUF_INIT;
5996                         strbuf_addstr(&sb, " rewrite ");
5997                         quote_c_style(p->two->path, &sb, NULL, 0);
5998                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5999                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
6000                                          sb.buf, sb.len, 0);
6001                         strbuf_release(&sb);
6002                 }
6003                 show_mode_change(opt, p, !p->score);
6004                 break;
6005         }
6006 }
6007
6008 struct patch_id_t {
6009         git_hash_ctx *ctx;
6010         int patchlen;
6011 };
6012
6013 static int remove_space(char *line, int len)
6014 {
6015         int i;
6016         char *dst = line;
6017         unsigned char c;
6018
6019         for (i = 0; i < len; i++)
6020                 if (!isspace((c = line[i])))
6021                         *dst++ = c;
6022
6023         return dst - line;
6024 }
6025
6026 void flush_one_hunk(struct object_id *result, git_hash_ctx *ctx)
6027 {
6028         unsigned char hash[GIT_MAX_RAWSZ];
6029         unsigned short carry = 0;
6030         int i;
6031
6032         the_hash_algo->final_fn(hash, ctx);
6033         the_hash_algo->init_fn(ctx);
6034         /* 20-byte sum, with carry */
6035         for (i = 0; i < the_hash_algo->rawsz; ++i) {
6036                 carry += result->hash[i] + hash[i];
6037                 result->hash[i] = carry;
6038                 carry >>= 8;
6039         }
6040 }
6041
6042 static void patch_id_consume(void *priv, char *line, unsigned long len)
6043 {
6044         struct patch_id_t *data = priv;
6045         int new_len;
6046
6047         new_len = remove_space(line, len);
6048
6049         the_hash_algo->update_fn(data->ctx, line, new_len);
6050         data->patchlen += new_len;
6051 }
6052
6053 static void patch_id_add_string(git_hash_ctx *ctx, const char *str)
6054 {
6055         the_hash_algo->update_fn(ctx, str, strlen(str));
6056 }
6057
6058 static void patch_id_add_mode(git_hash_ctx *ctx, unsigned mode)
6059 {
6060         /* large enough for 2^32 in octal */
6061         char buf[12];
6062         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
6063         the_hash_algo->update_fn(ctx, buf, len);
6064 }
6065
6066 /* returns 0 upon success, and writes result into oid */
6067 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6068 {
6069         struct diff_queue_struct *q = &diff_queued_diff;
6070         int i;
6071         git_hash_ctx ctx;
6072         struct patch_id_t data;
6073
6074         the_hash_algo->init_fn(&ctx);
6075         memset(&data, 0, sizeof(struct patch_id_t));
6076         data.ctx = &ctx;
6077         oidclr(oid);
6078
6079         for (i = 0; i < q->nr; i++) {
6080                 xpparam_t xpp;
6081                 xdemitconf_t xecfg;
6082                 mmfile_t mf1, mf2;
6083                 struct diff_filepair *p = q->queue[i];
6084                 int len1, len2;
6085
6086                 memset(&xpp, 0, sizeof(xpp));
6087                 memset(&xecfg, 0, sizeof(xecfg));
6088                 if (p->status == 0)
6089                         return error("internal diff status error");
6090                 if (p->status == DIFF_STATUS_UNKNOWN)
6091                         continue;
6092                 if (diff_unmodified_pair(p))
6093                         continue;
6094                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
6095                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
6096                         continue;
6097                 if (DIFF_PAIR_UNMERGED(p))
6098                         continue;
6099
6100                 diff_fill_oid_info(p->one, options->repo->index);
6101                 diff_fill_oid_info(p->two, options->repo->index);
6102
6103                 len1 = remove_space(p->one->path, strlen(p->one->path));
6104                 len2 = remove_space(p->two->path, strlen(p->two->path));
6105                 patch_id_add_string(&ctx, "diff--git");
6106                 patch_id_add_string(&ctx, "a/");
6107                 the_hash_algo->update_fn(&ctx, p->one->path, len1);
6108                 patch_id_add_string(&ctx, "b/");
6109                 the_hash_algo->update_fn(&ctx, p->two->path, len2);
6110
6111                 if (p->one->mode == 0) {
6112                         patch_id_add_string(&ctx, "newfilemode");
6113                         patch_id_add_mode(&ctx, p->two->mode);
6114                         patch_id_add_string(&ctx, "---/dev/null");
6115                         patch_id_add_string(&ctx, "+++b/");
6116                         the_hash_algo->update_fn(&ctx, p->two->path, len2);
6117                 } else if (p->two->mode == 0) {
6118                         patch_id_add_string(&ctx, "deletedfilemode");
6119                         patch_id_add_mode(&ctx, p->one->mode);
6120                         patch_id_add_string(&ctx, "---a/");
6121                         the_hash_algo->update_fn(&ctx, p->one->path, len1);
6122                         patch_id_add_string(&ctx, "+++/dev/null");
6123                 } else {
6124                         patch_id_add_string(&ctx, "---a/");
6125                         the_hash_algo->update_fn(&ctx, p->one->path, len1);
6126                         patch_id_add_string(&ctx, "+++b/");
6127                         the_hash_algo->update_fn(&ctx, p->two->path, len2);
6128                 }
6129
6130                 if (diff_header_only)
6131                         continue;
6132
6133                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
6134                     fill_mmfile(options->repo, &mf2, p->two) < 0)
6135                         return error("unable to read files to diff");
6136
6137                 if (diff_filespec_is_binary(options->repo, p->one) ||
6138                     diff_filespec_is_binary(options->repo, p->two)) {
6139                         the_hash_algo->update_fn(&ctx, oid_to_hex(&p->one->oid),
6140                                         the_hash_algo->hexsz);
6141                         the_hash_algo->update_fn(&ctx, oid_to_hex(&p->two->oid),
6142                                         the_hash_algo->hexsz);
6143                         continue;
6144                 }
6145
6146                 xpp.flags = 0;
6147                 xecfg.ctxlen = 3;
6148                 xecfg.flags = 0;
6149                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
6150                                   patch_id_consume, &data, &xpp, &xecfg))
6151                         return error("unable to generate patch-id diff for %s",
6152                                      p->one->path);
6153
6154                 if (stable)
6155                         flush_one_hunk(oid, &ctx);
6156         }
6157
6158         if (!stable)
6159                 the_hash_algo->final_fn(oid->hash, &ctx);
6160
6161         return 0;
6162 }
6163
6164 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6165 {
6166         struct diff_queue_struct *q = &diff_queued_diff;
6167         int i;
6168         int result = diff_get_patch_id(options, oid, diff_header_only, stable);
6169
6170         for (i = 0; i < q->nr; i++)
6171                 diff_free_filepair(q->queue[i]);
6172
6173         free(q->queue);
6174         DIFF_QUEUE_CLEAR(q);
6175
6176         return result;
6177 }
6178
6179 static int is_summary_empty(const struct diff_queue_struct *q)
6180 {
6181         int i;
6182
6183         for (i = 0; i < q->nr; i++) {
6184                 const struct diff_filepair *p = q->queue[i];
6185
6186                 switch (p->status) {
6187                 case DIFF_STATUS_DELETED:
6188                 case DIFF_STATUS_ADDED:
6189                 case DIFF_STATUS_COPIED:
6190                 case DIFF_STATUS_RENAMED:
6191                         return 0;
6192                 default:
6193                         if (p->score)
6194                                 return 0;
6195                         if (p->one->mode && p->two->mode &&
6196                             p->one->mode != p->two->mode)
6197                                 return 0;
6198                         break;
6199                 }
6200         }
6201         return 1;
6202 }
6203
6204 static const char rename_limit_warning[] =
6205 N_("inexact rename detection was skipped due to too many files.");
6206
6207 static const char degrade_cc_to_c_warning[] =
6208 N_("only found copies from modified paths due to too many files.");
6209
6210 static const char rename_limit_advice[] =
6211 N_("you may want to set your %s variable to at least "
6212    "%d and retry the command.");
6213
6214 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
6215 {
6216         fflush(stdout);
6217         if (degraded_cc)
6218                 warning(_(degrade_cc_to_c_warning));
6219         else if (needed)
6220                 warning(_(rename_limit_warning));
6221         else
6222                 return;
6223         if (0 < needed)
6224                 warning(_(rename_limit_advice), varname, needed);
6225 }
6226
6227 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
6228 {
6229         int i;
6230         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
6231         struct diff_queue_struct *q = &diff_queued_diff;
6232
6233         if (WSEH_NEW & WS_RULE_MASK)
6234                 BUG("WS rules bit mask overlaps with diff symbol flags");
6235
6236         if (o->color_moved)
6237                 o->emitted_symbols = &esm;
6238
6239         for (i = 0; i < q->nr; i++) {
6240                 struct diff_filepair *p = q->queue[i];
6241                 if (check_pair_status(p))
6242                         diff_flush_patch(p, o);
6243         }
6244
6245         if (o->emitted_symbols) {
6246                 if (o->color_moved) {
6247                         struct hashmap add_lines, del_lines;
6248
6249                         if (o->color_moved_ws_handling &
6250                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
6251                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
6252
6253                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
6254                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
6255
6256                         add_lines_to_move_detection(o, &add_lines, &del_lines);
6257                         mark_color_as_moved(o, &add_lines, &del_lines);
6258                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6259                                 dim_moved_lines(o);
6260
6261                         hashmap_free_entries(&add_lines, struct moved_entry,
6262                                                 ent);
6263                         hashmap_free_entries(&del_lines, struct moved_entry,
6264                                                 ent);
6265                 }
6266
6267                 for (i = 0; i < esm.nr; i++)
6268                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6269
6270                 for (i = 0; i < esm.nr; i++)
6271                         free((void *)esm.buf[i].line);
6272                 esm.nr = 0;
6273
6274                 o->emitted_symbols = NULL;
6275         }
6276 }
6277
6278 void diff_flush(struct diff_options *options)
6279 {
6280         struct diff_queue_struct *q = &diff_queued_diff;
6281         int i, output_format = options->output_format;
6282         int separator = 0;
6283         int dirstat_by_line = 0;
6284
6285         /*
6286          * Order: raw, stat, summary, patch
6287          * or:    name/name-status/checkdiff (other bits clear)
6288          */
6289         if (!q->nr)
6290                 goto free_queue;
6291
6292         if (output_format & (DIFF_FORMAT_RAW |
6293                              DIFF_FORMAT_NAME |
6294                              DIFF_FORMAT_NAME_STATUS |
6295                              DIFF_FORMAT_CHECKDIFF)) {
6296                 for (i = 0; i < q->nr; i++) {
6297                         struct diff_filepair *p = q->queue[i];
6298                         if (check_pair_status(p))
6299                                 flush_one_pair(p, options);
6300                 }
6301                 separator++;
6302         }
6303
6304         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6305                 dirstat_by_line = 1;
6306
6307         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6308             dirstat_by_line) {
6309                 struct diffstat_t diffstat;
6310
6311                 compute_diffstat(options, &diffstat, q);
6312                 if (output_format & DIFF_FORMAT_NUMSTAT)
6313                         show_numstat(&diffstat, options);
6314                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6315                         show_stats(&diffstat, options);
6316                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6317                         show_shortstats(&diffstat, options);
6318                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6319                         show_dirstat_by_line(&diffstat, options);
6320                 free_diffstat_info(&diffstat);
6321                 separator++;
6322         }
6323         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6324                 show_dirstat(options);
6325
6326         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6327                 for (i = 0; i < q->nr; i++) {
6328                         diff_summary(options, q->queue[i]);
6329                 }
6330                 separator++;
6331         }
6332
6333         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6334             options->flags.exit_with_status &&
6335             options->flags.diff_from_contents) {
6336                 /*
6337                  * run diff_flush_patch for the exit status. setting
6338                  * options->file to /dev/null should be safe, because we
6339                  * aren't supposed to produce any output anyway.
6340                  */
6341                 if (options->close_file)
6342                         fclose(options->file);
6343                 options->file = xfopen("/dev/null", "w");
6344                 options->close_file = 1;
6345                 options->color_moved = 0;
6346                 for (i = 0; i < q->nr; i++) {
6347                         struct diff_filepair *p = q->queue[i];
6348                         if (check_pair_status(p))
6349                                 diff_flush_patch(p, options);
6350                         if (options->found_changes)
6351                                 break;
6352                 }
6353         }
6354
6355         if (output_format & DIFF_FORMAT_PATCH) {
6356                 if (separator) {
6357                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6358                         if (options->stat_sep)
6359                                 /* attach patch instead of inline */
6360                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6361                                                  NULL, 0, 0);
6362                 }
6363
6364                 diff_flush_patch_all_file_pairs(options);
6365         }
6366
6367         if (output_format & DIFF_FORMAT_CALLBACK)
6368                 options->format_callback(q, options, options->format_callback_data);
6369
6370         for (i = 0; i < q->nr; i++)
6371                 diff_free_filepair(q->queue[i]);
6372 free_queue:
6373         free(q->queue);
6374         DIFF_QUEUE_CLEAR(q);
6375         if (options->close_file)
6376                 fclose(options->file);
6377
6378         /*
6379          * Report the content-level differences with HAS_CHANGES;
6380          * diff_addremove/diff_change does not set the bit when
6381          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6382          */
6383         if (options->flags.diff_from_contents) {
6384                 if (options->found_changes)
6385                         options->flags.has_changes = 1;
6386                 else
6387                         options->flags.has_changes = 0;
6388         }
6389 }
6390
6391 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6392 {
6393         return (((p->status == DIFF_STATUS_MODIFIED) &&
6394                  ((p->score &&
6395                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6396                   (!p->score &&
6397                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6398                 ((p->status != DIFF_STATUS_MODIFIED) &&
6399                  filter_bit_tst(p->status, options)));
6400 }
6401
6402 static void diffcore_apply_filter(struct diff_options *options)
6403 {
6404         int i;
6405         struct diff_queue_struct *q = &diff_queued_diff;
6406         struct diff_queue_struct outq;
6407
6408         DIFF_QUEUE_CLEAR(&outq);
6409
6410         if (!options->filter)
6411                 return;
6412
6413         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6414                 int found;
6415                 for (i = found = 0; !found && i < q->nr; i++) {
6416                         if (match_filter(options, q->queue[i]))
6417                                 found++;
6418                 }
6419                 if (found)
6420                         return;
6421
6422                 /* otherwise we will clear the whole queue
6423                  * by copying the empty outq at the end of this
6424                  * function, but first clear the current entries
6425                  * in the queue.
6426                  */
6427                 for (i = 0; i < q->nr; i++)
6428                         diff_free_filepair(q->queue[i]);
6429         }
6430         else {
6431                 /* Only the matching ones */
6432                 for (i = 0; i < q->nr; i++) {
6433                         struct diff_filepair *p = q->queue[i];
6434                         if (match_filter(options, p))
6435                                 diff_q(&outq, p);
6436                         else
6437                                 diff_free_filepair(p);
6438                 }
6439         }
6440         free(q->queue);
6441         *q = outq;
6442 }
6443
6444 /* Check whether two filespecs with the same mode and size are identical */
6445 static int diff_filespec_is_identical(struct repository *r,
6446                                       struct diff_filespec *one,
6447                                       struct diff_filespec *two)
6448 {
6449         if (S_ISGITLINK(one->mode))
6450                 return 0;
6451         if (diff_populate_filespec(r, one, NULL))
6452                 return 0;
6453         if (diff_populate_filespec(r, two, NULL))
6454                 return 0;
6455         return !memcmp(one->data, two->data, one->size);
6456 }
6457
6458 static int diff_filespec_check_stat_unmatch(struct repository *r,
6459                                             struct diff_filepair *p)
6460 {
6461         struct diff_populate_filespec_options dpf_options = {
6462                 .check_size_only = 1,
6463                 .missing_object_cb = diff_queued_diff_prefetch,
6464                 .missing_object_data = r,
6465         };
6466
6467         if (p->done_skip_stat_unmatch)
6468                 return p->skip_stat_unmatch_result;
6469
6470         p->done_skip_stat_unmatch = 1;
6471         p->skip_stat_unmatch_result = 0;
6472         /*
6473          * 1. Entries that come from stat info dirtiness
6474          *    always have both sides (iow, not create/delete),
6475          *    one side of the object name is unknown, with
6476          *    the same mode and size.  Keep the ones that
6477          *    do not match these criteria.  They have real
6478          *    differences.
6479          *
6480          * 2. At this point, the file is known to be modified,
6481          *    with the same mode and size, and the object
6482          *    name of one side is unknown.  Need to inspect
6483          *    the identical contents.
6484          */
6485         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6486             !DIFF_FILE_VALID(p->two) ||
6487             (p->one->oid_valid && p->two->oid_valid) ||
6488             (p->one->mode != p->two->mode) ||
6489             diff_populate_filespec(r, p->one, &dpf_options) ||
6490             diff_populate_filespec(r, p->two, &dpf_options) ||
6491             (p->one->size != p->two->size) ||
6492             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6493                 p->skip_stat_unmatch_result = 1;
6494         return p->skip_stat_unmatch_result;
6495 }
6496
6497 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6498 {
6499         int i;
6500         struct diff_queue_struct *q = &diff_queued_diff;
6501         struct diff_queue_struct outq;
6502         DIFF_QUEUE_CLEAR(&outq);
6503
6504         for (i = 0; i < q->nr; i++) {
6505                 struct diff_filepair *p = q->queue[i];
6506
6507                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6508                         diff_q(&outq, p);
6509                 else {
6510                         /*
6511                          * The caller can subtract 1 from skip_stat_unmatch
6512                          * to determine how many paths were dirty only
6513                          * due to stat info mismatch.
6514                          */
6515                         if (!diffopt->flags.no_index)
6516                                 diffopt->skip_stat_unmatch++;
6517                         diff_free_filepair(p);
6518                 }
6519         }
6520         free(q->queue);
6521         *q = outq;
6522 }
6523
6524 static int diffnamecmp(const void *a_, const void *b_)
6525 {
6526         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6527         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6528         const char *name_a, *name_b;
6529
6530         name_a = a->one ? a->one->path : a->two->path;
6531         name_b = b->one ? b->one->path : b->two->path;
6532         return strcmp(name_a, name_b);
6533 }
6534
6535 void diffcore_fix_diff_index(void)
6536 {
6537         struct diff_queue_struct *q = &diff_queued_diff;
6538         QSORT(q->queue, q->nr, diffnamecmp);
6539 }
6540
6541 void diff_add_if_missing(struct repository *r,
6542                          struct oid_array *to_fetch,
6543                          const struct diff_filespec *filespec)
6544 {
6545         if (filespec && filespec->oid_valid &&
6546             !S_ISGITLINK(filespec->mode) &&
6547             oid_object_info_extended(r, &filespec->oid, NULL,
6548                                      OBJECT_INFO_FOR_PREFETCH))
6549                 oid_array_append(to_fetch, &filespec->oid);
6550 }
6551
6552 void diff_queued_diff_prefetch(void *repository)
6553 {
6554         struct repository *repo = repository;
6555         int i;
6556         struct diff_queue_struct *q = &diff_queued_diff;
6557         struct oid_array to_fetch = OID_ARRAY_INIT;
6558
6559         for (i = 0; i < q->nr; i++) {
6560                 struct diff_filepair *p = q->queue[i];
6561                 diff_add_if_missing(repo, &to_fetch, p->one);
6562                 diff_add_if_missing(repo, &to_fetch, p->two);
6563         }
6564
6565         /*
6566          * NEEDSWORK: Consider deduplicating the OIDs sent.
6567          */
6568         promisor_remote_get_direct(repo, to_fetch.oid, to_fetch.nr);
6569
6570         oid_array_clear(&to_fetch);
6571 }
6572
6573 void diffcore_std(struct diff_options *options)
6574 {
6575         int output_formats_to_prefetch = DIFF_FORMAT_DIFFSTAT |
6576                 DIFF_FORMAT_NUMSTAT |
6577                 DIFF_FORMAT_PATCH |
6578                 DIFF_FORMAT_SHORTSTAT |
6579                 DIFF_FORMAT_DIRSTAT;
6580
6581         /*
6582          * Check if the user requested a blob-data-requiring diff output and/or
6583          * break-rewrite detection (which requires blob data). If yes, prefetch
6584          * the diff pairs.
6585          *
6586          * If no prefetching occurs, diffcore_rename() will prefetch if it
6587          * decides that it needs inexact rename detection.
6588          */
6589         if (options->repo == the_repository && has_promisor_remote() &&
6590             (options->output_format & output_formats_to_prefetch ||
6591              options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
6592                 diff_queued_diff_prefetch(options->repo);
6593
6594         /* NOTE please keep the following in sync with diff_tree_combined() */
6595         if (options->skip_stat_unmatch)
6596                 diffcore_skip_stat_unmatch(options);
6597         if (!options->found_follow) {
6598                 /* See try_to_follow_renames() in tree-diff.c */
6599                 if (options->break_opt != -1)
6600                         diffcore_break(options->repo,
6601                                        options->break_opt);
6602                 if (options->detect_rename)
6603                         diffcore_rename(options);
6604                 if (options->break_opt != -1)
6605                         diffcore_merge_broken();
6606         }
6607         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6608                 diffcore_pickaxe(options);
6609         if (options->orderfile)
6610                 diffcore_order(options->orderfile);
6611         if (!options->found_follow)
6612                 /* See try_to_follow_renames() in tree-diff.c */
6613                 diff_resolve_rename_copy();
6614         diffcore_apply_filter(options);
6615
6616         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6617                 options->flags.has_changes = 1;
6618         else
6619                 options->flags.has_changes = 0;
6620
6621         options->found_follow = 0;
6622 }
6623
6624 int diff_result_code(struct diff_options *opt, int status)
6625 {
6626         int result = 0;
6627
6628         diff_warn_rename_limit("diff.renameLimit",
6629                                opt->needed_rename_limit,
6630                                opt->degraded_cc_to_c);
6631         if (!opt->flags.exit_with_status &&
6632             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6633                 return status;
6634         if (opt->flags.exit_with_status &&
6635             opt->flags.has_changes)
6636                 result |= 01;
6637         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6638             opt->flags.check_failed)
6639                 result |= 02;
6640         return result;
6641 }
6642
6643 int diff_can_quit_early(struct diff_options *opt)
6644 {
6645         return (opt->flags.quick &&
6646                 !opt->filter &&
6647                 opt->flags.has_changes);
6648 }
6649
6650 /*
6651  * Shall changes to this submodule be ignored?
6652  *
6653  * Submodule changes can be configured to be ignored separately for each path,
6654  * but that configuration can be overridden from the command line.
6655  */
6656 static int is_submodule_ignored(const char *path, struct diff_options *options)
6657 {
6658         int ignored = 0;
6659         struct diff_flags orig_flags = options->flags;
6660         if (!options->flags.override_submodule_config)
6661                 set_diffopt_flags_from_submodule_config(options, path);
6662         if (options->flags.ignore_submodules)
6663                 ignored = 1;
6664         options->flags = orig_flags;
6665         return ignored;
6666 }
6667
6668 void compute_diffstat(struct diff_options *options,
6669                       struct diffstat_t *diffstat,
6670                       struct diff_queue_struct *q)
6671 {
6672         int i;
6673
6674         memset(diffstat, 0, sizeof(struct diffstat_t));
6675         for (i = 0; i < q->nr; i++) {
6676                 struct diff_filepair *p = q->queue[i];
6677                 if (check_pair_status(p))
6678                         diff_flush_stat(p, options, diffstat);
6679         }
6680 }
6681
6682 void diff_addremove(struct diff_options *options,
6683                     int addremove, unsigned mode,
6684                     const struct object_id *oid,
6685                     int oid_valid,
6686                     const char *concatpath, unsigned dirty_submodule)
6687 {
6688         struct diff_filespec *one, *two;
6689
6690         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6691                 return;
6692
6693         /* This may look odd, but it is a preparation for
6694          * feeding "there are unchanged files which should
6695          * not produce diffs, but when you are doing copy
6696          * detection you would need them, so here they are"
6697          * entries to the diff-core.  They will be prefixed
6698          * with something like '=' or '*' (I haven't decided
6699          * which but should not make any difference).
6700          * Feeding the same new and old to diff_change()
6701          * also has the same effect.
6702          * Before the final output happens, they are pruned after
6703          * merged into rename/copy pairs as appropriate.
6704          */
6705         if (options->flags.reverse_diff)
6706                 addremove = (addremove == '+' ? '-' :
6707                              addremove == '-' ? '+' : addremove);
6708
6709         if (options->prefix &&
6710             strncmp(concatpath, options->prefix, options->prefix_length))
6711                 return;
6712
6713         one = alloc_filespec(concatpath);
6714         two = alloc_filespec(concatpath);
6715
6716         if (addremove != '+')
6717                 fill_filespec(one, oid, oid_valid, mode);
6718         if (addremove != '-') {
6719                 fill_filespec(two, oid, oid_valid, mode);
6720                 two->dirty_submodule = dirty_submodule;
6721         }
6722
6723         diff_queue(&diff_queued_diff, one, two);
6724         if (!options->flags.diff_from_contents)
6725                 options->flags.has_changes = 1;
6726 }
6727
6728 void diff_change(struct diff_options *options,
6729                  unsigned old_mode, unsigned new_mode,
6730                  const struct object_id *old_oid,
6731                  const struct object_id *new_oid,
6732                  int old_oid_valid, int new_oid_valid,
6733                  const char *concatpath,
6734                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6735 {
6736         struct diff_filespec *one, *two;
6737         struct diff_filepair *p;
6738
6739         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6740             is_submodule_ignored(concatpath, options))
6741                 return;
6742
6743         if (options->flags.reverse_diff) {
6744                 SWAP(old_mode, new_mode);
6745                 SWAP(old_oid, new_oid);
6746                 SWAP(old_oid_valid, new_oid_valid);
6747                 SWAP(old_dirty_submodule, new_dirty_submodule);
6748         }
6749
6750         if (options->prefix &&
6751             strncmp(concatpath, options->prefix, options->prefix_length))
6752                 return;
6753
6754         one = alloc_filespec(concatpath);
6755         two = alloc_filespec(concatpath);
6756         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6757         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6758         one->dirty_submodule = old_dirty_submodule;
6759         two->dirty_submodule = new_dirty_submodule;
6760         p = diff_queue(&diff_queued_diff, one, two);
6761
6762         if (options->flags.diff_from_contents)
6763                 return;
6764
6765         if (options->flags.quick && options->skip_stat_unmatch &&
6766             !diff_filespec_check_stat_unmatch(options->repo, p)) {
6767                 diff_free_filespec_data(p->one);
6768                 diff_free_filespec_data(p->two);
6769                 return;
6770         }
6771
6772         options->flags.has_changes = 1;
6773 }
6774
6775 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6776 {
6777         struct diff_filepair *pair;
6778         struct diff_filespec *one, *two;
6779
6780         if (options->prefix &&
6781             strncmp(path, options->prefix, options->prefix_length))
6782                 return NULL;
6783
6784         one = alloc_filespec(path);
6785         two = alloc_filespec(path);
6786         pair = diff_queue(&diff_queued_diff, one, two);
6787         pair->is_unmerged = 1;
6788         return pair;
6789 }
6790
6791 static char *run_textconv(struct repository *r,
6792                           const char *pgm,
6793                           struct diff_filespec *spec,
6794                           size_t *outsize)
6795 {
6796         struct diff_tempfile *temp;
6797         const char *argv[3];
6798         const char **arg = argv;
6799         struct child_process child = CHILD_PROCESS_INIT;
6800         struct strbuf buf = STRBUF_INIT;
6801         int err = 0;
6802
6803         temp = prepare_temp_file(r, spec->path, spec);
6804         *arg++ = pgm;
6805         *arg++ = temp->name;
6806         *arg = NULL;
6807
6808         child.use_shell = 1;
6809         child.argv = argv;
6810         child.out = -1;
6811         if (start_command(&child)) {
6812                 remove_tempfile();
6813                 return NULL;
6814         }
6815
6816         if (strbuf_read(&buf, child.out, 0) < 0)
6817                 err = error("error reading from textconv command '%s'", pgm);
6818         close(child.out);
6819
6820         if (finish_command(&child) || err) {
6821                 strbuf_release(&buf);
6822                 remove_tempfile();
6823                 return NULL;
6824         }
6825         remove_tempfile();
6826
6827         return strbuf_detach(&buf, outsize);
6828 }
6829
6830 size_t fill_textconv(struct repository *r,
6831                      struct userdiff_driver *driver,
6832                      struct diff_filespec *df,
6833                      char **outbuf)
6834 {
6835         size_t size;
6836
6837         if (!driver) {
6838                 if (!DIFF_FILE_VALID(df)) {
6839                         *outbuf = "";
6840                         return 0;
6841                 }
6842                 if (diff_populate_filespec(r, df, NULL))
6843                         die("unable to read files to diff");
6844                 *outbuf = df->data;
6845                 return df->size;
6846         }
6847
6848         if (!driver->textconv)
6849                 BUG("fill_textconv called with non-textconv driver");
6850
6851         if (driver->textconv_cache && df->oid_valid) {
6852                 *outbuf = notes_cache_get(driver->textconv_cache,
6853                                           &df->oid,
6854                                           &size);
6855                 if (*outbuf)
6856                         return size;
6857         }
6858
6859         *outbuf = run_textconv(r, driver->textconv, df, &size);
6860         if (!*outbuf)
6861                 die("unable to read files to diff");
6862
6863         if (driver->textconv_cache && df->oid_valid) {
6864                 /* ignore errors, as we might be in a readonly repository */
6865                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6866                                 size);
6867                 /*
6868                  * we could save up changes and flush them all at the end,
6869                  * but we would need an extra call after all diffing is done.
6870                  * Since generating a cache entry is the slow path anyway,
6871                  * this extra overhead probably isn't a big deal.
6872                  */
6873                 notes_cache_write(driver->textconv_cache);
6874         }
6875
6876         return size;
6877 }
6878
6879 int textconv_object(struct repository *r,
6880                     const char *path,
6881                     unsigned mode,
6882                     const struct object_id *oid,
6883                     int oid_valid,
6884                     char **buf,
6885                     unsigned long *buf_size)
6886 {
6887         struct diff_filespec *df;
6888         struct userdiff_driver *textconv;
6889
6890         df = alloc_filespec(path);
6891         fill_filespec(df, oid, oid_valid, mode);
6892         textconv = get_textconv(r, df);
6893         if (!textconv) {
6894                 free_filespec(df);
6895                 return 0;
6896         }
6897
6898         *buf_size = fill_textconv(r, textconv, df, buf);
6899         free_filespec(df);
6900         return 1;
6901 }
6902
6903 void setup_diff_pager(struct diff_options *opt)
6904 {
6905         /*
6906          * If the user asked for our exit code, then either they want --quiet
6907          * or --exit-code. We should definitely not bother with a pager in the
6908          * former case, as we will generate no output. Since we still properly
6909          * report our exit code even when a pager is run, we _could_ run a
6910          * pager with --exit-code. But since we have not done so historically,
6911          * and because it is easy to find people oneline advising "git diff
6912          * --exit-code" in hooks and other scripts, we do not do so.
6913          */
6914         if (!opt->flags.exit_with_status &&
6915             check_pager_config("diff") != 0)
6916                 setup_pager();
6917 }