Merge branch 'sg/travis-fixes' into maint
[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 "userdiff.h"
17 #include "submodule-config.h"
18 #include "submodule.h"
19 #include "hashmap.h"
20 #include "ll-merge.h"
21 #include "string-list.h"
22 #include "argv-array.h"
23 #include "graph.h"
24 #include "packfile.h"
25
26 #ifdef NO_FAST_WORKING_DIRECTORY
27 #define FAST_WORKING_DIRECTORY 0
28 #else
29 #define FAST_WORKING_DIRECTORY 1
30 #endif
31
32 static int diff_detect_rename_default;
33 static int diff_indent_heuristic = 1;
34 static int diff_rename_limit_default = 400;
35 static int diff_suppress_blank_empty;
36 static int diff_use_color_default = -1;
37 static int diff_color_moved_default;
38 static int diff_context_default = 3;
39 static int diff_interhunk_context_default;
40 static const char *diff_word_regex_cfg;
41 static const char *external_diff_cmd_cfg;
42 static const char *diff_order_file_cfg;
43 int diff_auto_refresh_index = 1;
44 static int diff_mnemonic_prefix;
45 static int diff_no_prefix;
46 static int diff_stat_graph_width;
47 static int diff_dirstat_permille_default = 30;
48 static struct diff_options default_diff_options;
49 static long diff_algorithm;
50 static unsigned ws_error_highlight_default = WSEH_NEW;
51
52 static char diff_colors[][COLOR_MAXLEN] = {
53         GIT_COLOR_RESET,
54         GIT_COLOR_NORMAL,       /* CONTEXT */
55         GIT_COLOR_BOLD,         /* METAINFO */
56         GIT_COLOR_CYAN,         /* FRAGINFO */
57         GIT_COLOR_RED,          /* OLD */
58         GIT_COLOR_GREEN,        /* NEW */
59         GIT_COLOR_YELLOW,       /* COMMIT */
60         GIT_COLOR_BG_RED,       /* WHITESPACE */
61         GIT_COLOR_NORMAL,       /* FUNCINFO */
62         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
63         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
64         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
65         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
66         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
67         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
68         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
69         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
70 };
71
72 static NORETURN void die_want_option(const char *option_name)
73 {
74         die(_("option '%s' requires a value"), option_name);
75 }
76
77 static int parse_diff_color_slot(const char *var)
78 {
79         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
80                 return DIFF_CONTEXT;
81         if (!strcasecmp(var, "meta"))
82                 return DIFF_METAINFO;
83         if (!strcasecmp(var, "frag"))
84                 return DIFF_FRAGINFO;
85         if (!strcasecmp(var, "old"))
86                 return DIFF_FILE_OLD;
87         if (!strcasecmp(var, "new"))
88                 return DIFF_FILE_NEW;
89         if (!strcasecmp(var, "commit"))
90                 return DIFF_COMMIT;
91         if (!strcasecmp(var, "whitespace"))
92                 return DIFF_WHITESPACE;
93         if (!strcasecmp(var, "func"))
94                 return DIFF_FUNCINFO;
95         if (!strcasecmp(var, "oldmoved"))
96                 return DIFF_FILE_OLD_MOVED;
97         if (!strcasecmp(var, "oldmovedalternative"))
98                 return DIFF_FILE_OLD_MOVED_ALT;
99         if (!strcasecmp(var, "oldmoveddimmed"))
100                 return DIFF_FILE_OLD_MOVED_DIM;
101         if (!strcasecmp(var, "oldmovedalternativedimmed"))
102                 return DIFF_FILE_OLD_MOVED_ALT_DIM;
103         if (!strcasecmp(var, "newmoved"))
104                 return DIFF_FILE_NEW_MOVED;
105         if (!strcasecmp(var, "newmovedalternative"))
106                 return DIFF_FILE_NEW_MOVED_ALT;
107         if (!strcasecmp(var, "newmoveddimmed"))
108                 return DIFF_FILE_NEW_MOVED_DIM;
109         if (!strcasecmp(var, "newmovedalternativedimmed"))
110                 return DIFF_FILE_NEW_MOVED_ALT_DIM;
111         return -1;
112 }
113
114 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
115                                 struct strbuf *errmsg)
116 {
117         char *params_copy = xstrdup(params_string);
118         struct string_list params = STRING_LIST_INIT_NODUP;
119         int ret = 0;
120         int i;
121
122         if (*params_copy)
123                 string_list_split_in_place(&params, params_copy, ',', -1);
124         for (i = 0; i < params.nr; i++) {
125                 const char *p = params.items[i].string;
126                 if (!strcmp(p, "changes")) {
127                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
128                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
129                 } else if (!strcmp(p, "lines")) {
130                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
131                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
132                 } else if (!strcmp(p, "files")) {
133                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
134                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
135                 } else if (!strcmp(p, "noncumulative")) {
136                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
137                 } else if (!strcmp(p, "cumulative")) {
138                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
139                 } else if (isdigit(*p)) {
140                         char *end;
141                         int permille = strtoul(p, &end, 10) * 10;
142                         if (*end == '.' && isdigit(*++end)) {
143                                 /* only use first digit */
144                                 permille += *end - '0';
145                                 /* .. and ignore any further digits */
146                                 while (isdigit(*++end))
147                                         ; /* nothing */
148                         }
149                         if (!*end)
150                                 options->dirstat_permille = permille;
151                         else {
152                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
153                                             p);
154                                 ret++;
155                         }
156                 } else {
157                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
158                         ret++;
159                 }
160
161         }
162         string_list_clear(&params, 0);
163         free(params_copy);
164         return ret;
165 }
166
167 static int parse_submodule_params(struct diff_options *options, const char *value)
168 {
169         if (!strcmp(value, "log"))
170                 options->submodule_format = DIFF_SUBMODULE_LOG;
171         else if (!strcmp(value, "short"))
172                 options->submodule_format = DIFF_SUBMODULE_SHORT;
173         else if (!strcmp(value, "diff"))
174                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
175         else
176                 return -1;
177         return 0;
178 }
179
180 static int git_config_rename(const char *var, const char *value)
181 {
182         if (!value)
183                 return DIFF_DETECT_RENAME;
184         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
185                 return  DIFF_DETECT_COPY;
186         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
187 }
188
189 long parse_algorithm_value(const char *value)
190 {
191         if (!value)
192                 return -1;
193         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
194                 return 0;
195         else if (!strcasecmp(value, "minimal"))
196                 return XDF_NEED_MINIMAL;
197         else if (!strcasecmp(value, "patience"))
198                 return XDF_PATIENCE_DIFF;
199         else if (!strcasecmp(value, "histogram"))
200                 return XDF_HISTOGRAM_DIFF;
201         return -1;
202 }
203
204 static int parse_one_token(const char **arg, const char *token)
205 {
206         const char *rest;
207         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
208                 *arg = rest;
209                 return 1;
210         }
211         return 0;
212 }
213
214 static int parse_ws_error_highlight(const char *arg)
215 {
216         const char *orig_arg = arg;
217         unsigned val = 0;
218
219         while (*arg) {
220                 if (parse_one_token(&arg, "none"))
221                         val = 0;
222                 else if (parse_one_token(&arg, "default"))
223                         val = WSEH_NEW;
224                 else if (parse_one_token(&arg, "all"))
225                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
226                 else if (parse_one_token(&arg, "new"))
227                         val |= WSEH_NEW;
228                 else if (parse_one_token(&arg, "old"))
229                         val |= WSEH_OLD;
230                 else if (parse_one_token(&arg, "context"))
231                         val |= WSEH_CONTEXT;
232                 else {
233                         return -1 - (int)(arg - orig_arg);
234                 }
235                 if (*arg)
236                         arg++;
237         }
238         return val;
239 }
240
241 /*
242  * These are to give UI layer defaults.
243  * The core-level commands such as git-diff-files should
244  * never be affected by the setting of diff.renames
245  * the user happens to have in the configuration file.
246  */
247 void init_diff_ui_defaults(void)
248 {
249         diff_detect_rename_default = 1;
250 }
251
252 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
253 {
254         if (!strcmp(var, "diff.indentheuristic"))
255                 diff_indent_heuristic = git_config_bool(var, value);
256         return 0;
257 }
258
259 static int parse_color_moved(const char *arg)
260 {
261         switch (git_parse_maybe_bool(arg)) {
262         case 0:
263                 return COLOR_MOVED_NO;
264         case 1:
265                 return COLOR_MOVED_DEFAULT;
266         default:
267                 break;
268         }
269
270         if (!strcmp(arg, "no"))
271                 return COLOR_MOVED_NO;
272         else if (!strcmp(arg, "plain"))
273                 return COLOR_MOVED_PLAIN;
274         else if (!strcmp(arg, "zebra"))
275                 return COLOR_MOVED_ZEBRA;
276         else if (!strcmp(arg, "default"))
277                 return COLOR_MOVED_DEFAULT;
278         else if (!strcmp(arg, "dimmed_zebra"))
279                 return COLOR_MOVED_ZEBRA_DIM;
280         else
281                 return error(_("color moved setting must be one of 'no', 'default', 'zebra', 'dimmed_zebra', 'plain'"));
282 }
283
284 int git_diff_ui_config(const char *var, const char *value, void *cb)
285 {
286         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
287                 diff_use_color_default = git_config_colorbool(var, value);
288                 return 0;
289         }
290         if (!strcmp(var, "diff.colormoved")) {
291                 int cm = parse_color_moved(value);
292                 if (cm < 0)
293                         return -1;
294                 diff_color_moved_default = cm;
295                 return 0;
296         }
297         if (!strcmp(var, "diff.context")) {
298                 diff_context_default = git_config_int(var, value);
299                 if (diff_context_default < 0)
300                         return -1;
301                 return 0;
302         }
303         if (!strcmp(var, "diff.interhunkcontext")) {
304                 diff_interhunk_context_default = git_config_int(var, value);
305                 if (diff_interhunk_context_default < 0)
306                         return -1;
307                 return 0;
308         }
309         if (!strcmp(var, "diff.renames")) {
310                 diff_detect_rename_default = git_config_rename(var, value);
311                 return 0;
312         }
313         if (!strcmp(var, "diff.autorefreshindex")) {
314                 diff_auto_refresh_index = git_config_bool(var, value);
315                 return 0;
316         }
317         if (!strcmp(var, "diff.mnemonicprefix")) {
318                 diff_mnemonic_prefix = git_config_bool(var, value);
319                 return 0;
320         }
321         if (!strcmp(var, "diff.noprefix")) {
322                 diff_no_prefix = git_config_bool(var, value);
323                 return 0;
324         }
325         if (!strcmp(var, "diff.statgraphwidth")) {
326                 diff_stat_graph_width = git_config_int(var, value);
327                 return 0;
328         }
329         if (!strcmp(var, "diff.external"))
330                 return git_config_string(&external_diff_cmd_cfg, var, value);
331         if (!strcmp(var, "diff.wordregex"))
332                 return git_config_string(&diff_word_regex_cfg, var, value);
333         if (!strcmp(var, "diff.orderfile"))
334                 return git_config_pathname(&diff_order_file_cfg, var, value);
335
336         if (!strcmp(var, "diff.ignoresubmodules"))
337                 handle_ignore_submodules_arg(&default_diff_options, value);
338
339         if (!strcmp(var, "diff.submodule")) {
340                 if (parse_submodule_params(&default_diff_options, value))
341                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
342                                 value);
343                 return 0;
344         }
345
346         if (!strcmp(var, "diff.algorithm")) {
347                 diff_algorithm = parse_algorithm_value(value);
348                 if (diff_algorithm < 0)
349                         return -1;
350                 return 0;
351         }
352
353         if (!strcmp(var, "diff.wserrorhighlight")) {
354                 int val = parse_ws_error_highlight(value);
355                 if (val < 0)
356                         return -1;
357                 ws_error_highlight_default = val;
358                 return 0;
359         }
360
361         if (git_color_config(var, value, cb) < 0)
362                 return -1;
363
364         return git_diff_basic_config(var, value, cb);
365 }
366
367 int git_diff_basic_config(const char *var, const char *value, void *cb)
368 {
369         const char *name;
370
371         if (!strcmp(var, "diff.renamelimit")) {
372                 diff_rename_limit_default = git_config_int(var, value);
373                 return 0;
374         }
375
376         if (userdiff_config(var, value) < 0)
377                 return -1;
378
379         if (skip_prefix(var, "diff.color.", &name) ||
380             skip_prefix(var, "color.diff.", &name)) {
381                 int slot = parse_diff_color_slot(name);
382                 if (slot < 0)
383                         return 0;
384                 if (!value)
385                         return config_error_nonbool(var);
386                 return color_parse(value, diff_colors[slot]);
387         }
388
389         /* like GNU diff's --suppress-blank-empty option  */
390         if (!strcmp(var, "diff.suppressblankempty") ||
391                         /* for backwards compatibility */
392                         !strcmp(var, "diff.suppress-blank-empty")) {
393                 diff_suppress_blank_empty = git_config_bool(var, value);
394                 return 0;
395         }
396
397         if (!strcmp(var, "diff.dirstat")) {
398                 struct strbuf errmsg = STRBUF_INIT;
399                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
400                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
401                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
402                                 errmsg.buf);
403                 strbuf_release(&errmsg);
404                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
405                 return 0;
406         }
407
408         if (git_diff_heuristic_config(var, value, cb) < 0)
409                 return -1;
410
411         return git_default_config(var, value, cb);
412 }
413
414 static char *quote_two(const char *one, const char *two)
415 {
416         int need_one = quote_c_style(one, NULL, NULL, 1);
417         int need_two = quote_c_style(two, NULL, NULL, 1);
418         struct strbuf res = STRBUF_INIT;
419
420         if (need_one + need_two) {
421                 strbuf_addch(&res, '"');
422                 quote_c_style(one, &res, NULL, 1);
423                 quote_c_style(two, &res, NULL, 1);
424                 strbuf_addch(&res, '"');
425         } else {
426                 strbuf_addstr(&res, one);
427                 strbuf_addstr(&res, two);
428         }
429         return strbuf_detach(&res, NULL);
430 }
431
432 static const char *external_diff(void)
433 {
434         static const char *external_diff_cmd = NULL;
435         static int done_preparing = 0;
436
437         if (done_preparing)
438                 return external_diff_cmd;
439         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
440         if (!external_diff_cmd)
441                 external_diff_cmd = external_diff_cmd_cfg;
442         done_preparing = 1;
443         return external_diff_cmd;
444 }
445
446 /*
447  * Keep track of files used for diffing. Sometimes such an entry
448  * refers to a temporary file, sometimes to an existing file, and
449  * sometimes to "/dev/null".
450  */
451 static struct diff_tempfile {
452         /*
453          * filename external diff should read from, or NULL if this
454          * entry is currently not in use:
455          */
456         const char *name;
457
458         char hex[GIT_MAX_HEXSZ + 1];
459         char mode[10];
460
461         /*
462          * If this diff_tempfile instance refers to a temporary file,
463          * this tempfile object is used to manage its lifetime.
464          */
465         struct tempfile *tempfile;
466 } diff_temp[2];
467
468 struct emit_callback {
469         int color_diff;
470         unsigned ws_rule;
471         int blank_at_eof_in_preimage;
472         int blank_at_eof_in_postimage;
473         int lno_in_preimage;
474         int lno_in_postimage;
475         const char **label_path;
476         struct diff_words_data *diff_words;
477         struct diff_options *opt;
478         struct strbuf *header;
479 };
480
481 static int count_lines(const char *data, int size)
482 {
483         int count, ch, completely_empty = 1, nl_just_seen = 0;
484         count = 0;
485         while (0 < size--) {
486                 ch = *data++;
487                 if (ch == '\n') {
488                         count++;
489                         nl_just_seen = 1;
490                         completely_empty = 0;
491                 }
492                 else {
493                         nl_just_seen = 0;
494                         completely_empty = 0;
495                 }
496         }
497         if (completely_empty)
498                 return 0;
499         if (!nl_just_seen)
500                 count++; /* no trailing newline */
501         return count;
502 }
503
504 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
505 {
506         if (!DIFF_FILE_VALID(one)) {
507                 mf->ptr = (char *)""; /* does not matter */
508                 mf->size = 0;
509                 return 0;
510         }
511         else if (diff_populate_filespec(one, 0))
512                 return -1;
513
514         mf->ptr = one->data;
515         mf->size = one->size;
516         return 0;
517 }
518
519 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
520 static unsigned long diff_filespec_size(struct diff_filespec *one)
521 {
522         if (!DIFF_FILE_VALID(one))
523                 return 0;
524         diff_populate_filespec(one, CHECK_SIZE_ONLY);
525         return one->size;
526 }
527
528 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
529 {
530         char *ptr = mf->ptr;
531         long size = mf->size;
532         int cnt = 0;
533
534         if (!size)
535                 return cnt;
536         ptr += size - 1; /* pointing at the very end */
537         if (*ptr != '\n')
538                 ; /* incomplete line */
539         else
540                 ptr--; /* skip the last LF */
541         while (mf->ptr < ptr) {
542                 char *prev_eol;
543                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
544                         if (*prev_eol == '\n')
545                                 break;
546                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
547                         break;
548                 cnt++;
549                 ptr = prev_eol - 1;
550         }
551         return cnt;
552 }
553
554 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
555                                struct emit_callback *ecbdata)
556 {
557         int l1, l2, at;
558         unsigned ws_rule = ecbdata->ws_rule;
559         l1 = count_trailing_blank(mf1, ws_rule);
560         l2 = count_trailing_blank(mf2, ws_rule);
561         if (l2 <= l1) {
562                 ecbdata->blank_at_eof_in_preimage = 0;
563                 ecbdata->blank_at_eof_in_postimage = 0;
564                 return;
565         }
566         at = count_lines(mf1->ptr, mf1->size);
567         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
568
569         at = count_lines(mf2->ptr, mf2->size);
570         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
571 }
572
573 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
574                         int first, const char *line, int len)
575 {
576         int has_trailing_newline, has_trailing_carriage_return;
577         int nofirst;
578         FILE *file = o->file;
579
580         fputs(diff_line_prefix(o), file);
581
582         if (len == 0) {
583                 has_trailing_newline = (first == '\n');
584                 has_trailing_carriage_return = (!has_trailing_newline &&
585                                                 (first == '\r'));
586                 nofirst = has_trailing_newline || has_trailing_carriage_return;
587         } else {
588                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
589                 if (has_trailing_newline)
590                         len--;
591                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
592                 if (has_trailing_carriage_return)
593                         len--;
594                 nofirst = 0;
595         }
596
597         if (len || !nofirst) {
598                 fputs(set, file);
599                 if (!nofirst)
600                         fputc(first, file);
601                 fwrite(line, len, 1, file);
602                 fputs(reset, file);
603         }
604         if (has_trailing_carriage_return)
605                 fputc('\r', file);
606         if (has_trailing_newline)
607                 fputc('\n', file);
608 }
609
610 static void emit_line(struct diff_options *o, const char *set, const char *reset,
611                       const char *line, int len)
612 {
613         emit_line_0(o, set, reset, line[0], line+1, len-1);
614 }
615
616 enum diff_symbol {
617         DIFF_SYMBOL_BINARY_DIFF_HEADER,
618         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
619         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
620         DIFF_SYMBOL_BINARY_DIFF_BODY,
621         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
622         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
623         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
624         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
625         DIFF_SYMBOL_STATS_LINE,
626         DIFF_SYMBOL_WORD_DIFF,
627         DIFF_SYMBOL_STAT_SEP,
628         DIFF_SYMBOL_SUMMARY,
629         DIFF_SYMBOL_SUBMODULE_ADD,
630         DIFF_SYMBOL_SUBMODULE_DEL,
631         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
632         DIFF_SYMBOL_SUBMODULE_MODIFIED,
633         DIFF_SYMBOL_SUBMODULE_HEADER,
634         DIFF_SYMBOL_SUBMODULE_ERROR,
635         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
636         DIFF_SYMBOL_REWRITE_DIFF,
637         DIFF_SYMBOL_BINARY_FILES,
638         DIFF_SYMBOL_HEADER,
639         DIFF_SYMBOL_FILEPAIR_PLUS,
640         DIFF_SYMBOL_FILEPAIR_MINUS,
641         DIFF_SYMBOL_WORDS_PORCELAIN,
642         DIFF_SYMBOL_WORDS,
643         DIFF_SYMBOL_CONTEXT,
644         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
645         DIFF_SYMBOL_PLUS,
646         DIFF_SYMBOL_MINUS,
647         DIFF_SYMBOL_NO_LF_EOF,
648         DIFF_SYMBOL_CONTEXT_FRAGINFO,
649         DIFF_SYMBOL_CONTEXT_MARKER,
650         DIFF_SYMBOL_SEPARATOR
651 };
652 /*
653  * Flags for content lines:
654  * 0..12 are whitespace rules
655  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
656  * 16 is marking if the line is blank at EOF
657  */
658 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
659 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
660 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
661 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
662 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
663
664 /*
665  * This struct is used when we need to buffer the output of the diff output.
666  *
667  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
668  * into the pre/post image file. This pointer could be a union with the
669  * line pointer. By storing an offset into the file instead of the literal line,
670  * we can decrease the memory footprint for the buffered output. At first we
671  * may want to only have indirection for the content lines, but we could also
672  * enhance the state for emitting prefabricated lines, e.g. the similarity
673  * score line or hunk/file headers would only need to store a number or path
674  * and then the output can be constructed later on depending on state.
675  */
676 struct emitted_diff_symbol {
677         const char *line;
678         int len;
679         int flags;
680         enum diff_symbol s;
681 };
682 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
683
684 struct emitted_diff_symbols {
685         struct emitted_diff_symbol *buf;
686         int nr, alloc;
687 };
688 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
689
690 static void append_emitted_diff_symbol(struct diff_options *o,
691                                        struct emitted_diff_symbol *e)
692 {
693         struct emitted_diff_symbol *f;
694
695         ALLOC_GROW(o->emitted_symbols->buf,
696                    o->emitted_symbols->nr + 1,
697                    o->emitted_symbols->alloc);
698         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
699
700         memcpy(f, e, sizeof(struct emitted_diff_symbol));
701         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
702 }
703
704 struct moved_entry {
705         struct hashmap_entry ent;
706         const struct emitted_diff_symbol *es;
707         struct moved_entry *next_line;
708 };
709
710 static int next_byte(const char **cp, const char **endp,
711                      const struct diff_options *diffopt)
712 {
713         int retval;
714
715         if (*cp > *endp)
716                 return -1;
717
718         if (isspace(**cp)) {
719                 if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE_CHANGE)) {
720                         while (*cp < *endp && isspace(**cp))
721                                 (*cp)++;
722                         /*
723                          * After skipping a couple of whitespaces,
724                          * we still have to account for one space.
725                          */
726                         return (int)' ';
727                 }
728
729                 if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE)) {
730                         while (*cp < *endp && isspace(**cp))
731                                 (*cp)++;
732                         /* return the first non-ws character via the usual below */
733                 }
734         }
735
736         retval = (unsigned char)(**cp);
737         (*cp)++;
738         return retval;
739 }
740
741 static int moved_entry_cmp(const struct diff_options *diffopt,
742                            const struct moved_entry *a,
743                            const struct moved_entry *b,
744                            const void *keydata)
745 {
746         const char *ap = a->es->line, *ae = a->es->line + a->es->len;
747         const char *bp = b->es->line, *be = b->es->line + b->es->len;
748
749         if (!(diffopt->xdl_opts & XDF_WHITESPACE_FLAGS))
750                 return a->es->len != b->es->len  || memcmp(ap, bp, a->es->len);
751
752         if (DIFF_XDL_TST(diffopt, IGNORE_WHITESPACE_AT_EOL)) {
753                 while (ae > ap && isspace(*ae))
754                         ae--;
755                 while (be > bp && isspace(*be))
756                         be--;
757         }
758
759         while (1) {
760                 int ca, cb;
761                 ca = next_byte(&ap, &ae, diffopt);
762                 cb = next_byte(&bp, &be, diffopt);
763                 if (ca != cb)
764                         return 1;
765                 if (ca < 0)
766                         return 0;
767         }
768 }
769
770 static unsigned get_string_hash(struct emitted_diff_symbol *es, struct diff_options *o)
771 {
772         if (o->xdl_opts & XDF_WHITESPACE_FLAGS) {
773                 static struct strbuf sb = STRBUF_INIT;
774                 const char *ap = es->line, *ae = es->line + es->len;
775                 int c;
776
777                 strbuf_reset(&sb);
778                 while (ae > ap && isspace(*ae))
779                         ae--;
780                 while ((c = next_byte(&ap, &ae, o)) > 0)
781                         strbuf_addch(&sb, c);
782
783                 return memhash(sb.buf, sb.len);
784         } else {
785                 return memhash(es->line, es->len);
786         }
787 }
788
789 static struct moved_entry *prepare_entry(struct diff_options *o,
790                                          int line_no)
791 {
792         struct moved_entry *ret = xmalloc(sizeof(*ret));
793         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
794
795         ret->ent.hash = get_string_hash(l, o);
796         ret->es = l;
797         ret->next_line = NULL;
798
799         return ret;
800 }
801
802 static void add_lines_to_move_detection(struct diff_options *o,
803                                         struct hashmap *add_lines,
804                                         struct hashmap *del_lines)
805 {
806         struct moved_entry *prev_line = NULL;
807
808         int n;
809         for (n = 0; n < o->emitted_symbols->nr; n++) {
810                 struct hashmap *hm;
811                 struct moved_entry *key;
812
813                 switch (o->emitted_symbols->buf[n].s) {
814                 case DIFF_SYMBOL_PLUS:
815                         hm = add_lines;
816                         break;
817                 case DIFF_SYMBOL_MINUS:
818                         hm = del_lines;
819                         break;
820                 default:
821                         prev_line = NULL;
822                         continue;
823                 }
824
825                 key = prepare_entry(o, n);
826                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
827                         prev_line->next_line = key;
828
829                 hashmap_add(hm, key);
830                 prev_line = key;
831         }
832 }
833
834 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
835                                          int pmb_nr)
836 {
837         int lp, rp;
838
839         /* Shrink the set of potential block to the remaining running */
840         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
841                 while (lp < pmb_nr && pmb[lp])
842                         lp++;
843                 /* lp points at the first NULL now */
844
845                 while (rp > -1 && !pmb[rp])
846                         rp--;
847                 /* rp points at the last non-NULL */
848
849                 if (lp < pmb_nr && rp > -1 && lp < rp) {
850                         pmb[lp] = pmb[rp];
851                         pmb[rp] = NULL;
852                         rp--;
853                         lp++;
854                 }
855         }
856
857         /* Remember the number of running sets */
858         return rp + 1;
859 }
860
861 /*
862  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
863  *
864  * Otherwise, if the last block has fewer alphanumeric characters than
865  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
866  * that block.
867  *
868  * The last block consists of the (n - block_length)'th line up to but not
869  * including the nth line.
870  *
871  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
872  * Think of a way to unify them.
873  */
874 static void adjust_last_block(struct diff_options *o, int n, int block_length)
875 {
876         int i, alnum_count = 0;
877         if (o->color_moved == COLOR_MOVED_PLAIN)
878                 return;
879         for (i = 1; i < block_length + 1; i++) {
880                 const char *c = o->emitted_symbols->buf[n - i].line;
881                 for (; *c; c++) {
882                         if (!isalnum(*c))
883                                 continue;
884                         alnum_count++;
885                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
886                                 return;
887                 }
888         }
889         for (i = 1; i < block_length + 1; i++)
890                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
891 }
892
893 /* Find blocks of moved code, delegate actual coloring decision to helper */
894 static void mark_color_as_moved(struct diff_options *o,
895                                 struct hashmap *add_lines,
896                                 struct hashmap *del_lines)
897 {
898         struct moved_entry **pmb = NULL; /* potentially moved blocks */
899         int pmb_nr = 0, pmb_alloc = 0;
900         int n, flipped_block = 1, block_length = 0;
901
902
903         for (n = 0; n < o->emitted_symbols->nr; n++) {
904                 struct hashmap *hm = NULL;
905                 struct moved_entry *key;
906                 struct moved_entry *match = NULL;
907                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
908                 int i;
909
910                 switch (l->s) {
911                 case DIFF_SYMBOL_PLUS:
912                         hm = del_lines;
913                         key = prepare_entry(o, n);
914                         match = hashmap_get(hm, key, o);
915                         free(key);
916                         break;
917                 case DIFF_SYMBOL_MINUS:
918                         hm = add_lines;
919                         key = prepare_entry(o, n);
920                         match = hashmap_get(hm, key, o);
921                         free(key);
922                         break;
923                 default:
924                         flipped_block = 1;
925                 }
926
927                 if (!match) {
928                         adjust_last_block(o, n, block_length);
929                         pmb_nr = 0;
930                         block_length = 0;
931                         continue;
932                 }
933
934                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
935
936                 if (o->color_moved == COLOR_MOVED_PLAIN)
937                         continue;
938
939                 /* Check any potential block runs, advance each or nullify */
940                 for (i = 0; i < pmb_nr; i++) {
941                         struct moved_entry *p = pmb[i];
942                         struct moved_entry *pnext = (p && p->next_line) ?
943                                         p->next_line : NULL;
944                         if (pnext && !hm->cmpfn(o, pnext, match, NULL)) {
945                                 pmb[i] = p->next_line;
946                         } else {
947                                 pmb[i] = NULL;
948                         }
949                 }
950
951                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
952
953                 if (pmb_nr == 0) {
954                         /*
955                          * The current line is the start of a new block.
956                          * Setup the set of potential blocks.
957                          */
958                         for (; match; match = hashmap_get_next(hm, match)) {
959                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
960                                 pmb[pmb_nr++] = match;
961                         }
962
963                         flipped_block = (flipped_block + 1) % 2;
964
965                         adjust_last_block(o, n, block_length);
966                         block_length = 0;
967                 }
968
969                 block_length++;
970
971                 if (flipped_block)
972                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
973         }
974         adjust_last_block(o, n, block_length);
975
976         free(pmb);
977 }
978
979 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
980   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
981 static void dim_moved_lines(struct diff_options *o)
982 {
983         int n;
984         for (n = 0; n < o->emitted_symbols->nr; n++) {
985                 struct emitted_diff_symbol *prev = (n != 0) ?
986                                 &o->emitted_symbols->buf[n - 1] : NULL;
987                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
988                 struct emitted_diff_symbol *next =
989                                 (n < o->emitted_symbols->nr - 1) ?
990                                 &o->emitted_symbols->buf[n + 1] : NULL;
991
992                 /* Not a plus or minus line? */
993                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
994                         continue;
995
996                 /* Not a moved line? */
997                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
998                         continue;
999
1000                 /*
1001                  * If prev or next are not a plus or minus line,
1002                  * pretend they don't exist
1003                  */
1004                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1005                             prev->s != DIFF_SYMBOL_MINUS)
1006                         prev = NULL;
1007                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1008                             next->s != DIFF_SYMBOL_MINUS)
1009                         next = NULL;
1010
1011                 /* Inside a block? */
1012                 if ((prev &&
1013                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1014                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1015                     (next &&
1016                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1017                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1018                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1019                         continue;
1020                 }
1021
1022                 /* Check if we are at an interesting bound: */
1023                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1024                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1025                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1026                         continue;
1027                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1028                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1029                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1030                         continue;
1031
1032                 /*
1033                  * The boundary to prev and next are not interesting,
1034                  * so this line is not interesting as a whole
1035                  */
1036                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1037         }
1038 }
1039
1040 static void emit_line_ws_markup(struct diff_options *o,
1041                                 const char *set, const char *reset,
1042                                 const char *line, int len, char sign,
1043                                 unsigned ws_rule, int blank_at_eof)
1044 {
1045         const char *ws = NULL;
1046
1047         if (o->ws_error_highlight & ws_rule) {
1048                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1049                 if (!*ws)
1050                         ws = NULL;
1051         }
1052
1053         if (!ws)
1054                 emit_line_0(o, set, reset, sign, line, len);
1055         else if (blank_at_eof)
1056                 /* Blank line at EOF - paint '+' as well */
1057                 emit_line_0(o, ws, reset, sign, line, len);
1058         else {
1059                 /* Emit just the prefix, then the rest. */
1060                 emit_line_0(o, set, reset, sign, "", 0);
1061                 ws_check_emit(line, len, ws_rule,
1062                               o->file, set, reset, ws);
1063         }
1064 }
1065
1066 static void emit_diff_symbol_from_struct(struct diff_options *o,
1067                                          struct emitted_diff_symbol *eds)
1068 {
1069         static const char *nneof = " No newline at end of file\n";
1070         const char *context, *reset, *set, *meta, *fraginfo;
1071         struct strbuf sb = STRBUF_INIT;
1072
1073         enum diff_symbol s = eds->s;
1074         const char *line = eds->line;
1075         int len = eds->len;
1076         unsigned flags = eds->flags;
1077
1078         switch (s) {
1079         case DIFF_SYMBOL_NO_LF_EOF:
1080                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1081                 reset = diff_get_color_opt(o, DIFF_RESET);
1082                 putc('\n', o->file);
1083                 emit_line_0(o, context, reset, '\\',
1084                             nneof, strlen(nneof));
1085                 break;
1086         case DIFF_SYMBOL_SUBMODULE_HEADER:
1087         case DIFF_SYMBOL_SUBMODULE_ERROR:
1088         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1089         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1090         case DIFF_SYMBOL_SUMMARY:
1091         case DIFF_SYMBOL_STATS_LINE:
1092         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1093         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1094                 emit_line(o, "", "", line, len);
1095                 break;
1096         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1097         case DIFF_SYMBOL_CONTEXT_MARKER:
1098                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1099                 reset = diff_get_color_opt(o, DIFF_RESET);
1100                 emit_line(o, context, reset, line, len);
1101                 break;
1102         case DIFF_SYMBOL_SEPARATOR:
1103                 fprintf(o->file, "%s%c",
1104                         diff_line_prefix(o),
1105                         o->line_termination);
1106                 break;
1107         case DIFF_SYMBOL_CONTEXT:
1108                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1109                 reset = diff_get_color_opt(o, DIFF_RESET);
1110                 emit_line_ws_markup(o, set, reset, line, len, ' ',
1111                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1112                 break;
1113         case DIFF_SYMBOL_PLUS:
1114                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1115                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1116                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1117                 case DIFF_SYMBOL_MOVED_LINE |
1118                      DIFF_SYMBOL_MOVED_LINE_ALT |
1119                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1120                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1121                         break;
1122                 case DIFF_SYMBOL_MOVED_LINE |
1123                      DIFF_SYMBOL_MOVED_LINE_ALT:
1124                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1125                         break;
1126                 case DIFF_SYMBOL_MOVED_LINE |
1127                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1128                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1129                         break;
1130                 case DIFF_SYMBOL_MOVED_LINE:
1131                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1132                         break;
1133                 default:
1134                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1135                 }
1136                 reset = diff_get_color_opt(o, DIFF_RESET);
1137                 emit_line_ws_markup(o, set, reset, line, len, '+',
1138                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1139                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1140                 break;
1141         case DIFF_SYMBOL_MINUS:
1142                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1143                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1144                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1145                 case DIFF_SYMBOL_MOVED_LINE |
1146                      DIFF_SYMBOL_MOVED_LINE_ALT |
1147                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1148                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1149                         break;
1150                 case DIFF_SYMBOL_MOVED_LINE |
1151                      DIFF_SYMBOL_MOVED_LINE_ALT:
1152                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1153                         break;
1154                 case DIFF_SYMBOL_MOVED_LINE |
1155                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1156                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1157                         break;
1158                 case DIFF_SYMBOL_MOVED_LINE:
1159                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1160                         break;
1161                 default:
1162                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1163                 }
1164                 reset = diff_get_color_opt(o, DIFF_RESET);
1165                 emit_line_ws_markup(o, set, reset, line, len, '-',
1166                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1167                 break;
1168         case DIFF_SYMBOL_WORDS_PORCELAIN:
1169                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1170                 reset = diff_get_color_opt(o, DIFF_RESET);
1171                 emit_line(o, context, reset, line, len);
1172                 fputs("~\n", o->file);
1173                 break;
1174         case DIFF_SYMBOL_WORDS:
1175                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1176                 reset = diff_get_color_opt(o, DIFF_RESET);
1177                 /*
1178                  * Skip the prefix character, if any.  With
1179                  * diff_suppress_blank_empty, there may be
1180                  * none.
1181                  */
1182                 if (line[0] != '\n') {
1183                         line++;
1184                         len--;
1185                 }
1186                 emit_line(o, context, reset, line, len);
1187                 break;
1188         case DIFF_SYMBOL_FILEPAIR_PLUS:
1189                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1190                 reset = diff_get_color_opt(o, DIFF_RESET);
1191                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1192                         line, reset,
1193                         strchr(line, ' ') ? "\t" : "");
1194                 break;
1195         case DIFF_SYMBOL_FILEPAIR_MINUS:
1196                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1197                 reset = diff_get_color_opt(o, DIFF_RESET);
1198                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1199                         line, reset,
1200                         strchr(line, ' ') ? "\t" : "");
1201                 break;
1202         case DIFF_SYMBOL_BINARY_FILES:
1203         case DIFF_SYMBOL_HEADER:
1204                 fprintf(o->file, "%s", line);
1205                 break;
1206         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1207                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1208                 break;
1209         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1210                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1211                 break;
1212         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1213                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1214                 break;
1215         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1216                 fputs(diff_line_prefix(o), o->file);
1217                 fputc('\n', o->file);
1218                 break;
1219         case DIFF_SYMBOL_REWRITE_DIFF:
1220                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1221                 reset = diff_get_color_opt(o, DIFF_RESET);
1222                 emit_line(o, fraginfo, reset, line, len);
1223                 break;
1224         case DIFF_SYMBOL_SUBMODULE_ADD:
1225                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1226                 reset = diff_get_color_opt(o, DIFF_RESET);
1227                 emit_line(o, set, reset, line, len);
1228                 break;
1229         case DIFF_SYMBOL_SUBMODULE_DEL:
1230                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1231                 reset = diff_get_color_opt(o, DIFF_RESET);
1232                 emit_line(o, set, reset, line, len);
1233                 break;
1234         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1235                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1236                         diff_line_prefix(o), line);
1237                 break;
1238         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1239                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1240                         diff_line_prefix(o), line);
1241                 break;
1242         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1243                 emit_line(o, "", "", " 0 files changed\n",
1244                           strlen(" 0 files changed\n"));
1245                 break;
1246         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1247                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1248                 break;
1249         case DIFF_SYMBOL_WORD_DIFF:
1250                 fprintf(o->file, "%.*s", len, line);
1251                 break;
1252         case DIFF_SYMBOL_STAT_SEP:
1253                 fputs(o->stat_sep, o->file);
1254                 break;
1255         default:
1256                 die("BUG: unknown diff symbol");
1257         }
1258         strbuf_release(&sb);
1259 }
1260
1261 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1262                              const char *line, int len, unsigned flags)
1263 {
1264         struct emitted_diff_symbol e = {line, len, flags, s};
1265
1266         if (o->emitted_symbols)
1267                 append_emitted_diff_symbol(o, &e);
1268         else
1269                 emit_diff_symbol_from_struct(o, &e);
1270 }
1271
1272 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1273 {
1274         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1275 }
1276
1277 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1278 {
1279         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1280 }
1281
1282 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1283 {
1284         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1285                          path, strlen(path), 0);
1286 }
1287
1288 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1289 {
1290         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1291                          path, strlen(path), 0);
1292 }
1293
1294 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1295 {
1296         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1297                          header, strlen(header), 0);
1298 }
1299
1300 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1301 {
1302         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1303 }
1304
1305 void diff_emit_submodule_pipethrough(struct diff_options *o,
1306                                      const char *line, int len)
1307 {
1308         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1309 }
1310
1311 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1312 {
1313         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1314               ecbdata->blank_at_eof_in_preimage &&
1315               ecbdata->blank_at_eof_in_postimage &&
1316               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1317               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1318                 return 0;
1319         return ws_blank_line(line, len, ecbdata->ws_rule);
1320 }
1321
1322 static void emit_add_line(const char *reset,
1323                           struct emit_callback *ecbdata,
1324                           const char *line, int len)
1325 {
1326         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1327         if (new_blank_line_at_eof(ecbdata, line, len))
1328                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1329
1330         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1331 }
1332
1333 static void emit_del_line(const char *reset,
1334                           struct emit_callback *ecbdata,
1335                           const char *line, int len)
1336 {
1337         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1338         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1339 }
1340
1341 static void emit_context_line(const char *reset,
1342                               struct emit_callback *ecbdata,
1343                               const char *line, int len)
1344 {
1345         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1346         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1347 }
1348
1349 static void emit_hunk_header(struct emit_callback *ecbdata,
1350                              const char *line, int len)
1351 {
1352         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1353         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1354         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1355         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1356         static const char atat[2] = { '@', '@' };
1357         const char *cp, *ep;
1358         struct strbuf msgbuf = STRBUF_INIT;
1359         int org_len = len;
1360         int i = 1;
1361
1362         /*
1363          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1364          * it always is at least 10 bytes long.
1365          */
1366         if (len < 10 ||
1367             memcmp(line, atat, 2) ||
1368             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1369                 emit_diff_symbol(ecbdata->opt,
1370                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1371                 return;
1372         }
1373         ep += 2; /* skip over @@ */
1374
1375         /* The hunk header in fraginfo color */
1376         strbuf_addstr(&msgbuf, frag);
1377         strbuf_add(&msgbuf, line, ep - line);
1378         strbuf_addstr(&msgbuf, reset);
1379
1380         /*
1381          * trailing "\r\n"
1382          */
1383         for ( ; i < 3; i++)
1384                 if (line[len - i] == '\r' || line[len - i] == '\n')
1385                         len--;
1386
1387         /* blank before the func header */
1388         for (cp = ep; ep - line < len; ep++)
1389                 if (*ep != ' ' && *ep != '\t')
1390                         break;
1391         if (ep != cp) {
1392                 strbuf_addstr(&msgbuf, context);
1393                 strbuf_add(&msgbuf, cp, ep - cp);
1394                 strbuf_addstr(&msgbuf, reset);
1395         }
1396
1397         if (ep < line + len) {
1398                 strbuf_addstr(&msgbuf, func);
1399                 strbuf_add(&msgbuf, ep, line + len - ep);
1400                 strbuf_addstr(&msgbuf, reset);
1401         }
1402
1403         strbuf_add(&msgbuf, line + len, org_len - len);
1404         strbuf_complete_line(&msgbuf);
1405         emit_diff_symbol(ecbdata->opt,
1406                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1407         strbuf_release(&msgbuf);
1408 }
1409
1410 static struct diff_tempfile *claim_diff_tempfile(void) {
1411         int i;
1412         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1413                 if (!diff_temp[i].name)
1414                         return diff_temp + i;
1415         die("BUG: diff is failing to clean up its tempfiles");
1416 }
1417
1418 static void remove_tempfile(void)
1419 {
1420         int i;
1421         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1422                 if (is_tempfile_active(diff_temp[i].tempfile))
1423                         delete_tempfile(&diff_temp[i].tempfile);
1424                 diff_temp[i].name = NULL;
1425         }
1426 }
1427
1428 static void add_line_count(struct strbuf *out, int count)
1429 {
1430         switch (count) {
1431         case 0:
1432                 strbuf_addstr(out, "0,0");
1433                 break;
1434         case 1:
1435                 strbuf_addstr(out, "1");
1436                 break;
1437         default:
1438                 strbuf_addf(out, "1,%d", count);
1439                 break;
1440         }
1441 }
1442
1443 static void emit_rewrite_lines(struct emit_callback *ecb,
1444                                int prefix, const char *data, int size)
1445 {
1446         const char *endp = NULL;
1447         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1448
1449         while (0 < size) {
1450                 int len;
1451
1452                 endp = memchr(data, '\n', size);
1453                 len = endp ? (endp - data + 1) : size;
1454                 if (prefix != '+') {
1455                         ecb->lno_in_preimage++;
1456                         emit_del_line(reset, ecb, data, len);
1457                 } else {
1458                         ecb->lno_in_postimage++;
1459                         emit_add_line(reset, ecb, data, len);
1460                 }
1461                 size -= len;
1462                 data += len;
1463         }
1464         if (!endp)
1465                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1466 }
1467
1468 static void emit_rewrite_diff(const char *name_a,
1469                               const char *name_b,
1470                               struct diff_filespec *one,
1471                               struct diff_filespec *two,
1472                               struct userdiff_driver *textconv_one,
1473                               struct userdiff_driver *textconv_two,
1474                               struct diff_options *o)
1475 {
1476         int lc_a, lc_b;
1477         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1478         const char *a_prefix, *b_prefix;
1479         char *data_one, *data_two;
1480         size_t size_one, size_two;
1481         struct emit_callback ecbdata;
1482         struct strbuf out = STRBUF_INIT;
1483
1484         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
1485                 a_prefix = o->b_prefix;
1486                 b_prefix = o->a_prefix;
1487         } else {
1488                 a_prefix = o->a_prefix;
1489                 b_prefix = o->b_prefix;
1490         }
1491
1492         name_a += (*name_a == '/');
1493         name_b += (*name_b == '/');
1494
1495         strbuf_reset(&a_name);
1496         strbuf_reset(&b_name);
1497         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1498         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1499
1500         size_one = fill_textconv(textconv_one, one, &data_one);
1501         size_two = fill_textconv(textconv_two, two, &data_two);
1502
1503         memset(&ecbdata, 0, sizeof(ecbdata));
1504         ecbdata.color_diff = want_color(o->use_color);
1505         ecbdata.ws_rule = whitespace_rule(name_b);
1506         ecbdata.opt = o;
1507         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1508                 mmfile_t mf1, mf2;
1509                 mf1.ptr = (char *)data_one;
1510                 mf2.ptr = (char *)data_two;
1511                 mf1.size = size_one;
1512                 mf2.size = size_two;
1513                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1514         }
1515         ecbdata.lno_in_preimage = 1;
1516         ecbdata.lno_in_postimage = 1;
1517
1518         lc_a = count_lines(data_one, size_one);
1519         lc_b = count_lines(data_two, size_two);
1520
1521         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1522                          a_name.buf, a_name.len, 0);
1523         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1524                          b_name.buf, b_name.len, 0);
1525
1526         strbuf_addstr(&out, "@@ -");
1527         if (!o->irreversible_delete)
1528                 add_line_count(&out, lc_a);
1529         else
1530                 strbuf_addstr(&out, "?,?");
1531         strbuf_addstr(&out, " +");
1532         add_line_count(&out, lc_b);
1533         strbuf_addstr(&out, " @@\n");
1534         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1535         strbuf_release(&out);
1536
1537         if (lc_a && !o->irreversible_delete)
1538                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1539         if (lc_b)
1540                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1541         if (textconv_one)
1542                 free((char *)data_one);
1543         if (textconv_two)
1544                 free((char *)data_two);
1545 }
1546
1547 struct diff_words_buffer {
1548         mmfile_t text;
1549         unsigned long alloc;
1550         struct diff_words_orig {
1551                 const char *begin, *end;
1552         } *orig;
1553         int orig_nr, orig_alloc;
1554 };
1555
1556 static void diff_words_append(char *line, unsigned long len,
1557                 struct diff_words_buffer *buffer)
1558 {
1559         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1560         line++;
1561         len--;
1562         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1563         buffer->text.size += len;
1564         buffer->text.ptr[buffer->text.size] = '\0';
1565 }
1566
1567 struct diff_words_style_elem {
1568         const char *prefix;
1569         const char *suffix;
1570         const char *color; /* NULL; filled in by the setup code if
1571                             * color is enabled */
1572 };
1573
1574 struct diff_words_style {
1575         enum diff_words_type type;
1576         struct diff_words_style_elem new, old, ctx;
1577         const char *newline;
1578 };
1579
1580 static struct diff_words_style diff_words_styles[] = {
1581         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1582         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1583         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1584 };
1585
1586 struct diff_words_data {
1587         struct diff_words_buffer minus, plus;
1588         const char *current_plus;
1589         int last_minus;
1590         struct diff_options *opt;
1591         regex_t *word_regex;
1592         enum diff_words_type type;
1593         struct diff_words_style *style;
1594 };
1595
1596 static int fn_out_diff_words_write_helper(struct diff_options *o,
1597                                           struct diff_words_style_elem *st_el,
1598                                           const char *newline,
1599                                           size_t count, const char *buf)
1600 {
1601         int print = 0;
1602         struct strbuf sb = STRBUF_INIT;
1603
1604         while (count) {
1605                 char *p = memchr(buf, '\n', count);
1606                 if (print)
1607                         strbuf_addstr(&sb, diff_line_prefix(o));
1608
1609                 if (p != buf) {
1610                         const char *reset = st_el->color && *st_el->color ?
1611                                             GIT_COLOR_RESET : NULL;
1612                         if (st_el->color && *st_el->color)
1613                                 strbuf_addstr(&sb, st_el->color);
1614                         strbuf_addstr(&sb, st_el->prefix);
1615                         strbuf_add(&sb, buf, p ? p - buf : count);
1616                         strbuf_addstr(&sb, st_el->suffix);
1617                         if (reset)
1618                                 strbuf_addstr(&sb, reset);
1619                 }
1620                 if (!p)
1621                         goto out;
1622
1623                 strbuf_addstr(&sb, newline);
1624                 count -= p + 1 - buf;
1625                 buf = p + 1;
1626                 print = 1;
1627                 if (count) {
1628                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1629                                          sb.buf, sb.len, 0);
1630                         strbuf_reset(&sb);
1631                 }
1632         }
1633
1634 out:
1635         if (sb.len)
1636                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1637                                  sb.buf, sb.len, 0);
1638         strbuf_release(&sb);
1639         return 0;
1640 }
1641
1642 /*
1643  * '--color-words' algorithm can be described as:
1644  *
1645  *   1. collect the minus/plus lines of a diff hunk, divided into
1646  *      minus-lines and plus-lines;
1647  *
1648  *   2. break both minus-lines and plus-lines into words and
1649  *      place them into two mmfile_t with one word for each line;
1650  *
1651  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1652  *
1653  * And for the common parts of the both file, we output the plus side text.
1654  * diff_words->current_plus is used to trace the current position of the plus file
1655  * which printed. diff_words->last_minus is used to trace the last minus word
1656  * printed.
1657  *
1658  * For '--graph' to work with '--color-words', we need to output the graph prefix
1659  * on each line of color words output. Generally, there are two conditions on
1660  * which we should output the prefix.
1661  *
1662  *   1. diff_words->last_minus == 0 &&
1663  *      diff_words->current_plus == diff_words->plus.text.ptr
1664  *
1665  *      that is: the plus text must start as a new line, and if there is no minus
1666  *      word printed, a graph prefix must be printed.
1667  *
1668  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1669  *      *(diff_words->current_plus - 1) == '\n'
1670  *
1671  *      that is: a graph prefix must be printed following a '\n'
1672  */
1673 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1674 {
1675         if ((diff_words->last_minus == 0 &&
1676                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1677                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1678                 *(diff_words->current_plus - 1) == '\n')) {
1679                 return 1;
1680         } else {
1681                 return 0;
1682         }
1683 }
1684
1685 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1686 {
1687         struct diff_words_data *diff_words = priv;
1688         struct diff_words_style *style = diff_words->style;
1689         int minus_first, minus_len, plus_first, plus_len;
1690         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1691         struct diff_options *opt = diff_words->opt;
1692         const char *line_prefix;
1693
1694         if (line[0] != '@' || parse_hunk_header(line, len,
1695                         &minus_first, &minus_len, &plus_first, &plus_len))
1696                 return;
1697
1698         assert(opt);
1699         line_prefix = diff_line_prefix(opt);
1700
1701         /* POSIX requires that first be decremented by one if len == 0... */
1702         if (minus_len) {
1703                 minus_begin = diff_words->minus.orig[minus_first].begin;
1704                 minus_end =
1705                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1706         } else
1707                 minus_begin = minus_end =
1708                         diff_words->minus.orig[minus_first].end;
1709
1710         if (plus_len) {
1711                 plus_begin = diff_words->plus.orig[plus_first].begin;
1712                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1713         } else
1714                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1715
1716         if (color_words_output_graph_prefix(diff_words)) {
1717                 fputs(line_prefix, diff_words->opt->file);
1718         }
1719         if (diff_words->current_plus != plus_begin) {
1720                 fn_out_diff_words_write_helper(diff_words->opt,
1721                                 &style->ctx, style->newline,
1722                                 plus_begin - diff_words->current_plus,
1723                                 diff_words->current_plus);
1724         }
1725         if (minus_begin != minus_end) {
1726                 fn_out_diff_words_write_helper(diff_words->opt,
1727                                 &style->old, style->newline,
1728                                 minus_end - minus_begin, minus_begin);
1729         }
1730         if (plus_begin != plus_end) {
1731                 fn_out_diff_words_write_helper(diff_words->opt,
1732                                 &style->new, style->newline,
1733                                 plus_end - plus_begin, plus_begin);
1734         }
1735
1736         diff_words->current_plus = plus_end;
1737         diff_words->last_minus = minus_first;
1738 }
1739
1740 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1741 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1742                 int *begin, int *end)
1743 {
1744         if (word_regex && *begin < buffer->size) {
1745                 regmatch_t match[1];
1746                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1747                                  buffer->size - *begin, 1, match, 0)) {
1748                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1749                                         '\n', match[0].rm_eo - match[0].rm_so);
1750                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1751                         *begin += match[0].rm_so;
1752                         return *begin >= *end;
1753                 }
1754                 return -1;
1755         }
1756
1757         /* find the next word */
1758         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1759                 (*begin)++;
1760         if (*begin >= buffer->size)
1761                 return -1;
1762
1763         /* find the end of the word */
1764         *end = *begin + 1;
1765         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1766                 (*end)++;
1767
1768         return 0;
1769 }
1770
1771 /*
1772  * This function splits the words in buffer->text, stores the list with
1773  * newline separator into out, and saves the offsets of the original words
1774  * in buffer->orig.
1775  */
1776 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1777                 regex_t *word_regex)
1778 {
1779         int i, j;
1780         long alloc = 0;
1781
1782         out->size = 0;
1783         out->ptr = NULL;
1784
1785         /* fake an empty "0th" word */
1786         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1787         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1788         buffer->orig_nr = 1;
1789
1790         for (i = 0; i < buffer->text.size; i++) {
1791                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1792                         return;
1793
1794                 /* store original boundaries */
1795                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1796                                 buffer->orig_alloc);
1797                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1798                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1799                 buffer->orig_nr++;
1800
1801                 /* store one word */
1802                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1803                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1804                 out->ptr[out->size + j - i] = '\n';
1805                 out->size += j - i + 1;
1806
1807                 i = j - 1;
1808         }
1809 }
1810
1811 /* this executes the word diff on the accumulated buffers */
1812 static void diff_words_show(struct diff_words_data *diff_words)
1813 {
1814         xpparam_t xpp;
1815         xdemitconf_t xecfg;
1816         mmfile_t minus, plus;
1817         struct diff_words_style *style = diff_words->style;
1818
1819         struct diff_options *opt = diff_words->opt;
1820         const char *line_prefix;
1821
1822         assert(opt);
1823         line_prefix = diff_line_prefix(opt);
1824
1825         /* special case: only removal */
1826         if (!diff_words->plus.text.size) {
1827                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1828                                  line_prefix, strlen(line_prefix), 0);
1829                 fn_out_diff_words_write_helper(diff_words->opt,
1830                         &style->old, style->newline,
1831                         diff_words->minus.text.size,
1832                         diff_words->minus.text.ptr);
1833                 diff_words->minus.text.size = 0;
1834                 return;
1835         }
1836
1837         diff_words->current_plus = diff_words->plus.text.ptr;
1838         diff_words->last_minus = 0;
1839
1840         memset(&xpp, 0, sizeof(xpp));
1841         memset(&xecfg, 0, sizeof(xecfg));
1842         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1843         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1844         xpp.flags = 0;
1845         /* as only the hunk header will be parsed, we need a 0-context */
1846         xecfg.ctxlen = 0;
1847         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1848                           &xpp, &xecfg))
1849                 die("unable to generate word diff");
1850         free(minus.ptr);
1851         free(plus.ptr);
1852         if (diff_words->current_plus != diff_words->plus.text.ptr +
1853                         diff_words->plus.text.size) {
1854                 if (color_words_output_graph_prefix(diff_words))
1855                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1856                                          line_prefix, strlen(line_prefix), 0);
1857                 fn_out_diff_words_write_helper(diff_words->opt,
1858                         &style->ctx, style->newline,
1859                         diff_words->plus.text.ptr + diff_words->plus.text.size
1860                         - diff_words->current_plus, diff_words->current_plus);
1861         }
1862         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1863 }
1864
1865 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1866 static void diff_words_flush(struct emit_callback *ecbdata)
1867 {
1868         struct diff_options *wo = ecbdata->diff_words->opt;
1869
1870         if (ecbdata->diff_words->minus.text.size ||
1871             ecbdata->diff_words->plus.text.size)
1872                 diff_words_show(ecbdata->diff_words);
1873
1874         if (wo->emitted_symbols) {
1875                 struct diff_options *o = ecbdata->opt;
1876                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
1877                 int i;
1878
1879                 /*
1880                  * NEEDSWORK:
1881                  * Instead of appending each, concat all words to a line?
1882                  */
1883                 for (i = 0; i < wol->nr; i++)
1884                         append_emitted_diff_symbol(o, &wol->buf[i]);
1885
1886                 for (i = 0; i < wol->nr; i++)
1887                         free((void *)wol->buf[i].line);
1888
1889                 wol->nr = 0;
1890         }
1891 }
1892
1893 static void diff_filespec_load_driver(struct diff_filespec *one)
1894 {
1895         /* Use already-loaded driver */
1896         if (one->driver)
1897                 return;
1898
1899         if (S_ISREG(one->mode))
1900                 one->driver = userdiff_find_by_path(one->path);
1901
1902         /* Fallback to default settings */
1903         if (!one->driver)
1904                 one->driver = userdiff_find_by_name("default");
1905 }
1906
1907 static const char *userdiff_word_regex(struct diff_filespec *one)
1908 {
1909         diff_filespec_load_driver(one);
1910         return one->driver->word_regex;
1911 }
1912
1913 static void init_diff_words_data(struct emit_callback *ecbdata,
1914                                  struct diff_options *orig_opts,
1915                                  struct diff_filespec *one,
1916                                  struct diff_filespec *two)
1917 {
1918         int i;
1919         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1920         memcpy(o, orig_opts, sizeof(struct diff_options));
1921
1922         ecbdata->diff_words =
1923                 xcalloc(1, sizeof(struct diff_words_data));
1924         ecbdata->diff_words->type = o->word_diff;
1925         ecbdata->diff_words->opt = o;
1926
1927         if (orig_opts->emitted_symbols)
1928                 o->emitted_symbols =
1929                         xcalloc(1, sizeof(struct emitted_diff_symbols));
1930
1931         if (!o->word_regex)
1932                 o->word_regex = userdiff_word_regex(one);
1933         if (!o->word_regex)
1934                 o->word_regex = userdiff_word_regex(two);
1935         if (!o->word_regex)
1936                 o->word_regex = diff_word_regex_cfg;
1937         if (o->word_regex) {
1938                 ecbdata->diff_words->word_regex = (regex_t *)
1939                         xmalloc(sizeof(regex_t));
1940                 if (regcomp(ecbdata->diff_words->word_regex,
1941                             o->word_regex,
1942                             REG_EXTENDED | REG_NEWLINE))
1943                         die ("Invalid regular expression: %s",
1944                              o->word_regex);
1945         }
1946         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1947                 if (o->word_diff == diff_words_styles[i].type) {
1948                         ecbdata->diff_words->style =
1949                                 &diff_words_styles[i];
1950                         break;
1951                 }
1952         }
1953         if (want_color(o->use_color)) {
1954                 struct diff_words_style *st = ecbdata->diff_words->style;
1955                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1956                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1957                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1958         }
1959 }
1960
1961 static void free_diff_words_data(struct emit_callback *ecbdata)
1962 {
1963         if (ecbdata->diff_words) {
1964                 diff_words_flush(ecbdata);
1965                 free (ecbdata->diff_words->opt->emitted_symbols);
1966                 free (ecbdata->diff_words->opt);
1967                 free (ecbdata->diff_words->minus.text.ptr);
1968                 free (ecbdata->diff_words->minus.orig);
1969                 free (ecbdata->diff_words->plus.text.ptr);
1970                 free (ecbdata->diff_words->plus.orig);
1971                 if (ecbdata->diff_words->word_regex) {
1972                         regfree(ecbdata->diff_words->word_regex);
1973                         free(ecbdata->diff_words->word_regex);
1974                 }
1975                 FREE_AND_NULL(ecbdata->diff_words);
1976         }
1977 }
1978
1979 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1980 {
1981         if (want_color(diff_use_color))
1982                 return diff_colors[ix];
1983         return "";
1984 }
1985
1986 const char *diff_line_prefix(struct diff_options *opt)
1987 {
1988         struct strbuf *msgbuf;
1989         if (!opt->output_prefix)
1990                 return "";
1991
1992         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1993         return msgbuf->buf;
1994 }
1995
1996 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1997 {
1998         const char *cp;
1999         unsigned long allot;
2000         size_t l = len;
2001
2002         cp = line;
2003         allot = l;
2004         while (0 < l) {
2005                 (void) utf8_width(&cp, &l);
2006                 if (!cp)
2007                         break; /* truncated in the middle? */
2008         }
2009         return allot - l;
2010 }
2011
2012 static void find_lno(const char *line, struct emit_callback *ecbdata)
2013 {
2014         const char *p;
2015         ecbdata->lno_in_preimage = 0;
2016         ecbdata->lno_in_postimage = 0;
2017         p = strchr(line, '-');
2018         if (!p)
2019                 return; /* cannot happen */
2020         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2021         p = strchr(p, '+');
2022         if (!p)
2023                 return; /* cannot happen */
2024         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2025 }
2026
2027 static void fn_out_consume(void *priv, char *line, unsigned long len)
2028 {
2029         struct emit_callback *ecbdata = priv;
2030         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2031         struct diff_options *o = ecbdata->opt;
2032
2033         o->found_changes = 1;
2034
2035         if (ecbdata->header) {
2036                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2037                                  ecbdata->header->buf, ecbdata->header->len, 0);
2038                 strbuf_reset(ecbdata->header);
2039                 ecbdata->header = NULL;
2040         }
2041
2042         if (ecbdata->label_path[0]) {
2043                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2044                                  ecbdata->label_path[0],
2045                                  strlen(ecbdata->label_path[0]), 0);
2046                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2047                                  ecbdata->label_path[1],
2048                                  strlen(ecbdata->label_path[1]), 0);
2049                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2050         }
2051
2052         if (diff_suppress_blank_empty
2053             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2054                 line[0] = '\n';
2055                 len = 1;
2056         }
2057
2058         if (line[0] == '@') {
2059                 if (ecbdata->diff_words)
2060                         diff_words_flush(ecbdata);
2061                 len = sane_truncate_line(ecbdata, line, len);
2062                 find_lno(line, ecbdata);
2063                 emit_hunk_header(ecbdata, line, len);
2064                 return;
2065         }
2066
2067         if (ecbdata->diff_words) {
2068                 enum diff_symbol s =
2069                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2070                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2071                 if (line[0] == '-') {
2072                         diff_words_append(line, len,
2073                                           &ecbdata->diff_words->minus);
2074                         return;
2075                 } else if (line[0] == '+') {
2076                         diff_words_append(line, len,
2077                                           &ecbdata->diff_words->plus);
2078                         return;
2079                 } else if (starts_with(line, "\\ ")) {
2080                         /*
2081                          * Eat the "no newline at eof" marker as if we
2082                          * saw a "+" or "-" line with nothing on it,
2083                          * and return without diff_words_flush() to
2084                          * defer processing. If this is the end of
2085                          * preimage, more "+" lines may come after it.
2086                          */
2087                         return;
2088                 }
2089                 diff_words_flush(ecbdata);
2090                 emit_diff_symbol(o, s, line, len, 0);
2091                 return;
2092         }
2093
2094         switch (line[0]) {
2095         case '+':
2096                 ecbdata->lno_in_postimage++;
2097                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2098                 break;
2099         case '-':
2100                 ecbdata->lno_in_preimage++;
2101                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2102                 break;
2103         case ' ':
2104                 ecbdata->lno_in_postimage++;
2105                 ecbdata->lno_in_preimage++;
2106                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2107                 break;
2108         default:
2109                 /* incomplete line at the end */
2110                 ecbdata->lno_in_preimage++;
2111                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2112                                  line, len, 0);
2113                 break;
2114         }
2115 }
2116
2117 static char *pprint_rename(const char *a, const char *b)
2118 {
2119         const char *old = a;
2120         const char *new = b;
2121         struct strbuf name = STRBUF_INIT;
2122         int pfx_length, sfx_length;
2123         int pfx_adjust_for_slash;
2124         int len_a = strlen(a);
2125         int len_b = strlen(b);
2126         int a_midlen, b_midlen;
2127         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2128         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2129
2130         if (qlen_a || qlen_b) {
2131                 quote_c_style(a, &name, NULL, 0);
2132                 strbuf_addstr(&name, " => ");
2133                 quote_c_style(b, &name, NULL, 0);
2134                 return strbuf_detach(&name, NULL);
2135         }
2136
2137         /* Find common prefix */
2138         pfx_length = 0;
2139         while (*old && *new && *old == *new) {
2140                 if (*old == '/')
2141                         pfx_length = old - a + 1;
2142                 old++;
2143                 new++;
2144         }
2145
2146         /* Find common suffix */
2147         old = a + len_a;
2148         new = b + len_b;
2149         sfx_length = 0;
2150         /*
2151          * If there is a common prefix, it must end in a slash.  In
2152          * that case we let this loop run 1 into the prefix to see the
2153          * same slash.
2154          *
2155          * If there is no common prefix, we cannot do this as it would
2156          * underrun the input strings.
2157          */
2158         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2159         while (a + pfx_length - pfx_adjust_for_slash <= old &&
2160                b + pfx_length - pfx_adjust_for_slash <= new &&
2161                *old == *new) {
2162                 if (*old == '/')
2163                         sfx_length = len_a - (old - a);
2164                 old--;
2165                 new--;
2166         }
2167
2168         /*
2169          * pfx{mid-a => mid-b}sfx
2170          * {pfx-a => pfx-b}sfx
2171          * pfx{sfx-a => sfx-b}
2172          * name-a => name-b
2173          */
2174         a_midlen = len_a - pfx_length - sfx_length;
2175         b_midlen = len_b - pfx_length - sfx_length;
2176         if (a_midlen < 0)
2177                 a_midlen = 0;
2178         if (b_midlen < 0)
2179                 b_midlen = 0;
2180
2181         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2182         if (pfx_length + sfx_length) {
2183                 strbuf_add(&name, a, pfx_length);
2184                 strbuf_addch(&name, '{');
2185         }
2186         strbuf_add(&name, a + pfx_length, a_midlen);
2187         strbuf_addstr(&name, " => ");
2188         strbuf_add(&name, b + pfx_length, b_midlen);
2189         if (pfx_length + sfx_length) {
2190                 strbuf_addch(&name, '}');
2191                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
2192         }
2193         return strbuf_detach(&name, NULL);
2194 }
2195
2196 struct diffstat_t {
2197         int nr;
2198         int alloc;
2199         struct diffstat_file {
2200                 char *from_name;
2201                 char *name;
2202                 char *print_name;
2203                 unsigned is_unmerged:1;
2204                 unsigned is_binary:1;
2205                 unsigned is_renamed:1;
2206                 unsigned is_interesting:1;
2207                 uintmax_t added, deleted;
2208         } **files;
2209 };
2210
2211 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2212                                           const char *name_a,
2213                                           const char *name_b)
2214 {
2215         struct diffstat_file *x;
2216         x = xcalloc(1, sizeof(*x));
2217         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2218         diffstat->files[diffstat->nr++] = x;
2219         if (name_b) {
2220                 x->from_name = xstrdup(name_a);
2221                 x->name = xstrdup(name_b);
2222                 x->is_renamed = 1;
2223         }
2224         else {
2225                 x->from_name = NULL;
2226                 x->name = xstrdup(name_a);
2227         }
2228         return x;
2229 }
2230
2231 static void diffstat_consume(void *priv, char *line, unsigned long len)
2232 {
2233         struct diffstat_t *diffstat = priv;
2234         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2235
2236         if (line[0] == '+')
2237                 x->added++;
2238         else if (line[0] == '-')
2239                 x->deleted++;
2240 }
2241
2242 const char mime_boundary_leader[] = "------------";
2243
2244 static int scale_linear(int it, int width, int max_change)
2245 {
2246         if (!it)
2247                 return 0;
2248         /*
2249          * make sure that at least one '-' or '+' is printed if
2250          * there is any change to this path. The easiest way is to
2251          * scale linearly as if the alloted width is one column shorter
2252          * than it is, and then add 1 to the result.
2253          */
2254         return 1 + (it * (width - 1) / max_change);
2255 }
2256
2257 static void show_graph(struct strbuf *out, char ch, int cnt,
2258                        const char *set, const char *reset)
2259 {
2260         if (cnt <= 0)
2261                 return;
2262         strbuf_addstr(out, set);
2263         strbuf_addchars(out, ch, cnt);
2264         strbuf_addstr(out, reset);
2265 }
2266
2267 static void fill_print_name(struct diffstat_file *file)
2268 {
2269         char *pname;
2270
2271         if (file->print_name)
2272                 return;
2273
2274         if (!file->is_renamed) {
2275                 struct strbuf buf = STRBUF_INIT;
2276                 if (quote_c_style(file->name, &buf, NULL, 0)) {
2277                         pname = strbuf_detach(&buf, NULL);
2278                 } else {
2279                         pname = file->name;
2280                         strbuf_release(&buf);
2281                 }
2282         } else {
2283                 pname = pprint_rename(file->from_name, file->name);
2284         }
2285         file->print_name = pname;
2286 }
2287
2288 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2289                 int files, int insertions, int deletions)
2290 {
2291         struct strbuf sb = STRBUF_INIT;
2292
2293         if (!files) {
2294                 assert(insertions == 0 && deletions == 0);
2295                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2296                                  NULL, 0, 0);
2297                 return;
2298         }
2299
2300         strbuf_addf(&sb,
2301                     (files == 1) ? " %d file changed" : " %d files changed",
2302                     files);
2303
2304         /*
2305          * For binary diff, the caller may want to print "x files
2306          * changed" with insertions == 0 && deletions == 0.
2307          *
2308          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2309          * is probably less confusing (i.e skip over "2 files changed
2310          * but nothing about added/removed lines? Is this a bug in Git?").
2311          */
2312         if (insertions || deletions == 0) {
2313                 strbuf_addf(&sb,
2314                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2315                             insertions);
2316         }
2317
2318         if (deletions || insertions == 0) {
2319                 strbuf_addf(&sb,
2320                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2321                             deletions);
2322         }
2323         strbuf_addch(&sb, '\n');
2324         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2325                          sb.buf, sb.len, 0);
2326         strbuf_release(&sb);
2327 }
2328
2329 void print_stat_summary(FILE *fp, int files,
2330                         int insertions, int deletions)
2331 {
2332         struct diff_options o;
2333         memset(&o, 0, sizeof(o));
2334         o.file = fp;
2335
2336         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2337 }
2338
2339 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2340 {
2341         int i, len, add, del, adds = 0, dels = 0;
2342         uintmax_t max_change = 0, max_len = 0;
2343         int total_files = data->nr, count;
2344         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2345         const char *reset, *add_c, *del_c;
2346         int extra_shown = 0;
2347         const char *line_prefix = diff_line_prefix(options);
2348         struct strbuf out = STRBUF_INIT;
2349
2350         if (data->nr == 0)
2351                 return;
2352
2353         count = options->stat_count ? options->stat_count : data->nr;
2354
2355         reset = diff_get_color_opt(options, DIFF_RESET);
2356         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2357         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2358
2359         /*
2360          * Find the longest filename and max number of changes
2361          */
2362         for (i = 0; (i < count) && (i < data->nr); i++) {
2363                 struct diffstat_file *file = data->files[i];
2364                 uintmax_t change = file->added + file->deleted;
2365
2366                 if (!file->is_interesting && (change == 0)) {
2367                         count++; /* not shown == room for one more */
2368                         continue;
2369                 }
2370                 fill_print_name(file);
2371                 len = strlen(file->print_name);
2372                 if (max_len < len)
2373                         max_len = len;
2374
2375                 if (file->is_unmerged) {
2376                         /* "Unmerged" is 8 characters */
2377                         bin_width = bin_width < 8 ? 8 : bin_width;
2378                         continue;
2379                 }
2380                 if (file->is_binary) {
2381                         /* "Bin XXX -> YYY bytes" */
2382                         int w = 14 + decimal_width(file->added)
2383                                 + decimal_width(file->deleted);
2384                         bin_width = bin_width < w ? w : bin_width;
2385                         /* Display change counts aligned with "Bin" */
2386                         number_width = 3;
2387                         continue;
2388                 }
2389
2390                 if (max_change < change)
2391                         max_change = change;
2392         }
2393         count = i; /* where we can stop scanning in data->files[] */
2394
2395         /*
2396          * We have width = stat_width or term_columns() columns total.
2397          * We want a maximum of min(max_len, stat_name_width) for the name part.
2398          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2399          * We also need 1 for " " and 4 + decimal_width(max_change)
2400          * for " | NNNN " and one the empty column at the end, altogether
2401          * 6 + decimal_width(max_change).
2402          *
2403          * If there's not enough space, we will use the smaller of
2404          * stat_name_width (if set) and 5/8*width for the filename,
2405          * and the rest for constant elements + graph part, but no more
2406          * than stat_graph_width for the graph part.
2407          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2408          * for the standard terminal size).
2409          *
2410          * In other words: stat_width limits the maximum width, and
2411          * stat_name_width fixes the maximum width of the filename,
2412          * and is also used to divide available columns if there
2413          * aren't enough.
2414          *
2415          * Binary files are displayed with "Bin XXX -> YYY bytes"
2416          * instead of the change count and graph. This part is treated
2417          * similarly to the graph part, except that it is not
2418          * "scaled". If total width is too small to accommodate the
2419          * guaranteed minimum width of the filename part and the
2420          * separators and this message, this message will "overflow"
2421          * making the line longer than the maximum width.
2422          */
2423
2424         if (options->stat_width == -1)
2425                 width = term_columns() - strlen(line_prefix);
2426         else
2427                 width = options->stat_width ? options->stat_width : 80;
2428         number_width = decimal_width(max_change) > number_width ?
2429                 decimal_width(max_change) : number_width;
2430
2431         if (options->stat_graph_width == -1)
2432                 options->stat_graph_width = diff_stat_graph_width;
2433
2434         /*
2435          * Guarantee 3/8*16==6 for the graph part
2436          * and 5/8*16==10 for the filename part
2437          */
2438         if (width < 16 + 6 + number_width)
2439                 width = 16 + 6 + number_width;
2440
2441         /*
2442          * First assign sizes that are wanted, ignoring available width.
2443          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2444          * starting from "XXX" should fit in graph_width.
2445          */
2446         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2447         if (options->stat_graph_width &&
2448             options->stat_graph_width < graph_width)
2449                 graph_width = options->stat_graph_width;
2450
2451         name_width = (options->stat_name_width > 0 &&
2452                       options->stat_name_width < max_len) ?
2453                 options->stat_name_width : max_len;
2454
2455         /*
2456          * Adjust adjustable widths not to exceed maximum width
2457          */
2458         if (name_width + number_width + 6 + graph_width > width) {
2459                 if (graph_width > width * 3/8 - number_width - 6) {
2460                         graph_width = width * 3/8 - number_width - 6;
2461                         if (graph_width < 6)
2462                                 graph_width = 6;
2463                 }
2464
2465                 if (options->stat_graph_width &&
2466                     graph_width > options->stat_graph_width)
2467                         graph_width = options->stat_graph_width;
2468                 if (name_width > width - number_width - 6 - graph_width)
2469                         name_width = width - number_width - 6 - graph_width;
2470                 else
2471                         graph_width = width - number_width - 6 - name_width;
2472         }
2473
2474         /*
2475          * From here name_width is the width of the name area,
2476          * and graph_width is the width of the graph area.
2477          * max_change is used to scale graph properly.
2478          */
2479         for (i = 0; i < count; i++) {
2480                 const char *prefix = "";
2481                 struct diffstat_file *file = data->files[i];
2482                 char *name = file->print_name;
2483                 uintmax_t added = file->added;
2484                 uintmax_t deleted = file->deleted;
2485                 int name_len;
2486
2487                 if (!file->is_interesting && (added + deleted == 0))
2488                         continue;
2489
2490                 /*
2491                  * "scale" the filename
2492                  */
2493                 len = name_width;
2494                 name_len = strlen(name);
2495                 if (name_width < name_len) {
2496                         char *slash;
2497                         prefix = "...";
2498                         len -= 3;
2499                         name += name_len - len;
2500                         slash = strchr(name, '/');
2501                         if (slash)
2502                                 name = slash;
2503                 }
2504
2505                 if (file->is_binary) {
2506                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2507                         strbuf_addf(&out, " %*s", number_width, "Bin");
2508                         if (!added && !deleted) {
2509                                 strbuf_addch(&out, '\n');
2510                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2511                                                  out.buf, out.len, 0);
2512                                 strbuf_reset(&out);
2513                                 continue;
2514                         }
2515                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2516                                 del_c, deleted, reset);
2517                         strbuf_addstr(&out, " -> ");
2518                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2519                                 add_c, added, reset);
2520                         strbuf_addstr(&out, " bytes\n");
2521                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2522                                          out.buf, out.len, 0);
2523                         strbuf_reset(&out);
2524                         continue;
2525                 }
2526                 else if (file->is_unmerged) {
2527                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2528                         strbuf_addstr(&out, " Unmerged\n");
2529                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2530                                          out.buf, out.len, 0);
2531                         strbuf_reset(&out);
2532                         continue;
2533                 }
2534
2535                 /*
2536                  * scale the add/delete
2537                  */
2538                 add = added;
2539                 del = deleted;
2540
2541                 if (graph_width <= max_change) {
2542                         int total = scale_linear(add + del, graph_width, max_change);
2543                         if (total < 2 && add && del)
2544                                 /* width >= 2 due to the sanity check */
2545                                 total = 2;
2546                         if (add < del) {
2547                                 add = scale_linear(add, graph_width, max_change);
2548                                 del = total - add;
2549                         } else {
2550                                 del = scale_linear(del, graph_width, max_change);
2551                                 add = total - del;
2552                         }
2553                 }
2554                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2555                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2556                         number_width, added + deleted,
2557                         added + deleted ? " " : "");
2558                 show_graph(&out, '+', add, add_c, reset);
2559                 show_graph(&out, '-', del, del_c, reset);
2560                 strbuf_addch(&out, '\n');
2561                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2562                                  out.buf, out.len, 0);
2563                 strbuf_reset(&out);
2564         }
2565
2566         for (i = 0; i < data->nr; i++) {
2567                 struct diffstat_file *file = data->files[i];
2568                 uintmax_t added = file->added;
2569                 uintmax_t deleted = file->deleted;
2570
2571                 if (file->is_unmerged ||
2572                     (!file->is_interesting && (added + deleted == 0))) {
2573                         total_files--;
2574                         continue;
2575                 }
2576
2577                 if (!file->is_binary) {
2578                         adds += added;
2579                         dels += deleted;
2580                 }
2581                 if (i < count)
2582                         continue;
2583                 if (!extra_shown)
2584                         emit_diff_symbol(options,
2585                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2586                                          NULL, 0, 0);
2587                 extra_shown = 1;
2588         }
2589
2590         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2591         strbuf_release(&out);
2592 }
2593
2594 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2595 {
2596         int i, adds = 0, dels = 0, total_files = data->nr;
2597
2598         if (data->nr == 0)
2599                 return;
2600
2601         for (i = 0; i < data->nr; i++) {
2602                 int added = data->files[i]->added;
2603                 int deleted = data->files[i]->deleted;
2604
2605                 if (data->files[i]->is_unmerged ||
2606                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2607                         total_files--;
2608                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2609                         adds += added;
2610                         dels += deleted;
2611                 }
2612         }
2613         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2614 }
2615
2616 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2617 {
2618         int i;
2619
2620         if (data->nr == 0)
2621                 return;
2622
2623         for (i = 0; i < data->nr; i++) {
2624                 struct diffstat_file *file = data->files[i];
2625
2626                 fprintf(options->file, "%s", diff_line_prefix(options));
2627
2628                 if (file->is_binary)
2629                         fprintf(options->file, "-\t-\t");
2630                 else
2631                         fprintf(options->file,
2632                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2633                                 file->added, file->deleted);
2634                 if (options->line_termination) {
2635                         fill_print_name(file);
2636                         if (!file->is_renamed)
2637                                 write_name_quoted(file->name, options->file,
2638                                                   options->line_termination);
2639                         else {
2640                                 fputs(file->print_name, options->file);
2641                                 putc(options->line_termination, options->file);
2642                         }
2643                 } else {
2644                         if (file->is_renamed) {
2645                                 putc('\0', options->file);
2646                                 write_name_quoted(file->from_name, options->file, '\0');
2647                         }
2648                         write_name_quoted(file->name, options->file, '\0');
2649                 }
2650         }
2651 }
2652
2653 struct dirstat_file {
2654         const char *name;
2655         unsigned long changed;
2656 };
2657
2658 struct dirstat_dir {
2659         struct dirstat_file *files;
2660         int alloc, nr, permille, cumulative;
2661 };
2662
2663 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2664                 unsigned long changed, const char *base, int baselen)
2665 {
2666         unsigned long this_dir = 0;
2667         unsigned int sources = 0;
2668         const char *line_prefix = diff_line_prefix(opt);
2669
2670         while (dir->nr) {
2671                 struct dirstat_file *f = dir->files;
2672                 int namelen = strlen(f->name);
2673                 unsigned long this;
2674                 char *slash;
2675
2676                 if (namelen < baselen)
2677                         break;
2678                 if (memcmp(f->name, base, baselen))
2679                         break;
2680                 slash = strchr(f->name + baselen, '/');
2681                 if (slash) {
2682                         int newbaselen = slash + 1 - f->name;
2683                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2684                         sources++;
2685                 } else {
2686                         this = f->changed;
2687                         dir->files++;
2688                         dir->nr--;
2689                         sources += 2;
2690                 }
2691                 this_dir += this;
2692         }
2693
2694         /*
2695          * We don't report dirstat's for
2696          *  - the top level
2697          *  - or cases where everything came from a single directory
2698          *    under this directory (sources == 1).
2699          */
2700         if (baselen && sources != 1) {
2701                 if (this_dir) {
2702                         int permille = this_dir * 1000 / changed;
2703                         if (permille >= dir->permille) {
2704                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2705                                         permille / 10, permille % 10, baselen, base);
2706                                 if (!dir->cumulative)
2707                                         return 0;
2708                         }
2709                 }
2710         }
2711         return this_dir;
2712 }
2713
2714 static int dirstat_compare(const void *_a, const void *_b)
2715 {
2716         const struct dirstat_file *a = _a;
2717         const struct dirstat_file *b = _b;
2718         return strcmp(a->name, b->name);
2719 }
2720
2721 static void show_dirstat(struct diff_options *options)
2722 {
2723         int i;
2724         unsigned long changed;
2725         struct dirstat_dir dir;
2726         struct diff_queue_struct *q = &diff_queued_diff;
2727
2728         dir.files = NULL;
2729         dir.alloc = 0;
2730         dir.nr = 0;
2731         dir.permille = options->dirstat_permille;
2732         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2733
2734         changed = 0;
2735         for (i = 0; i < q->nr; i++) {
2736                 struct diff_filepair *p = q->queue[i];
2737                 const char *name;
2738                 unsigned long copied, added, damage;
2739                 int content_changed;
2740
2741                 name = p->two->path ? p->two->path : p->one->path;
2742
2743                 if (p->one->oid_valid && p->two->oid_valid)
2744                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2745                 else
2746                         content_changed = 1;
2747
2748                 if (!content_changed) {
2749                         /*
2750                          * The SHA1 has not changed, so pre-/post-content is
2751                          * identical. We can therefore skip looking at the
2752                          * file contents altogether.
2753                          */
2754                         damage = 0;
2755                         goto found_damage;
2756                 }
2757
2758                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2759                         /*
2760                          * In --dirstat-by-file mode, we don't really need to
2761                          * look at the actual file contents at all.
2762                          * The fact that the SHA1 changed is enough for us to
2763                          * add this file to the list of results
2764                          * (with each file contributing equal damage).
2765                          */
2766                         damage = 1;
2767                         goto found_damage;
2768                 }
2769
2770                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2771                         diff_populate_filespec(p->one, 0);
2772                         diff_populate_filespec(p->two, 0);
2773                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2774                                                &copied, &added);
2775                         diff_free_filespec_data(p->one);
2776                         diff_free_filespec_data(p->two);
2777                 } else if (DIFF_FILE_VALID(p->one)) {
2778                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2779                         copied = added = 0;
2780                         diff_free_filespec_data(p->one);
2781                 } else if (DIFF_FILE_VALID(p->two)) {
2782                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2783                         copied = 0;
2784                         added = p->two->size;
2785                         diff_free_filespec_data(p->two);
2786                 } else
2787                         continue;
2788
2789                 /*
2790                  * Original minus copied is the removed material,
2791                  * added is the new material.  They are both damages
2792                  * made to the preimage.
2793                  * If the resulting damage is zero, we know that
2794                  * diffcore_count_changes() considers the two entries to
2795                  * be identical, but since content_changed is true, we
2796                  * know that there must have been _some_ kind of change,
2797                  * so we force all entries to have damage > 0.
2798                  */
2799                 damage = (p->one->size - copied) + added;
2800                 if (!damage)
2801                         damage = 1;
2802
2803 found_damage:
2804                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2805                 dir.files[dir.nr].name = name;
2806                 dir.files[dir.nr].changed = damage;
2807                 changed += damage;
2808                 dir.nr++;
2809         }
2810
2811         /* This can happen even with many files, if everything was renames */
2812         if (!changed)
2813                 return;
2814
2815         /* Show all directories with more than x% of the changes */
2816         QSORT(dir.files, dir.nr, dirstat_compare);
2817         gather_dirstat(options, &dir, changed, "", 0);
2818 }
2819
2820 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2821 {
2822         int i;
2823         unsigned long changed;
2824         struct dirstat_dir dir;
2825
2826         if (data->nr == 0)
2827                 return;
2828
2829         dir.files = NULL;
2830         dir.alloc = 0;
2831         dir.nr = 0;
2832         dir.permille = options->dirstat_permille;
2833         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2834
2835         changed = 0;
2836         for (i = 0; i < data->nr; i++) {
2837                 struct diffstat_file *file = data->files[i];
2838                 unsigned long damage = file->added + file->deleted;
2839                 if (file->is_binary)
2840                         /*
2841                          * binary files counts bytes, not lines. Must find some
2842                          * way to normalize binary bytes vs. textual lines.
2843                          * The following heuristic assumes that there are 64
2844                          * bytes per "line".
2845                          * This is stupid and ugly, but very cheap...
2846                          */
2847                         damage = DIV_ROUND_UP(damage, 64);
2848                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2849                 dir.files[dir.nr].name = file->name;
2850                 dir.files[dir.nr].changed = damage;
2851                 changed += damage;
2852                 dir.nr++;
2853         }
2854
2855         /* This can happen even with many files, if everything was renames */
2856         if (!changed)
2857                 return;
2858
2859         /* Show all directories with more than x% of the changes */
2860         QSORT(dir.files, dir.nr, dirstat_compare);
2861         gather_dirstat(options, &dir, changed, "", 0);
2862 }
2863
2864 static void free_diffstat_info(struct diffstat_t *diffstat)
2865 {
2866         int i;
2867         for (i = 0; i < diffstat->nr; i++) {
2868                 struct diffstat_file *f = diffstat->files[i];
2869                 if (f->name != f->print_name)
2870                         free(f->print_name);
2871                 free(f->name);
2872                 free(f->from_name);
2873                 free(f);
2874         }
2875         free(diffstat->files);
2876 }
2877
2878 struct checkdiff_t {
2879         const char *filename;
2880         int lineno;
2881         int conflict_marker_size;
2882         struct diff_options *o;
2883         unsigned ws_rule;
2884         unsigned status;
2885 };
2886
2887 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2888 {
2889         char firstchar;
2890         int cnt;
2891
2892         if (len < marker_size + 1)
2893                 return 0;
2894         firstchar = line[0];
2895         switch (firstchar) {
2896         case '=': case '>': case '<': case '|':
2897                 break;
2898         default:
2899                 return 0;
2900         }
2901         for (cnt = 1; cnt < marker_size; cnt++)
2902                 if (line[cnt] != firstchar)
2903                         return 0;
2904         /* line[1] thru line[marker_size-1] are same as firstchar */
2905         if (len < marker_size + 1 || !isspace(line[marker_size]))
2906                 return 0;
2907         return 1;
2908 }
2909
2910 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2911 {
2912         struct checkdiff_t *data = priv;
2913         int marker_size = data->conflict_marker_size;
2914         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2915         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2916         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2917         char *err;
2918         const char *line_prefix;
2919
2920         assert(data->o);
2921         line_prefix = diff_line_prefix(data->o);
2922
2923         if (line[0] == '+') {
2924                 unsigned bad;
2925                 data->lineno++;
2926                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2927                         data->status |= 1;
2928                         fprintf(data->o->file,
2929                                 "%s%s:%d: leftover conflict marker\n",
2930                                 line_prefix, data->filename, data->lineno);
2931                 }
2932                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2933                 if (!bad)
2934                         return;
2935                 data->status |= bad;
2936                 err = whitespace_error_string(bad);
2937                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2938                         line_prefix, data->filename, data->lineno, err);
2939                 free(err);
2940                 emit_line(data->o, set, reset, line, 1);
2941                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2942                               data->o->file, set, reset, ws);
2943         } else if (line[0] == ' ') {
2944                 data->lineno++;
2945         } else if (line[0] == '@') {
2946                 char *plus = strchr(line, '+');
2947                 if (plus)
2948                         data->lineno = strtol(plus, NULL, 10) - 1;
2949                 else
2950                         die("invalid diff");
2951         }
2952 }
2953
2954 static unsigned char *deflate_it(char *data,
2955                                  unsigned long size,
2956                                  unsigned long *result_size)
2957 {
2958         int bound;
2959         unsigned char *deflated;
2960         git_zstream stream;
2961
2962         git_deflate_init(&stream, zlib_compression_level);
2963         bound = git_deflate_bound(&stream, size);
2964         deflated = xmalloc(bound);
2965         stream.next_out = deflated;
2966         stream.avail_out = bound;
2967
2968         stream.next_in = (unsigned char *)data;
2969         stream.avail_in = size;
2970         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2971                 ; /* nothing */
2972         git_deflate_end(&stream);
2973         *result_size = stream.total_out;
2974         return deflated;
2975 }
2976
2977 static void emit_binary_diff_body(struct diff_options *o,
2978                                   mmfile_t *one, mmfile_t *two)
2979 {
2980         void *cp;
2981         void *delta;
2982         void *deflated;
2983         void *data;
2984         unsigned long orig_size;
2985         unsigned long delta_size;
2986         unsigned long deflate_size;
2987         unsigned long data_size;
2988
2989         /* We could do deflated delta, or we could do just deflated two,
2990          * whichever is smaller.
2991          */
2992         delta = NULL;
2993         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2994         if (one->size && two->size) {
2995                 delta = diff_delta(one->ptr, one->size,
2996                                    two->ptr, two->size,
2997                                    &delta_size, deflate_size);
2998                 if (delta) {
2999                         void *to_free = delta;
3000                         orig_size = delta_size;
3001                         delta = deflate_it(delta, delta_size, &delta_size);
3002                         free(to_free);
3003                 }
3004         }
3005
3006         if (delta && delta_size < deflate_size) {
3007                 char *s = xstrfmt("%lu", orig_size);
3008                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3009                                  s, strlen(s), 0);
3010                 free(s);
3011                 free(deflated);
3012                 data = delta;
3013                 data_size = delta_size;
3014         } else {
3015                 char *s = xstrfmt("%lu", two->size);
3016                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3017                                  s, strlen(s), 0);
3018                 free(s);
3019                 free(delta);
3020                 data = deflated;
3021                 data_size = deflate_size;
3022         }
3023
3024         /* emit data encoded in base85 */
3025         cp = data;
3026         while (data_size) {
3027                 int len;
3028                 int bytes = (52 < data_size) ? 52 : data_size;
3029                 char line[71];
3030                 data_size -= bytes;
3031                 if (bytes <= 26)
3032                         line[0] = bytes + 'A' - 1;
3033                 else
3034                         line[0] = bytes - 26 + 'a' - 1;
3035                 encode_85(line + 1, cp, bytes);
3036                 cp = (char *) cp + bytes;
3037
3038                 len = strlen(line);
3039                 line[len++] = '\n';
3040                 line[len] = '\0';
3041
3042                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3043                                  line, len, 0);
3044         }
3045         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3046         free(data);
3047 }
3048
3049 static void emit_binary_diff(struct diff_options *o,
3050                              mmfile_t *one, mmfile_t *two)
3051 {
3052         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3053         emit_binary_diff_body(o, one, two);
3054         emit_binary_diff_body(o, two, one);
3055 }
3056
3057 int diff_filespec_is_binary(struct diff_filespec *one)
3058 {
3059         if (one->is_binary == -1) {
3060                 diff_filespec_load_driver(one);
3061                 if (one->driver->binary != -1)
3062                         one->is_binary = one->driver->binary;
3063                 else {
3064                         if (!one->data && DIFF_FILE_VALID(one))
3065                                 diff_populate_filespec(one, CHECK_BINARY);
3066                         if (one->is_binary == -1 && one->data)
3067                                 one->is_binary = buffer_is_binary(one->data,
3068                                                 one->size);
3069                         if (one->is_binary == -1)
3070                                 one->is_binary = 0;
3071                 }
3072         }
3073         return one->is_binary;
3074 }
3075
3076 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3077 {
3078         diff_filespec_load_driver(one);
3079         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3080 }
3081
3082 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3083 {
3084         if (!options->a_prefix)
3085                 options->a_prefix = a;
3086         if (!options->b_prefix)
3087                 options->b_prefix = b;
3088 }
3089
3090 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3091 {
3092         if (!DIFF_FILE_VALID(one))
3093                 return NULL;
3094
3095         diff_filespec_load_driver(one);
3096         return userdiff_get_textconv(one->driver);
3097 }
3098
3099 static void builtin_diff(const char *name_a,
3100                          const char *name_b,
3101                          struct diff_filespec *one,
3102                          struct diff_filespec *two,
3103                          const char *xfrm_msg,
3104                          int must_show_header,
3105                          struct diff_options *o,
3106                          int complete_rewrite)
3107 {
3108         mmfile_t mf1, mf2;
3109         const char *lbl[2];
3110         char *a_one, *b_two;
3111         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3112         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3113         const char *a_prefix, *b_prefix;
3114         struct userdiff_driver *textconv_one = NULL;
3115         struct userdiff_driver *textconv_two = NULL;
3116         struct strbuf header = STRBUF_INIT;
3117         const char *line_prefix = diff_line_prefix(o);
3118
3119         diff_set_mnemonic_prefix(o, "a/", "b/");
3120         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
3121                 a_prefix = o->b_prefix;
3122                 b_prefix = o->a_prefix;
3123         } else {
3124                 a_prefix = o->a_prefix;
3125                 b_prefix = o->b_prefix;
3126         }
3127
3128         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3129             (!one->mode || S_ISGITLINK(one->mode)) &&
3130             (!two->mode || S_ISGITLINK(two->mode))) {
3131                 show_submodule_summary(o, one->path ? one->path : two->path,
3132                                 &one->oid, &two->oid,
3133                                 two->dirty_submodule);
3134                 return;
3135         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3136                    (!one->mode || S_ISGITLINK(one->mode)) &&
3137                    (!two->mode || S_ISGITLINK(two->mode))) {
3138                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3139                                 &one->oid, &two->oid,
3140                                 two->dirty_submodule);
3141                 return;
3142         }
3143
3144         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
3145                 textconv_one = get_textconv(one);
3146                 textconv_two = get_textconv(two);
3147         }
3148
3149         /* Never use a non-valid filename anywhere if at all possible */
3150         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3151         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3152
3153         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3154         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3155         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3156         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3157         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3158         if (lbl[0][0] == '/') {
3159                 /* /dev/null */
3160                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3161                 if (xfrm_msg)
3162                         strbuf_addstr(&header, xfrm_msg);
3163                 must_show_header = 1;
3164         }
3165         else if (lbl[1][0] == '/') {
3166                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3167                 if (xfrm_msg)
3168                         strbuf_addstr(&header, xfrm_msg);
3169                 must_show_header = 1;
3170         }
3171         else {
3172                 if (one->mode != two->mode) {
3173                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3174                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3175                         must_show_header = 1;
3176                 }
3177                 if (xfrm_msg)
3178                         strbuf_addstr(&header, xfrm_msg);
3179
3180                 /*
3181                  * we do not run diff between different kind
3182                  * of objects.
3183                  */
3184                 if ((one->mode ^ two->mode) & S_IFMT)
3185                         goto free_ab_and_return;
3186                 if (complete_rewrite &&
3187                     (textconv_one || !diff_filespec_is_binary(one)) &&
3188                     (textconv_two || !diff_filespec_is_binary(two))) {
3189                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3190                                          header.buf, header.len, 0);
3191                         strbuf_reset(&header);
3192                         emit_rewrite_diff(name_a, name_b, one, two,
3193                                                 textconv_one, textconv_two, o);
3194                         o->found_changes = 1;
3195                         goto free_ab_and_return;
3196                 }
3197         }
3198
3199         if (o->irreversible_delete && lbl[1][0] == '/') {
3200                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3201                                  header.len, 0);
3202                 strbuf_reset(&header);
3203                 goto free_ab_and_return;
3204         } else if (!DIFF_OPT_TST(o, TEXT) &&
3205             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3206               (!textconv_two && diff_filespec_is_binary(two)) )) {
3207                 struct strbuf sb = STRBUF_INIT;
3208                 if (!one->data && !two->data &&
3209                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3210                     !DIFF_OPT_TST(o, BINARY)) {
3211                         if (!oidcmp(&one->oid, &two->oid)) {
3212                                 if (must_show_header)
3213                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3214                                                          header.buf, header.len,
3215                                                          0);
3216                                 goto free_ab_and_return;
3217                         }
3218                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3219                                          header.buf, header.len, 0);
3220                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3221                                     diff_line_prefix(o), lbl[0], lbl[1]);
3222                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3223                                          sb.buf, sb.len, 0);
3224                         strbuf_release(&sb);
3225                         goto free_ab_and_return;
3226                 }
3227                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3228                         die("unable to read files to diff");
3229                 /* Quite common confusing case */
3230                 if (mf1.size == mf2.size &&
3231                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3232                         if (must_show_header)
3233                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3234                                                  header.buf, header.len, 0);
3235                         goto free_ab_and_return;
3236                 }
3237                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3238                 strbuf_reset(&header);
3239                 if (DIFF_OPT_TST(o, BINARY))
3240                         emit_binary_diff(o, &mf1, &mf2);
3241                 else {
3242                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3243                                     diff_line_prefix(o), lbl[0], lbl[1]);
3244                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3245                                          sb.buf, sb.len, 0);
3246                         strbuf_release(&sb);
3247                 }
3248                 o->found_changes = 1;
3249         } else {
3250                 /* Crazy xdl interfaces.. */
3251                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3252                 const char *v;
3253                 xpparam_t xpp;
3254                 xdemitconf_t xecfg;
3255                 struct emit_callback ecbdata;
3256                 const struct userdiff_funcname *pe;
3257
3258                 if (must_show_header) {
3259                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3260                                          header.buf, header.len, 0);
3261                         strbuf_reset(&header);
3262                 }
3263
3264                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3265                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3266
3267                 pe = diff_funcname_pattern(one);
3268                 if (!pe)
3269                         pe = diff_funcname_pattern(two);
3270
3271                 memset(&xpp, 0, sizeof(xpp));
3272                 memset(&xecfg, 0, sizeof(xecfg));
3273                 memset(&ecbdata, 0, sizeof(ecbdata));
3274                 ecbdata.label_path = lbl;
3275                 ecbdata.color_diff = want_color(o->use_color);
3276                 ecbdata.ws_rule = whitespace_rule(name_b);
3277                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3278                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3279                 ecbdata.opt = o;
3280                 ecbdata.header = header.len ? &header : NULL;
3281                 xpp.flags = o->xdl_opts;
3282                 xecfg.ctxlen = o->context;
3283                 xecfg.interhunkctxlen = o->interhunkcontext;
3284                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3285                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
3286                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3287                 if (pe)
3288                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3289                 if (!diffopts)
3290                         ;
3291                 else if (skip_prefix(diffopts, "--unified=", &v))
3292                         xecfg.ctxlen = strtoul(v, NULL, 10);
3293                 else if (skip_prefix(diffopts, "-u", &v))
3294                         xecfg.ctxlen = strtoul(v, NULL, 10);
3295                 if (o->word_diff)
3296                         init_diff_words_data(&ecbdata, o, one, two);
3297                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3298                                   &xpp, &xecfg))
3299                         die("unable to generate diff for %s", one->path);
3300                 if (o->word_diff)
3301                         free_diff_words_data(&ecbdata);
3302                 if (textconv_one)
3303                         free(mf1.ptr);
3304                 if (textconv_two)
3305                         free(mf2.ptr);
3306                 xdiff_clear_find_func(&xecfg);
3307         }
3308
3309  free_ab_and_return:
3310         strbuf_release(&header);
3311         diff_free_filespec_data(one);
3312         diff_free_filespec_data(two);
3313         free(a_one);
3314         free(b_two);
3315         return;
3316 }
3317
3318 static void builtin_diffstat(const char *name_a, const char *name_b,
3319                              struct diff_filespec *one,
3320                              struct diff_filespec *two,
3321                              struct diffstat_t *diffstat,
3322                              struct diff_options *o,
3323                              struct diff_filepair *p)
3324 {
3325         mmfile_t mf1, mf2;
3326         struct diffstat_file *data;
3327         int same_contents;
3328         int complete_rewrite = 0;
3329
3330         if (!DIFF_PAIR_UNMERGED(p)) {
3331                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3332                         complete_rewrite = 1;
3333         }
3334
3335         data = diffstat_add(diffstat, name_a, name_b);
3336         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3337
3338         if (!one || !two) {
3339                 data->is_unmerged = 1;
3340                 return;
3341         }
3342
3343         same_contents = !oidcmp(&one->oid, &two->oid);
3344
3345         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3346                 data->is_binary = 1;
3347                 if (same_contents) {
3348                         data->added = 0;
3349                         data->deleted = 0;
3350                 } else {
3351                         data->added = diff_filespec_size(two);
3352                         data->deleted = diff_filespec_size(one);
3353                 }
3354         }
3355
3356         else if (complete_rewrite) {
3357                 diff_populate_filespec(one, 0);
3358                 diff_populate_filespec(two, 0);
3359                 data->deleted = count_lines(one->data, one->size);
3360                 data->added = count_lines(two->data, two->size);
3361         }
3362
3363         else if (!same_contents) {
3364                 /* Crazy xdl interfaces.. */
3365                 xpparam_t xpp;
3366                 xdemitconf_t xecfg;
3367
3368                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3369                         die("unable to read files to diff");
3370
3371                 memset(&xpp, 0, sizeof(xpp));
3372                 memset(&xecfg, 0, sizeof(xecfg));
3373                 xpp.flags = o->xdl_opts;
3374                 xecfg.ctxlen = o->context;
3375                 xecfg.interhunkctxlen = o->interhunkcontext;
3376                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3377                                   &xpp, &xecfg))
3378                         die("unable to generate diffstat for %s", one->path);
3379         }
3380
3381         diff_free_filespec_data(one);
3382         diff_free_filespec_data(two);
3383 }
3384
3385 static void builtin_checkdiff(const char *name_a, const char *name_b,
3386                               const char *attr_path,
3387                               struct diff_filespec *one,
3388                               struct diff_filespec *two,
3389                               struct diff_options *o)
3390 {
3391         mmfile_t mf1, mf2;
3392         struct checkdiff_t data;
3393
3394         if (!two)
3395                 return;
3396
3397         memset(&data, 0, sizeof(data));
3398         data.filename = name_b ? name_b : name_a;
3399         data.lineno = 0;
3400         data.o = o;
3401         data.ws_rule = whitespace_rule(attr_path);
3402         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3403
3404         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3405                 die("unable to read files to diff");
3406
3407         /*
3408          * All the other codepaths check both sides, but not checking
3409          * the "old" side here is deliberate.  We are checking the newly
3410          * introduced changes, and as long as the "new" side is text, we
3411          * can and should check what it introduces.
3412          */
3413         if (diff_filespec_is_binary(two))
3414                 goto free_and_return;
3415         else {
3416                 /* Crazy xdl interfaces.. */
3417                 xpparam_t xpp;
3418                 xdemitconf_t xecfg;
3419
3420                 memset(&xpp, 0, sizeof(xpp));
3421                 memset(&xecfg, 0, sizeof(xecfg));
3422                 xecfg.ctxlen = 1; /* at least one context line */
3423                 xpp.flags = 0;
3424                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3425                                   &xpp, &xecfg))
3426                         die("unable to generate checkdiff for %s", one->path);
3427
3428                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3429                         struct emit_callback ecbdata;
3430                         int blank_at_eof;
3431
3432                         ecbdata.ws_rule = data.ws_rule;
3433                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3434                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3435
3436                         if (blank_at_eof) {
3437                                 static char *err;
3438                                 if (!err)
3439                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3440                                 fprintf(o->file, "%s:%d: %s.\n",
3441                                         data.filename, blank_at_eof, err);
3442                                 data.status = 1; /* report errors */
3443                         }
3444                 }
3445         }
3446  free_and_return:
3447         diff_free_filespec_data(one);
3448         diff_free_filespec_data(two);
3449         if (data.status)
3450                 DIFF_OPT_SET(o, CHECK_FAILED);
3451 }
3452
3453 struct diff_filespec *alloc_filespec(const char *path)
3454 {
3455         struct diff_filespec *spec;
3456
3457         FLEXPTR_ALLOC_STR(spec, path, path);
3458         spec->count = 1;
3459         spec->is_binary = -1;
3460         return spec;
3461 }
3462
3463 void free_filespec(struct diff_filespec *spec)
3464 {
3465         if (!--spec->count) {
3466                 diff_free_filespec_data(spec);
3467                 free(spec);
3468         }
3469 }
3470
3471 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3472                    int oid_valid, unsigned short mode)
3473 {
3474         if (mode) {
3475                 spec->mode = canon_mode(mode);
3476                 oidcpy(&spec->oid, oid);
3477                 spec->oid_valid = oid_valid;
3478         }
3479 }
3480
3481 /*
3482  * Given a name and sha1 pair, if the index tells us the file in
3483  * the work tree has that object contents, return true, so that
3484  * prepare_temp_file() does not have to inflate and extract.
3485  */
3486 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3487 {
3488         const struct cache_entry *ce;
3489         struct stat st;
3490         int pos, len;
3491
3492         /*
3493          * We do not read the cache ourselves here, because the
3494          * benchmark with my previous version that always reads cache
3495          * shows that it makes things worse for diff-tree comparing
3496          * two linux-2.6 kernel trees in an already checked out work
3497          * tree.  This is because most diff-tree comparisons deal with
3498          * only a small number of files, while reading the cache is
3499          * expensive for a large project, and its cost outweighs the
3500          * savings we get by not inflating the object to a temporary
3501          * file.  Practically, this code only helps when we are used
3502          * by diff-cache --cached, which does read the cache before
3503          * calling us.
3504          */
3505         if (!active_cache)
3506                 return 0;
3507
3508         /* We want to avoid the working directory if our caller
3509          * doesn't need the data in a normal file, this system
3510          * is rather slow with its stat/open/mmap/close syscalls,
3511          * and the object is contained in a pack file.  The pack
3512          * is probably already open and will be faster to obtain
3513          * the data through than the working directory.  Loose
3514          * objects however would tend to be slower as they need
3515          * to be individually opened and inflated.
3516          */
3517         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
3518                 return 0;
3519
3520         /*
3521          * Similarly, if we'd have to convert the file contents anyway, that
3522          * makes the optimization not worthwhile.
3523          */
3524         if (!want_file && would_convert_to_git(&the_index, name))
3525                 return 0;
3526
3527         len = strlen(name);
3528         pos = cache_name_pos(name, len);
3529         if (pos < 0)
3530                 return 0;
3531         ce = active_cache[pos];
3532
3533         /*
3534          * This is not the sha1 we are looking for, or
3535          * unreusable because it is not a regular file.
3536          */
3537         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3538                 return 0;
3539
3540         /*
3541          * If ce is marked as "assume unchanged", there is no
3542          * guarantee that work tree matches what we are looking for.
3543          */
3544         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3545                 return 0;
3546
3547         /*
3548          * If ce matches the file in the work tree, we can reuse it.
3549          */
3550         if (ce_uptodate(ce) ||
3551             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3552                 return 1;
3553
3554         return 0;
3555 }
3556
3557 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3558 {
3559         struct strbuf buf = STRBUF_INIT;
3560         char *dirty = "";
3561
3562         /* Are we looking at the work tree? */
3563         if (s->dirty_submodule)
3564                 dirty = "-dirty";
3565
3566         strbuf_addf(&buf, "Subproject commit %s%s\n",
3567                     oid_to_hex(&s->oid), dirty);
3568         s->size = buf.len;
3569         if (size_only) {
3570                 s->data = NULL;
3571                 strbuf_release(&buf);
3572         } else {
3573                 s->data = strbuf_detach(&buf, NULL);
3574                 s->should_free = 1;
3575         }
3576         return 0;
3577 }
3578
3579 /*
3580  * While doing rename detection and pickaxe operation, we may need to
3581  * grab the data for the blob (or file) for our own in-core comparison.
3582  * diff_filespec has data and size fields for this purpose.
3583  */
3584 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3585 {
3586         int size_only = flags & CHECK_SIZE_ONLY;
3587         int err = 0;
3588         /*
3589          * demote FAIL to WARN to allow inspecting the situation
3590          * instead of refusing.
3591          */
3592         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
3593                                     ? SAFE_CRLF_WARN
3594                                     : safe_crlf);
3595
3596         if (!DIFF_FILE_VALID(s))
3597                 die("internal error: asking to populate invalid file.");
3598         if (S_ISDIR(s->mode))
3599                 return -1;
3600
3601         if (s->data)
3602                 return 0;
3603
3604         if (size_only && 0 < s->size)
3605                 return 0;
3606
3607         if (S_ISGITLINK(s->mode))
3608                 return diff_populate_gitlink(s, size_only);
3609
3610         if (!s->oid_valid ||
3611             reuse_worktree_file(s->path, &s->oid, 0)) {
3612                 struct strbuf buf = STRBUF_INIT;
3613                 struct stat st;
3614                 int fd;
3615
3616                 if (lstat(s->path, &st) < 0) {
3617                         if (errno == ENOENT) {
3618                         err_empty:
3619                                 err = -1;
3620                         empty:
3621                                 s->data = (char *)"";
3622                                 s->size = 0;
3623                                 return err;
3624                         }
3625                 }
3626                 s->size = xsize_t(st.st_size);
3627                 if (!s->size)
3628                         goto empty;
3629                 if (S_ISLNK(st.st_mode)) {
3630                         struct strbuf sb = STRBUF_INIT;
3631
3632                         if (strbuf_readlink(&sb, s->path, s->size))
3633                                 goto err_empty;
3634                         s->size = sb.len;
3635                         s->data = strbuf_detach(&sb, NULL);
3636                         s->should_free = 1;
3637                         return 0;
3638                 }
3639
3640                 /*
3641                  * Even if the caller would be happy with getting
3642                  * only the size, we cannot return early at this
3643                  * point if the path requires us to run the content
3644                  * conversion.
3645                  */
3646                 if (size_only && !would_convert_to_git(&the_index, s->path))
3647                         return 0;
3648
3649                 /*
3650                  * Note: this check uses xsize_t(st.st_size) that may
3651                  * not be the true size of the blob after it goes
3652                  * through convert_to_git().  This may not strictly be
3653                  * correct, but the whole point of big_file_threshold
3654                  * and is_binary check being that we want to avoid
3655                  * opening the file and inspecting the contents, this
3656                  * is probably fine.
3657                  */
3658                 if ((flags & CHECK_BINARY) &&
3659                     s->size > big_file_threshold && s->is_binary == -1) {
3660                         s->is_binary = 1;
3661                         return 0;
3662                 }
3663                 fd = open(s->path, O_RDONLY);
3664                 if (fd < 0)
3665                         goto err_empty;
3666                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3667                 close(fd);
3668                 s->should_munmap = 1;
3669
3670                 /*
3671                  * Convert from working tree format to canonical git format
3672                  */
3673                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, crlf_warn)) {
3674                         size_t size = 0;
3675                         munmap(s->data, s->size);
3676                         s->should_munmap = 0;
3677                         s->data = strbuf_detach(&buf, &size);
3678                         s->size = size;
3679                         s->should_free = 1;
3680                 }
3681         }
3682         else {
3683                 enum object_type type;
3684                 if (size_only || (flags & CHECK_BINARY)) {
3685                         type = sha1_object_info(s->oid.hash, &s->size);
3686                         if (type < 0)
3687                                 die("unable to read %s",
3688                                     oid_to_hex(&s->oid));
3689                         if (size_only)
3690                                 return 0;
3691                         if (s->size > big_file_threshold && s->is_binary == -1) {
3692                                 s->is_binary = 1;
3693                                 return 0;
3694                         }
3695                 }
3696                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
3697                 if (!s->data)
3698                         die("unable to read %s", oid_to_hex(&s->oid));
3699                 s->should_free = 1;
3700         }
3701         return 0;
3702 }
3703
3704 void diff_free_filespec_blob(struct diff_filespec *s)
3705 {
3706         if (s->should_free)
3707                 free(s->data);
3708         else if (s->should_munmap)
3709                 munmap(s->data, s->size);
3710
3711         if (s->should_free || s->should_munmap) {
3712                 s->should_free = s->should_munmap = 0;
3713                 s->data = NULL;
3714         }
3715 }
3716
3717 void diff_free_filespec_data(struct diff_filespec *s)
3718 {
3719         diff_free_filespec_blob(s);
3720         FREE_AND_NULL(s->cnt_data);
3721 }
3722
3723 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3724                            void *blob,
3725                            unsigned long size,
3726                            const struct object_id *oid,
3727                            int mode)
3728 {
3729         struct strbuf buf = STRBUF_INIT;
3730         struct strbuf template = STRBUF_INIT;
3731         char *path_dup = xstrdup(path);
3732         const char *base = basename(path_dup);
3733
3734         /* Generate "XXXXXX_basename.ext" */
3735         strbuf_addstr(&template, "XXXXXX_");
3736         strbuf_addstr(&template, base);
3737
3738         temp->tempfile = mks_tempfile_ts(template.buf, strlen(base) + 1);
3739         if (!temp->tempfile)
3740                 die_errno("unable to create temp-file");
3741         if (convert_to_working_tree(path,
3742                         (const char *)blob, (size_t)size, &buf)) {
3743                 blob = buf.buf;
3744                 size = buf.len;
3745         }
3746         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
3747             close_tempfile_gently(temp->tempfile))
3748                 die_errno("unable to write temp-file");
3749         temp->name = get_tempfile_path(temp->tempfile);
3750         oid_to_hex_r(temp->hex, oid);
3751         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3752         strbuf_release(&buf);
3753         strbuf_release(&template);
3754         free(path_dup);
3755 }
3756
3757 static struct diff_tempfile *prepare_temp_file(const char *name,
3758                 struct diff_filespec *one)
3759 {
3760         struct diff_tempfile *temp = claim_diff_tempfile();
3761
3762         if (!DIFF_FILE_VALID(one)) {
3763         not_a_valid_file:
3764                 /* A '-' entry produces this for file-2, and
3765                  * a '+' entry produces this for file-1.
3766                  */
3767                 temp->name = "/dev/null";
3768                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3769                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3770                 return temp;
3771         }
3772
3773         if (!S_ISGITLINK(one->mode) &&
3774             (!one->oid_valid ||
3775              reuse_worktree_file(name, &one->oid, 1))) {
3776                 struct stat st;
3777                 if (lstat(name, &st) < 0) {
3778                         if (errno == ENOENT)
3779                                 goto not_a_valid_file;
3780                         die_errno("stat(%s)", name);
3781                 }
3782                 if (S_ISLNK(st.st_mode)) {
3783                         struct strbuf sb = STRBUF_INIT;
3784                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3785                                 die_errno("readlink(%s)", name);
3786                         prep_temp_blob(name, temp, sb.buf, sb.len,
3787                                        (one->oid_valid ?
3788                                         &one->oid : &null_oid),
3789                                        (one->oid_valid ?
3790                                         one->mode : S_IFLNK));
3791                         strbuf_release(&sb);
3792                 }
3793                 else {
3794                         /* we can borrow from the file in the work tree */
3795                         temp->name = name;
3796                         if (!one->oid_valid)
3797                                 oid_to_hex_r(temp->hex, &null_oid);
3798                         else
3799                                 oid_to_hex_r(temp->hex, &one->oid);
3800                         /* Even though we may sometimes borrow the
3801                          * contents from the work tree, we always want
3802                          * one->mode.  mode is trustworthy even when
3803                          * !(one->oid_valid), as long as
3804                          * DIFF_FILE_VALID(one).
3805                          */
3806                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3807                 }
3808                 return temp;
3809         }
3810         else {
3811                 if (diff_populate_filespec(one, 0))
3812                         die("cannot read data blob for %s", one->path);
3813                 prep_temp_blob(name, temp, one->data, one->size,
3814                                &one->oid, one->mode);
3815         }
3816         return temp;
3817 }
3818
3819 static void add_external_diff_name(struct argv_array *argv,
3820                                    const char *name,
3821                                    struct diff_filespec *df)
3822 {
3823         struct diff_tempfile *temp = prepare_temp_file(name, df);
3824         argv_array_push(argv, temp->name);
3825         argv_array_push(argv, temp->hex);
3826         argv_array_push(argv, temp->mode);
3827 }
3828
3829 /* An external diff command takes:
3830  *
3831  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3832  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3833  *
3834  */
3835 static void run_external_diff(const char *pgm,
3836                               const char *name,
3837                               const char *other,
3838                               struct diff_filespec *one,
3839                               struct diff_filespec *two,
3840                               const char *xfrm_msg,
3841                               int complete_rewrite,
3842                               struct diff_options *o)
3843 {
3844         struct argv_array argv = ARGV_ARRAY_INIT;
3845         struct argv_array env = ARGV_ARRAY_INIT;
3846         struct diff_queue_struct *q = &diff_queued_diff;
3847
3848         argv_array_push(&argv, pgm);
3849         argv_array_push(&argv, name);
3850
3851         if (one && two) {
3852                 add_external_diff_name(&argv, name, one);
3853                 if (!other)
3854                         add_external_diff_name(&argv, name, two);
3855                 else {
3856                         add_external_diff_name(&argv, other, two);
3857                         argv_array_push(&argv, other);
3858                         argv_array_push(&argv, xfrm_msg);
3859                 }
3860         }
3861
3862         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3863         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3864
3865         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3866                 die(_("external diff died, stopping at %s"), name);
3867
3868         remove_tempfile();
3869         argv_array_clear(&argv);
3870         argv_array_clear(&env);
3871 }
3872
3873 static int similarity_index(struct diff_filepair *p)
3874 {
3875         return p->score * 100 / MAX_SCORE;
3876 }
3877
3878 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3879 {
3880         if (startup_info->have_repository)
3881                 return find_unique_abbrev(oid->hash, abbrev);
3882         else {
3883                 char *hex = oid_to_hex(oid);
3884                 if (abbrev < 0)
3885                         abbrev = FALLBACK_DEFAULT_ABBREV;
3886                 if (abbrev > GIT_SHA1_HEXSZ)
3887                         die("BUG: oid abbreviation out of range: %d", abbrev);
3888                 if (abbrev)
3889                         hex[abbrev] = '\0';
3890                 return hex;
3891         }
3892 }
3893
3894 static void fill_metainfo(struct strbuf *msg,
3895                           const char *name,
3896                           const char *other,
3897                           struct diff_filespec *one,
3898                           struct diff_filespec *two,
3899                           struct diff_options *o,
3900                           struct diff_filepair *p,
3901                           int *must_show_header,
3902                           int use_color)
3903 {
3904         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3905         const char *reset = diff_get_color(use_color, DIFF_RESET);
3906         const char *line_prefix = diff_line_prefix(o);
3907
3908         *must_show_header = 1;
3909         strbuf_init(msg, PATH_MAX * 2 + 300);
3910         switch (p->status) {
3911         case DIFF_STATUS_COPIED:
3912                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3913                             line_prefix, set, similarity_index(p));
3914                 strbuf_addf(msg, "%s\n%s%scopy from ",
3915                             reset,  line_prefix, set);
3916                 quote_c_style(name, msg, NULL, 0);
3917                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3918                 quote_c_style(other, msg, NULL, 0);
3919                 strbuf_addf(msg, "%s\n", reset);
3920                 break;
3921         case DIFF_STATUS_RENAMED:
3922                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3923                             line_prefix, set, similarity_index(p));
3924                 strbuf_addf(msg, "%s\n%s%srename from ",
3925                             reset, line_prefix, set);
3926                 quote_c_style(name, msg, NULL, 0);
3927                 strbuf_addf(msg, "%s\n%s%srename to ",
3928                             reset, line_prefix, set);
3929                 quote_c_style(other, msg, NULL, 0);
3930                 strbuf_addf(msg, "%s\n", reset);
3931                 break;
3932         case DIFF_STATUS_MODIFIED:
3933                 if (p->score) {
3934                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3935                                     line_prefix,
3936                                     set, similarity_index(p), reset);
3937                         break;
3938                 }
3939                 /* fallthru */
3940         default:
3941                 *must_show_header = 0;
3942         }
3943         if (one && two && oidcmp(&one->oid, &two->oid)) {
3944                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3945
3946                 if (DIFF_OPT_TST(o, BINARY)) {
3947                         mmfile_t mf;
3948                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3949                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3950                                 abbrev = 40;
3951                 }
3952                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3953                             diff_abbrev_oid(&one->oid, abbrev),
3954                             diff_abbrev_oid(&two->oid, abbrev));
3955                 if (one->mode == two->mode)
3956                         strbuf_addf(msg, " %06o", one->mode);
3957                 strbuf_addf(msg, "%s\n", reset);
3958         }
3959 }
3960
3961 static void run_diff_cmd(const char *pgm,
3962                          const char *name,
3963                          const char *other,
3964                          const char *attr_path,
3965                          struct diff_filespec *one,
3966                          struct diff_filespec *two,
3967                          struct strbuf *msg,
3968                          struct diff_options *o,
3969                          struct diff_filepair *p)
3970 {
3971         const char *xfrm_msg = NULL;
3972         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3973         int must_show_header = 0;
3974
3975
3976         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3977                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3978                 if (drv && drv->external)
3979                         pgm = drv->external;
3980         }
3981
3982         if (msg) {
3983                 /*
3984                  * don't use colors when the header is intended for an
3985                  * external diff driver
3986                  */
3987                 fill_metainfo(msg, name, other, one, two, o, p,
3988                               &must_show_header,
3989                               want_color(o->use_color) && !pgm);
3990                 xfrm_msg = msg->len ? msg->buf : NULL;
3991         }
3992
3993         if (pgm) {
3994                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3995                                   complete_rewrite, o);
3996                 return;
3997         }
3998         if (one && two)
3999                 builtin_diff(name, other ? other : name,
4000                              one, two, xfrm_msg, must_show_header,
4001                              o, complete_rewrite);
4002         else
4003                 fprintf(o->file, "* Unmerged path %s\n", name);
4004 }
4005
4006 static void diff_fill_oid_info(struct diff_filespec *one)
4007 {
4008         if (DIFF_FILE_VALID(one)) {
4009                 if (!one->oid_valid) {
4010                         struct stat st;
4011                         if (one->is_stdin) {
4012                                 oidclr(&one->oid);
4013                                 return;
4014                         }
4015                         if (lstat(one->path, &st) < 0)
4016                                 die_errno("stat '%s'", one->path);
4017                         if (index_path(&one->oid, one->path, &st, 0))
4018                                 die("cannot hash %s", one->path);
4019                 }
4020         }
4021         else
4022                 oidclr(&one->oid);
4023 }
4024
4025 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4026 {
4027         /* Strip the prefix but do not molest /dev/null and absolute paths */
4028         if (*namep && **namep != '/') {
4029                 *namep += prefix_length;
4030                 if (**namep == '/')
4031                         ++*namep;
4032         }
4033         if (*otherp && **otherp != '/') {
4034                 *otherp += prefix_length;
4035                 if (**otherp == '/')
4036                         ++*otherp;
4037         }
4038 }
4039
4040 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4041 {
4042         const char *pgm = external_diff();
4043         struct strbuf msg;
4044         struct diff_filespec *one = p->one;
4045         struct diff_filespec *two = p->two;
4046         const char *name;
4047         const char *other;
4048         const char *attr_path;
4049
4050         name  = one->path;
4051         other = (strcmp(name, two->path) ? two->path : NULL);
4052         attr_path = name;
4053         if (o->prefix_length)
4054                 strip_prefix(o->prefix_length, &name, &other);
4055
4056         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
4057                 pgm = NULL;
4058
4059         if (DIFF_PAIR_UNMERGED(p)) {
4060                 run_diff_cmd(pgm, name, NULL, attr_path,
4061                              NULL, NULL, NULL, o, p);
4062                 return;
4063         }
4064
4065         diff_fill_oid_info(one);
4066         diff_fill_oid_info(two);
4067
4068         if (!pgm &&
4069             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4070             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4071                 /*
4072                  * a filepair that changes between file and symlink
4073                  * needs to be split into deletion and creation.
4074                  */
4075                 struct diff_filespec *null = alloc_filespec(two->path);
4076                 run_diff_cmd(NULL, name, other, attr_path,
4077                              one, null, &msg, o, p);
4078                 free(null);
4079                 strbuf_release(&msg);
4080
4081                 null = alloc_filespec(one->path);
4082                 run_diff_cmd(NULL, name, other, attr_path,
4083                              null, two, &msg, o, p);
4084                 free(null);
4085         }
4086         else
4087                 run_diff_cmd(pgm, name, other, attr_path,
4088                              one, two, &msg, o, p);
4089
4090         strbuf_release(&msg);
4091 }
4092
4093 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4094                          struct diffstat_t *diffstat)
4095 {
4096         const char *name;
4097         const char *other;
4098
4099         if (DIFF_PAIR_UNMERGED(p)) {
4100                 /* unmerged */
4101                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4102                 return;
4103         }
4104
4105         name = p->one->path;
4106         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4107
4108         if (o->prefix_length)
4109                 strip_prefix(o->prefix_length, &name, &other);
4110
4111         diff_fill_oid_info(p->one);
4112         diff_fill_oid_info(p->two);
4113
4114         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4115 }
4116
4117 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4118 {
4119         const char *name;
4120         const char *other;
4121         const char *attr_path;
4122
4123         if (DIFF_PAIR_UNMERGED(p)) {
4124                 /* unmerged */
4125                 return;
4126         }
4127
4128         name = p->one->path;
4129         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4130         attr_path = other ? other : name;
4131
4132         if (o->prefix_length)
4133                 strip_prefix(o->prefix_length, &name, &other);
4134
4135         diff_fill_oid_info(p->one);
4136         diff_fill_oid_info(p->two);
4137
4138         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4139 }
4140
4141 void diff_setup(struct diff_options *options)
4142 {
4143         memcpy(options, &default_diff_options, sizeof(*options));
4144
4145         options->file = stdout;
4146
4147         options->abbrev = DEFAULT_ABBREV;
4148         options->line_termination = '\n';
4149         options->break_opt = -1;
4150         options->rename_limit = -1;
4151         options->dirstat_permille = diff_dirstat_permille_default;
4152         options->context = diff_context_default;
4153         options->interhunkcontext = diff_interhunk_context_default;
4154         options->ws_error_highlight = ws_error_highlight_default;
4155         DIFF_OPT_SET(options, RENAME_EMPTY);
4156
4157         /* pathchange left =NULL by default */
4158         options->change = diff_change;
4159         options->add_remove = diff_addremove;
4160         options->use_color = diff_use_color_default;
4161         options->detect_rename = diff_detect_rename_default;
4162         options->xdl_opts |= diff_algorithm;
4163         if (diff_indent_heuristic)
4164                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4165
4166         options->orderfile = diff_order_file_cfg;
4167
4168         if (diff_no_prefix) {
4169                 options->a_prefix = options->b_prefix = "";
4170         } else if (!diff_mnemonic_prefix) {
4171                 options->a_prefix = "a/";
4172                 options->b_prefix = "b/";
4173         }
4174
4175         options->color_moved = diff_color_moved_default;
4176 }
4177
4178 void diff_setup_done(struct diff_options *options)
4179 {
4180         int count = 0;
4181
4182         if (options->set_default)
4183                 options->set_default(options);
4184
4185         if (options->output_format & DIFF_FORMAT_NAME)
4186                 count++;
4187         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
4188                 count++;
4189         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
4190                 count++;
4191         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
4192                 count++;
4193         if (count > 1)
4194                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4195
4196         /*
4197          * Most of the time we can say "there are changes"
4198          * only by checking if there are changed paths, but
4199          * --ignore-whitespace* options force us to look
4200          * inside contents.
4201          */
4202
4203         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
4204             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
4205             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
4206                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
4207         else
4208                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
4209
4210         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
4211                 options->detect_rename = DIFF_DETECT_COPY;
4212
4213         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
4214                 options->prefix = NULL;
4215         if (options->prefix)
4216                 options->prefix_length = strlen(options->prefix);
4217         else
4218                 options->prefix_length = 0;
4219
4220         if (options->output_format & (DIFF_FORMAT_NAME |
4221                                       DIFF_FORMAT_NAME_STATUS |
4222                                       DIFF_FORMAT_CHECKDIFF |
4223                                       DIFF_FORMAT_NO_OUTPUT))
4224                 options->output_format &= ~(DIFF_FORMAT_RAW |
4225                                             DIFF_FORMAT_NUMSTAT |
4226                                             DIFF_FORMAT_DIFFSTAT |
4227                                             DIFF_FORMAT_SHORTSTAT |
4228                                             DIFF_FORMAT_DIRSTAT |
4229                                             DIFF_FORMAT_SUMMARY |
4230                                             DIFF_FORMAT_PATCH);
4231
4232         /*
4233          * These cases always need recursive; we do not drop caller-supplied
4234          * recursive bits for other formats here.
4235          */
4236         if (options->output_format & (DIFF_FORMAT_PATCH |
4237                                       DIFF_FORMAT_NUMSTAT |
4238                                       DIFF_FORMAT_DIFFSTAT |
4239                                       DIFF_FORMAT_SHORTSTAT |
4240                                       DIFF_FORMAT_DIRSTAT |
4241                                       DIFF_FORMAT_SUMMARY |
4242                                       DIFF_FORMAT_CHECKDIFF))
4243                 DIFF_OPT_SET(options, RECURSIVE);
4244         /*
4245          * Also pickaxe would not work very well if you do not say recursive
4246          */
4247         if (options->pickaxe)
4248                 DIFF_OPT_SET(options, RECURSIVE);
4249         /*
4250          * When patches are generated, submodules diffed against the work tree
4251          * must be checked for dirtiness too so it can be shown in the output
4252          */
4253         if (options->output_format & DIFF_FORMAT_PATCH)
4254                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
4255
4256         if (options->detect_rename && options->rename_limit < 0)
4257                 options->rename_limit = diff_rename_limit_default;
4258         if (options->setup & DIFF_SETUP_USE_CACHE) {
4259                 if (!active_cache)
4260                         /* read-cache does not die even when it fails
4261                          * so it is safe for us to do this here.  Also
4262                          * it does not smudge active_cache or active_nr
4263                          * when it fails, so we do not have to worry about
4264                          * cleaning it up ourselves either.
4265                          */
4266                         read_cache();
4267         }
4268         if (40 < options->abbrev)
4269                 options->abbrev = 40; /* full */
4270
4271         /*
4272          * It does not make sense to show the first hit we happened
4273          * to have found.  It does not make sense not to return with
4274          * exit code in such a case either.
4275          */
4276         if (DIFF_OPT_TST(options, QUICK)) {
4277                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4278                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
4279         }
4280
4281         options->diff_path_counter = 0;
4282
4283         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
4284                 die(_("--follow requires exactly one pathspec"));
4285
4286         if (!options->use_color || external_diff())
4287                 options->color_moved = 0;
4288 }
4289
4290 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4291 {
4292         char c, *eq;
4293         int len;
4294
4295         if (*arg != '-')
4296                 return 0;
4297         c = *++arg;
4298         if (!c)
4299                 return 0;
4300         if (c == arg_short) {
4301                 c = *++arg;
4302                 if (!c)
4303                         return 1;
4304                 if (val && isdigit(c)) {
4305                         char *end;
4306                         int n = strtoul(arg, &end, 10);
4307                         if (*end)
4308                                 return 0;
4309                         *val = n;
4310                         return 1;
4311                 }
4312                 return 0;
4313         }
4314         if (c != '-')
4315                 return 0;
4316         arg++;
4317         eq = strchrnul(arg, '=');
4318         len = eq - arg;
4319         if (!len || strncmp(arg, arg_long, len))
4320                 return 0;
4321         if (*eq) {
4322                 int n;
4323                 char *end;
4324                 if (!isdigit(*++eq))
4325                         return 0;
4326                 n = strtoul(eq, &end, 10);
4327                 if (*end)
4328                         return 0;
4329                 *val = n;
4330         }
4331         return 1;
4332 }
4333
4334 static int diff_scoreopt_parse(const char *opt);
4335
4336 static inline int short_opt(char opt, const char **argv,
4337                             const char **optarg)
4338 {
4339         const char *arg = argv[0];
4340         if (arg[0] != '-' || arg[1] != opt)
4341                 return 0;
4342         if (arg[2] != '\0') {
4343                 *optarg = arg + 2;
4344                 return 1;
4345         }
4346         if (!argv[1])
4347                 die("Option '%c' requires a value", opt);
4348         *optarg = argv[1];
4349         return 2;
4350 }
4351
4352 int parse_long_opt(const char *opt, const char **argv,
4353                    const char **optarg)
4354 {
4355         const char *arg = argv[0];
4356         if (!skip_prefix(arg, "--", &arg))
4357                 return 0;
4358         if (!skip_prefix(arg, opt, &arg))
4359                 return 0;
4360         if (*arg == '=') { /* stuck form: --option=value */
4361                 *optarg = arg + 1;
4362                 return 1;
4363         }
4364         if (*arg != '\0')
4365                 return 0;
4366         /* separate form: --option value */
4367         if (!argv[1])
4368                 die("Option '--%s' requires a value", opt);
4369         *optarg = argv[1];
4370         return 2;
4371 }
4372
4373 static int stat_opt(struct diff_options *options, const char **av)
4374 {
4375         const char *arg = av[0];
4376         char *end;
4377         int width = options->stat_width;
4378         int name_width = options->stat_name_width;
4379         int graph_width = options->stat_graph_width;
4380         int count = options->stat_count;
4381         int argcount = 1;
4382
4383         if (!skip_prefix(arg, "--stat", &arg))
4384                 die("BUG: stat option does not begin with --stat: %s", arg);
4385         end = (char *)arg;
4386
4387         switch (*arg) {
4388         case '-':
4389                 if (skip_prefix(arg, "-width", &arg)) {
4390                         if (*arg == '=')
4391                                 width = strtoul(arg + 1, &end, 10);
4392                         else if (!*arg && !av[1])
4393                                 die_want_option("--stat-width");
4394                         else if (!*arg) {
4395                                 width = strtoul(av[1], &end, 10);
4396                                 argcount = 2;
4397                         }
4398                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4399                         if (*arg == '=')
4400                                 name_width = strtoul(arg + 1, &end, 10);
4401                         else if (!*arg && !av[1])
4402                                 die_want_option("--stat-name-width");
4403                         else if (!*arg) {
4404                                 name_width = strtoul(av[1], &end, 10);
4405                                 argcount = 2;
4406                         }
4407                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4408                         if (*arg == '=')
4409                                 graph_width = strtoul(arg + 1, &end, 10);
4410                         else if (!*arg && !av[1])
4411                                 die_want_option("--stat-graph-width");
4412                         else if (!*arg) {
4413                                 graph_width = strtoul(av[1], &end, 10);
4414                                 argcount = 2;
4415                         }
4416                 } else if (skip_prefix(arg, "-count", &arg)) {
4417                         if (*arg == '=')
4418                                 count = strtoul(arg + 1, &end, 10);
4419                         else if (!*arg && !av[1])
4420                                 die_want_option("--stat-count");
4421                         else if (!*arg) {
4422                                 count = strtoul(av[1], &end, 10);
4423                                 argcount = 2;
4424                         }
4425                 }
4426                 break;
4427         case '=':
4428                 width = strtoul(arg+1, &end, 10);
4429                 if (*end == ',')
4430                         name_width = strtoul(end+1, &end, 10);
4431                 if (*end == ',')
4432                         count = strtoul(end+1, &end, 10);
4433         }
4434
4435         /* Important! This checks all the error cases! */
4436         if (*end)
4437                 return 0;
4438         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4439         options->stat_name_width = name_width;
4440         options->stat_graph_width = graph_width;
4441         options->stat_width = width;
4442         options->stat_count = count;
4443         return argcount;
4444 }
4445
4446 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4447 {
4448         struct strbuf errmsg = STRBUF_INIT;
4449         if (parse_dirstat_params(options, params, &errmsg))
4450                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4451                     errmsg.buf);
4452         strbuf_release(&errmsg);
4453         /*
4454          * The caller knows a dirstat-related option is given from the command
4455          * line; allow it to say "return this_function();"
4456          */
4457         options->output_format |= DIFF_FORMAT_DIRSTAT;
4458         return 1;
4459 }
4460
4461 static int parse_submodule_opt(struct diff_options *options, const char *value)
4462 {
4463         if (parse_submodule_params(options, value))
4464                 die(_("Failed to parse --submodule option parameter: '%s'"),
4465                         value);
4466         return 1;
4467 }
4468
4469 static const char diff_status_letters[] = {
4470         DIFF_STATUS_ADDED,
4471         DIFF_STATUS_COPIED,
4472         DIFF_STATUS_DELETED,
4473         DIFF_STATUS_MODIFIED,
4474         DIFF_STATUS_RENAMED,
4475         DIFF_STATUS_TYPE_CHANGED,
4476         DIFF_STATUS_UNKNOWN,
4477         DIFF_STATUS_UNMERGED,
4478         DIFF_STATUS_FILTER_AON,
4479         DIFF_STATUS_FILTER_BROKEN,
4480         '\0',
4481 };
4482
4483 static unsigned int filter_bit['Z' + 1];
4484
4485 static void prepare_filter_bits(void)
4486 {
4487         int i;
4488
4489         if (!filter_bit[DIFF_STATUS_ADDED]) {
4490                 for (i = 0; diff_status_letters[i]; i++)
4491                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4492         }
4493 }
4494
4495 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4496 {
4497         return opt->filter & filter_bit[(int) status];
4498 }
4499
4500 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4501 {
4502         int i, optch;
4503
4504         prepare_filter_bits();
4505
4506         /*
4507          * If there is a negation e.g. 'd' in the input, and we haven't
4508          * initialized the filter field with another --diff-filter, start
4509          * from full set of bits, except for AON.
4510          */
4511         if (!opt->filter) {
4512                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4513                         if (optch < 'a' || 'z' < optch)
4514                                 continue;
4515                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4516                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4517                         break;
4518                 }
4519         }
4520
4521         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4522                 unsigned int bit;
4523                 int negate;
4524
4525                 if ('a' <= optch && optch <= 'z') {
4526                         negate = 1;
4527                         optch = toupper(optch);
4528                 } else {
4529                         negate = 0;
4530                 }
4531
4532                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4533                 if (!bit)
4534                         return optarg[i];
4535                 if (negate)
4536                         opt->filter &= ~bit;
4537                 else
4538                         opt->filter |= bit;
4539         }
4540         return 0;
4541 }
4542
4543 static void enable_patch_output(int *fmt) {
4544         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4545         *fmt |= DIFF_FORMAT_PATCH;
4546 }
4547
4548 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4549 {
4550         int val = parse_ws_error_highlight(arg);
4551
4552         if (val < 0) {
4553                 error("unknown value after ws-error-highlight=%.*s",
4554                       -1 - val, arg);
4555                 return 0;
4556         }
4557         opt->ws_error_highlight = val;
4558         return 1;
4559 }
4560
4561 int diff_opt_parse(struct diff_options *options,
4562                    const char **av, int ac, const char *prefix)
4563 {
4564         const char *arg = av[0];
4565         const char *optarg;
4566         int argcount;
4567
4568         if (!prefix)
4569                 prefix = "";
4570
4571         /* Output format options */
4572         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4573             || opt_arg(arg, 'U', "unified", &options->context))
4574                 enable_patch_output(&options->output_format);
4575         else if (!strcmp(arg, "--raw"))
4576                 options->output_format |= DIFF_FORMAT_RAW;
4577         else if (!strcmp(arg, "--patch-with-raw")) {
4578                 enable_patch_output(&options->output_format);
4579                 options->output_format |= DIFF_FORMAT_RAW;
4580         } else if (!strcmp(arg, "--numstat"))
4581                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4582         else if (!strcmp(arg, "--shortstat"))
4583                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4584         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
4585                 return parse_dirstat_opt(options, "");
4586         else if (skip_prefix(arg, "-X", &arg))
4587                 return parse_dirstat_opt(options, arg);
4588         else if (skip_prefix(arg, "--dirstat=", &arg))
4589                 return parse_dirstat_opt(options, arg);
4590         else if (!strcmp(arg, "--cumulative"))
4591                 return parse_dirstat_opt(options, "cumulative");
4592         else if (!strcmp(arg, "--dirstat-by-file"))
4593                 return parse_dirstat_opt(options, "files");
4594         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
4595                 parse_dirstat_opt(options, "files");
4596                 return parse_dirstat_opt(options, arg);
4597         }
4598         else if (!strcmp(arg, "--check"))
4599                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4600         else if (!strcmp(arg, "--summary"))
4601                 options->output_format |= DIFF_FORMAT_SUMMARY;
4602         else if (!strcmp(arg, "--patch-with-stat")) {
4603                 enable_patch_output(&options->output_format);
4604                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4605         } else if (!strcmp(arg, "--name-only"))
4606                 options->output_format |= DIFF_FORMAT_NAME;
4607         else if (!strcmp(arg, "--name-status"))
4608                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4609         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4610                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4611         else if (starts_with(arg, "--stat"))
4612                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4613                 return stat_opt(options, av);
4614
4615         /* renames options */
4616         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
4617                  !strcmp(arg, "--break-rewrites")) {
4618                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4619                         return error("invalid argument to -B: %s", arg+2);
4620         }
4621         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
4622                  !strcmp(arg, "--find-renames")) {
4623                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4624                         return error("invalid argument to -M: %s", arg+2);
4625                 options->detect_rename = DIFF_DETECT_RENAME;
4626         }
4627         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4628                 options->irreversible_delete = 1;
4629         }
4630         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
4631                  !strcmp(arg, "--find-copies")) {
4632                 if (options->detect_rename == DIFF_DETECT_COPY)
4633                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
4634                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4635                         return error("invalid argument to -C: %s", arg+2);
4636                 options->detect_rename = DIFF_DETECT_COPY;
4637         }
4638         else if (!strcmp(arg, "--no-renames"))
4639                 options->detect_rename = 0;
4640         else if (!strcmp(arg, "--rename-empty"))
4641                 DIFF_OPT_SET(options, RENAME_EMPTY);
4642         else if (!strcmp(arg, "--no-rename-empty"))
4643                 DIFF_OPT_CLR(options, RENAME_EMPTY);
4644         else if (!strcmp(arg, "--relative"))
4645                 DIFF_OPT_SET(options, RELATIVE_NAME);
4646         else if (skip_prefix(arg, "--relative=", &arg)) {
4647                 DIFF_OPT_SET(options, RELATIVE_NAME);
4648                 options->prefix = arg;
4649         }
4650
4651         /* xdiff options */
4652         else if (!strcmp(arg, "--minimal"))
4653                 DIFF_XDL_SET(options, NEED_MINIMAL);
4654         else if (!strcmp(arg, "--no-minimal"))
4655                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4656         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4657                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4658         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4659                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4660         else if (!strcmp(arg, "--ignore-space-at-eol"))
4661                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4662         else if (!strcmp(arg, "--ignore-blank-lines"))
4663                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4664         else if (!strcmp(arg, "--indent-heuristic"))
4665                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4666         else if (!strcmp(arg, "--no-indent-heuristic"))
4667                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4668         else if (!strcmp(arg, "--patience"))
4669                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4670         else if (!strcmp(arg, "--histogram"))
4671                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4672         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4673                 long value = parse_algorithm_value(optarg);
4674                 if (value < 0)
4675                         return error("option diff-algorithm accepts \"myers\", "
4676                                      "\"minimal\", \"patience\" and \"histogram\"");
4677                 /* clear out previous settings */
4678                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4679                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4680                 options->xdl_opts |= value;
4681                 return argcount;
4682         }
4683
4684         /* flags options */
4685         else if (!strcmp(arg, "--binary")) {
4686                 enable_patch_output(&options->output_format);
4687                 DIFF_OPT_SET(options, BINARY);
4688         }
4689         else if (!strcmp(arg, "--full-index"))
4690                 DIFF_OPT_SET(options, FULL_INDEX);
4691         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4692                 DIFF_OPT_SET(options, TEXT);
4693         else if (!strcmp(arg, "-R"))
4694                 DIFF_OPT_SET(options, REVERSE_DIFF);
4695         else if (!strcmp(arg, "--find-copies-harder"))
4696                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
4697         else if (!strcmp(arg, "--follow"))
4698                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
4699         else if (!strcmp(arg, "--no-follow")) {
4700                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
4701                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
4702         } else if (!strcmp(arg, "--color"))
4703                 options->use_color = 1;
4704         else if (skip_prefix(arg, "--color=", &arg)) {
4705                 int value = git_config_colorbool(NULL, arg);
4706                 if (value < 0)
4707                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4708                 options->use_color = value;
4709         }
4710         else if (!strcmp(arg, "--no-color"))
4711                 options->use_color = 0;
4712         else if (!strcmp(arg, "--color-moved")) {
4713                 if (diff_color_moved_default)
4714                         options->color_moved = diff_color_moved_default;
4715                 if (options->color_moved == COLOR_MOVED_NO)
4716                         options->color_moved = COLOR_MOVED_DEFAULT;
4717         } else if (!strcmp(arg, "--no-color-moved"))
4718                 options->color_moved = COLOR_MOVED_NO;
4719         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4720                 int cm = parse_color_moved(arg);
4721                 if (cm < 0)
4722                         die("bad --color-moved argument: %s", arg);
4723                 options->color_moved = cm;
4724         } else if (!strcmp(arg, "--color-words")) {
4725                 options->use_color = 1;
4726                 options->word_diff = DIFF_WORDS_COLOR;
4727         }
4728         else if (skip_prefix(arg, "--color-words=", &arg)) {
4729                 options->use_color = 1;
4730                 options->word_diff = DIFF_WORDS_COLOR;
4731                 options->word_regex = arg;
4732         }
4733         else if (!strcmp(arg, "--word-diff")) {
4734                 if (options->word_diff == DIFF_WORDS_NONE)
4735                         options->word_diff = DIFF_WORDS_PLAIN;
4736         }
4737         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4738                 if (!strcmp(arg, "plain"))
4739                         options->word_diff = DIFF_WORDS_PLAIN;
4740                 else if (!strcmp(arg, "color")) {
4741                         options->use_color = 1;
4742                         options->word_diff = DIFF_WORDS_COLOR;
4743                 }
4744                 else if (!strcmp(arg, "porcelain"))
4745                         options->word_diff = DIFF_WORDS_PORCELAIN;
4746                 else if (!strcmp(arg, "none"))
4747                         options->word_diff = DIFF_WORDS_NONE;
4748                 else
4749                         die("bad --word-diff argument: %s", arg);
4750         }
4751         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4752                 if (options->word_diff == DIFF_WORDS_NONE)
4753                         options->word_diff = DIFF_WORDS_PLAIN;
4754                 options->word_regex = optarg;
4755                 return argcount;
4756         }
4757         else if (!strcmp(arg, "--exit-code"))
4758                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
4759         else if (!strcmp(arg, "--quiet"))
4760                 DIFF_OPT_SET(options, QUICK);
4761         else if (!strcmp(arg, "--ext-diff"))
4762                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
4763         else if (!strcmp(arg, "--no-ext-diff"))
4764                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
4765         else if (!strcmp(arg, "--textconv"))
4766                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
4767         else if (!strcmp(arg, "--no-textconv"))
4768                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
4769         else if (!strcmp(arg, "--ignore-submodules")) {
4770                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
4771                 handle_ignore_submodules_arg(options, "all");
4772         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
4773                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
4774                 handle_ignore_submodules_arg(options, arg);
4775         } else if (!strcmp(arg, "--submodule"))
4776                 options->submodule_format = DIFF_SUBMODULE_LOG;
4777         else if (skip_prefix(arg, "--submodule=", &arg))
4778                 return parse_submodule_opt(options, arg);
4779         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
4780                 return parse_ws_error_highlight_opt(options, arg);
4781         else if (!strcmp(arg, "--ita-invisible-in-index"))
4782                 options->ita_invisible_in_index = 1;
4783         else if (!strcmp(arg, "--ita-visible-in-index"))
4784                 options->ita_invisible_in_index = 0;
4785
4786         /* misc options */
4787         else if (!strcmp(arg, "-z"))
4788                 options->line_termination = 0;
4789         else if ((argcount = short_opt('l', av, &optarg))) {
4790                 options->rename_limit = strtoul(optarg, NULL, 10);
4791                 return argcount;
4792         }
4793         else if ((argcount = short_opt('S', av, &optarg))) {
4794                 options->pickaxe = optarg;
4795                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4796                 return argcount;
4797         } else if ((argcount = short_opt('G', av, &optarg))) {
4798                 options->pickaxe = optarg;
4799                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4800                 return argcount;
4801         }
4802         else if (!strcmp(arg, "--pickaxe-all"))
4803                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4804         else if (!strcmp(arg, "--pickaxe-regex"))
4805                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4806         else if ((argcount = short_opt('O', av, &optarg))) {
4807                 options->orderfile = prefix_filename(prefix, optarg);
4808                 return argcount;
4809         }
4810         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4811                 int offending = parse_diff_filter_opt(optarg, options);
4812                 if (offending)
4813                         die("unknown change class '%c' in --diff-filter=%s",
4814                             offending, optarg);
4815                 return argcount;
4816         }
4817         else if (!strcmp(arg, "--no-abbrev"))
4818                 options->abbrev = 0;
4819         else if (!strcmp(arg, "--abbrev"))
4820                 options->abbrev = DEFAULT_ABBREV;
4821         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4822                 options->abbrev = strtoul(arg, NULL, 10);
4823                 if (options->abbrev < MINIMUM_ABBREV)
4824                         options->abbrev = MINIMUM_ABBREV;
4825                 else if (40 < options->abbrev)
4826                         options->abbrev = 40;
4827         }
4828         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4829                 options->a_prefix = optarg;
4830                 return argcount;
4831         }
4832         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4833                 options->line_prefix = optarg;
4834                 options->line_prefix_length = strlen(options->line_prefix);
4835                 graph_setup_line_prefix(options);
4836                 return argcount;
4837         }
4838         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4839                 options->b_prefix = optarg;
4840                 return argcount;
4841         }
4842         else if (!strcmp(arg, "--no-prefix"))
4843                 options->a_prefix = options->b_prefix = "";
4844         else if (opt_arg(arg, '\0', "inter-hunk-context",
4845                          &options->interhunkcontext))
4846                 ;
4847         else if (!strcmp(arg, "-W"))
4848                 DIFF_OPT_SET(options, FUNCCONTEXT);
4849         else if (!strcmp(arg, "--function-context"))
4850                 DIFF_OPT_SET(options, FUNCCONTEXT);
4851         else if (!strcmp(arg, "--no-function-context"))
4852                 DIFF_OPT_CLR(options, FUNCCONTEXT);
4853         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4854                 char *path = prefix_filename(prefix, optarg);
4855                 options->file = xfopen(path, "w");
4856                 options->close_file = 1;
4857                 if (options->use_color != GIT_COLOR_ALWAYS)
4858                         options->use_color = GIT_COLOR_NEVER;
4859                 free(path);
4860                 return argcount;
4861         } else
4862                 return 0;
4863         return 1;
4864 }
4865
4866 int parse_rename_score(const char **cp_p)
4867 {
4868         unsigned long num, scale;
4869         int ch, dot;
4870         const char *cp = *cp_p;
4871
4872         num = 0;
4873         scale = 1;
4874         dot = 0;
4875         for (;;) {
4876                 ch = *cp;
4877                 if ( !dot && ch == '.' ) {
4878                         scale = 1;
4879                         dot = 1;
4880                 } else if ( ch == '%' ) {
4881                         scale = dot ? scale*100 : 100;
4882                         cp++;   /* % is always at the end */
4883                         break;
4884                 } else if ( ch >= '0' && ch <= '9' ) {
4885                         if ( scale < 100000 ) {
4886                                 scale *= 10;
4887                                 num = (num*10) + (ch-'0');
4888                         }
4889                 } else {
4890                         break;
4891                 }
4892                 cp++;
4893         }
4894         *cp_p = cp;
4895
4896         /* user says num divided by scale and we say internally that
4897          * is MAX_SCORE * num / scale.
4898          */
4899         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4900 }
4901
4902 static int diff_scoreopt_parse(const char *opt)
4903 {
4904         int opt1, opt2, cmd;
4905
4906         if (*opt++ != '-')
4907                 return -1;
4908         cmd = *opt++;
4909         if (cmd == '-') {
4910                 /* convert the long-form arguments into short-form versions */
4911                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4912                         if (*opt == 0 || *opt++ == '=')
4913                                 cmd = 'B';
4914                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4915                         if (*opt == 0 || *opt++ == '=')
4916                                 cmd = 'C';
4917                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4918                         if (*opt == 0 || *opt++ == '=')
4919                                 cmd = 'M';
4920                 }
4921         }
4922         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4923                 return -1; /* that is not a -M, -C, or -B option */
4924
4925         opt1 = parse_rename_score(&opt);
4926         if (cmd != 'B')
4927                 opt2 = 0;
4928         else {
4929                 if (*opt == 0)
4930                         opt2 = 0;
4931                 else if (*opt != '/')
4932                         return -1; /* we expect -B80/99 or -B80 */
4933                 else {
4934                         opt++;
4935                         opt2 = parse_rename_score(&opt);
4936                 }
4937         }
4938         if (*opt != 0)
4939                 return -1;
4940         return opt1 | (opt2 << 16);
4941 }
4942
4943 struct diff_queue_struct diff_queued_diff;
4944
4945 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4946 {
4947         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4948         queue->queue[queue->nr++] = dp;
4949 }
4950
4951 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4952                                  struct diff_filespec *one,
4953                                  struct diff_filespec *two)
4954 {
4955         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4956         dp->one = one;
4957         dp->two = two;
4958         if (queue)
4959                 diff_q(queue, dp);
4960         return dp;
4961 }
4962
4963 void diff_free_filepair(struct diff_filepair *p)
4964 {
4965         free_filespec(p->one);
4966         free_filespec(p->two);
4967         free(p);
4968 }
4969
4970 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4971 {
4972         int abblen;
4973         const char *abbrev;
4974
4975         if (len == GIT_SHA1_HEXSZ)
4976                 return oid_to_hex(oid);
4977
4978         abbrev = diff_abbrev_oid(oid, len);
4979         abblen = strlen(abbrev);
4980
4981         /*
4982          * In well-behaved cases, where the abbbreviated result is the
4983          * same as the requested length, append three dots after the
4984          * abbreviation (hence the whole logic is limited to the case
4985          * where abblen < 37); when the actual abbreviated result is a
4986          * bit longer than the requested length, we reduce the number
4987          * of dots so that they match the well-behaved ones.  However,
4988          * if the actual abbreviation is longer than the requested
4989          * length by more than three, we give up on aligning, and add
4990          * three dots anyway, to indicate that the output is not the
4991          * full object name.  Yes, this may be suboptimal, but this
4992          * appears only in "diff --raw --abbrev" output and it is not
4993          * worth the effort to change it now.  Note that this would
4994          * likely to work fine when the automatic sizing of default
4995          * abbreviation length is used--we would be fed -1 in "len" in
4996          * that case, and will end up always appending three-dots, but
4997          * the automatic sizing is supposed to give abblen that ensures
4998          * uniqueness across all objects (statistically speaking).
4999          */
5000         if (abblen < GIT_SHA1_HEXSZ - 3) {
5001                 static char hex[GIT_MAX_HEXSZ + 1];
5002                 if (len < abblen && abblen <= len + 2)
5003                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5004                 else
5005                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5006                 return hex;
5007         }
5008
5009         return oid_to_hex(oid);
5010 }
5011
5012 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5013 {
5014         int line_termination = opt->line_termination;
5015         int inter_name_termination = line_termination ? '\t' : '\0';
5016
5017         fprintf(opt->file, "%s", diff_line_prefix(opt));
5018         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5019                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5020                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5021                 fprintf(opt->file, "%s ",
5022                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5023         }
5024         if (p->score) {
5025                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5026                         inter_name_termination);
5027         } else {
5028                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5029         }
5030
5031         if (p->status == DIFF_STATUS_COPIED ||
5032             p->status == DIFF_STATUS_RENAMED) {
5033                 const char *name_a, *name_b;
5034                 name_a = p->one->path;
5035                 name_b = p->two->path;
5036                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5037                 write_name_quoted(name_a, opt->file, inter_name_termination);
5038                 write_name_quoted(name_b, opt->file, line_termination);
5039         } else {
5040                 const char *name_a, *name_b;
5041                 name_a = p->one->mode ? p->one->path : p->two->path;
5042                 name_b = NULL;
5043                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5044                 write_name_quoted(name_a, opt->file, line_termination);
5045         }
5046 }
5047
5048 int diff_unmodified_pair(struct diff_filepair *p)
5049 {
5050         /* This function is written stricter than necessary to support
5051          * the currently implemented transformers, but the idea is to
5052          * let transformers to produce diff_filepairs any way they want,
5053          * and filter and clean them up here before producing the output.
5054          */
5055         struct diff_filespec *one = p->one, *two = p->two;
5056
5057         if (DIFF_PAIR_UNMERGED(p))
5058                 return 0; /* unmerged is interesting */
5059
5060         /* deletion, addition, mode or type change
5061          * and rename are all interesting.
5062          */
5063         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5064             DIFF_PAIR_MODE_CHANGED(p) ||
5065             strcmp(one->path, two->path))
5066                 return 0;
5067
5068         /* both are valid and point at the same path.  that is, we are
5069          * dealing with a change.
5070          */
5071         if (one->oid_valid && two->oid_valid &&
5072             !oidcmp(&one->oid, &two->oid) &&
5073             !one->dirty_submodule && !two->dirty_submodule)
5074                 return 1; /* no change */
5075         if (!one->oid_valid && !two->oid_valid)
5076                 return 1; /* both look at the same file on the filesystem. */
5077         return 0;
5078 }
5079
5080 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5081 {
5082         if (diff_unmodified_pair(p))
5083                 return;
5084
5085         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5086             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5087                 return; /* no tree diffs in patch format */
5088
5089         run_diff(p, o);
5090 }
5091
5092 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5093                             struct diffstat_t *diffstat)
5094 {
5095         if (diff_unmodified_pair(p))
5096                 return;
5097
5098         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5099             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5100                 return; /* no useful stat for tree diffs */
5101
5102         run_diffstat(p, o, diffstat);
5103 }
5104
5105 static void diff_flush_checkdiff(struct diff_filepair *p,
5106                 struct diff_options *o)
5107 {
5108         if (diff_unmodified_pair(p))
5109                 return;
5110
5111         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5112             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5113                 return; /* nothing to check in tree diffs */
5114
5115         run_checkdiff(p, o);
5116 }
5117
5118 int diff_queue_is_empty(void)
5119 {
5120         struct diff_queue_struct *q = &diff_queued_diff;
5121         int i;
5122         for (i = 0; i < q->nr; i++)
5123                 if (!diff_unmodified_pair(q->queue[i]))
5124                         return 0;
5125         return 1;
5126 }
5127
5128 #if DIFF_DEBUG
5129 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5130 {
5131         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5132                 x, one ? one : "",
5133                 s->path,
5134                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5135                 s->mode,
5136                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5137         fprintf(stderr, "queue[%d] %s size %lu\n",
5138                 x, one ? one : "",
5139                 s->size);
5140 }
5141
5142 void diff_debug_filepair(const struct diff_filepair *p, int i)
5143 {
5144         diff_debug_filespec(p->one, i, "one");
5145         diff_debug_filespec(p->two, i, "two");
5146         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5147                 p->score, p->status ? p->status : '?',
5148                 p->one->rename_used, p->broken_pair);
5149 }
5150
5151 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5152 {
5153         int i;
5154         if (msg)
5155                 fprintf(stderr, "%s\n", msg);
5156         fprintf(stderr, "q->nr = %d\n", q->nr);
5157         for (i = 0; i < q->nr; i++) {
5158                 struct diff_filepair *p = q->queue[i];
5159                 diff_debug_filepair(p, i);
5160         }
5161 }
5162 #endif
5163
5164 static void diff_resolve_rename_copy(void)
5165 {
5166         int i;
5167         struct diff_filepair *p;
5168         struct diff_queue_struct *q = &diff_queued_diff;
5169
5170         diff_debug_queue("resolve-rename-copy", q);
5171
5172         for (i = 0; i < q->nr; i++) {
5173                 p = q->queue[i];
5174                 p->status = 0; /* undecided */
5175                 if (DIFF_PAIR_UNMERGED(p))
5176                         p->status = DIFF_STATUS_UNMERGED;
5177                 else if (!DIFF_FILE_VALID(p->one))
5178                         p->status = DIFF_STATUS_ADDED;
5179                 else if (!DIFF_FILE_VALID(p->two))
5180                         p->status = DIFF_STATUS_DELETED;
5181                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5182                         p->status = DIFF_STATUS_TYPE_CHANGED;
5183
5184                 /* from this point on, we are dealing with a pair
5185                  * whose both sides are valid and of the same type, i.e.
5186                  * either in-place edit or rename/copy edit.
5187                  */
5188                 else if (DIFF_PAIR_RENAME(p)) {
5189                         /*
5190                          * A rename might have re-connected a broken
5191                          * pair up, causing the pathnames to be the
5192                          * same again. If so, that's not a rename at
5193                          * all, just a modification..
5194                          *
5195                          * Otherwise, see if this source was used for
5196                          * multiple renames, in which case we decrement
5197                          * the count, and call it a copy.
5198                          */
5199                         if (!strcmp(p->one->path, p->two->path))
5200                                 p->status = DIFF_STATUS_MODIFIED;
5201                         else if (--p->one->rename_used > 0)
5202                                 p->status = DIFF_STATUS_COPIED;
5203                         else
5204                                 p->status = DIFF_STATUS_RENAMED;
5205                 }
5206                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5207                          p->one->mode != p->two->mode ||
5208                          p->one->dirty_submodule ||
5209                          p->two->dirty_submodule ||
5210                          is_null_oid(&p->one->oid))
5211                         p->status = DIFF_STATUS_MODIFIED;
5212                 else {
5213                         /* This is a "no-change" entry and should not
5214                          * happen anymore, but prepare for broken callers.
5215                          */
5216                         error("feeding unmodified %s to diffcore",
5217                               p->one->path);
5218                         p->status = DIFF_STATUS_UNKNOWN;
5219                 }
5220         }
5221         diff_debug_queue("resolve-rename-copy done", q);
5222 }
5223
5224 static int check_pair_status(struct diff_filepair *p)
5225 {
5226         switch (p->status) {
5227         case DIFF_STATUS_UNKNOWN:
5228                 return 0;
5229         case 0:
5230                 die("internal error in diff-resolve-rename-copy");
5231         default:
5232                 return 1;
5233         }
5234 }
5235
5236 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5237 {
5238         int fmt = opt->output_format;
5239
5240         if (fmt & DIFF_FORMAT_CHECKDIFF)
5241                 diff_flush_checkdiff(p, opt);
5242         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5243                 diff_flush_raw(p, opt);
5244         else if (fmt & DIFF_FORMAT_NAME) {
5245                 const char *name_a, *name_b;
5246                 name_a = p->two->path;
5247                 name_b = NULL;
5248                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5249                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5250                 write_name_quoted(name_a, opt->file, opt->line_termination);
5251         }
5252 }
5253
5254 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5255 {
5256         struct strbuf sb = STRBUF_INIT;
5257         if (fs->mode)
5258                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5259         else
5260                 strbuf_addf(&sb, " %s ", newdelete);
5261
5262         quote_c_style(fs->path, &sb, NULL, 0);
5263         strbuf_addch(&sb, '\n');
5264         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5265                          sb.buf, sb.len, 0);
5266         strbuf_release(&sb);
5267 }
5268
5269 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5270                 int show_name)
5271 {
5272         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5273                 struct strbuf sb = STRBUF_INIT;
5274                 strbuf_addf(&sb, " mode change %06o => %06o",
5275                             p->one->mode, p->two->mode);
5276                 if (show_name) {
5277                         strbuf_addch(&sb, ' ');
5278                         quote_c_style(p->two->path, &sb, NULL, 0);
5279                 }
5280                 strbuf_addch(&sb, '\n');
5281                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5282                                  sb.buf, sb.len, 0);
5283                 strbuf_release(&sb);
5284         }
5285 }
5286
5287 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5288                 struct diff_filepair *p)
5289 {
5290         struct strbuf sb = STRBUF_INIT;
5291         char *names = pprint_rename(p->one->path, p->two->path);
5292         strbuf_addf(&sb, " %s %s (%d%%)\n",
5293                         renamecopy, names, similarity_index(p));
5294         free(names);
5295         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5296                                  sb.buf, sb.len, 0);
5297         show_mode_change(opt, p, 0);
5298         strbuf_release(&sb);
5299 }
5300
5301 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5302 {
5303         switch(p->status) {
5304         case DIFF_STATUS_DELETED:
5305                 show_file_mode_name(opt, "delete", p->one);
5306                 break;
5307         case DIFF_STATUS_ADDED:
5308                 show_file_mode_name(opt, "create", p->two);
5309                 break;
5310         case DIFF_STATUS_COPIED:
5311                 show_rename_copy(opt, "copy", p);
5312                 break;
5313         case DIFF_STATUS_RENAMED:
5314                 show_rename_copy(opt, "rename", p);
5315                 break;
5316         default:
5317                 if (p->score) {
5318                         struct strbuf sb = STRBUF_INIT;
5319                         strbuf_addstr(&sb, " rewrite ");
5320                         quote_c_style(p->two->path, &sb, NULL, 0);
5321                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5322                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5323                                          sb.buf, sb.len, 0);
5324                         strbuf_release(&sb);
5325                 }
5326                 show_mode_change(opt, p, !p->score);
5327                 break;
5328         }
5329 }
5330
5331 struct patch_id_t {
5332         git_SHA_CTX *ctx;
5333         int patchlen;
5334 };
5335
5336 static int remove_space(char *line, int len)
5337 {
5338         int i;
5339         char *dst = line;
5340         unsigned char c;
5341
5342         for (i = 0; i < len; i++)
5343                 if (!isspace((c = line[i])))
5344                         *dst++ = c;
5345
5346         return dst - line;
5347 }
5348
5349 static void patch_id_consume(void *priv, char *line, unsigned long len)
5350 {
5351         struct patch_id_t *data = priv;
5352         int new_len;
5353
5354         /* Ignore line numbers when computing the SHA1 of the patch */
5355         if (starts_with(line, "@@ -"))
5356                 return;
5357
5358         new_len = remove_space(line, len);
5359
5360         git_SHA1_Update(data->ctx, line, new_len);
5361         data->patchlen += new_len;
5362 }
5363
5364 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5365 {
5366         git_SHA1_Update(ctx, str, strlen(str));
5367 }
5368
5369 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5370 {
5371         /* large enough for 2^32 in octal */
5372         char buf[12];
5373         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5374         git_SHA1_Update(ctx, buf, len);
5375 }
5376
5377 /* returns 0 upon success, and writes result into sha1 */
5378 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5379 {
5380         struct diff_queue_struct *q = &diff_queued_diff;
5381         int i;
5382         git_SHA_CTX ctx;
5383         struct patch_id_t data;
5384
5385         git_SHA1_Init(&ctx);
5386         memset(&data, 0, sizeof(struct patch_id_t));
5387         data.ctx = &ctx;
5388
5389         for (i = 0; i < q->nr; i++) {
5390                 xpparam_t xpp;
5391                 xdemitconf_t xecfg;
5392                 mmfile_t mf1, mf2;
5393                 struct diff_filepair *p = q->queue[i];
5394                 int len1, len2;
5395
5396                 memset(&xpp, 0, sizeof(xpp));
5397                 memset(&xecfg, 0, sizeof(xecfg));
5398                 if (p->status == 0)
5399                         return error("internal diff status error");
5400                 if (p->status == DIFF_STATUS_UNKNOWN)
5401                         continue;
5402                 if (diff_unmodified_pair(p))
5403                         continue;
5404                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5405                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5406                         continue;
5407                 if (DIFF_PAIR_UNMERGED(p))
5408                         continue;
5409
5410                 diff_fill_oid_info(p->one);
5411                 diff_fill_oid_info(p->two);
5412
5413                 len1 = remove_space(p->one->path, strlen(p->one->path));
5414                 len2 = remove_space(p->two->path, strlen(p->two->path));
5415                 patch_id_add_string(&ctx, "diff--git");
5416                 patch_id_add_string(&ctx, "a/");
5417                 git_SHA1_Update(&ctx, p->one->path, len1);
5418                 patch_id_add_string(&ctx, "b/");
5419                 git_SHA1_Update(&ctx, p->two->path, len2);
5420
5421                 if (p->one->mode == 0) {
5422                         patch_id_add_string(&ctx, "newfilemode");
5423                         patch_id_add_mode(&ctx, p->two->mode);
5424                         patch_id_add_string(&ctx, "---/dev/null");
5425                         patch_id_add_string(&ctx, "+++b/");
5426                         git_SHA1_Update(&ctx, p->two->path, len2);
5427                 } else if (p->two->mode == 0) {
5428                         patch_id_add_string(&ctx, "deletedfilemode");
5429                         patch_id_add_mode(&ctx, p->one->mode);
5430                         patch_id_add_string(&ctx, "---a/");
5431                         git_SHA1_Update(&ctx, p->one->path, len1);
5432                         patch_id_add_string(&ctx, "+++/dev/null");
5433                 } else {
5434                         patch_id_add_string(&ctx, "---a/");
5435                         git_SHA1_Update(&ctx, p->one->path, len1);
5436                         patch_id_add_string(&ctx, "+++b/");
5437                         git_SHA1_Update(&ctx, p->two->path, len2);
5438                 }
5439
5440                 if (diff_header_only)
5441                         continue;
5442
5443                 if (fill_mmfile(&mf1, p->one) < 0 ||
5444                     fill_mmfile(&mf2, p->two) < 0)
5445                         return error("unable to read files to diff");
5446
5447                 if (diff_filespec_is_binary(p->one) ||
5448                     diff_filespec_is_binary(p->two)) {
5449                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5450                                         GIT_SHA1_HEXSZ);
5451                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5452                                         GIT_SHA1_HEXSZ);
5453                         continue;
5454                 }
5455
5456                 xpp.flags = 0;
5457                 xecfg.ctxlen = 3;
5458                 xecfg.flags = 0;
5459                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5460                                   &xpp, &xecfg))
5461                         return error("unable to generate patch-id diff for %s",
5462                                      p->one->path);
5463         }
5464
5465         git_SHA1_Final(oid->hash, &ctx);
5466         return 0;
5467 }
5468
5469 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5470 {
5471         struct diff_queue_struct *q = &diff_queued_diff;
5472         int i;
5473         int result = diff_get_patch_id(options, oid, diff_header_only);
5474
5475         for (i = 0; i < q->nr; i++)
5476                 diff_free_filepair(q->queue[i]);
5477
5478         free(q->queue);
5479         DIFF_QUEUE_CLEAR(q);
5480
5481         return result;
5482 }
5483
5484 static int is_summary_empty(const struct diff_queue_struct *q)
5485 {
5486         int i;
5487
5488         for (i = 0; i < q->nr; i++) {
5489                 const struct diff_filepair *p = q->queue[i];
5490
5491                 switch (p->status) {
5492                 case DIFF_STATUS_DELETED:
5493                 case DIFF_STATUS_ADDED:
5494                 case DIFF_STATUS_COPIED:
5495                 case DIFF_STATUS_RENAMED:
5496                         return 0;
5497                 default:
5498                         if (p->score)
5499                                 return 0;
5500                         if (p->one->mode && p->two->mode &&
5501                             p->one->mode != p->two->mode)
5502                                 return 0;
5503                         break;
5504                 }
5505         }
5506         return 1;
5507 }
5508
5509 static const char rename_limit_warning[] =
5510 N_("inexact rename detection was skipped due to too many files.");
5511
5512 static const char degrade_cc_to_c_warning[] =
5513 N_("only found copies from modified paths due to too many files.");
5514
5515 static const char rename_limit_advice[] =
5516 N_("you may want to set your %s variable to at least "
5517    "%d and retry the command.");
5518
5519 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5520 {
5521         if (degraded_cc)
5522                 warning(_(degrade_cc_to_c_warning));
5523         else if (needed)
5524                 warning(_(rename_limit_warning));
5525         else
5526                 return;
5527         if (0 < needed && needed < 32767)
5528                 warning(_(rename_limit_advice), varname, needed);
5529 }
5530
5531 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5532 {
5533         int i;
5534         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5535         struct diff_queue_struct *q = &diff_queued_diff;
5536
5537         if (WSEH_NEW & WS_RULE_MASK)
5538                 die("BUG: WS rules bit mask overlaps with diff symbol flags");
5539
5540         if (o->color_moved)
5541                 o->emitted_symbols = &esm;
5542
5543         for (i = 0; i < q->nr; i++) {
5544                 struct diff_filepair *p = q->queue[i];
5545                 if (check_pair_status(p))
5546                         diff_flush_patch(p, o);
5547         }
5548
5549         if (o->emitted_symbols) {
5550                 if (o->color_moved) {
5551                         struct hashmap add_lines, del_lines;
5552
5553                         hashmap_init(&del_lines,
5554                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5555                         hashmap_init(&add_lines,
5556                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5557
5558                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5559                         mark_color_as_moved(o, &add_lines, &del_lines);
5560                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5561                                 dim_moved_lines(o);
5562
5563                         hashmap_free(&add_lines, 0);
5564                         hashmap_free(&del_lines, 0);
5565                 }
5566
5567                 for (i = 0; i < esm.nr; i++)
5568                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5569
5570                 for (i = 0; i < esm.nr; i++)
5571                         free((void *)esm.buf[i].line);
5572         }
5573         esm.nr = 0;
5574 }
5575
5576 void diff_flush(struct diff_options *options)
5577 {
5578         struct diff_queue_struct *q = &diff_queued_diff;
5579         int i, output_format = options->output_format;
5580         int separator = 0;
5581         int dirstat_by_line = 0;
5582
5583         /*
5584          * Order: raw, stat, summary, patch
5585          * or:    name/name-status/checkdiff (other bits clear)
5586          */
5587         if (!q->nr)
5588                 goto free_queue;
5589
5590         if (output_format & (DIFF_FORMAT_RAW |
5591                              DIFF_FORMAT_NAME |
5592                              DIFF_FORMAT_NAME_STATUS |
5593                              DIFF_FORMAT_CHECKDIFF)) {
5594                 for (i = 0; i < q->nr; i++) {
5595                         struct diff_filepair *p = q->queue[i];
5596                         if (check_pair_status(p))
5597                                 flush_one_pair(p, options);
5598                 }
5599                 separator++;
5600         }
5601
5602         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
5603                 dirstat_by_line = 1;
5604
5605         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5606             dirstat_by_line) {
5607                 struct diffstat_t diffstat;
5608
5609                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5610                 for (i = 0; i < q->nr; i++) {
5611                         struct diff_filepair *p = q->queue[i];
5612                         if (check_pair_status(p))
5613                                 diff_flush_stat(p, options, &diffstat);
5614                 }
5615                 if (output_format & DIFF_FORMAT_NUMSTAT)
5616                         show_numstat(&diffstat, options);
5617                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5618                         show_stats(&diffstat, options);
5619                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5620                         show_shortstats(&diffstat, options);
5621                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5622                         show_dirstat_by_line(&diffstat, options);
5623                 free_diffstat_info(&diffstat);
5624                 separator++;
5625         }
5626         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5627                 show_dirstat(options);
5628
5629         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5630                 for (i = 0; i < q->nr; i++) {
5631                         diff_summary(options, q->queue[i]);
5632                 }
5633                 separator++;
5634         }
5635
5636         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5637             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
5638             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
5639                 /*
5640                  * run diff_flush_patch for the exit status. setting
5641                  * options->file to /dev/null should be safe, because we
5642                  * aren't supposed to produce any output anyway.
5643                  */
5644                 if (options->close_file)
5645                         fclose(options->file);
5646                 options->file = xfopen("/dev/null", "w");
5647                 options->close_file = 1;
5648                 options->color_moved = 0;
5649                 for (i = 0; i < q->nr; i++) {
5650                         struct diff_filepair *p = q->queue[i];
5651                         if (check_pair_status(p))
5652                                 diff_flush_patch(p, options);
5653                         if (options->found_changes)
5654                                 break;
5655                 }
5656         }
5657
5658         if (output_format & DIFF_FORMAT_PATCH) {
5659                 if (separator) {
5660                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5661                         if (options->stat_sep)
5662                                 /* attach patch instead of inline */
5663                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5664                                                  NULL, 0, 0);
5665                 }
5666
5667                 diff_flush_patch_all_file_pairs(options);
5668         }
5669
5670         if (output_format & DIFF_FORMAT_CALLBACK)
5671                 options->format_callback(q, options, options->format_callback_data);
5672
5673         for (i = 0; i < q->nr; i++)
5674                 diff_free_filepair(q->queue[i]);
5675 free_queue:
5676         free(q->queue);
5677         DIFF_QUEUE_CLEAR(q);
5678         if (options->close_file)
5679                 fclose(options->file);
5680
5681         /*
5682          * Report the content-level differences with HAS_CHANGES;
5683          * diff_addremove/diff_change does not set the bit when
5684          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5685          */
5686         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
5687                 if (options->found_changes)
5688                         DIFF_OPT_SET(options, HAS_CHANGES);
5689                 else
5690                         DIFF_OPT_CLR(options, HAS_CHANGES);
5691         }
5692 }
5693
5694 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5695 {
5696         return (((p->status == DIFF_STATUS_MODIFIED) &&
5697                  ((p->score &&
5698                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5699                   (!p->score &&
5700                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5701                 ((p->status != DIFF_STATUS_MODIFIED) &&
5702                  filter_bit_tst(p->status, options)));
5703 }
5704
5705 static void diffcore_apply_filter(struct diff_options *options)
5706 {
5707         int i;
5708         struct diff_queue_struct *q = &diff_queued_diff;
5709         struct diff_queue_struct outq;
5710
5711         DIFF_QUEUE_CLEAR(&outq);
5712
5713         if (!options->filter)
5714                 return;
5715
5716         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5717                 int found;
5718                 for (i = found = 0; !found && i < q->nr; i++) {
5719                         if (match_filter(options, q->queue[i]))
5720                                 found++;
5721                 }
5722                 if (found)
5723                         return;
5724
5725                 /* otherwise we will clear the whole queue
5726                  * by copying the empty outq at the end of this
5727                  * function, but first clear the current entries
5728                  * in the queue.
5729                  */
5730                 for (i = 0; i < q->nr; i++)
5731                         diff_free_filepair(q->queue[i]);
5732         }
5733         else {
5734                 /* Only the matching ones */
5735                 for (i = 0; i < q->nr; i++) {
5736                         struct diff_filepair *p = q->queue[i];
5737                         if (match_filter(options, p))
5738                                 diff_q(&outq, p);
5739                         else
5740                                 diff_free_filepair(p);
5741                 }
5742         }
5743         free(q->queue);
5744         *q = outq;
5745 }
5746
5747 /* Check whether two filespecs with the same mode and size are identical */
5748 static int diff_filespec_is_identical(struct diff_filespec *one,
5749                                       struct diff_filespec *two)
5750 {
5751         if (S_ISGITLINK(one->mode))
5752                 return 0;
5753         if (diff_populate_filespec(one, 0))
5754                 return 0;
5755         if (diff_populate_filespec(two, 0))
5756                 return 0;
5757         return !memcmp(one->data, two->data, one->size);
5758 }
5759
5760 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
5761 {
5762         if (p->done_skip_stat_unmatch)
5763                 return p->skip_stat_unmatch_result;
5764
5765         p->done_skip_stat_unmatch = 1;
5766         p->skip_stat_unmatch_result = 0;
5767         /*
5768          * 1. Entries that come from stat info dirtiness
5769          *    always have both sides (iow, not create/delete),
5770          *    one side of the object name is unknown, with
5771          *    the same mode and size.  Keep the ones that
5772          *    do not match these criteria.  They have real
5773          *    differences.
5774          *
5775          * 2. At this point, the file is known to be modified,
5776          *    with the same mode and size, and the object
5777          *    name of one side is unknown.  Need to inspect
5778          *    the identical contents.
5779          */
5780         if (!DIFF_FILE_VALID(p->one) || /* (1) */
5781             !DIFF_FILE_VALID(p->two) ||
5782             (p->one->oid_valid && p->two->oid_valid) ||
5783             (p->one->mode != p->two->mode) ||
5784             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
5785             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
5786             (p->one->size != p->two->size) ||
5787             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
5788                 p->skip_stat_unmatch_result = 1;
5789         return p->skip_stat_unmatch_result;
5790 }
5791
5792 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
5793 {
5794         int i;
5795         struct diff_queue_struct *q = &diff_queued_diff;
5796         struct diff_queue_struct outq;
5797         DIFF_QUEUE_CLEAR(&outq);
5798
5799         for (i = 0; i < q->nr; i++) {
5800                 struct diff_filepair *p = q->queue[i];
5801
5802                 if (diff_filespec_check_stat_unmatch(p))
5803                         diff_q(&outq, p);
5804                 else {
5805                         /*
5806                          * The caller can subtract 1 from skip_stat_unmatch
5807                          * to determine how many paths were dirty only
5808                          * due to stat info mismatch.
5809                          */
5810                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
5811                                 diffopt->skip_stat_unmatch++;
5812                         diff_free_filepair(p);
5813                 }
5814         }
5815         free(q->queue);
5816         *q = outq;
5817 }
5818
5819 static int diffnamecmp(const void *a_, const void *b_)
5820 {
5821         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
5822         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
5823         const char *name_a, *name_b;
5824
5825         name_a = a->one ? a->one->path : a->two->path;
5826         name_b = b->one ? b->one->path : b->two->path;
5827         return strcmp(name_a, name_b);
5828 }
5829
5830 void diffcore_fix_diff_index(struct diff_options *options)
5831 {
5832         struct diff_queue_struct *q = &diff_queued_diff;
5833         QSORT(q->queue, q->nr, diffnamecmp);
5834 }
5835
5836 void diffcore_std(struct diff_options *options)
5837 {
5838         /* NOTE please keep the following in sync with diff_tree_combined() */
5839         if (options->skip_stat_unmatch)
5840                 diffcore_skip_stat_unmatch(options);
5841         if (!options->found_follow) {
5842                 /* See try_to_follow_renames() in tree-diff.c */
5843                 if (options->break_opt != -1)
5844                         diffcore_break(options->break_opt);
5845                 if (options->detect_rename)
5846                         diffcore_rename(options);
5847                 if (options->break_opt != -1)
5848                         diffcore_merge_broken();
5849         }
5850         if (options->pickaxe)
5851                 diffcore_pickaxe(options);
5852         if (options->orderfile)
5853                 diffcore_order(options->orderfile);
5854         if (!options->found_follow)
5855                 /* See try_to_follow_renames() in tree-diff.c */
5856                 diff_resolve_rename_copy();
5857         diffcore_apply_filter(options);
5858
5859         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5860                 DIFF_OPT_SET(options, HAS_CHANGES);
5861         else
5862                 DIFF_OPT_CLR(options, HAS_CHANGES);
5863
5864         options->found_follow = 0;
5865 }
5866
5867 int diff_result_code(struct diff_options *opt, int status)
5868 {
5869         int result = 0;
5870
5871         diff_warn_rename_limit("diff.renameLimit",
5872                                opt->needed_rename_limit,
5873                                opt->degraded_cc_to_c);
5874         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5875             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5876                 return status;
5877         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5878             DIFF_OPT_TST(opt, HAS_CHANGES))
5879                 result |= 01;
5880         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5881             DIFF_OPT_TST(opt, CHECK_FAILED))
5882                 result |= 02;
5883         return result;
5884 }
5885
5886 int diff_can_quit_early(struct diff_options *opt)
5887 {
5888         return (DIFF_OPT_TST(opt, QUICK) &&
5889                 !opt->filter &&
5890                 DIFF_OPT_TST(opt, HAS_CHANGES));
5891 }
5892
5893 /*
5894  * Shall changes to this submodule be ignored?
5895  *
5896  * Submodule changes can be configured to be ignored separately for each path,
5897  * but that configuration can be overridden from the command line.
5898  */
5899 static int is_submodule_ignored(const char *path, struct diff_options *options)
5900 {
5901         int ignored = 0;
5902         unsigned orig_flags = options->flags;
5903         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
5904                 set_diffopt_flags_from_submodule_config(options, path);
5905         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
5906                 ignored = 1;
5907         options->flags = orig_flags;
5908         return ignored;
5909 }
5910
5911 void diff_addremove(struct diff_options *options,
5912                     int addremove, unsigned mode,
5913                     const struct object_id *oid,
5914                     int oid_valid,
5915                     const char *concatpath, unsigned dirty_submodule)
5916 {
5917         struct diff_filespec *one, *two;
5918
5919         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5920                 return;
5921
5922         /* This may look odd, but it is a preparation for
5923          * feeding "there are unchanged files which should
5924          * not produce diffs, but when you are doing copy
5925          * detection you would need them, so here they are"
5926          * entries to the diff-core.  They will be prefixed
5927          * with something like '=' or '*' (I haven't decided
5928          * which but should not make any difference).
5929          * Feeding the same new and old to diff_change()
5930          * also has the same effect.
5931          * Before the final output happens, they are pruned after
5932          * merged into rename/copy pairs as appropriate.
5933          */
5934         if (DIFF_OPT_TST(options, REVERSE_DIFF))
5935                 addremove = (addremove == '+' ? '-' :
5936                              addremove == '-' ? '+' : addremove);
5937
5938         if (options->prefix &&
5939             strncmp(concatpath, options->prefix, options->prefix_length))
5940                 return;
5941
5942         one = alloc_filespec(concatpath);
5943         two = alloc_filespec(concatpath);
5944
5945         if (addremove != '+')
5946                 fill_filespec(one, oid, oid_valid, mode);
5947         if (addremove != '-') {
5948                 fill_filespec(two, oid, oid_valid, mode);
5949                 two->dirty_submodule = dirty_submodule;
5950         }
5951
5952         diff_queue(&diff_queued_diff, one, two);
5953         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5954                 DIFF_OPT_SET(options, HAS_CHANGES);
5955 }
5956
5957 void diff_change(struct diff_options *options,
5958                  unsigned old_mode, unsigned new_mode,
5959                  const struct object_id *old_oid,
5960                  const struct object_id *new_oid,
5961                  int old_oid_valid, int new_oid_valid,
5962                  const char *concatpath,
5963                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5964 {
5965         struct diff_filespec *one, *two;
5966         struct diff_filepair *p;
5967
5968         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5969             is_submodule_ignored(concatpath, options))
5970                 return;
5971
5972         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5973                 SWAP(old_mode, new_mode);
5974                 SWAP(old_oid, new_oid);
5975                 SWAP(old_oid_valid, new_oid_valid);
5976                 SWAP(old_dirty_submodule, new_dirty_submodule);
5977         }
5978
5979         if (options->prefix &&
5980             strncmp(concatpath, options->prefix, options->prefix_length))
5981                 return;
5982
5983         one = alloc_filespec(concatpath);
5984         two = alloc_filespec(concatpath);
5985         fill_filespec(one, old_oid, old_oid_valid, old_mode);
5986         fill_filespec(two, new_oid, new_oid_valid, new_mode);
5987         one->dirty_submodule = old_dirty_submodule;
5988         two->dirty_submodule = new_dirty_submodule;
5989         p = diff_queue(&diff_queued_diff, one, two);
5990
5991         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5992                 return;
5993
5994         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5995             !diff_filespec_check_stat_unmatch(p))
5996                 return;
5997
5998         DIFF_OPT_SET(options, HAS_CHANGES);
5999 }
6000
6001 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6002 {
6003         struct diff_filepair *pair;
6004         struct diff_filespec *one, *two;
6005
6006         if (options->prefix &&
6007             strncmp(path, options->prefix, options->prefix_length))
6008                 return NULL;
6009
6010         one = alloc_filespec(path);
6011         two = alloc_filespec(path);
6012         pair = diff_queue(&diff_queued_diff, one, two);
6013         pair->is_unmerged = 1;
6014         return pair;
6015 }
6016
6017 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
6018                 size_t *outsize)
6019 {
6020         struct diff_tempfile *temp;
6021         const char *argv[3];
6022         const char **arg = argv;
6023         struct child_process child = CHILD_PROCESS_INIT;
6024         struct strbuf buf = STRBUF_INIT;
6025         int err = 0;
6026
6027         temp = prepare_temp_file(spec->path, spec);
6028         *arg++ = pgm;
6029         *arg++ = temp->name;
6030         *arg = NULL;
6031
6032         child.use_shell = 1;
6033         child.argv = argv;
6034         child.out = -1;
6035         if (start_command(&child)) {
6036                 remove_tempfile();
6037                 return NULL;
6038         }
6039
6040         if (strbuf_read(&buf, child.out, 0) < 0)
6041                 err = error("error reading from textconv command '%s'", pgm);
6042         close(child.out);
6043
6044         if (finish_command(&child) || err) {
6045                 strbuf_release(&buf);
6046                 remove_tempfile();
6047                 return NULL;
6048         }
6049         remove_tempfile();
6050
6051         return strbuf_detach(&buf, outsize);
6052 }
6053
6054 size_t fill_textconv(struct userdiff_driver *driver,
6055                      struct diff_filespec *df,
6056                      char **outbuf)
6057 {
6058         size_t size;
6059
6060         if (!driver) {
6061                 if (!DIFF_FILE_VALID(df)) {
6062                         *outbuf = "";
6063                         return 0;
6064                 }
6065                 if (diff_populate_filespec(df, 0))
6066                         die("unable to read files to diff");
6067                 *outbuf = df->data;
6068                 return df->size;
6069         }
6070
6071         if (!driver->textconv)
6072                 die("BUG: fill_textconv called with non-textconv driver");
6073
6074         if (driver->textconv_cache && df->oid_valid) {
6075                 *outbuf = notes_cache_get(driver->textconv_cache,
6076                                           &df->oid,
6077                                           &size);
6078                 if (*outbuf)
6079                         return size;
6080         }
6081
6082         *outbuf = run_textconv(driver->textconv, df, &size);
6083         if (!*outbuf)
6084                 die("unable to read files to diff");
6085
6086         if (driver->textconv_cache && df->oid_valid) {
6087                 /* ignore errors, as we might be in a readonly repository */
6088                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6089                                 size);
6090                 /*
6091                  * we could save up changes and flush them all at the end,
6092                  * but we would need an extra call after all diffing is done.
6093                  * Since generating a cache entry is the slow path anyway,
6094                  * this extra overhead probably isn't a big deal.
6095                  */
6096                 notes_cache_write(driver->textconv_cache);
6097         }
6098
6099         return size;
6100 }
6101
6102 int textconv_object(const char *path,
6103                     unsigned mode,
6104                     const struct object_id *oid,
6105                     int oid_valid,
6106                     char **buf,
6107                     unsigned long *buf_size)
6108 {
6109         struct diff_filespec *df;
6110         struct userdiff_driver *textconv;
6111
6112         df = alloc_filespec(path);
6113         fill_filespec(df, oid, oid_valid, mode);
6114         textconv = get_textconv(df);
6115         if (!textconv) {
6116                 free_filespec(df);
6117                 return 0;
6118         }
6119
6120         *buf_size = fill_textconv(textconv, df, buf);
6121         free_filespec(df);
6122         return 1;
6123 }
6124
6125 void setup_diff_pager(struct diff_options *opt)
6126 {
6127         /*
6128          * If the user asked for our exit code, then either they want --quiet
6129          * or --exit-code. We should definitely not bother with a pager in the
6130          * former case, as we will generate no output. Since we still properly
6131          * report our exit code even when a pager is run, we _could_ run a
6132          * pager with --exit-code. But since we have not done so historically,
6133          * and because it is easy to find people oneline advising "git diff
6134          * --exit-code" in hooks and other scripts, we do not do so.
6135          */
6136         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
6137             check_pager_config("diff") != 0)
6138                 setup_pager();
6139 }