Merge branch 'va/i18n-perl-scripts'
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "tempfile.h"
6 #include "quote.h"
7 #include "diff.h"
8 #include "diffcore.h"
9 #include "delta.h"
10 #include "xdiff-interface.h"
11 #include "color.h"
12 #include "attr.h"
13 #include "run-command.h"
14 #include "utf8.h"
15 #include "userdiff.h"
16 #include "submodule-config.h"
17 #include "submodule.h"
18 #include "ll-merge.h"
19 #include "string-list.h"
20 #include "argv-array.h"
21 #include "graph.h"
22
23 #ifdef NO_FAST_WORKING_DIRECTORY
24 #define FAST_WORKING_DIRECTORY 0
25 #else
26 #define FAST_WORKING_DIRECTORY 1
27 #endif
28
29 static int diff_detect_rename_default;
30 static int diff_indent_heuristic; /* experimental */
31 static int diff_compaction_heuristic; /* experimental */
32 static int diff_rename_limit_default = 400;
33 static int diff_suppress_blank_empty;
34 static int diff_use_color_default = -1;
35 static int diff_context_default = 3;
36 static const char *diff_word_regex_cfg;
37 static const char *external_diff_cmd_cfg;
38 static const char *diff_order_file_cfg;
39 int diff_auto_refresh_index = 1;
40 static int diff_mnemonic_prefix;
41 static int diff_no_prefix;
42 static int diff_stat_graph_width;
43 static int diff_dirstat_permille_default = 30;
44 static struct diff_options default_diff_options;
45 static long diff_algorithm;
46 static unsigned ws_error_highlight_default = WSEH_NEW;
47
48 static char diff_colors[][COLOR_MAXLEN] = {
49         GIT_COLOR_RESET,
50         GIT_COLOR_NORMAL,       /* CONTEXT */
51         GIT_COLOR_BOLD,         /* METAINFO */
52         GIT_COLOR_CYAN,         /* FRAGINFO */
53         GIT_COLOR_RED,          /* OLD */
54         GIT_COLOR_GREEN,        /* NEW */
55         GIT_COLOR_YELLOW,       /* COMMIT */
56         GIT_COLOR_BG_RED,       /* WHITESPACE */
57         GIT_COLOR_NORMAL,       /* FUNCINFO */
58 };
59
60 static NORETURN void die_want_option(const char *option_name)
61 {
62         die(_("option '%s' requires a value"), option_name);
63 }
64
65 static int parse_diff_color_slot(const char *var)
66 {
67         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
68                 return DIFF_CONTEXT;
69         if (!strcasecmp(var, "meta"))
70                 return DIFF_METAINFO;
71         if (!strcasecmp(var, "frag"))
72                 return DIFF_FRAGINFO;
73         if (!strcasecmp(var, "old"))
74                 return DIFF_FILE_OLD;
75         if (!strcasecmp(var, "new"))
76                 return DIFF_FILE_NEW;
77         if (!strcasecmp(var, "commit"))
78                 return DIFF_COMMIT;
79         if (!strcasecmp(var, "whitespace"))
80                 return DIFF_WHITESPACE;
81         if (!strcasecmp(var, "func"))
82                 return DIFF_FUNCINFO;
83         return -1;
84 }
85
86 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
87                                 struct strbuf *errmsg)
88 {
89         char *params_copy = xstrdup(params_string);
90         struct string_list params = STRING_LIST_INIT_NODUP;
91         int ret = 0;
92         int i;
93
94         if (*params_copy)
95                 string_list_split_in_place(&params, params_copy, ',', -1);
96         for (i = 0; i < params.nr; i++) {
97                 const char *p = params.items[i].string;
98                 if (!strcmp(p, "changes")) {
99                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
100                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
101                 } else if (!strcmp(p, "lines")) {
102                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
103                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
104                 } else if (!strcmp(p, "files")) {
105                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
106                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
107                 } else if (!strcmp(p, "noncumulative")) {
108                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
109                 } else if (!strcmp(p, "cumulative")) {
110                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
111                 } else if (isdigit(*p)) {
112                         char *end;
113                         int permille = strtoul(p, &end, 10) * 10;
114                         if (*end == '.' && isdigit(*++end)) {
115                                 /* only use first digit */
116                                 permille += *end - '0';
117                                 /* .. and ignore any further digits */
118                                 while (isdigit(*++end))
119                                         ; /* nothing */
120                         }
121                         if (!*end)
122                                 options->dirstat_permille = permille;
123                         else {
124                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
125                                             p);
126                                 ret++;
127                         }
128                 } else {
129                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
130                         ret++;
131                 }
132
133         }
134         string_list_clear(&params, 0);
135         free(params_copy);
136         return ret;
137 }
138
139 static int parse_submodule_params(struct diff_options *options, const char *value)
140 {
141         if (!strcmp(value, "log"))
142                 options->submodule_format = DIFF_SUBMODULE_LOG;
143         else if (!strcmp(value, "short"))
144                 options->submodule_format = DIFF_SUBMODULE_SHORT;
145         else if (!strcmp(value, "diff"))
146                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
147         else
148                 return -1;
149         return 0;
150 }
151
152 static int git_config_rename(const char *var, const char *value)
153 {
154         if (!value)
155                 return DIFF_DETECT_RENAME;
156         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
157                 return  DIFF_DETECT_COPY;
158         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
159 }
160
161 long parse_algorithm_value(const char *value)
162 {
163         if (!value)
164                 return -1;
165         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
166                 return 0;
167         else if (!strcasecmp(value, "minimal"))
168                 return XDF_NEED_MINIMAL;
169         else if (!strcasecmp(value, "patience"))
170                 return XDF_PATIENCE_DIFF;
171         else if (!strcasecmp(value, "histogram"))
172                 return XDF_HISTOGRAM_DIFF;
173         return -1;
174 }
175
176 static int parse_one_token(const char **arg, const char *token)
177 {
178         const char *rest;
179         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
180                 *arg = rest;
181                 return 1;
182         }
183         return 0;
184 }
185
186 static int parse_ws_error_highlight(const char *arg)
187 {
188         const char *orig_arg = arg;
189         unsigned val = 0;
190
191         while (*arg) {
192                 if (parse_one_token(&arg, "none"))
193                         val = 0;
194                 else if (parse_one_token(&arg, "default"))
195                         val = WSEH_NEW;
196                 else if (parse_one_token(&arg, "all"))
197                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
198                 else if (parse_one_token(&arg, "new"))
199                         val |= WSEH_NEW;
200                 else if (parse_one_token(&arg, "old"))
201                         val |= WSEH_OLD;
202                 else if (parse_one_token(&arg, "context"))
203                         val |= WSEH_CONTEXT;
204                 else {
205                         return -1 - (int)(arg - orig_arg);
206                 }
207                 if (*arg)
208                         arg++;
209         }
210         return val;
211 }
212
213 /*
214  * These are to give UI layer defaults.
215  * The core-level commands such as git-diff-files should
216  * never be affected by the setting of diff.renames
217  * the user happens to have in the configuration file.
218  */
219 void init_diff_ui_defaults(void)
220 {
221         diff_detect_rename_default = 1;
222 }
223
224 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
225 {
226         if (!strcmp(var, "diff.indentheuristic")) {
227                 diff_indent_heuristic = git_config_bool(var, value);
228                 if (diff_indent_heuristic)
229                         diff_compaction_heuristic = 0;
230         }
231         if (!strcmp(var, "diff.compactionheuristic")) {
232                 diff_compaction_heuristic = git_config_bool(var, value);
233                 if (diff_compaction_heuristic)
234                         diff_indent_heuristic = 0;
235         }
236         return 0;
237 }
238
239 int git_diff_ui_config(const char *var, const char *value, void *cb)
240 {
241         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
242                 diff_use_color_default = git_config_colorbool(var, value);
243                 return 0;
244         }
245         if (!strcmp(var, "diff.context")) {
246                 diff_context_default = git_config_int(var, value);
247                 if (diff_context_default < 0)
248                         return -1;
249                 return 0;
250         }
251         if (!strcmp(var, "diff.renames")) {
252                 diff_detect_rename_default = git_config_rename(var, value);
253                 return 0;
254         }
255         if (!strcmp(var, "diff.autorefreshindex")) {
256                 diff_auto_refresh_index = git_config_bool(var, value);
257                 return 0;
258         }
259         if (!strcmp(var, "diff.mnemonicprefix")) {
260                 diff_mnemonic_prefix = git_config_bool(var, value);
261                 return 0;
262         }
263         if (!strcmp(var, "diff.noprefix")) {
264                 diff_no_prefix = git_config_bool(var, value);
265                 return 0;
266         }
267         if (!strcmp(var, "diff.statgraphwidth")) {
268                 diff_stat_graph_width = git_config_int(var, value);
269                 return 0;
270         }
271         if (!strcmp(var, "diff.external"))
272                 return git_config_string(&external_diff_cmd_cfg, var, value);
273         if (!strcmp(var, "diff.wordregex"))
274                 return git_config_string(&diff_word_regex_cfg, var, value);
275         if (!strcmp(var, "diff.orderfile"))
276                 return git_config_pathname(&diff_order_file_cfg, var, value);
277
278         if (!strcmp(var, "diff.ignoresubmodules"))
279                 handle_ignore_submodules_arg(&default_diff_options, value);
280
281         if (!strcmp(var, "diff.submodule")) {
282                 if (parse_submodule_params(&default_diff_options, value))
283                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
284                                 value);
285                 return 0;
286         }
287
288         if (!strcmp(var, "diff.algorithm")) {
289                 diff_algorithm = parse_algorithm_value(value);
290                 if (diff_algorithm < 0)
291                         return -1;
292                 return 0;
293         }
294
295         if (git_diff_heuristic_config(var, value, cb) < 0)
296                 return -1;
297
298         if (!strcmp(var, "diff.wserrorhighlight")) {
299                 int val = parse_ws_error_highlight(value);
300                 if (val < 0)
301                         return -1;
302                 ws_error_highlight_default = val;
303                 return 0;
304         }
305
306         if (git_color_config(var, value, cb) < 0)
307                 return -1;
308
309         return git_diff_basic_config(var, value, cb);
310 }
311
312 int git_diff_basic_config(const char *var, const char *value, void *cb)
313 {
314         const char *name;
315
316         if (!strcmp(var, "diff.renamelimit")) {
317                 diff_rename_limit_default = git_config_int(var, value);
318                 return 0;
319         }
320
321         if (userdiff_config(var, value) < 0)
322                 return -1;
323
324         if (skip_prefix(var, "diff.color.", &name) ||
325             skip_prefix(var, "color.diff.", &name)) {
326                 int slot = parse_diff_color_slot(name);
327                 if (slot < 0)
328                         return 0;
329                 if (!value)
330                         return config_error_nonbool(var);
331                 return color_parse(value, diff_colors[slot]);
332         }
333
334         /* like GNU diff's --suppress-blank-empty option  */
335         if (!strcmp(var, "diff.suppressblankempty") ||
336                         /* for backwards compatibility */
337                         !strcmp(var, "diff.suppress-blank-empty")) {
338                 diff_suppress_blank_empty = git_config_bool(var, value);
339                 return 0;
340         }
341
342         if (!strcmp(var, "diff.dirstat")) {
343                 struct strbuf errmsg = STRBUF_INIT;
344                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
345                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
346                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
347                                 errmsg.buf);
348                 strbuf_release(&errmsg);
349                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
350                 return 0;
351         }
352
353         if (starts_with(var, "submodule."))
354                 return parse_submodule_config_option(var, value);
355
356         return git_default_config(var, value, cb);
357 }
358
359 static char *quote_two(const char *one, const char *two)
360 {
361         int need_one = quote_c_style(one, NULL, NULL, 1);
362         int need_two = quote_c_style(two, NULL, NULL, 1);
363         struct strbuf res = STRBUF_INIT;
364
365         if (need_one + need_two) {
366                 strbuf_addch(&res, '"');
367                 quote_c_style(one, &res, NULL, 1);
368                 quote_c_style(two, &res, NULL, 1);
369                 strbuf_addch(&res, '"');
370         } else {
371                 strbuf_addstr(&res, one);
372                 strbuf_addstr(&res, two);
373         }
374         return strbuf_detach(&res, NULL);
375 }
376
377 static const char *external_diff(void)
378 {
379         static const char *external_diff_cmd = NULL;
380         static int done_preparing = 0;
381
382         if (done_preparing)
383                 return external_diff_cmd;
384         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
385         if (!external_diff_cmd)
386                 external_diff_cmd = external_diff_cmd_cfg;
387         done_preparing = 1;
388         return external_diff_cmd;
389 }
390
391 /*
392  * Keep track of files used for diffing. Sometimes such an entry
393  * refers to a temporary file, sometimes to an existing file, and
394  * sometimes to "/dev/null".
395  */
396 static struct diff_tempfile {
397         /*
398          * filename external diff should read from, or NULL if this
399          * entry is currently not in use:
400          */
401         const char *name;
402
403         char hex[GIT_SHA1_HEXSZ + 1];
404         char mode[10];
405
406         /*
407          * If this diff_tempfile instance refers to a temporary file,
408          * this tempfile object is used to manage its lifetime.
409          */
410         struct tempfile tempfile;
411 } diff_temp[2];
412
413 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
414
415 struct emit_callback {
416         int color_diff;
417         unsigned ws_rule;
418         int blank_at_eof_in_preimage;
419         int blank_at_eof_in_postimage;
420         int lno_in_preimage;
421         int lno_in_postimage;
422         sane_truncate_fn truncate;
423         const char **label_path;
424         struct diff_words_data *diff_words;
425         struct diff_options *opt;
426         struct strbuf *header;
427 };
428
429 static int count_lines(const char *data, int size)
430 {
431         int count, ch, completely_empty = 1, nl_just_seen = 0;
432         count = 0;
433         while (0 < size--) {
434                 ch = *data++;
435                 if (ch == '\n') {
436                         count++;
437                         nl_just_seen = 1;
438                         completely_empty = 0;
439                 }
440                 else {
441                         nl_just_seen = 0;
442                         completely_empty = 0;
443                 }
444         }
445         if (completely_empty)
446                 return 0;
447         if (!nl_just_seen)
448                 count++; /* no trailing newline */
449         return count;
450 }
451
452 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
453 {
454         if (!DIFF_FILE_VALID(one)) {
455                 mf->ptr = (char *)""; /* does not matter */
456                 mf->size = 0;
457                 return 0;
458         }
459         else if (diff_populate_filespec(one, 0))
460                 return -1;
461
462         mf->ptr = one->data;
463         mf->size = one->size;
464         return 0;
465 }
466
467 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
468 static unsigned long diff_filespec_size(struct diff_filespec *one)
469 {
470         if (!DIFF_FILE_VALID(one))
471                 return 0;
472         diff_populate_filespec(one, CHECK_SIZE_ONLY);
473         return one->size;
474 }
475
476 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
477 {
478         char *ptr = mf->ptr;
479         long size = mf->size;
480         int cnt = 0;
481
482         if (!size)
483                 return cnt;
484         ptr += size - 1; /* pointing at the very end */
485         if (*ptr != '\n')
486                 ; /* incomplete line */
487         else
488                 ptr--; /* skip the last LF */
489         while (mf->ptr < ptr) {
490                 char *prev_eol;
491                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
492                         if (*prev_eol == '\n')
493                                 break;
494                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
495                         break;
496                 cnt++;
497                 ptr = prev_eol - 1;
498         }
499         return cnt;
500 }
501
502 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
503                                struct emit_callback *ecbdata)
504 {
505         int l1, l2, at;
506         unsigned ws_rule = ecbdata->ws_rule;
507         l1 = count_trailing_blank(mf1, ws_rule);
508         l2 = count_trailing_blank(mf2, ws_rule);
509         if (l2 <= l1) {
510                 ecbdata->blank_at_eof_in_preimage = 0;
511                 ecbdata->blank_at_eof_in_postimage = 0;
512                 return;
513         }
514         at = count_lines(mf1->ptr, mf1->size);
515         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
516
517         at = count_lines(mf2->ptr, mf2->size);
518         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
519 }
520
521 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
522                         int first, const char *line, int len)
523 {
524         int has_trailing_newline, has_trailing_carriage_return;
525         int nofirst;
526         FILE *file = o->file;
527
528         fputs(diff_line_prefix(o), file);
529
530         if (len == 0) {
531                 has_trailing_newline = (first == '\n');
532                 has_trailing_carriage_return = (!has_trailing_newline &&
533                                                 (first == '\r'));
534                 nofirst = has_trailing_newline || has_trailing_carriage_return;
535         } else {
536                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
537                 if (has_trailing_newline)
538                         len--;
539                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
540                 if (has_trailing_carriage_return)
541                         len--;
542                 nofirst = 0;
543         }
544
545         if (len || !nofirst) {
546                 fputs(set, file);
547                 if (!nofirst)
548                         fputc(first, file);
549                 fwrite(line, len, 1, file);
550                 fputs(reset, file);
551         }
552         if (has_trailing_carriage_return)
553                 fputc('\r', file);
554         if (has_trailing_newline)
555                 fputc('\n', file);
556 }
557
558 static void emit_line(struct diff_options *o, const char *set, const char *reset,
559                       const char *line, int len)
560 {
561         emit_line_0(o, set, reset, line[0], line+1, len-1);
562 }
563
564 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
565 {
566         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
567               ecbdata->blank_at_eof_in_preimage &&
568               ecbdata->blank_at_eof_in_postimage &&
569               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
570               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
571                 return 0;
572         return ws_blank_line(line, len, ecbdata->ws_rule);
573 }
574
575 static void emit_line_checked(const char *reset,
576                               struct emit_callback *ecbdata,
577                               const char *line, int len,
578                               enum color_diff color,
579                               unsigned ws_error_highlight,
580                               char sign)
581 {
582         const char *set = diff_get_color(ecbdata->color_diff, color);
583         const char *ws = NULL;
584
585         if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
586                 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
587                 if (!*ws)
588                         ws = NULL;
589         }
590
591         if (!ws)
592                 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
593         else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
594                 /* Blank line at EOF - paint '+' as well */
595                 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
596         else {
597                 /* Emit just the prefix, then the rest. */
598                 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
599                 ws_check_emit(line, len, ecbdata->ws_rule,
600                               ecbdata->opt->file, set, reset, ws);
601         }
602 }
603
604 static void emit_add_line(const char *reset,
605                           struct emit_callback *ecbdata,
606                           const char *line, int len)
607 {
608         emit_line_checked(reset, ecbdata, line, len,
609                           DIFF_FILE_NEW, WSEH_NEW, '+');
610 }
611
612 static void emit_del_line(const char *reset,
613                           struct emit_callback *ecbdata,
614                           const char *line, int len)
615 {
616         emit_line_checked(reset, ecbdata, line, len,
617                           DIFF_FILE_OLD, WSEH_OLD, '-');
618 }
619
620 static void emit_context_line(const char *reset,
621                               struct emit_callback *ecbdata,
622                               const char *line, int len)
623 {
624         emit_line_checked(reset, ecbdata, line, len,
625                           DIFF_CONTEXT, WSEH_CONTEXT, ' ');
626 }
627
628 static void emit_hunk_header(struct emit_callback *ecbdata,
629                              const char *line, int len)
630 {
631         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
632         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
633         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
634         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
635         static const char atat[2] = { '@', '@' };
636         const char *cp, *ep;
637         struct strbuf msgbuf = STRBUF_INIT;
638         int org_len = len;
639         int i = 1;
640
641         /*
642          * As a hunk header must begin with "@@ -<old>, +<new> @@",
643          * it always is at least 10 bytes long.
644          */
645         if (len < 10 ||
646             memcmp(line, atat, 2) ||
647             !(ep = memmem(line + 2, len - 2, atat, 2))) {
648                 emit_line(ecbdata->opt, context, reset, line, len);
649                 return;
650         }
651         ep += 2; /* skip over @@ */
652
653         /* The hunk header in fraginfo color */
654         strbuf_addstr(&msgbuf, frag);
655         strbuf_add(&msgbuf, line, ep - line);
656         strbuf_addstr(&msgbuf, reset);
657
658         /*
659          * trailing "\r\n"
660          */
661         for ( ; i < 3; i++)
662                 if (line[len - i] == '\r' || line[len - i] == '\n')
663                         len--;
664
665         /* blank before the func header */
666         for (cp = ep; ep - line < len; ep++)
667                 if (*ep != ' ' && *ep != '\t')
668                         break;
669         if (ep != cp) {
670                 strbuf_addstr(&msgbuf, context);
671                 strbuf_add(&msgbuf, cp, ep - cp);
672                 strbuf_addstr(&msgbuf, reset);
673         }
674
675         if (ep < line + len) {
676                 strbuf_addstr(&msgbuf, func);
677                 strbuf_add(&msgbuf, ep, line + len - ep);
678                 strbuf_addstr(&msgbuf, reset);
679         }
680
681         strbuf_add(&msgbuf, line + len, org_len - len);
682         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
683         strbuf_release(&msgbuf);
684 }
685
686 static struct diff_tempfile *claim_diff_tempfile(void) {
687         int i;
688         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
689                 if (!diff_temp[i].name)
690                         return diff_temp + i;
691         die("BUG: diff is failing to clean up its tempfiles");
692 }
693
694 static void remove_tempfile(void)
695 {
696         int i;
697         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
698                 if (is_tempfile_active(&diff_temp[i].tempfile))
699                         delete_tempfile(&diff_temp[i].tempfile);
700                 diff_temp[i].name = NULL;
701         }
702 }
703
704 static void print_line_count(FILE *file, int count)
705 {
706         switch (count) {
707         case 0:
708                 fprintf(file, "0,0");
709                 break;
710         case 1:
711                 fprintf(file, "1");
712                 break;
713         default:
714                 fprintf(file, "1,%d", count);
715                 break;
716         }
717 }
718
719 static void emit_rewrite_lines(struct emit_callback *ecb,
720                                int prefix, const char *data, int size)
721 {
722         const char *endp = NULL;
723         static const char *nneof = " No newline at end of file\n";
724         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
725
726         while (0 < size) {
727                 int len;
728
729                 endp = memchr(data, '\n', size);
730                 len = endp ? (endp - data + 1) : size;
731                 if (prefix != '+') {
732                         ecb->lno_in_preimage++;
733                         emit_del_line(reset, ecb, data, len);
734                 } else {
735                         ecb->lno_in_postimage++;
736                         emit_add_line(reset, ecb, data, len);
737                 }
738                 size -= len;
739                 data += len;
740         }
741         if (!endp) {
742                 const char *context = diff_get_color(ecb->color_diff,
743                                                      DIFF_CONTEXT);
744                 putc('\n', ecb->opt->file);
745                 emit_line_0(ecb->opt, context, reset, '\\',
746                             nneof, strlen(nneof));
747         }
748 }
749
750 static void emit_rewrite_diff(const char *name_a,
751                               const char *name_b,
752                               struct diff_filespec *one,
753                               struct diff_filespec *two,
754                               struct userdiff_driver *textconv_one,
755                               struct userdiff_driver *textconv_two,
756                               struct diff_options *o)
757 {
758         int lc_a, lc_b;
759         const char *name_a_tab, *name_b_tab;
760         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
761         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
762         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
763         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
764         const char *a_prefix, *b_prefix;
765         char *data_one, *data_two;
766         size_t size_one, size_two;
767         struct emit_callback ecbdata;
768         const char *line_prefix = diff_line_prefix(o);
769
770         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
771                 a_prefix = o->b_prefix;
772                 b_prefix = o->a_prefix;
773         } else {
774                 a_prefix = o->a_prefix;
775                 b_prefix = o->b_prefix;
776         }
777
778         name_a += (*name_a == '/');
779         name_b += (*name_b == '/');
780         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
781         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
782
783         strbuf_reset(&a_name);
784         strbuf_reset(&b_name);
785         quote_two_c_style(&a_name, a_prefix, name_a, 0);
786         quote_two_c_style(&b_name, b_prefix, name_b, 0);
787
788         size_one = fill_textconv(textconv_one, one, &data_one);
789         size_two = fill_textconv(textconv_two, two, &data_two);
790
791         memset(&ecbdata, 0, sizeof(ecbdata));
792         ecbdata.color_diff = want_color(o->use_color);
793         ecbdata.ws_rule = whitespace_rule(name_b);
794         ecbdata.opt = o;
795         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
796                 mmfile_t mf1, mf2;
797                 mf1.ptr = (char *)data_one;
798                 mf2.ptr = (char *)data_two;
799                 mf1.size = size_one;
800                 mf2.size = size_two;
801                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
802         }
803         ecbdata.lno_in_preimage = 1;
804         ecbdata.lno_in_postimage = 1;
805
806         lc_a = count_lines(data_one, size_one);
807         lc_b = count_lines(data_two, size_two);
808         fprintf(o->file,
809                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
810                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
811                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
812                 line_prefix, fraginfo);
813         if (!o->irreversible_delete)
814                 print_line_count(o->file, lc_a);
815         else
816                 fprintf(o->file, "?,?");
817         fprintf(o->file, " +");
818         print_line_count(o->file, lc_b);
819         fprintf(o->file, " @@%s\n", reset);
820         if (lc_a && !o->irreversible_delete)
821                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
822         if (lc_b)
823                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
824         if (textconv_one)
825                 free((char *)data_one);
826         if (textconv_two)
827                 free((char *)data_two);
828 }
829
830 struct diff_words_buffer {
831         mmfile_t text;
832         long alloc;
833         struct diff_words_orig {
834                 const char *begin, *end;
835         } *orig;
836         int orig_nr, orig_alloc;
837 };
838
839 static void diff_words_append(char *line, unsigned long len,
840                 struct diff_words_buffer *buffer)
841 {
842         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
843         line++;
844         len--;
845         memcpy(buffer->text.ptr + buffer->text.size, line, len);
846         buffer->text.size += len;
847         buffer->text.ptr[buffer->text.size] = '\0';
848 }
849
850 struct diff_words_style_elem {
851         const char *prefix;
852         const char *suffix;
853         const char *color; /* NULL; filled in by the setup code if
854                             * color is enabled */
855 };
856
857 struct diff_words_style {
858         enum diff_words_type type;
859         struct diff_words_style_elem new, old, ctx;
860         const char *newline;
861 };
862
863 static struct diff_words_style diff_words_styles[] = {
864         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
865         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
866         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
867 };
868
869 struct diff_words_data {
870         struct diff_words_buffer minus, plus;
871         const char *current_plus;
872         int last_minus;
873         struct diff_options *opt;
874         regex_t *word_regex;
875         enum diff_words_type type;
876         struct diff_words_style *style;
877 };
878
879 static int fn_out_diff_words_write_helper(FILE *fp,
880                                           struct diff_words_style_elem *st_el,
881                                           const char *newline,
882                                           size_t count, const char *buf,
883                                           const char *line_prefix)
884 {
885         int print = 0;
886
887         while (count) {
888                 char *p = memchr(buf, '\n', count);
889                 if (print)
890                         fputs(line_prefix, fp);
891                 if (p != buf) {
892                         if (st_el->color && fputs(st_el->color, fp) < 0)
893                                 return -1;
894                         if (fputs(st_el->prefix, fp) < 0 ||
895                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
896                             fputs(st_el->suffix, fp) < 0)
897                                 return -1;
898                         if (st_el->color && *st_el->color
899                             && fputs(GIT_COLOR_RESET, fp) < 0)
900                                 return -1;
901                 }
902                 if (!p)
903                         return 0;
904                 if (fputs(newline, fp) < 0)
905                         return -1;
906                 count -= p + 1 - buf;
907                 buf = p + 1;
908                 print = 1;
909         }
910         return 0;
911 }
912
913 /*
914  * '--color-words' algorithm can be described as:
915  *
916  *   1. collect a the minus/plus lines of a diff hunk, divided into
917  *      minus-lines and plus-lines;
918  *
919  *   2. break both minus-lines and plus-lines into words and
920  *      place them into two mmfile_t with one word for each line;
921  *
922  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
923  *
924  * And for the common parts of the both file, we output the plus side text.
925  * diff_words->current_plus is used to trace the current position of the plus file
926  * which printed. diff_words->last_minus is used to trace the last minus word
927  * printed.
928  *
929  * For '--graph' to work with '--color-words', we need to output the graph prefix
930  * on each line of color words output. Generally, there are two conditions on
931  * which we should output the prefix.
932  *
933  *   1. diff_words->last_minus == 0 &&
934  *      diff_words->current_plus == diff_words->plus.text.ptr
935  *
936  *      that is: the plus text must start as a new line, and if there is no minus
937  *      word printed, a graph prefix must be printed.
938  *
939  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
940  *      *(diff_words->current_plus - 1) == '\n'
941  *
942  *      that is: a graph prefix must be printed following a '\n'
943  */
944 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
945 {
946         if ((diff_words->last_minus == 0 &&
947                 diff_words->current_plus == diff_words->plus.text.ptr) ||
948                 (diff_words->current_plus > diff_words->plus.text.ptr &&
949                 *(diff_words->current_plus - 1) == '\n')) {
950                 return 1;
951         } else {
952                 return 0;
953         }
954 }
955
956 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
957 {
958         struct diff_words_data *diff_words = priv;
959         struct diff_words_style *style = diff_words->style;
960         int minus_first, minus_len, plus_first, plus_len;
961         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
962         struct diff_options *opt = diff_words->opt;
963         const char *line_prefix;
964
965         if (line[0] != '@' || parse_hunk_header(line, len,
966                         &minus_first, &minus_len, &plus_first, &plus_len))
967                 return;
968
969         assert(opt);
970         line_prefix = diff_line_prefix(opt);
971
972         /* POSIX requires that first be decremented by one if len == 0... */
973         if (minus_len) {
974                 minus_begin = diff_words->minus.orig[minus_first].begin;
975                 minus_end =
976                         diff_words->minus.orig[minus_first + minus_len - 1].end;
977         } else
978                 minus_begin = minus_end =
979                         diff_words->minus.orig[minus_first].end;
980
981         if (plus_len) {
982                 plus_begin = diff_words->plus.orig[plus_first].begin;
983                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
984         } else
985                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
986
987         if (color_words_output_graph_prefix(diff_words)) {
988                 fputs(line_prefix, diff_words->opt->file);
989         }
990         if (diff_words->current_plus != plus_begin) {
991                 fn_out_diff_words_write_helper(diff_words->opt->file,
992                                 &style->ctx, style->newline,
993                                 plus_begin - diff_words->current_plus,
994                                 diff_words->current_plus, line_prefix);
995                 if (*(plus_begin - 1) == '\n')
996                         fputs(line_prefix, diff_words->opt->file);
997         }
998         if (minus_begin != minus_end) {
999                 fn_out_diff_words_write_helper(diff_words->opt->file,
1000                                 &style->old, style->newline,
1001                                 minus_end - minus_begin, minus_begin,
1002                                 line_prefix);
1003         }
1004         if (plus_begin != plus_end) {
1005                 fn_out_diff_words_write_helper(diff_words->opt->file,
1006                                 &style->new, style->newline,
1007                                 plus_end - plus_begin, plus_begin,
1008                                 line_prefix);
1009         }
1010
1011         diff_words->current_plus = plus_end;
1012         diff_words->last_minus = minus_first;
1013 }
1014
1015 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1016 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1017                 int *begin, int *end)
1018 {
1019         if (word_regex && *begin < buffer->size) {
1020                 regmatch_t match[1];
1021                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1022                                  buffer->size - *begin, 1, match, 0)) {
1023                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1024                                         '\n', match[0].rm_eo - match[0].rm_so);
1025                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1026                         *begin += match[0].rm_so;
1027                         return *begin >= *end;
1028                 }
1029                 return -1;
1030         }
1031
1032         /* find the next word */
1033         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1034                 (*begin)++;
1035         if (*begin >= buffer->size)
1036                 return -1;
1037
1038         /* find the end of the word */
1039         *end = *begin + 1;
1040         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1041                 (*end)++;
1042
1043         return 0;
1044 }
1045
1046 /*
1047  * This function splits the words in buffer->text, stores the list with
1048  * newline separator into out, and saves the offsets of the original words
1049  * in buffer->orig.
1050  */
1051 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1052                 regex_t *word_regex)
1053 {
1054         int i, j;
1055         long alloc = 0;
1056
1057         out->size = 0;
1058         out->ptr = NULL;
1059
1060         /* fake an empty "0th" word */
1061         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1062         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1063         buffer->orig_nr = 1;
1064
1065         for (i = 0; i < buffer->text.size; i++) {
1066                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1067                         return;
1068
1069                 /* store original boundaries */
1070                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1071                                 buffer->orig_alloc);
1072                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1073                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1074                 buffer->orig_nr++;
1075
1076                 /* store one word */
1077                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1078                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1079                 out->ptr[out->size + j - i] = '\n';
1080                 out->size += j - i + 1;
1081
1082                 i = j - 1;
1083         }
1084 }
1085
1086 /* this executes the word diff on the accumulated buffers */
1087 static void diff_words_show(struct diff_words_data *diff_words)
1088 {
1089         xpparam_t xpp;
1090         xdemitconf_t xecfg;
1091         mmfile_t minus, plus;
1092         struct diff_words_style *style = diff_words->style;
1093
1094         struct diff_options *opt = diff_words->opt;
1095         const char *line_prefix;
1096
1097         assert(opt);
1098         line_prefix = diff_line_prefix(opt);
1099
1100         /* special case: only removal */
1101         if (!diff_words->plus.text.size) {
1102                 fputs(line_prefix, diff_words->opt->file);
1103                 fn_out_diff_words_write_helper(diff_words->opt->file,
1104                         &style->old, style->newline,
1105                         diff_words->minus.text.size,
1106                         diff_words->minus.text.ptr, line_prefix);
1107                 diff_words->minus.text.size = 0;
1108                 return;
1109         }
1110
1111         diff_words->current_plus = diff_words->plus.text.ptr;
1112         diff_words->last_minus = 0;
1113
1114         memset(&xpp, 0, sizeof(xpp));
1115         memset(&xecfg, 0, sizeof(xecfg));
1116         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1117         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1118         xpp.flags = 0;
1119         /* as only the hunk header will be parsed, we need a 0-context */
1120         xecfg.ctxlen = 0;
1121         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1122                           &xpp, &xecfg))
1123                 die("unable to generate word diff");
1124         free(minus.ptr);
1125         free(plus.ptr);
1126         if (diff_words->current_plus != diff_words->plus.text.ptr +
1127                         diff_words->plus.text.size) {
1128                 if (color_words_output_graph_prefix(diff_words))
1129                         fputs(line_prefix, diff_words->opt->file);
1130                 fn_out_diff_words_write_helper(diff_words->opt->file,
1131                         &style->ctx, style->newline,
1132                         diff_words->plus.text.ptr + diff_words->plus.text.size
1133                         - diff_words->current_plus, diff_words->current_plus,
1134                         line_prefix);
1135         }
1136         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1137 }
1138
1139 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1140 static void diff_words_flush(struct emit_callback *ecbdata)
1141 {
1142         if (ecbdata->diff_words->minus.text.size ||
1143             ecbdata->diff_words->plus.text.size)
1144                 diff_words_show(ecbdata->diff_words);
1145 }
1146
1147 static void diff_filespec_load_driver(struct diff_filespec *one)
1148 {
1149         /* Use already-loaded driver */
1150         if (one->driver)
1151                 return;
1152
1153         if (S_ISREG(one->mode))
1154                 one->driver = userdiff_find_by_path(one->path);
1155
1156         /* Fallback to default settings */
1157         if (!one->driver)
1158                 one->driver = userdiff_find_by_name("default");
1159 }
1160
1161 static const char *userdiff_word_regex(struct diff_filespec *one)
1162 {
1163         diff_filespec_load_driver(one);
1164         return one->driver->word_regex;
1165 }
1166
1167 static void init_diff_words_data(struct emit_callback *ecbdata,
1168                                  struct diff_options *orig_opts,
1169                                  struct diff_filespec *one,
1170                                  struct diff_filespec *two)
1171 {
1172         int i;
1173         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1174         memcpy(o, orig_opts, sizeof(struct diff_options));
1175
1176         ecbdata->diff_words =
1177                 xcalloc(1, sizeof(struct diff_words_data));
1178         ecbdata->diff_words->type = o->word_diff;
1179         ecbdata->diff_words->opt = o;
1180         if (!o->word_regex)
1181                 o->word_regex = userdiff_word_regex(one);
1182         if (!o->word_regex)
1183                 o->word_regex = userdiff_word_regex(two);
1184         if (!o->word_regex)
1185                 o->word_regex = diff_word_regex_cfg;
1186         if (o->word_regex) {
1187                 ecbdata->diff_words->word_regex = (regex_t *)
1188                         xmalloc(sizeof(regex_t));
1189                 if (regcomp(ecbdata->diff_words->word_regex,
1190                             o->word_regex,
1191                             REG_EXTENDED | REG_NEWLINE))
1192                         die ("Invalid regular expression: %s",
1193                              o->word_regex);
1194         }
1195         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1196                 if (o->word_diff == diff_words_styles[i].type) {
1197                         ecbdata->diff_words->style =
1198                                 &diff_words_styles[i];
1199                         break;
1200                 }
1201         }
1202         if (want_color(o->use_color)) {
1203                 struct diff_words_style *st = ecbdata->diff_words->style;
1204                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1205                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1206                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1207         }
1208 }
1209
1210 static void free_diff_words_data(struct emit_callback *ecbdata)
1211 {
1212         if (ecbdata->diff_words) {
1213                 diff_words_flush(ecbdata);
1214                 free (ecbdata->diff_words->opt);
1215                 free (ecbdata->diff_words->minus.text.ptr);
1216                 free (ecbdata->diff_words->minus.orig);
1217                 free (ecbdata->diff_words->plus.text.ptr);
1218                 free (ecbdata->diff_words->plus.orig);
1219                 if (ecbdata->diff_words->word_regex) {
1220                         regfree(ecbdata->diff_words->word_regex);
1221                         free(ecbdata->diff_words->word_regex);
1222                 }
1223                 free(ecbdata->diff_words);
1224                 ecbdata->diff_words = NULL;
1225         }
1226 }
1227
1228 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1229 {
1230         if (want_color(diff_use_color))
1231                 return diff_colors[ix];
1232         return "";
1233 }
1234
1235 const char *diff_line_prefix(struct diff_options *opt)
1236 {
1237         struct strbuf *msgbuf;
1238         if (!opt->output_prefix)
1239                 return "";
1240
1241         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1242         return msgbuf->buf;
1243 }
1244
1245 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1246 {
1247         const char *cp;
1248         unsigned long allot;
1249         size_t l = len;
1250
1251         if (ecb->truncate)
1252                 return ecb->truncate(line, len);
1253         cp = line;
1254         allot = l;
1255         while (0 < l) {
1256                 (void) utf8_width(&cp, &l);
1257                 if (!cp)
1258                         break; /* truncated in the middle? */
1259         }
1260         return allot - l;
1261 }
1262
1263 static void find_lno(const char *line, struct emit_callback *ecbdata)
1264 {
1265         const char *p;
1266         ecbdata->lno_in_preimage = 0;
1267         ecbdata->lno_in_postimage = 0;
1268         p = strchr(line, '-');
1269         if (!p)
1270                 return; /* cannot happen */
1271         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1272         p = strchr(p, '+');
1273         if (!p)
1274                 return; /* cannot happen */
1275         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1276 }
1277
1278 static void fn_out_consume(void *priv, char *line, unsigned long len)
1279 {
1280         struct emit_callback *ecbdata = priv;
1281         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1282         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1283         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1284         struct diff_options *o = ecbdata->opt;
1285         const char *line_prefix = diff_line_prefix(o);
1286
1287         o->found_changes = 1;
1288
1289         if (ecbdata->header) {
1290                 fprintf(o->file, "%s", ecbdata->header->buf);
1291                 strbuf_reset(ecbdata->header);
1292                 ecbdata->header = NULL;
1293         }
1294
1295         if (ecbdata->label_path[0]) {
1296                 const char *name_a_tab, *name_b_tab;
1297
1298                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1299                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1300
1301                 fprintf(o->file, "%s%s--- %s%s%s\n",
1302                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1303                 fprintf(o->file, "%s%s+++ %s%s%s\n",
1304                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1305                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1306         }
1307
1308         if (diff_suppress_blank_empty
1309             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1310                 line[0] = '\n';
1311                 len = 1;
1312         }
1313
1314         if (line[0] == '@') {
1315                 if (ecbdata->diff_words)
1316                         diff_words_flush(ecbdata);
1317                 len = sane_truncate_line(ecbdata, line, len);
1318                 find_lno(line, ecbdata);
1319                 emit_hunk_header(ecbdata, line, len);
1320                 if (line[len-1] != '\n')
1321                         putc('\n', o->file);
1322                 return;
1323         }
1324
1325         if (ecbdata->diff_words) {
1326                 if (line[0] == '-') {
1327                         diff_words_append(line, len,
1328                                           &ecbdata->diff_words->minus);
1329                         return;
1330                 } else if (line[0] == '+') {
1331                         diff_words_append(line, len,
1332                                           &ecbdata->diff_words->plus);
1333                         return;
1334                 } else if (starts_with(line, "\\ ")) {
1335                         /*
1336                          * Eat the "no newline at eof" marker as if we
1337                          * saw a "+" or "-" line with nothing on it,
1338                          * and return without diff_words_flush() to
1339                          * defer processing. If this is the end of
1340                          * preimage, more "+" lines may come after it.
1341                          */
1342                         return;
1343                 }
1344                 diff_words_flush(ecbdata);
1345                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1346                         emit_line(o, context, reset, line, len);
1347                         fputs("~\n", o->file);
1348                 } else {
1349                         /*
1350                          * Skip the prefix character, if any.  With
1351                          * diff_suppress_blank_empty, there may be
1352                          * none.
1353                          */
1354                         if (line[0] != '\n') {
1355                               line++;
1356                               len--;
1357                         }
1358                         emit_line(o, context, reset, line, len);
1359                 }
1360                 return;
1361         }
1362
1363         switch (line[0]) {
1364         case '+':
1365                 ecbdata->lno_in_postimage++;
1366                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1367                 break;
1368         case '-':
1369                 ecbdata->lno_in_preimage++;
1370                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1371                 break;
1372         case ' ':
1373                 ecbdata->lno_in_postimage++;
1374                 ecbdata->lno_in_preimage++;
1375                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1376                 break;
1377         default:
1378                 /* incomplete line at the end */
1379                 ecbdata->lno_in_preimage++;
1380                 emit_line(o, diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1381                           reset, line, len);
1382                 break;
1383         }
1384 }
1385
1386 static char *pprint_rename(const char *a, const char *b)
1387 {
1388         const char *old = a;
1389         const char *new = b;
1390         struct strbuf name = STRBUF_INIT;
1391         int pfx_length, sfx_length;
1392         int pfx_adjust_for_slash;
1393         int len_a = strlen(a);
1394         int len_b = strlen(b);
1395         int a_midlen, b_midlen;
1396         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1397         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1398
1399         if (qlen_a || qlen_b) {
1400                 quote_c_style(a, &name, NULL, 0);
1401                 strbuf_addstr(&name, " => ");
1402                 quote_c_style(b, &name, NULL, 0);
1403                 return strbuf_detach(&name, NULL);
1404         }
1405
1406         /* Find common prefix */
1407         pfx_length = 0;
1408         while (*old && *new && *old == *new) {
1409                 if (*old == '/')
1410                         pfx_length = old - a + 1;
1411                 old++;
1412                 new++;
1413         }
1414
1415         /* Find common suffix */
1416         old = a + len_a;
1417         new = b + len_b;
1418         sfx_length = 0;
1419         /*
1420          * If there is a common prefix, it must end in a slash.  In
1421          * that case we let this loop run 1 into the prefix to see the
1422          * same slash.
1423          *
1424          * If there is no common prefix, we cannot do this as it would
1425          * underrun the input strings.
1426          */
1427         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1428         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1429                b + pfx_length - pfx_adjust_for_slash <= new &&
1430                *old == *new) {
1431                 if (*old == '/')
1432                         sfx_length = len_a - (old - a);
1433                 old--;
1434                 new--;
1435         }
1436
1437         /*
1438          * pfx{mid-a => mid-b}sfx
1439          * {pfx-a => pfx-b}sfx
1440          * pfx{sfx-a => sfx-b}
1441          * name-a => name-b
1442          */
1443         a_midlen = len_a - pfx_length - sfx_length;
1444         b_midlen = len_b - pfx_length - sfx_length;
1445         if (a_midlen < 0)
1446                 a_midlen = 0;
1447         if (b_midlen < 0)
1448                 b_midlen = 0;
1449
1450         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1451         if (pfx_length + sfx_length) {
1452                 strbuf_add(&name, a, pfx_length);
1453                 strbuf_addch(&name, '{');
1454         }
1455         strbuf_add(&name, a + pfx_length, a_midlen);
1456         strbuf_addstr(&name, " => ");
1457         strbuf_add(&name, b + pfx_length, b_midlen);
1458         if (pfx_length + sfx_length) {
1459                 strbuf_addch(&name, '}');
1460                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1461         }
1462         return strbuf_detach(&name, NULL);
1463 }
1464
1465 struct diffstat_t {
1466         int nr;
1467         int alloc;
1468         struct diffstat_file {
1469                 char *from_name;
1470                 char *name;
1471                 char *print_name;
1472                 unsigned is_unmerged:1;
1473                 unsigned is_binary:1;
1474                 unsigned is_renamed:1;
1475                 unsigned is_interesting:1;
1476                 uintmax_t added, deleted;
1477         } **files;
1478 };
1479
1480 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1481                                           const char *name_a,
1482                                           const char *name_b)
1483 {
1484         struct diffstat_file *x;
1485         x = xcalloc(1, sizeof(*x));
1486         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1487         diffstat->files[diffstat->nr++] = x;
1488         if (name_b) {
1489                 x->from_name = xstrdup(name_a);
1490                 x->name = xstrdup(name_b);
1491                 x->is_renamed = 1;
1492         }
1493         else {
1494                 x->from_name = NULL;
1495                 x->name = xstrdup(name_a);
1496         }
1497         return x;
1498 }
1499
1500 static void diffstat_consume(void *priv, char *line, unsigned long len)
1501 {
1502         struct diffstat_t *diffstat = priv;
1503         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1504
1505         if (line[0] == '+')
1506                 x->added++;
1507         else if (line[0] == '-')
1508                 x->deleted++;
1509 }
1510
1511 const char mime_boundary_leader[] = "------------";
1512
1513 static int scale_linear(int it, int width, int max_change)
1514 {
1515         if (!it)
1516                 return 0;
1517         /*
1518          * make sure that at least one '-' or '+' is printed if
1519          * there is any change to this path. The easiest way is to
1520          * scale linearly as if the alloted width is one column shorter
1521          * than it is, and then add 1 to the result.
1522          */
1523         return 1 + (it * (width - 1) / max_change);
1524 }
1525
1526 static void show_name(FILE *file,
1527                       const char *prefix, const char *name, int len)
1528 {
1529         fprintf(file, " %s%-*s |", prefix, len, name);
1530 }
1531
1532 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1533 {
1534         if (cnt <= 0)
1535                 return;
1536         fprintf(file, "%s", set);
1537         while (cnt--)
1538                 putc(ch, file);
1539         fprintf(file, "%s", reset);
1540 }
1541
1542 static void fill_print_name(struct diffstat_file *file)
1543 {
1544         char *pname;
1545
1546         if (file->print_name)
1547                 return;
1548
1549         if (!file->is_renamed) {
1550                 struct strbuf buf = STRBUF_INIT;
1551                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1552                         pname = strbuf_detach(&buf, NULL);
1553                 } else {
1554                         pname = file->name;
1555                         strbuf_release(&buf);
1556                 }
1557         } else {
1558                 pname = pprint_rename(file->from_name, file->name);
1559         }
1560         file->print_name = pname;
1561 }
1562
1563 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1564 {
1565         struct strbuf sb = STRBUF_INIT;
1566         int ret;
1567
1568         if (!files) {
1569                 assert(insertions == 0 && deletions == 0);
1570                 return fprintf(fp, "%s\n", " 0 files changed");
1571         }
1572
1573         strbuf_addf(&sb,
1574                     (files == 1) ? " %d file changed" : " %d files changed",
1575                     files);
1576
1577         /*
1578          * For binary diff, the caller may want to print "x files
1579          * changed" with insertions == 0 && deletions == 0.
1580          *
1581          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1582          * is probably less confusing (i.e skip over "2 files changed
1583          * but nothing about added/removed lines? Is this a bug in Git?").
1584          */
1585         if (insertions || deletions == 0) {
1586                 strbuf_addf(&sb,
1587                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1588                             insertions);
1589         }
1590
1591         if (deletions || insertions == 0) {
1592                 strbuf_addf(&sb,
1593                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1594                             deletions);
1595         }
1596         strbuf_addch(&sb, '\n');
1597         ret = fputs(sb.buf, fp);
1598         strbuf_release(&sb);
1599         return ret;
1600 }
1601
1602 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1603 {
1604         int i, len, add, del, adds = 0, dels = 0;
1605         uintmax_t max_change = 0, max_len = 0;
1606         int total_files = data->nr, count;
1607         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1608         const char *reset, *add_c, *del_c;
1609         const char *line_prefix = "";
1610         int extra_shown = 0;
1611
1612         if (data->nr == 0)
1613                 return;
1614
1615         line_prefix = diff_line_prefix(options);
1616         count = options->stat_count ? options->stat_count : data->nr;
1617
1618         reset = diff_get_color_opt(options, DIFF_RESET);
1619         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1620         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1621
1622         /*
1623          * Find the longest filename and max number of changes
1624          */
1625         for (i = 0; (i < count) && (i < data->nr); i++) {
1626                 struct diffstat_file *file = data->files[i];
1627                 uintmax_t change = file->added + file->deleted;
1628
1629                 if (!file->is_interesting && (change == 0)) {
1630                         count++; /* not shown == room for one more */
1631                         continue;
1632                 }
1633                 fill_print_name(file);
1634                 len = strlen(file->print_name);
1635                 if (max_len < len)
1636                         max_len = len;
1637
1638                 if (file->is_unmerged) {
1639                         /* "Unmerged" is 8 characters */
1640                         bin_width = bin_width < 8 ? 8 : bin_width;
1641                         continue;
1642                 }
1643                 if (file->is_binary) {
1644                         /* "Bin XXX -> YYY bytes" */
1645                         int w = 14 + decimal_width(file->added)
1646                                 + decimal_width(file->deleted);
1647                         bin_width = bin_width < w ? w : bin_width;
1648                         /* Display change counts aligned with "Bin" */
1649                         number_width = 3;
1650                         continue;
1651                 }
1652
1653                 if (max_change < change)
1654                         max_change = change;
1655         }
1656         count = i; /* where we can stop scanning in data->files[] */
1657
1658         /*
1659          * We have width = stat_width or term_columns() columns total.
1660          * We want a maximum of min(max_len, stat_name_width) for the name part.
1661          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1662          * We also need 1 for " " and 4 + decimal_width(max_change)
1663          * for " | NNNN " and one the empty column at the end, altogether
1664          * 6 + decimal_width(max_change).
1665          *
1666          * If there's not enough space, we will use the smaller of
1667          * stat_name_width (if set) and 5/8*width for the filename,
1668          * and the rest for constant elements + graph part, but no more
1669          * than stat_graph_width for the graph part.
1670          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1671          * for the standard terminal size).
1672          *
1673          * In other words: stat_width limits the maximum width, and
1674          * stat_name_width fixes the maximum width of the filename,
1675          * and is also used to divide available columns if there
1676          * aren't enough.
1677          *
1678          * Binary files are displayed with "Bin XXX -> YYY bytes"
1679          * instead of the change count and graph. This part is treated
1680          * similarly to the graph part, except that it is not
1681          * "scaled". If total width is too small to accommodate the
1682          * guaranteed minimum width of the filename part and the
1683          * separators and this message, this message will "overflow"
1684          * making the line longer than the maximum width.
1685          */
1686
1687         if (options->stat_width == -1)
1688                 width = term_columns() - strlen(line_prefix);
1689         else
1690                 width = options->stat_width ? options->stat_width : 80;
1691         number_width = decimal_width(max_change) > number_width ?
1692                 decimal_width(max_change) : number_width;
1693
1694         if (options->stat_graph_width == -1)
1695                 options->stat_graph_width = diff_stat_graph_width;
1696
1697         /*
1698          * Guarantee 3/8*16==6 for the graph part
1699          * and 5/8*16==10 for the filename part
1700          */
1701         if (width < 16 + 6 + number_width)
1702                 width = 16 + 6 + number_width;
1703
1704         /*
1705          * First assign sizes that are wanted, ignoring available width.
1706          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1707          * starting from "XXX" should fit in graph_width.
1708          */
1709         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1710         if (options->stat_graph_width &&
1711             options->stat_graph_width < graph_width)
1712                 graph_width = options->stat_graph_width;
1713
1714         name_width = (options->stat_name_width > 0 &&
1715                       options->stat_name_width < max_len) ?
1716                 options->stat_name_width : max_len;
1717
1718         /*
1719          * Adjust adjustable widths not to exceed maximum width
1720          */
1721         if (name_width + number_width + 6 + graph_width > width) {
1722                 if (graph_width > width * 3/8 - number_width - 6) {
1723                         graph_width = width * 3/8 - number_width - 6;
1724                         if (graph_width < 6)
1725                                 graph_width = 6;
1726                 }
1727
1728                 if (options->stat_graph_width &&
1729                     graph_width > options->stat_graph_width)
1730                         graph_width = options->stat_graph_width;
1731                 if (name_width > width - number_width - 6 - graph_width)
1732                         name_width = width - number_width - 6 - graph_width;
1733                 else
1734                         graph_width = width - number_width - 6 - name_width;
1735         }
1736
1737         /*
1738          * From here name_width is the width of the name area,
1739          * and graph_width is the width of the graph area.
1740          * max_change is used to scale graph properly.
1741          */
1742         for (i = 0; i < count; i++) {
1743                 const char *prefix = "";
1744                 struct diffstat_file *file = data->files[i];
1745                 char *name = file->print_name;
1746                 uintmax_t added = file->added;
1747                 uintmax_t deleted = file->deleted;
1748                 int name_len;
1749
1750                 if (!file->is_interesting && (added + deleted == 0))
1751                         continue;
1752
1753                 /*
1754                  * "scale" the filename
1755                  */
1756                 len = name_width;
1757                 name_len = strlen(name);
1758                 if (name_width < name_len) {
1759                         char *slash;
1760                         prefix = "...";
1761                         len -= 3;
1762                         name += name_len - len;
1763                         slash = strchr(name, '/');
1764                         if (slash)
1765                                 name = slash;
1766                 }
1767
1768                 if (file->is_binary) {
1769                         fprintf(options->file, "%s", line_prefix);
1770                         show_name(options->file, prefix, name, len);
1771                         fprintf(options->file, " %*s", number_width, "Bin");
1772                         if (!added && !deleted) {
1773                                 putc('\n', options->file);
1774                                 continue;
1775                         }
1776                         fprintf(options->file, " %s%"PRIuMAX"%s",
1777                                 del_c, deleted, reset);
1778                         fprintf(options->file, " -> ");
1779                         fprintf(options->file, "%s%"PRIuMAX"%s",
1780                                 add_c, added, reset);
1781                         fprintf(options->file, " bytes");
1782                         fprintf(options->file, "\n");
1783                         continue;
1784                 }
1785                 else if (file->is_unmerged) {
1786                         fprintf(options->file, "%s", line_prefix);
1787                         show_name(options->file, prefix, name, len);
1788                         fprintf(options->file, " Unmerged\n");
1789                         continue;
1790                 }
1791
1792                 /*
1793                  * scale the add/delete
1794                  */
1795                 add = added;
1796                 del = deleted;
1797
1798                 if (graph_width <= max_change) {
1799                         int total = scale_linear(add + del, graph_width, max_change);
1800                         if (total < 2 && add && del)
1801                                 /* width >= 2 due to the sanity check */
1802                                 total = 2;
1803                         if (add < del) {
1804                                 add = scale_linear(add, graph_width, max_change);
1805                                 del = total - add;
1806                         } else {
1807                                 del = scale_linear(del, graph_width, max_change);
1808                                 add = total - del;
1809                         }
1810                 }
1811                 fprintf(options->file, "%s", line_prefix);
1812                 show_name(options->file, prefix, name, len);
1813                 fprintf(options->file, " %*"PRIuMAX"%s",
1814                         number_width, added + deleted,
1815                         added + deleted ? " " : "");
1816                 show_graph(options->file, '+', add, add_c, reset);
1817                 show_graph(options->file, '-', del, del_c, reset);
1818                 fprintf(options->file, "\n");
1819         }
1820
1821         for (i = 0; i < data->nr; i++) {
1822                 struct diffstat_file *file = data->files[i];
1823                 uintmax_t added = file->added;
1824                 uintmax_t deleted = file->deleted;
1825
1826                 if (file->is_unmerged ||
1827                     (!file->is_interesting && (added + deleted == 0))) {
1828                         total_files--;
1829                         continue;
1830                 }
1831
1832                 if (!file->is_binary) {
1833                         adds += added;
1834                         dels += deleted;
1835                 }
1836                 if (i < count)
1837                         continue;
1838                 if (!extra_shown)
1839                         fprintf(options->file, "%s ...\n", line_prefix);
1840                 extra_shown = 1;
1841         }
1842         fprintf(options->file, "%s", line_prefix);
1843         print_stat_summary(options->file, total_files, adds, dels);
1844 }
1845
1846 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1847 {
1848         int i, adds = 0, dels = 0, total_files = data->nr;
1849
1850         if (data->nr == 0)
1851                 return;
1852
1853         for (i = 0; i < data->nr; i++) {
1854                 int added = data->files[i]->added;
1855                 int deleted= data->files[i]->deleted;
1856
1857                 if (data->files[i]->is_unmerged ||
1858                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1859                         total_files--;
1860                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1861                         adds += added;
1862                         dels += deleted;
1863                 }
1864         }
1865         fprintf(options->file, "%s", diff_line_prefix(options));
1866         print_stat_summary(options->file, total_files, adds, dels);
1867 }
1868
1869 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1870 {
1871         int i;
1872
1873         if (data->nr == 0)
1874                 return;
1875
1876         for (i = 0; i < data->nr; i++) {
1877                 struct diffstat_file *file = data->files[i];
1878
1879                 fprintf(options->file, "%s", diff_line_prefix(options));
1880
1881                 if (file->is_binary)
1882                         fprintf(options->file, "-\t-\t");
1883                 else
1884                         fprintf(options->file,
1885                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1886                                 file->added, file->deleted);
1887                 if (options->line_termination) {
1888                         fill_print_name(file);
1889                         if (!file->is_renamed)
1890                                 write_name_quoted(file->name, options->file,
1891                                                   options->line_termination);
1892                         else {
1893                                 fputs(file->print_name, options->file);
1894                                 putc(options->line_termination, options->file);
1895                         }
1896                 } else {
1897                         if (file->is_renamed) {
1898                                 putc('\0', options->file);
1899                                 write_name_quoted(file->from_name, options->file, '\0');
1900                         }
1901                         write_name_quoted(file->name, options->file, '\0');
1902                 }
1903         }
1904 }
1905
1906 struct dirstat_file {
1907         const char *name;
1908         unsigned long changed;
1909 };
1910
1911 struct dirstat_dir {
1912         struct dirstat_file *files;
1913         int alloc, nr, permille, cumulative;
1914 };
1915
1916 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1917                 unsigned long changed, const char *base, int baselen)
1918 {
1919         unsigned long this_dir = 0;
1920         unsigned int sources = 0;
1921         const char *line_prefix = diff_line_prefix(opt);
1922
1923         while (dir->nr) {
1924                 struct dirstat_file *f = dir->files;
1925                 int namelen = strlen(f->name);
1926                 unsigned long this;
1927                 char *slash;
1928
1929                 if (namelen < baselen)
1930                         break;
1931                 if (memcmp(f->name, base, baselen))
1932                         break;
1933                 slash = strchr(f->name + baselen, '/');
1934                 if (slash) {
1935                         int newbaselen = slash + 1 - f->name;
1936                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1937                         sources++;
1938                 } else {
1939                         this = f->changed;
1940                         dir->files++;
1941                         dir->nr--;
1942                         sources += 2;
1943                 }
1944                 this_dir += this;
1945         }
1946
1947         /*
1948          * We don't report dirstat's for
1949          *  - the top level
1950          *  - or cases where everything came from a single directory
1951          *    under this directory (sources == 1).
1952          */
1953         if (baselen && sources != 1) {
1954                 if (this_dir) {
1955                         int permille = this_dir * 1000 / changed;
1956                         if (permille >= dir->permille) {
1957                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1958                                         permille / 10, permille % 10, baselen, base);
1959                                 if (!dir->cumulative)
1960                                         return 0;
1961                         }
1962                 }
1963         }
1964         return this_dir;
1965 }
1966
1967 static int dirstat_compare(const void *_a, const void *_b)
1968 {
1969         const struct dirstat_file *a = _a;
1970         const struct dirstat_file *b = _b;
1971         return strcmp(a->name, b->name);
1972 }
1973
1974 static void show_dirstat(struct diff_options *options)
1975 {
1976         int i;
1977         unsigned long changed;
1978         struct dirstat_dir dir;
1979         struct diff_queue_struct *q = &diff_queued_diff;
1980
1981         dir.files = NULL;
1982         dir.alloc = 0;
1983         dir.nr = 0;
1984         dir.permille = options->dirstat_permille;
1985         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1986
1987         changed = 0;
1988         for (i = 0; i < q->nr; i++) {
1989                 struct diff_filepair *p = q->queue[i];
1990                 const char *name;
1991                 unsigned long copied, added, damage;
1992                 int content_changed;
1993
1994                 name = p->two->path ? p->two->path : p->one->path;
1995
1996                 if (p->one->oid_valid && p->two->oid_valid)
1997                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
1998                 else
1999                         content_changed = 1;
2000
2001                 if (!content_changed) {
2002                         /*
2003                          * The SHA1 has not changed, so pre-/post-content is
2004                          * identical. We can therefore skip looking at the
2005                          * file contents altogether.
2006                          */
2007                         damage = 0;
2008                         goto found_damage;
2009                 }
2010
2011                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2012                         /*
2013                          * In --dirstat-by-file mode, we don't really need to
2014                          * look at the actual file contents at all.
2015                          * The fact that the SHA1 changed is enough for us to
2016                          * add this file to the list of results
2017                          * (with each file contributing equal damage).
2018                          */
2019                         damage = 1;
2020                         goto found_damage;
2021                 }
2022
2023                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2024                         diff_populate_filespec(p->one, 0);
2025                         diff_populate_filespec(p->two, 0);
2026                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2027                                                &copied, &added);
2028                         diff_free_filespec_data(p->one);
2029                         diff_free_filespec_data(p->two);
2030                 } else if (DIFF_FILE_VALID(p->one)) {
2031                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2032                         copied = added = 0;
2033                         diff_free_filespec_data(p->one);
2034                 } else if (DIFF_FILE_VALID(p->two)) {
2035                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2036                         copied = 0;
2037                         added = p->two->size;
2038                         diff_free_filespec_data(p->two);
2039                 } else
2040                         continue;
2041
2042                 /*
2043                  * Original minus copied is the removed material,
2044                  * added is the new material.  They are both damages
2045                  * made to the preimage.
2046                  * If the resulting damage is zero, we know that
2047                  * diffcore_count_changes() considers the two entries to
2048                  * be identical, but since content_changed is true, we
2049                  * know that there must have been _some_ kind of change,
2050                  * so we force all entries to have damage > 0.
2051                  */
2052                 damage = (p->one->size - copied) + added;
2053                 if (!damage)
2054                         damage = 1;
2055
2056 found_damage:
2057                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2058                 dir.files[dir.nr].name = name;
2059                 dir.files[dir.nr].changed = damage;
2060                 changed += damage;
2061                 dir.nr++;
2062         }
2063
2064         /* This can happen even with many files, if everything was renames */
2065         if (!changed)
2066                 return;
2067
2068         /* Show all directories with more than x% of the changes */
2069         QSORT(dir.files, dir.nr, dirstat_compare);
2070         gather_dirstat(options, &dir, changed, "", 0);
2071 }
2072
2073 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2074 {
2075         int i;
2076         unsigned long changed;
2077         struct dirstat_dir dir;
2078
2079         if (data->nr == 0)
2080                 return;
2081
2082         dir.files = NULL;
2083         dir.alloc = 0;
2084         dir.nr = 0;
2085         dir.permille = options->dirstat_permille;
2086         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2087
2088         changed = 0;
2089         for (i = 0; i < data->nr; i++) {
2090                 struct diffstat_file *file = data->files[i];
2091                 unsigned long damage = file->added + file->deleted;
2092                 if (file->is_binary)
2093                         /*
2094                          * binary files counts bytes, not lines. Must find some
2095                          * way to normalize binary bytes vs. textual lines.
2096                          * The following heuristic assumes that there are 64
2097                          * bytes per "line".
2098                          * This is stupid and ugly, but very cheap...
2099                          */
2100                         damage = (damage + 63) / 64;
2101                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2102                 dir.files[dir.nr].name = file->name;
2103                 dir.files[dir.nr].changed = damage;
2104                 changed += damage;
2105                 dir.nr++;
2106         }
2107
2108         /* This can happen even with many files, if everything was renames */
2109         if (!changed)
2110                 return;
2111
2112         /* Show all directories with more than x% of the changes */
2113         QSORT(dir.files, dir.nr, dirstat_compare);
2114         gather_dirstat(options, &dir, changed, "", 0);
2115 }
2116
2117 static void free_diffstat_info(struct diffstat_t *diffstat)
2118 {
2119         int i;
2120         for (i = 0; i < diffstat->nr; i++) {
2121                 struct diffstat_file *f = diffstat->files[i];
2122                 if (f->name != f->print_name)
2123                         free(f->print_name);
2124                 free(f->name);
2125                 free(f->from_name);
2126                 free(f);
2127         }
2128         free(diffstat->files);
2129 }
2130
2131 struct checkdiff_t {
2132         const char *filename;
2133         int lineno;
2134         int conflict_marker_size;
2135         struct diff_options *o;
2136         unsigned ws_rule;
2137         unsigned status;
2138 };
2139
2140 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2141 {
2142         char firstchar;
2143         int cnt;
2144
2145         if (len < marker_size + 1)
2146                 return 0;
2147         firstchar = line[0];
2148         switch (firstchar) {
2149         case '=': case '>': case '<': case '|':
2150                 break;
2151         default:
2152                 return 0;
2153         }
2154         for (cnt = 1; cnt < marker_size; cnt++)
2155                 if (line[cnt] != firstchar)
2156                         return 0;
2157         /* line[1] thru line[marker_size-1] are same as firstchar */
2158         if (len < marker_size + 1 || !isspace(line[marker_size]))
2159                 return 0;
2160         return 1;
2161 }
2162
2163 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2164 {
2165         struct checkdiff_t *data = priv;
2166         int marker_size = data->conflict_marker_size;
2167         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2168         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2169         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2170         char *err;
2171         const char *line_prefix;
2172
2173         assert(data->o);
2174         line_prefix = diff_line_prefix(data->o);
2175
2176         if (line[0] == '+') {
2177                 unsigned bad;
2178                 data->lineno++;
2179                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2180                         data->status |= 1;
2181                         fprintf(data->o->file,
2182                                 "%s%s:%d: leftover conflict marker\n",
2183                                 line_prefix, data->filename, data->lineno);
2184                 }
2185                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2186                 if (!bad)
2187                         return;
2188                 data->status |= bad;
2189                 err = whitespace_error_string(bad);
2190                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2191                         line_prefix, data->filename, data->lineno, err);
2192                 free(err);
2193                 emit_line(data->o, set, reset, line, 1);
2194                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2195                               data->o->file, set, reset, ws);
2196         } else if (line[0] == ' ') {
2197                 data->lineno++;
2198         } else if (line[0] == '@') {
2199                 char *plus = strchr(line, '+');
2200                 if (plus)
2201                         data->lineno = strtol(plus, NULL, 10) - 1;
2202                 else
2203                         die("invalid diff");
2204         }
2205 }
2206
2207 static unsigned char *deflate_it(char *data,
2208                                  unsigned long size,
2209                                  unsigned long *result_size)
2210 {
2211         int bound;
2212         unsigned char *deflated;
2213         git_zstream stream;
2214
2215         git_deflate_init(&stream, zlib_compression_level);
2216         bound = git_deflate_bound(&stream, size);
2217         deflated = xmalloc(bound);
2218         stream.next_out = deflated;
2219         stream.avail_out = bound;
2220
2221         stream.next_in = (unsigned char *)data;
2222         stream.avail_in = size;
2223         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2224                 ; /* nothing */
2225         git_deflate_end(&stream);
2226         *result_size = stream.total_out;
2227         return deflated;
2228 }
2229
2230 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2231                                   const char *prefix)
2232 {
2233         void *cp;
2234         void *delta;
2235         void *deflated;
2236         void *data;
2237         unsigned long orig_size;
2238         unsigned long delta_size;
2239         unsigned long deflate_size;
2240         unsigned long data_size;
2241
2242         /* We could do deflated delta, or we could do just deflated two,
2243          * whichever is smaller.
2244          */
2245         delta = NULL;
2246         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2247         if (one->size && two->size) {
2248                 delta = diff_delta(one->ptr, one->size,
2249                                    two->ptr, two->size,
2250                                    &delta_size, deflate_size);
2251                 if (delta) {
2252                         void *to_free = delta;
2253                         orig_size = delta_size;
2254                         delta = deflate_it(delta, delta_size, &delta_size);
2255                         free(to_free);
2256                 }
2257         }
2258
2259         if (delta && delta_size < deflate_size) {
2260                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2261                 free(deflated);
2262                 data = delta;
2263                 data_size = delta_size;
2264         }
2265         else {
2266                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2267                 free(delta);
2268                 data = deflated;
2269                 data_size = deflate_size;
2270         }
2271
2272         /* emit data encoded in base85 */
2273         cp = data;
2274         while (data_size) {
2275                 int bytes = (52 < data_size) ? 52 : data_size;
2276                 char line[70];
2277                 data_size -= bytes;
2278                 if (bytes <= 26)
2279                         line[0] = bytes + 'A' - 1;
2280                 else
2281                         line[0] = bytes - 26 + 'a' - 1;
2282                 encode_85(line + 1, cp, bytes);
2283                 cp = (char *) cp + bytes;
2284                 fprintf(file, "%s", prefix);
2285                 fputs(line, file);
2286                 fputc('\n', file);
2287         }
2288         fprintf(file, "%s\n", prefix);
2289         free(data);
2290 }
2291
2292 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2293                              const char *prefix)
2294 {
2295         fprintf(file, "%sGIT binary patch\n", prefix);
2296         emit_binary_diff_body(file, one, two, prefix);
2297         emit_binary_diff_body(file, two, one, prefix);
2298 }
2299
2300 int diff_filespec_is_binary(struct diff_filespec *one)
2301 {
2302         if (one->is_binary == -1) {
2303                 diff_filespec_load_driver(one);
2304                 if (one->driver->binary != -1)
2305                         one->is_binary = one->driver->binary;
2306                 else {
2307                         if (!one->data && DIFF_FILE_VALID(one))
2308                                 diff_populate_filespec(one, CHECK_BINARY);
2309                         if (one->is_binary == -1 && one->data)
2310                                 one->is_binary = buffer_is_binary(one->data,
2311                                                 one->size);
2312                         if (one->is_binary == -1)
2313                                 one->is_binary = 0;
2314                 }
2315         }
2316         return one->is_binary;
2317 }
2318
2319 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2320 {
2321         diff_filespec_load_driver(one);
2322         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2323 }
2324
2325 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2326 {
2327         if (!options->a_prefix)
2328                 options->a_prefix = a;
2329         if (!options->b_prefix)
2330                 options->b_prefix = b;
2331 }
2332
2333 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2334 {
2335         if (!DIFF_FILE_VALID(one))
2336                 return NULL;
2337
2338         diff_filespec_load_driver(one);
2339         return userdiff_get_textconv(one->driver);
2340 }
2341
2342 static void builtin_diff(const char *name_a,
2343                          const char *name_b,
2344                          struct diff_filespec *one,
2345                          struct diff_filespec *two,
2346                          const char *xfrm_msg,
2347                          int must_show_header,
2348                          struct diff_options *o,
2349                          int complete_rewrite)
2350 {
2351         mmfile_t mf1, mf2;
2352         const char *lbl[2];
2353         char *a_one, *b_two;
2354         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2355         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2356         const char *a_prefix, *b_prefix;
2357         struct userdiff_driver *textconv_one = NULL;
2358         struct userdiff_driver *textconv_two = NULL;
2359         struct strbuf header = STRBUF_INIT;
2360         const char *line_prefix = diff_line_prefix(o);
2361
2362         diff_set_mnemonic_prefix(o, "a/", "b/");
2363         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2364                 a_prefix = o->b_prefix;
2365                 b_prefix = o->a_prefix;
2366         } else {
2367                 a_prefix = o->a_prefix;
2368                 b_prefix = o->b_prefix;
2369         }
2370
2371         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
2372             (!one->mode || S_ISGITLINK(one->mode)) &&
2373             (!two->mode || S_ISGITLINK(two->mode))) {
2374                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2375                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2376                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2377                                 line_prefix,
2378                                 &one->oid, &two->oid,
2379                                 two->dirty_submodule,
2380                                 meta, del, add, reset);
2381                 return;
2382         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
2383                    (!one->mode || S_ISGITLINK(one->mode)) &&
2384                    (!two->mode || S_ISGITLINK(two->mode))) {
2385                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2386                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2387                 show_submodule_inline_diff(o->file, one->path ? one->path : two->path,
2388                                 line_prefix,
2389                                 &one->oid, &two->oid,
2390                                 two->dirty_submodule,
2391                                 meta, del, add, reset, o);
2392                 return;
2393         }
2394
2395         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2396                 textconv_one = get_textconv(one);
2397                 textconv_two = get_textconv(two);
2398         }
2399
2400         /* Never use a non-valid filename anywhere if at all possible */
2401         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2402         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2403
2404         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2405         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2406         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2407         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2408         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2409         if (lbl[0][0] == '/') {
2410                 /* /dev/null */
2411                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2412                 if (xfrm_msg)
2413                         strbuf_addstr(&header, xfrm_msg);
2414                 must_show_header = 1;
2415         }
2416         else if (lbl[1][0] == '/') {
2417                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2418                 if (xfrm_msg)
2419                         strbuf_addstr(&header, xfrm_msg);
2420                 must_show_header = 1;
2421         }
2422         else {
2423                 if (one->mode != two->mode) {
2424                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2425                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2426                         must_show_header = 1;
2427                 }
2428                 if (xfrm_msg)
2429                         strbuf_addstr(&header, xfrm_msg);
2430
2431                 /*
2432                  * we do not run diff between different kind
2433                  * of objects.
2434                  */
2435                 if ((one->mode ^ two->mode) & S_IFMT)
2436                         goto free_ab_and_return;
2437                 if (complete_rewrite &&
2438                     (textconv_one || !diff_filespec_is_binary(one)) &&
2439                     (textconv_two || !diff_filespec_is_binary(two))) {
2440                         fprintf(o->file, "%s", header.buf);
2441                         strbuf_reset(&header);
2442                         emit_rewrite_diff(name_a, name_b, one, two,
2443                                                 textconv_one, textconv_two, o);
2444                         o->found_changes = 1;
2445                         goto free_ab_and_return;
2446                 }
2447         }
2448
2449         if (o->irreversible_delete && lbl[1][0] == '/') {
2450                 fprintf(o->file, "%s", header.buf);
2451                 strbuf_reset(&header);
2452                 goto free_ab_and_return;
2453         } else if (!DIFF_OPT_TST(o, TEXT) &&
2454             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2455               (!textconv_two && diff_filespec_is_binary(two)) )) {
2456                 if (!one->data && !two->data &&
2457                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2458                     !DIFF_OPT_TST(o, BINARY)) {
2459                         if (!oidcmp(&one->oid, &two->oid)) {
2460                                 if (must_show_header)
2461                                         fprintf(o->file, "%s", header.buf);
2462                                 goto free_ab_and_return;
2463                         }
2464                         fprintf(o->file, "%s", header.buf);
2465                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2466                                 line_prefix, lbl[0], lbl[1]);
2467                         goto free_ab_and_return;
2468                 }
2469                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2470                         die("unable to read files to diff");
2471                 /* Quite common confusing case */
2472                 if (mf1.size == mf2.size &&
2473                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2474                         if (must_show_header)
2475                                 fprintf(o->file, "%s", header.buf);
2476                         goto free_ab_and_return;
2477                 }
2478                 fprintf(o->file, "%s", header.buf);
2479                 strbuf_reset(&header);
2480                 if (DIFF_OPT_TST(o, BINARY))
2481                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2482                 else
2483                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2484                                 line_prefix, lbl[0], lbl[1]);
2485                 o->found_changes = 1;
2486         } else {
2487                 /* Crazy xdl interfaces.. */
2488                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2489                 const char *v;
2490                 xpparam_t xpp;
2491                 xdemitconf_t xecfg;
2492                 struct emit_callback ecbdata;
2493                 const struct userdiff_funcname *pe;
2494
2495                 if (must_show_header) {
2496                         fprintf(o->file, "%s", header.buf);
2497                         strbuf_reset(&header);
2498                 }
2499
2500                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2501                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2502
2503                 pe = diff_funcname_pattern(one);
2504                 if (!pe)
2505                         pe = diff_funcname_pattern(two);
2506
2507                 memset(&xpp, 0, sizeof(xpp));
2508                 memset(&xecfg, 0, sizeof(xecfg));
2509                 memset(&ecbdata, 0, sizeof(ecbdata));
2510                 ecbdata.label_path = lbl;
2511                 ecbdata.color_diff = want_color(o->use_color);
2512                 ecbdata.ws_rule = whitespace_rule(name_b);
2513                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2514                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2515                 ecbdata.opt = o;
2516                 ecbdata.header = header.len ? &header : NULL;
2517                 xpp.flags = o->xdl_opts;
2518                 xecfg.ctxlen = o->context;
2519                 xecfg.interhunkctxlen = o->interhunkcontext;
2520                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2521                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2522                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2523                 if (pe)
2524                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2525                 if (!diffopts)
2526                         ;
2527                 else if (skip_prefix(diffopts, "--unified=", &v))
2528                         xecfg.ctxlen = strtoul(v, NULL, 10);
2529                 else if (skip_prefix(diffopts, "-u", &v))
2530                         xecfg.ctxlen = strtoul(v, NULL, 10);
2531                 if (o->word_diff)
2532                         init_diff_words_data(&ecbdata, o, one, two);
2533                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2534                                   &xpp, &xecfg))
2535                         die("unable to generate diff for %s", one->path);
2536                 if (o->word_diff)
2537                         free_diff_words_data(&ecbdata);
2538                 if (textconv_one)
2539                         free(mf1.ptr);
2540                 if (textconv_two)
2541                         free(mf2.ptr);
2542                 xdiff_clear_find_func(&xecfg);
2543         }
2544
2545  free_ab_and_return:
2546         strbuf_release(&header);
2547         diff_free_filespec_data(one);
2548         diff_free_filespec_data(two);
2549         free(a_one);
2550         free(b_two);
2551         return;
2552 }
2553
2554 static void builtin_diffstat(const char *name_a, const char *name_b,
2555                              struct diff_filespec *one,
2556                              struct diff_filespec *two,
2557                              struct diffstat_t *diffstat,
2558                              struct diff_options *o,
2559                              struct diff_filepair *p)
2560 {
2561         mmfile_t mf1, mf2;
2562         struct diffstat_file *data;
2563         int same_contents;
2564         int complete_rewrite = 0;
2565
2566         if (!DIFF_PAIR_UNMERGED(p)) {
2567                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2568                         complete_rewrite = 1;
2569         }
2570
2571         data = diffstat_add(diffstat, name_a, name_b);
2572         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2573
2574         if (!one || !two) {
2575                 data->is_unmerged = 1;
2576                 return;
2577         }
2578
2579         same_contents = !oidcmp(&one->oid, &two->oid);
2580
2581         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2582                 data->is_binary = 1;
2583                 if (same_contents) {
2584                         data->added = 0;
2585                         data->deleted = 0;
2586                 } else {
2587                         data->added = diff_filespec_size(two);
2588                         data->deleted = diff_filespec_size(one);
2589                 }
2590         }
2591
2592         else if (complete_rewrite) {
2593                 diff_populate_filespec(one, 0);
2594                 diff_populate_filespec(two, 0);
2595                 data->deleted = count_lines(one->data, one->size);
2596                 data->added = count_lines(two->data, two->size);
2597         }
2598
2599         else if (!same_contents) {
2600                 /* Crazy xdl interfaces.. */
2601                 xpparam_t xpp;
2602                 xdemitconf_t xecfg;
2603
2604                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2605                         die("unable to read files to diff");
2606
2607                 memset(&xpp, 0, sizeof(xpp));
2608                 memset(&xecfg, 0, sizeof(xecfg));
2609                 xpp.flags = o->xdl_opts;
2610                 xecfg.ctxlen = o->context;
2611                 xecfg.interhunkctxlen = o->interhunkcontext;
2612                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2613                                   &xpp, &xecfg))
2614                         die("unable to generate diffstat for %s", one->path);
2615         }
2616
2617         diff_free_filespec_data(one);
2618         diff_free_filespec_data(two);
2619 }
2620
2621 static void builtin_checkdiff(const char *name_a, const char *name_b,
2622                               const char *attr_path,
2623                               struct diff_filespec *one,
2624                               struct diff_filespec *two,
2625                               struct diff_options *o)
2626 {
2627         mmfile_t mf1, mf2;
2628         struct checkdiff_t data;
2629
2630         if (!two)
2631                 return;
2632
2633         memset(&data, 0, sizeof(data));
2634         data.filename = name_b ? name_b : name_a;
2635         data.lineno = 0;
2636         data.o = o;
2637         data.ws_rule = whitespace_rule(attr_path);
2638         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2639
2640         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2641                 die("unable to read files to diff");
2642
2643         /*
2644          * All the other codepaths check both sides, but not checking
2645          * the "old" side here is deliberate.  We are checking the newly
2646          * introduced changes, and as long as the "new" side is text, we
2647          * can and should check what it introduces.
2648          */
2649         if (diff_filespec_is_binary(two))
2650                 goto free_and_return;
2651         else {
2652                 /* Crazy xdl interfaces.. */
2653                 xpparam_t xpp;
2654                 xdemitconf_t xecfg;
2655
2656                 memset(&xpp, 0, sizeof(xpp));
2657                 memset(&xecfg, 0, sizeof(xecfg));
2658                 xecfg.ctxlen = 1; /* at least one context line */
2659                 xpp.flags = 0;
2660                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2661                                   &xpp, &xecfg))
2662                         die("unable to generate checkdiff for %s", one->path);
2663
2664                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2665                         struct emit_callback ecbdata;
2666                         int blank_at_eof;
2667
2668                         ecbdata.ws_rule = data.ws_rule;
2669                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2670                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2671
2672                         if (blank_at_eof) {
2673                                 static char *err;
2674                                 if (!err)
2675                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2676                                 fprintf(o->file, "%s:%d: %s.\n",
2677                                         data.filename, blank_at_eof, err);
2678                                 data.status = 1; /* report errors */
2679                         }
2680                 }
2681         }
2682  free_and_return:
2683         diff_free_filespec_data(one);
2684         diff_free_filespec_data(two);
2685         if (data.status)
2686                 DIFF_OPT_SET(o, CHECK_FAILED);
2687 }
2688
2689 struct diff_filespec *alloc_filespec(const char *path)
2690 {
2691         struct diff_filespec *spec;
2692
2693         FLEXPTR_ALLOC_STR(spec, path, path);
2694         spec->count = 1;
2695         spec->is_binary = -1;
2696         return spec;
2697 }
2698
2699 void free_filespec(struct diff_filespec *spec)
2700 {
2701         if (!--spec->count) {
2702                 diff_free_filespec_data(spec);
2703                 free(spec);
2704         }
2705 }
2706
2707 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2708                    int sha1_valid, unsigned short mode)
2709 {
2710         if (mode) {
2711                 spec->mode = canon_mode(mode);
2712                 hashcpy(spec->oid.hash, sha1);
2713                 spec->oid_valid = sha1_valid;
2714         }
2715 }
2716
2717 /*
2718  * Given a name and sha1 pair, if the index tells us the file in
2719  * the work tree has that object contents, return true, so that
2720  * prepare_temp_file() does not have to inflate and extract.
2721  */
2722 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2723 {
2724         const struct cache_entry *ce;
2725         struct stat st;
2726         int pos, len;
2727
2728         /*
2729          * We do not read the cache ourselves here, because the
2730          * benchmark with my previous version that always reads cache
2731          * shows that it makes things worse for diff-tree comparing
2732          * two linux-2.6 kernel trees in an already checked out work
2733          * tree.  This is because most diff-tree comparisons deal with
2734          * only a small number of files, while reading the cache is
2735          * expensive for a large project, and its cost outweighs the
2736          * savings we get by not inflating the object to a temporary
2737          * file.  Practically, this code only helps when we are used
2738          * by diff-cache --cached, which does read the cache before
2739          * calling us.
2740          */
2741         if (!active_cache)
2742                 return 0;
2743
2744         /* We want to avoid the working directory if our caller
2745          * doesn't need the data in a normal file, this system
2746          * is rather slow with its stat/open/mmap/close syscalls,
2747          * and the object is contained in a pack file.  The pack
2748          * is probably already open and will be faster to obtain
2749          * the data through than the working directory.  Loose
2750          * objects however would tend to be slower as they need
2751          * to be individually opened and inflated.
2752          */
2753         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2754                 return 0;
2755
2756         /*
2757          * Similarly, if we'd have to convert the file contents anyway, that
2758          * makes the optimization not worthwhile.
2759          */
2760         if (!want_file && would_convert_to_git(name))
2761                 return 0;
2762
2763         len = strlen(name);
2764         pos = cache_name_pos(name, len);
2765         if (pos < 0)
2766                 return 0;
2767         ce = active_cache[pos];
2768
2769         /*
2770          * This is not the sha1 we are looking for, or
2771          * unreusable because it is not a regular file.
2772          */
2773         if (hashcmp(sha1, ce->oid.hash) || !S_ISREG(ce->ce_mode))
2774                 return 0;
2775
2776         /*
2777          * If ce is marked as "assume unchanged", there is no
2778          * guarantee that work tree matches what we are looking for.
2779          */
2780         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2781                 return 0;
2782
2783         /*
2784          * If ce matches the file in the work tree, we can reuse it.
2785          */
2786         if (ce_uptodate(ce) ||
2787             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2788                 return 1;
2789
2790         return 0;
2791 }
2792
2793 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2794 {
2795         struct strbuf buf = STRBUF_INIT;
2796         char *dirty = "";
2797
2798         /* Are we looking at the work tree? */
2799         if (s->dirty_submodule)
2800                 dirty = "-dirty";
2801
2802         strbuf_addf(&buf, "Subproject commit %s%s\n",
2803                     oid_to_hex(&s->oid), dirty);
2804         s->size = buf.len;
2805         if (size_only) {
2806                 s->data = NULL;
2807                 strbuf_release(&buf);
2808         } else {
2809                 s->data = strbuf_detach(&buf, NULL);
2810                 s->should_free = 1;
2811         }
2812         return 0;
2813 }
2814
2815 /*
2816  * While doing rename detection and pickaxe operation, we may need to
2817  * grab the data for the blob (or file) for our own in-core comparison.
2818  * diff_filespec has data and size fields for this purpose.
2819  */
2820 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2821 {
2822         int size_only = flags & CHECK_SIZE_ONLY;
2823         int err = 0;
2824         /*
2825          * demote FAIL to WARN to allow inspecting the situation
2826          * instead of refusing.
2827          */
2828         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2829                                     ? SAFE_CRLF_WARN
2830                                     : safe_crlf);
2831
2832         if (!DIFF_FILE_VALID(s))
2833                 die("internal error: asking to populate invalid file.");
2834         if (S_ISDIR(s->mode))
2835                 return -1;
2836
2837         if (s->data)
2838                 return 0;
2839
2840         if (size_only && 0 < s->size)
2841                 return 0;
2842
2843         if (S_ISGITLINK(s->mode))
2844                 return diff_populate_gitlink(s, size_only);
2845
2846         if (!s->oid_valid ||
2847             reuse_worktree_file(s->path, s->oid.hash, 0)) {
2848                 struct strbuf buf = STRBUF_INIT;
2849                 struct stat st;
2850                 int fd;
2851
2852                 if (lstat(s->path, &st) < 0) {
2853                         if (errno == ENOENT) {
2854                         err_empty:
2855                                 err = -1;
2856                         empty:
2857                                 s->data = (char *)"";
2858                                 s->size = 0;
2859                                 return err;
2860                         }
2861                 }
2862                 s->size = xsize_t(st.st_size);
2863                 if (!s->size)
2864                         goto empty;
2865                 if (S_ISLNK(st.st_mode)) {
2866                         struct strbuf sb = STRBUF_INIT;
2867
2868                         if (strbuf_readlink(&sb, s->path, s->size))
2869                                 goto err_empty;
2870                         s->size = sb.len;
2871                         s->data = strbuf_detach(&sb, NULL);
2872                         s->should_free = 1;
2873                         return 0;
2874                 }
2875                 if (size_only)
2876                         return 0;
2877                 if ((flags & CHECK_BINARY) &&
2878                     s->size > big_file_threshold && s->is_binary == -1) {
2879                         s->is_binary = 1;
2880                         return 0;
2881                 }
2882                 fd = open(s->path, O_RDONLY);
2883                 if (fd < 0)
2884                         goto err_empty;
2885                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2886                 close(fd);
2887                 s->should_munmap = 1;
2888
2889                 /*
2890                  * Convert from working tree format to canonical git format
2891                  */
2892                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2893                         size_t size = 0;
2894                         munmap(s->data, s->size);
2895                         s->should_munmap = 0;
2896                         s->data = strbuf_detach(&buf, &size);
2897                         s->size = size;
2898                         s->should_free = 1;
2899                 }
2900         }
2901         else {
2902                 enum object_type type;
2903                 if (size_only || (flags & CHECK_BINARY)) {
2904                         type = sha1_object_info(s->oid.hash, &s->size);
2905                         if (type < 0)
2906                                 die("unable to read %s",
2907                                     oid_to_hex(&s->oid));
2908                         if (size_only)
2909                                 return 0;
2910                         if (s->size > big_file_threshold && s->is_binary == -1) {
2911                                 s->is_binary = 1;
2912                                 return 0;
2913                         }
2914                 }
2915                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
2916                 if (!s->data)
2917                         die("unable to read %s", oid_to_hex(&s->oid));
2918                 s->should_free = 1;
2919         }
2920         return 0;
2921 }
2922
2923 void diff_free_filespec_blob(struct diff_filespec *s)
2924 {
2925         if (s->should_free)
2926                 free(s->data);
2927         else if (s->should_munmap)
2928                 munmap(s->data, s->size);
2929
2930         if (s->should_free || s->should_munmap) {
2931                 s->should_free = s->should_munmap = 0;
2932                 s->data = NULL;
2933         }
2934 }
2935
2936 void diff_free_filespec_data(struct diff_filespec *s)
2937 {
2938         diff_free_filespec_blob(s);
2939         free(s->cnt_data);
2940         s->cnt_data = NULL;
2941 }
2942
2943 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2944                            void *blob,
2945                            unsigned long size,
2946                            const struct object_id *oid,
2947                            int mode)
2948 {
2949         int fd;
2950         struct strbuf buf = STRBUF_INIT;
2951         struct strbuf template = STRBUF_INIT;
2952         char *path_dup = xstrdup(path);
2953         const char *base = basename(path_dup);
2954
2955         /* Generate "XXXXXX_basename.ext" */
2956         strbuf_addstr(&template, "XXXXXX_");
2957         strbuf_addstr(&template, base);
2958
2959         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2960         if (fd < 0)
2961                 die_errno("unable to create temp-file");
2962         if (convert_to_working_tree(path,
2963                         (const char *)blob, (size_t)size, &buf)) {
2964                 blob = buf.buf;
2965                 size = buf.len;
2966         }
2967         if (write_in_full(fd, blob, size) != size)
2968                 die_errno("unable to write temp-file");
2969         close_tempfile(&temp->tempfile);
2970         temp->name = get_tempfile_path(&temp->tempfile);
2971         oid_to_hex_r(temp->hex, oid);
2972         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
2973         strbuf_release(&buf);
2974         strbuf_release(&template);
2975         free(path_dup);
2976 }
2977
2978 static struct diff_tempfile *prepare_temp_file(const char *name,
2979                 struct diff_filespec *one)
2980 {
2981         struct diff_tempfile *temp = claim_diff_tempfile();
2982
2983         if (!DIFF_FILE_VALID(one)) {
2984         not_a_valid_file:
2985                 /* A '-' entry produces this for file-2, and
2986                  * a '+' entry produces this for file-1.
2987                  */
2988                 temp->name = "/dev/null";
2989                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
2990                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
2991                 return temp;
2992         }
2993
2994         if (!S_ISGITLINK(one->mode) &&
2995             (!one->oid_valid ||
2996              reuse_worktree_file(name, one->oid.hash, 1))) {
2997                 struct stat st;
2998                 if (lstat(name, &st) < 0) {
2999                         if (errno == ENOENT)
3000                                 goto not_a_valid_file;
3001                         die_errno("stat(%s)", name);
3002                 }
3003                 if (S_ISLNK(st.st_mode)) {
3004                         struct strbuf sb = STRBUF_INIT;
3005                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3006                                 die_errno("readlink(%s)", name);
3007                         prep_temp_blob(name, temp, sb.buf, sb.len,
3008                                        (one->oid_valid ?
3009                                         &one->oid : &null_oid),
3010                                        (one->oid_valid ?
3011                                         one->mode : S_IFLNK));
3012                         strbuf_release(&sb);
3013                 }
3014                 else {
3015                         /* we can borrow from the file in the work tree */
3016                         temp->name = name;
3017                         if (!one->oid_valid)
3018                                 sha1_to_hex_r(temp->hex, null_sha1);
3019                         else
3020                                 sha1_to_hex_r(temp->hex, one->oid.hash);
3021                         /* Even though we may sometimes borrow the
3022                          * contents from the work tree, we always want
3023                          * one->mode.  mode is trustworthy even when
3024                          * !(one->sha1_valid), as long as
3025                          * DIFF_FILE_VALID(one).
3026                          */
3027                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3028                 }
3029                 return temp;
3030         }
3031         else {
3032                 if (diff_populate_filespec(one, 0))
3033                         die("cannot read data blob for %s", one->path);
3034                 prep_temp_blob(name, temp, one->data, one->size,
3035                                &one->oid, one->mode);
3036         }
3037         return temp;
3038 }
3039
3040 static void add_external_diff_name(struct argv_array *argv,
3041                                    const char *name,
3042                                    struct diff_filespec *df)
3043 {
3044         struct diff_tempfile *temp = prepare_temp_file(name, df);
3045         argv_array_push(argv, temp->name);
3046         argv_array_push(argv, temp->hex);
3047         argv_array_push(argv, temp->mode);
3048 }
3049
3050 /* An external diff command takes:
3051  *
3052  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3053  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3054  *
3055  */
3056 static void run_external_diff(const char *pgm,
3057                               const char *name,
3058                               const char *other,
3059                               struct diff_filespec *one,
3060                               struct diff_filespec *two,
3061                               const char *xfrm_msg,
3062                               int complete_rewrite,
3063                               struct diff_options *o)
3064 {
3065         struct argv_array argv = ARGV_ARRAY_INIT;
3066         struct argv_array env = ARGV_ARRAY_INIT;
3067         struct diff_queue_struct *q = &diff_queued_diff;
3068
3069         argv_array_push(&argv, pgm);
3070         argv_array_push(&argv, name);
3071
3072         if (one && two) {
3073                 add_external_diff_name(&argv, name, one);
3074                 if (!other)
3075                         add_external_diff_name(&argv, name, two);
3076                 else {
3077                         add_external_diff_name(&argv, other, two);
3078                         argv_array_push(&argv, other);
3079                         argv_array_push(&argv, xfrm_msg);
3080                 }
3081         }
3082
3083         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3084         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3085
3086         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3087                 die(_("external diff died, stopping at %s"), name);
3088
3089         remove_tempfile();
3090         argv_array_clear(&argv);
3091         argv_array_clear(&env);
3092 }
3093
3094 static int similarity_index(struct diff_filepair *p)
3095 {
3096         return p->score * 100 / MAX_SCORE;
3097 }
3098
3099 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3100 {
3101         if (startup_info->have_repository)
3102                 return find_unique_abbrev(oid->hash, abbrev);
3103         else {
3104                 char *hex = oid_to_hex(oid);
3105                 if (abbrev < 0)
3106                         abbrev = FALLBACK_DEFAULT_ABBREV;
3107                 if (abbrev > GIT_SHA1_HEXSZ)
3108                         die("BUG: oid abbreviation out of range: %d", abbrev);
3109                 if (abbrev)
3110                         hex[abbrev] = '\0';
3111                 return hex;
3112         }
3113 }
3114
3115 static void fill_metainfo(struct strbuf *msg,
3116                           const char *name,
3117                           const char *other,
3118                           struct diff_filespec *one,
3119                           struct diff_filespec *two,
3120                           struct diff_options *o,
3121                           struct diff_filepair *p,
3122                           int *must_show_header,
3123                           int use_color)
3124 {
3125         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3126         const char *reset = diff_get_color(use_color, DIFF_RESET);
3127         const char *line_prefix = diff_line_prefix(o);
3128
3129         *must_show_header = 1;
3130         strbuf_init(msg, PATH_MAX * 2 + 300);
3131         switch (p->status) {
3132         case DIFF_STATUS_COPIED:
3133                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3134                             line_prefix, set, similarity_index(p));
3135                 strbuf_addf(msg, "%s\n%s%scopy from ",
3136                             reset,  line_prefix, set);
3137                 quote_c_style(name, msg, NULL, 0);
3138                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3139                 quote_c_style(other, msg, NULL, 0);
3140                 strbuf_addf(msg, "%s\n", reset);
3141                 break;
3142         case DIFF_STATUS_RENAMED:
3143                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3144                             line_prefix, set, similarity_index(p));
3145                 strbuf_addf(msg, "%s\n%s%srename from ",
3146                             reset, line_prefix, set);
3147                 quote_c_style(name, msg, NULL, 0);
3148                 strbuf_addf(msg, "%s\n%s%srename to ",
3149                             reset, line_prefix, set);
3150                 quote_c_style(other, msg, NULL, 0);
3151                 strbuf_addf(msg, "%s\n", reset);
3152                 break;
3153         case DIFF_STATUS_MODIFIED:
3154                 if (p->score) {
3155                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3156                                     line_prefix,
3157                                     set, similarity_index(p), reset);
3158                         break;
3159                 }
3160                 /* fallthru */
3161         default:
3162                 *must_show_header = 0;
3163         }
3164         if (one && two && oidcmp(&one->oid, &two->oid)) {
3165                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3166
3167                 if (DIFF_OPT_TST(o, BINARY)) {
3168                         mmfile_t mf;
3169                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3170                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3171                                 abbrev = 40;
3172                 }
3173                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3174                             diff_abbrev_oid(&one->oid, abbrev),
3175                             diff_abbrev_oid(&two->oid, abbrev));
3176                 if (one->mode == two->mode)
3177                         strbuf_addf(msg, " %06o", one->mode);
3178                 strbuf_addf(msg, "%s\n", reset);
3179         }
3180 }
3181
3182 static void run_diff_cmd(const char *pgm,
3183                          const char *name,
3184                          const char *other,
3185                          const char *attr_path,
3186                          struct diff_filespec *one,
3187                          struct diff_filespec *two,
3188                          struct strbuf *msg,
3189                          struct diff_options *o,
3190                          struct diff_filepair *p)
3191 {
3192         const char *xfrm_msg = NULL;
3193         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3194         int must_show_header = 0;
3195
3196
3197         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3198                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3199                 if (drv && drv->external)
3200                         pgm = drv->external;
3201         }
3202
3203         if (msg) {
3204                 /*
3205                  * don't use colors when the header is intended for an
3206                  * external diff driver
3207                  */
3208                 fill_metainfo(msg, name, other, one, two, o, p,
3209                               &must_show_header,
3210                               want_color(o->use_color) && !pgm);
3211                 xfrm_msg = msg->len ? msg->buf : NULL;
3212         }
3213
3214         if (pgm) {
3215                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3216                                   complete_rewrite, o);
3217                 return;
3218         }
3219         if (one && two)
3220                 builtin_diff(name, other ? other : name,
3221                              one, two, xfrm_msg, must_show_header,
3222                              o, complete_rewrite);
3223         else
3224                 fprintf(o->file, "* Unmerged path %s\n", name);
3225 }
3226
3227 static void diff_fill_sha1_info(struct diff_filespec *one)
3228 {
3229         if (DIFF_FILE_VALID(one)) {
3230                 if (!one->oid_valid) {
3231                         struct stat st;
3232                         if (one->is_stdin) {
3233                                 oidclr(&one->oid);
3234                                 return;
3235                         }
3236                         if (lstat(one->path, &st) < 0)
3237                                 die_errno("stat '%s'", one->path);
3238                         if (index_path(one->oid.hash, one->path, &st, 0))
3239                                 die("cannot hash %s", one->path);
3240                 }
3241         }
3242         else
3243                 oidclr(&one->oid);
3244 }
3245
3246 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3247 {
3248         /* Strip the prefix but do not molest /dev/null and absolute paths */
3249         if (*namep && **namep != '/') {
3250                 *namep += prefix_length;
3251                 if (**namep == '/')
3252                         ++*namep;
3253         }
3254         if (*otherp && **otherp != '/') {
3255                 *otherp += prefix_length;
3256                 if (**otherp == '/')
3257                         ++*otherp;
3258         }
3259 }
3260
3261 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3262 {
3263         const char *pgm = external_diff();
3264         struct strbuf msg;
3265         struct diff_filespec *one = p->one;
3266         struct diff_filespec *two = p->two;
3267         const char *name;
3268         const char *other;
3269         const char *attr_path;
3270
3271         name  = p->one->path;
3272         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3273         attr_path = name;
3274         if (o->prefix_length)
3275                 strip_prefix(o->prefix_length, &name, &other);
3276
3277         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3278                 pgm = NULL;
3279
3280         if (DIFF_PAIR_UNMERGED(p)) {
3281                 run_diff_cmd(pgm, name, NULL, attr_path,
3282                              NULL, NULL, NULL, o, p);
3283                 return;
3284         }
3285
3286         diff_fill_sha1_info(one);
3287         diff_fill_sha1_info(two);
3288
3289         if (!pgm &&
3290             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3291             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3292                 /*
3293                  * a filepair that changes between file and symlink
3294                  * needs to be split into deletion and creation.
3295                  */
3296                 struct diff_filespec *null = alloc_filespec(two->path);
3297                 run_diff_cmd(NULL, name, other, attr_path,
3298                              one, null, &msg, o, p);
3299                 free(null);
3300                 strbuf_release(&msg);
3301
3302                 null = alloc_filespec(one->path);
3303                 run_diff_cmd(NULL, name, other, attr_path,
3304                              null, two, &msg, o, p);
3305                 free(null);
3306         }
3307         else
3308                 run_diff_cmd(pgm, name, other, attr_path,
3309                              one, two, &msg, o, p);
3310
3311         strbuf_release(&msg);
3312 }
3313
3314 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3315                          struct diffstat_t *diffstat)
3316 {
3317         const char *name;
3318         const char *other;
3319
3320         if (DIFF_PAIR_UNMERGED(p)) {
3321                 /* unmerged */
3322                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3323                 return;
3324         }
3325
3326         name = p->one->path;
3327         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3328
3329         if (o->prefix_length)
3330                 strip_prefix(o->prefix_length, &name, &other);
3331
3332         diff_fill_sha1_info(p->one);
3333         diff_fill_sha1_info(p->two);
3334
3335         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3336 }
3337
3338 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3339 {
3340         const char *name;
3341         const char *other;
3342         const char *attr_path;
3343
3344         if (DIFF_PAIR_UNMERGED(p)) {
3345                 /* unmerged */
3346                 return;
3347         }
3348
3349         name = p->one->path;
3350         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3351         attr_path = other ? other : name;
3352
3353         if (o->prefix_length)
3354                 strip_prefix(o->prefix_length, &name, &other);
3355
3356         diff_fill_sha1_info(p->one);
3357         diff_fill_sha1_info(p->two);
3358
3359         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3360 }
3361
3362 void diff_setup(struct diff_options *options)
3363 {
3364         memcpy(options, &default_diff_options, sizeof(*options));
3365
3366         options->file = stdout;
3367
3368         options->abbrev = DEFAULT_ABBREV;
3369         options->line_termination = '\n';
3370         options->break_opt = -1;
3371         options->rename_limit = -1;
3372         options->dirstat_permille = diff_dirstat_permille_default;
3373         options->context = diff_context_default;
3374         options->ws_error_highlight = ws_error_highlight_default;
3375         DIFF_OPT_SET(options, RENAME_EMPTY);
3376
3377         /* pathchange left =NULL by default */
3378         options->change = diff_change;
3379         options->add_remove = diff_addremove;
3380         options->use_color = diff_use_color_default;
3381         options->detect_rename = diff_detect_rename_default;
3382         options->xdl_opts |= diff_algorithm;
3383         if (diff_indent_heuristic)
3384                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3385         else if (diff_compaction_heuristic)
3386                 DIFF_XDL_SET(options, COMPACTION_HEURISTIC);
3387
3388         options->orderfile = diff_order_file_cfg;
3389
3390         if (diff_no_prefix) {
3391                 options->a_prefix = options->b_prefix = "";
3392         } else if (!diff_mnemonic_prefix) {
3393                 options->a_prefix = "a/";
3394                 options->b_prefix = "b/";
3395         }
3396 }
3397
3398 void diff_setup_done(struct diff_options *options)
3399 {
3400         int count = 0;
3401
3402         if (options->set_default)
3403                 options->set_default(options);
3404
3405         if (options->output_format & DIFF_FORMAT_NAME)
3406                 count++;
3407         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3408                 count++;
3409         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3410                 count++;
3411         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3412                 count++;
3413         if (count > 1)
3414                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
3415
3416         /*
3417          * Most of the time we can say "there are changes"
3418          * only by checking if there are changed paths, but
3419          * --ignore-whitespace* options force us to look
3420          * inside contents.
3421          */
3422
3423         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3424             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3425             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3426                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3427         else
3428                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3429
3430         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3431                 options->detect_rename = DIFF_DETECT_COPY;
3432
3433         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3434                 options->prefix = NULL;
3435         if (options->prefix)
3436                 options->prefix_length = strlen(options->prefix);
3437         else
3438                 options->prefix_length = 0;
3439
3440         if (options->output_format & (DIFF_FORMAT_NAME |
3441                                       DIFF_FORMAT_NAME_STATUS |
3442                                       DIFF_FORMAT_CHECKDIFF |
3443                                       DIFF_FORMAT_NO_OUTPUT))
3444                 options->output_format &= ~(DIFF_FORMAT_RAW |
3445                                             DIFF_FORMAT_NUMSTAT |
3446                                             DIFF_FORMAT_DIFFSTAT |
3447                                             DIFF_FORMAT_SHORTSTAT |
3448                                             DIFF_FORMAT_DIRSTAT |
3449                                             DIFF_FORMAT_SUMMARY |
3450                                             DIFF_FORMAT_PATCH);
3451
3452         /*
3453          * These cases always need recursive; we do not drop caller-supplied
3454          * recursive bits for other formats here.
3455          */
3456         if (options->output_format & (DIFF_FORMAT_PATCH |
3457                                       DIFF_FORMAT_NUMSTAT |
3458                                       DIFF_FORMAT_DIFFSTAT |
3459                                       DIFF_FORMAT_SHORTSTAT |
3460                                       DIFF_FORMAT_DIRSTAT |
3461                                       DIFF_FORMAT_SUMMARY |
3462                                       DIFF_FORMAT_CHECKDIFF))
3463                 DIFF_OPT_SET(options, RECURSIVE);
3464         /*
3465          * Also pickaxe would not work very well if you do not say recursive
3466          */
3467         if (options->pickaxe)
3468                 DIFF_OPT_SET(options, RECURSIVE);
3469         /*
3470          * When patches are generated, submodules diffed against the work tree
3471          * must be checked for dirtiness too so it can be shown in the output
3472          */
3473         if (options->output_format & DIFF_FORMAT_PATCH)
3474                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3475
3476         if (options->detect_rename && options->rename_limit < 0)
3477                 options->rename_limit = diff_rename_limit_default;
3478         if (options->setup & DIFF_SETUP_USE_CACHE) {
3479                 if (!active_cache)
3480                         /* read-cache does not die even when it fails
3481                          * so it is safe for us to do this here.  Also
3482                          * it does not smudge active_cache or active_nr
3483                          * when it fails, so we do not have to worry about
3484                          * cleaning it up ourselves either.
3485                          */
3486                         read_cache();
3487         }
3488         if (40 < options->abbrev)
3489                 options->abbrev = 40; /* full */
3490
3491         /*
3492          * It does not make sense to show the first hit we happened
3493          * to have found.  It does not make sense not to return with
3494          * exit code in such a case either.
3495          */
3496         if (DIFF_OPT_TST(options, QUICK)) {
3497                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3498                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3499         }
3500
3501         options->diff_path_counter = 0;
3502
3503         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3504                 die(_("--follow requires exactly one pathspec"));
3505 }
3506
3507 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3508 {
3509         char c, *eq;
3510         int len;
3511
3512         if (*arg != '-')
3513                 return 0;
3514         c = *++arg;
3515         if (!c)
3516                 return 0;
3517         if (c == arg_short) {
3518                 c = *++arg;
3519                 if (!c)
3520                         return 1;
3521                 if (val && isdigit(c)) {
3522                         char *end;
3523                         int n = strtoul(arg, &end, 10);
3524                         if (*end)
3525                                 return 0;
3526                         *val = n;
3527                         return 1;
3528                 }
3529                 return 0;
3530         }
3531         if (c != '-')
3532                 return 0;
3533         arg++;
3534         eq = strchrnul(arg, '=');
3535         len = eq - arg;
3536         if (!len || strncmp(arg, arg_long, len))
3537                 return 0;
3538         if (*eq) {
3539                 int n;
3540                 char *end;
3541                 if (!isdigit(*++eq))
3542                         return 0;
3543                 n = strtoul(eq, &end, 10);
3544                 if (*end)
3545                         return 0;
3546                 *val = n;
3547         }
3548         return 1;
3549 }
3550
3551 static int diff_scoreopt_parse(const char *opt);
3552
3553 static inline int short_opt(char opt, const char **argv,
3554                             const char **optarg)
3555 {
3556         const char *arg = argv[0];
3557         if (arg[0] != '-' || arg[1] != opt)
3558                 return 0;
3559         if (arg[2] != '\0') {
3560                 *optarg = arg + 2;
3561                 return 1;
3562         }
3563         if (!argv[1])
3564                 die("Option '%c' requires a value", opt);
3565         *optarg = argv[1];
3566         return 2;
3567 }
3568
3569 int parse_long_opt(const char *opt, const char **argv,
3570                    const char **optarg)
3571 {
3572         const char *arg = argv[0];
3573         if (!skip_prefix(arg, "--", &arg))
3574                 return 0;
3575         if (!skip_prefix(arg, opt, &arg))
3576                 return 0;
3577         if (*arg == '=') { /* stuck form: --option=value */
3578                 *optarg = arg + 1;
3579                 return 1;
3580         }
3581         if (*arg != '\0')
3582                 return 0;
3583         /* separate form: --option value */
3584         if (!argv[1])
3585                 die("Option '--%s' requires a value", opt);
3586         *optarg = argv[1];
3587         return 2;
3588 }
3589
3590 static int stat_opt(struct diff_options *options, const char **av)
3591 {
3592         const char *arg = av[0];
3593         char *end;
3594         int width = options->stat_width;
3595         int name_width = options->stat_name_width;
3596         int graph_width = options->stat_graph_width;
3597         int count = options->stat_count;
3598         int argcount = 1;
3599
3600         if (!skip_prefix(arg, "--stat", &arg))
3601                 die("BUG: stat option does not begin with --stat: %s", arg);
3602         end = (char *)arg;
3603
3604         switch (*arg) {
3605         case '-':
3606                 if (skip_prefix(arg, "-width", &arg)) {
3607                         if (*arg == '=')
3608                                 width = strtoul(arg + 1, &end, 10);
3609                         else if (!*arg && !av[1])
3610                                 die_want_option("--stat-width");
3611                         else if (!*arg) {
3612                                 width = strtoul(av[1], &end, 10);
3613                                 argcount = 2;
3614                         }
3615                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3616                         if (*arg == '=')
3617                                 name_width = strtoul(arg + 1, &end, 10);
3618                         else if (!*arg && !av[1])
3619                                 die_want_option("--stat-name-width");
3620                         else if (!*arg) {
3621                                 name_width = strtoul(av[1], &end, 10);
3622                                 argcount = 2;
3623                         }
3624                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3625                         if (*arg == '=')
3626                                 graph_width = strtoul(arg + 1, &end, 10);
3627                         else if (!*arg && !av[1])
3628                                 die_want_option("--stat-graph-width");
3629                         else if (!*arg) {
3630                                 graph_width = strtoul(av[1], &end, 10);
3631                                 argcount = 2;
3632                         }
3633                 } else if (skip_prefix(arg, "-count", &arg)) {
3634                         if (*arg == '=')
3635                                 count = strtoul(arg + 1, &end, 10);
3636                         else if (!*arg && !av[1])
3637                                 die_want_option("--stat-count");
3638                         else if (!*arg) {
3639                                 count = strtoul(av[1], &end, 10);
3640                                 argcount = 2;
3641                         }
3642                 }
3643                 break;
3644         case '=':
3645                 width = strtoul(arg+1, &end, 10);
3646                 if (*end == ',')
3647                         name_width = strtoul(end+1, &end, 10);
3648                 if (*end == ',')
3649                         count = strtoul(end+1, &end, 10);
3650         }
3651
3652         /* Important! This checks all the error cases! */
3653         if (*end)
3654                 return 0;
3655         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3656         options->stat_name_width = name_width;
3657         options->stat_graph_width = graph_width;
3658         options->stat_width = width;
3659         options->stat_count = count;
3660         return argcount;
3661 }
3662
3663 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3664 {
3665         struct strbuf errmsg = STRBUF_INIT;
3666         if (parse_dirstat_params(options, params, &errmsg))
3667                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3668                     errmsg.buf);
3669         strbuf_release(&errmsg);
3670         /*
3671          * The caller knows a dirstat-related option is given from the command
3672          * line; allow it to say "return this_function();"
3673          */
3674         options->output_format |= DIFF_FORMAT_DIRSTAT;
3675         return 1;
3676 }
3677
3678 static int parse_submodule_opt(struct diff_options *options, const char *value)
3679 {
3680         if (parse_submodule_params(options, value))
3681                 die(_("Failed to parse --submodule option parameter: '%s'"),
3682                         value);
3683         return 1;
3684 }
3685
3686 static const char diff_status_letters[] = {
3687         DIFF_STATUS_ADDED,
3688         DIFF_STATUS_COPIED,
3689         DIFF_STATUS_DELETED,
3690         DIFF_STATUS_MODIFIED,
3691         DIFF_STATUS_RENAMED,
3692         DIFF_STATUS_TYPE_CHANGED,
3693         DIFF_STATUS_UNKNOWN,
3694         DIFF_STATUS_UNMERGED,
3695         DIFF_STATUS_FILTER_AON,
3696         DIFF_STATUS_FILTER_BROKEN,
3697         '\0',
3698 };
3699
3700 static unsigned int filter_bit['Z' + 1];
3701
3702 static void prepare_filter_bits(void)
3703 {
3704         int i;
3705
3706         if (!filter_bit[DIFF_STATUS_ADDED]) {
3707                 for (i = 0; diff_status_letters[i]; i++)
3708                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3709         }
3710 }
3711
3712 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3713 {
3714         return opt->filter & filter_bit[(int) status];
3715 }
3716
3717 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3718 {
3719         int i, optch;
3720
3721         prepare_filter_bits();
3722
3723         /*
3724          * If there is a negation e.g. 'd' in the input, and we haven't
3725          * initialized the filter field with another --diff-filter, start
3726          * from full set of bits, except for AON.
3727          */
3728         if (!opt->filter) {
3729                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3730                         if (optch < 'a' || 'z' < optch)
3731                                 continue;
3732                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3733                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3734                         break;
3735                 }
3736         }
3737
3738         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3739                 unsigned int bit;
3740                 int negate;
3741
3742                 if ('a' <= optch && optch <= 'z') {
3743                         negate = 1;
3744                         optch = toupper(optch);
3745                 } else {
3746                         negate = 0;
3747                 }
3748
3749                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3750                 if (!bit)
3751                         return optarg[i];
3752                 if (negate)
3753                         opt->filter &= ~bit;
3754                 else
3755                         opt->filter |= bit;
3756         }
3757         return 0;
3758 }
3759
3760 static void enable_patch_output(int *fmt) {
3761         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3762         *fmt |= DIFF_FORMAT_PATCH;
3763 }
3764
3765 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
3766 {
3767         int val = parse_ws_error_highlight(arg);
3768
3769         if (val < 0) {
3770                 error("unknown value after ws-error-highlight=%.*s",
3771                       -1 - val, arg);
3772                 return 0;
3773         }
3774         opt->ws_error_highlight = val;
3775         return 1;
3776 }
3777
3778 int diff_opt_parse(struct diff_options *options,
3779                    const char **av, int ac, const char *prefix)
3780 {
3781         const char *arg = av[0];
3782         const char *optarg;
3783         int argcount;
3784
3785         if (!prefix)
3786                 prefix = "";
3787
3788         /* Output format options */
3789         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3790             || opt_arg(arg, 'U', "unified", &options->context))
3791                 enable_patch_output(&options->output_format);
3792         else if (!strcmp(arg, "--raw"))
3793                 options->output_format |= DIFF_FORMAT_RAW;
3794         else if (!strcmp(arg, "--patch-with-raw")) {
3795                 enable_patch_output(&options->output_format);
3796                 options->output_format |= DIFF_FORMAT_RAW;
3797         } else if (!strcmp(arg, "--numstat"))
3798                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3799         else if (!strcmp(arg, "--shortstat"))
3800                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3801         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3802                 return parse_dirstat_opt(options, "");
3803         else if (skip_prefix(arg, "-X", &arg))
3804                 return parse_dirstat_opt(options, arg);
3805         else if (skip_prefix(arg, "--dirstat=", &arg))
3806                 return parse_dirstat_opt(options, arg);
3807         else if (!strcmp(arg, "--cumulative"))
3808                 return parse_dirstat_opt(options, "cumulative");
3809         else if (!strcmp(arg, "--dirstat-by-file"))
3810                 return parse_dirstat_opt(options, "files");
3811         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3812                 parse_dirstat_opt(options, "files");
3813                 return parse_dirstat_opt(options, arg);
3814         }
3815         else if (!strcmp(arg, "--check"))
3816                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3817         else if (!strcmp(arg, "--summary"))
3818                 options->output_format |= DIFF_FORMAT_SUMMARY;
3819         else if (!strcmp(arg, "--patch-with-stat")) {
3820                 enable_patch_output(&options->output_format);
3821                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3822         } else if (!strcmp(arg, "--name-only"))
3823                 options->output_format |= DIFF_FORMAT_NAME;
3824         else if (!strcmp(arg, "--name-status"))
3825                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3826         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3827                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3828         else if (starts_with(arg, "--stat"))
3829                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3830                 return stat_opt(options, av);
3831
3832         /* renames options */
3833         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3834                  !strcmp(arg, "--break-rewrites")) {
3835                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3836                         return error("invalid argument to -B: %s", arg+2);
3837         }
3838         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3839                  !strcmp(arg, "--find-renames")) {
3840                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3841                         return error("invalid argument to -M: %s", arg+2);
3842                 options->detect_rename = DIFF_DETECT_RENAME;
3843         }
3844         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3845                 options->irreversible_delete = 1;
3846         }
3847         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3848                  !strcmp(arg, "--find-copies")) {
3849                 if (options->detect_rename == DIFF_DETECT_COPY)
3850                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3851                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3852                         return error("invalid argument to -C: %s", arg+2);
3853                 options->detect_rename = DIFF_DETECT_COPY;
3854         }
3855         else if (!strcmp(arg, "--no-renames"))
3856                 options->detect_rename = 0;
3857         else if (!strcmp(arg, "--rename-empty"))
3858                 DIFF_OPT_SET(options, RENAME_EMPTY);
3859         else if (!strcmp(arg, "--no-rename-empty"))
3860                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3861         else if (!strcmp(arg, "--relative"))
3862                 DIFF_OPT_SET(options, RELATIVE_NAME);
3863         else if (skip_prefix(arg, "--relative=", &arg)) {
3864                 DIFF_OPT_SET(options, RELATIVE_NAME);
3865                 options->prefix = arg;
3866         }
3867
3868         /* xdiff options */
3869         else if (!strcmp(arg, "--minimal"))
3870                 DIFF_XDL_SET(options, NEED_MINIMAL);
3871         else if (!strcmp(arg, "--no-minimal"))
3872                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3873         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3874                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3875         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3876                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3877         else if (!strcmp(arg, "--ignore-space-at-eol"))
3878                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3879         else if (!strcmp(arg, "--ignore-blank-lines"))
3880                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3881         else if (!strcmp(arg, "--indent-heuristic")) {
3882                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3883                 DIFF_XDL_CLR(options, COMPACTION_HEURISTIC);
3884         } else if (!strcmp(arg, "--no-indent-heuristic"))
3885                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
3886         else if (!strcmp(arg, "--compaction-heuristic")) {
3887                 DIFF_XDL_SET(options, COMPACTION_HEURISTIC);
3888                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
3889         } else if (!strcmp(arg, "--no-compaction-heuristic"))
3890                 DIFF_XDL_CLR(options, COMPACTION_HEURISTIC);
3891         else if (!strcmp(arg, "--patience"))
3892                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3893         else if (!strcmp(arg, "--histogram"))
3894                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3895         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3896                 long value = parse_algorithm_value(optarg);
3897                 if (value < 0)
3898                         return error("option diff-algorithm accepts \"myers\", "
3899                                      "\"minimal\", \"patience\" and \"histogram\"");
3900                 /* clear out previous settings */
3901                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3902                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3903                 options->xdl_opts |= value;
3904                 return argcount;
3905         }
3906
3907         /* flags options */
3908         else if (!strcmp(arg, "--binary")) {
3909                 enable_patch_output(&options->output_format);
3910                 DIFF_OPT_SET(options, BINARY);
3911         }
3912         else if (!strcmp(arg, "--full-index"))
3913                 DIFF_OPT_SET(options, FULL_INDEX);
3914         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3915                 DIFF_OPT_SET(options, TEXT);
3916         else if (!strcmp(arg, "-R"))
3917                 DIFF_OPT_SET(options, REVERSE_DIFF);
3918         else if (!strcmp(arg, "--find-copies-harder"))
3919                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3920         else if (!strcmp(arg, "--follow"))
3921                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3922         else if (!strcmp(arg, "--no-follow")) {
3923                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3924                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3925         } else if (!strcmp(arg, "--color"))
3926                 options->use_color = 1;
3927         else if (skip_prefix(arg, "--color=", &arg)) {
3928                 int value = git_config_colorbool(NULL, arg);
3929                 if (value < 0)
3930                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3931                 options->use_color = value;
3932         }
3933         else if (!strcmp(arg, "--no-color"))
3934                 options->use_color = 0;
3935         else if (!strcmp(arg, "--color-words")) {
3936                 options->use_color = 1;
3937                 options->word_diff = DIFF_WORDS_COLOR;
3938         }
3939         else if (skip_prefix(arg, "--color-words=", &arg)) {
3940                 options->use_color = 1;
3941                 options->word_diff = DIFF_WORDS_COLOR;
3942                 options->word_regex = arg;
3943         }
3944         else if (!strcmp(arg, "--word-diff")) {
3945                 if (options->word_diff == DIFF_WORDS_NONE)
3946                         options->word_diff = DIFF_WORDS_PLAIN;
3947         }
3948         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3949                 if (!strcmp(arg, "plain"))
3950                         options->word_diff = DIFF_WORDS_PLAIN;
3951                 else if (!strcmp(arg, "color")) {
3952                         options->use_color = 1;
3953                         options->word_diff = DIFF_WORDS_COLOR;
3954                 }
3955                 else if (!strcmp(arg, "porcelain"))
3956                         options->word_diff = DIFF_WORDS_PORCELAIN;
3957                 else if (!strcmp(arg, "none"))
3958                         options->word_diff = DIFF_WORDS_NONE;
3959                 else
3960                         die("bad --word-diff argument: %s", arg);
3961         }
3962         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3963                 if (options->word_diff == DIFF_WORDS_NONE)
3964                         options->word_diff = DIFF_WORDS_PLAIN;
3965                 options->word_regex = optarg;
3966                 return argcount;
3967         }
3968         else if (!strcmp(arg, "--exit-code"))
3969                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3970         else if (!strcmp(arg, "--quiet"))
3971                 DIFF_OPT_SET(options, QUICK);
3972         else if (!strcmp(arg, "--ext-diff"))
3973                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3974         else if (!strcmp(arg, "--no-ext-diff"))
3975                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3976         else if (!strcmp(arg, "--textconv"))
3977                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3978         else if (!strcmp(arg, "--no-textconv"))
3979                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3980         else if (!strcmp(arg, "--ignore-submodules")) {
3981                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3982                 handle_ignore_submodules_arg(options, "all");
3983         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3984                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3985                 handle_ignore_submodules_arg(options, arg);
3986         } else if (!strcmp(arg, "--submodule"))
3987                 options->submodule_format = DIFF_SUBMODULE_LOG;
3988         else if (skip_prefix(arg, "--submodule=", &arg))
3989                 return parse_submodule_opt(options, arg);
3990         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3991                 return parse_ws_error_highlight_opt(options, arg);
3992         else if (!strcmp(arg, "--ita-invisible-in-index"))
3993                 options->ita_invisible_in_index = 1;
3994         else if (!strcmp(arg, "--ita-visible-in-index"))
3995                 options->ita_invisible_in_index = 0;
3996
3997         /* misc options */
3998         else if (!strcmp(arg, "-z"))
3999                 options->line_termination = 0;
4000         else if ((argcount = short_opt('l', av, &optarg))) {
4001                 options->rename_limit = strtoul(optarg, NULL, 10);
4002                 return argcount;
4003         }
4004         else if ((argcount = short_opt('S', av, &optarg))) {
4005                 options->pickaxe = optarg;
4006                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4007                 return argcount;
4008         } else if ((argcount = short_opt('G', av, &optarg))) {
4009                 options->pickaxe = optarg;
4010                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4011                 return argcount;
4012         }
4013         else if (!strcmp(arg, "--pickaxe-all"))
4014                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4015         else if (!strcmp(arg, "--pickaxe-regex"))
4016                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4017         else if ((argcount = short_opt('O', av, &optarg))) {
4018                 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
4019                 options->orderfile = xstrdup(path);
4020                 return argcount;
4021         }
4022         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4023                 int offending = parse_diff_filter_opt(optarg, options);
4024                 if (offending)
4025                         die("unknown change class '%c' in --diff-filter=%s",
4026                             offending, optarg);
4027                 return argcount;
4028         }
4029         else if (!strcmp(arg, "--no-abbrev"))
4030                 options->abbrev = 0;
4031         else if (!strcmp(arg, "--abbrev"))
4032                 options->abbrev = DEFAULT_ABBREV;
4033         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4034                 options->abbrev = strtoul(arg, NULL, 10);
4035                 if (options->abbrev < MINIMUM_ABBREV)
4036                         options->abbrev = MINIMUM_ABBREV;
4037                 else if (40 < options->abbrev)
4038                         options->abbrev = 40;
4039         }
4040         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4041                 options->a_prefix = optarg;
4042                 return argcount;
4043         }
4044         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4045                 options->line_prefix = optarg;
4046                 options->line_prefix_length = strlen(options->line_prefix);
4047                 graph_setup_line_prefix(options);
4048                 return argcount;
4049         }
4050         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4051                 options->b_prefix = optarg;
4052                 return argcount;
4053         }
4054         else if (!strcmp(arg, "--no-prefix"))
4055                 options->a_prefix = options->b_prefix = "";
4056         else if (opt_arg(arg, '\0', "inter-hunk-context",
4057                          &options->interhunkcontext))
4058                 ;
4059         else if (!strcmp(arg, "-W"))
4060                 DIFF_OPT_SET(options, FUNCCONTEXT);
4061         else if (!strcmp(arg, "--function-context"))
4062                 DIFF_OPT_SET(options, FUNCCONTEXT);
4063         else if (!strcmp(arg, "--no-function-context"))
4064                 DIFF_OPT_CLR(options, FUNCCONTEXT);
4065         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4066                 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
4067                 options->file = fopen(path, "w");
4068                 if (!options->file)
4069                         die_errno("Could not open '%s'", path);
4070                 options->close_file = 1;
4071                 if (options->use_color != GIT_COLOR_ALWAYS)
4072                         options->use_color = GIT_COLOR_NEVER;
4073                 return argcount;
4074         } else
4075                 return 0;
4076         return 1;
4077 }
4078
4079 int parse_rename_score(const char **cp_p)
4080 {
4081         unsigned long num, scale;
4082         int ch, dot;
4083         const char *cp = *cp_p;
4084
4085         num = 0;
4086         scale = 1;
4087         dot = 0;
4088         for (;;) {
4089                 ch = *cp;
4090                 if ( !dot && ch == '.' ) {
4091                         scale = 1;
4092                         dot = 1;
4093                 } else if ( ch == '%' ) {
4094                         scale = dot ? scale*100 : 100;
4095                         cp++;   /* % is always at the end */
4096                         break;
4097                 } else if ( ch >= '0' && ch <= '9' ) {
4098                         if ( scale < 100000 ) {
4099                                 scale *= 10;
4100                                 num = (num*10) + (ch-'0');
4101                         }
4102                 } else {
4103                         break;
4104                 }
4105                 cp++;
4106         }
4107         *cp_p = cp;
4108
4109         /* user says num divided by scale and we say internally that
4110          * is MAX_SCORE * num / scale.
4111          */
4112         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4113 }
4114
4115 static int diff_scoreopt_parse(const char *opt)
4116 {
4117         int opt1, opt2, cmd;
4118
4119         if (*opt++ != '-')
4120                 return -1;
4121         cmd = *opt++;
4122         if (cmd == '-') {
4123                 /* convert the long-form arguments into short-form versions */
4124                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4125                         if (*opt == 0 || *opt++ == '=')
4126                                 cmd = 'B';
4127                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4128                         if (*opt == 0 || *opt++ == '=')
4129                                 cmd = 'C';
4130                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4131                         if (*opt == 0 || *opt++ == '=')
4132                                 cmd = 'M';
4133                 }
4134         }
4135         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4136                 return -1; /* that is not a -M, -C, or -B option */
4137
4138         opt1 = parse_rename_score(&opt);
4139         if (cmd != 'B')
4140                 opt2 = 0;
4141         else {
4142                 if (*opt == 0)
4143                         opt2 = 0;
4144                 else if (*opt != '/')
4145                         return -1; /* we expect -B80/99 or -B80 */
4146                 else {
4147                         opt++;
4148                         opt2 = parse_rename_score(&opt);
4149                 }
4150         }
4151         if (*opt != 0)
4152                 return -1;
4153         return opt1 | (opt2 << 16);
4154 }
4155
4156 struct diff_queue_struct diff_queued_diff;
4157
4158 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4159 {
4160         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4161         queue->queue[queue->nr++] = dp;
4162 }
4163
4164 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4165                                  struct diff_filespec *one,
4166                                  struct diff_filespec *two)
4167 {
4168         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4169         dp->one = one;
4170         dp->two = two;
4171         if (queue)
4172                 diff_q(queue, dp);
4173         return dp;
4174 }
4175
4176 void diff_free_filepair(struct diff_filepair *p)
4177 {
4178         free_filespec(p->one);
4179         free_filespec(p->two);
4180         free(p);
4181 }
4182
4183 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4184 {
4185         int abblen;
4186         const char *abbrev;
4187
4188         if (len == GIT_SHA1_HEXSZ)
4189                 return oid_to_hex(oid);
4190
4191         abbrev = diff_abbrev_oid(oid, len);
4192         abblen = strlen(abbrev);
4193
4194         /*
4195          * In well-behaved cases, where the abbbreviated result is the
4196          * same as the requested length, append three dots after the
4197          * abbreviation (hence the whole logic is limited to the case
4198          * where abblen < 37); when the actual abbreviated result is a
4199          * bit longer than the requested length, we reduce the number
4200          * of dots so that they match the well-behaved ones.  However,
4201          * if the actual abbreviation is longer than the requested
4202          * length by more than three, we give up on aligning, and add
4203          * three dots anyway, to indicate that the output is not the
4204          * full object name.  Yes, this may be suboptimal, but this
4205          * appears only in "diff --raw --abbrev" output and it is not
4206          * worth the effort to change it now.  Note that this would
4207          * likely to work fine when the automatic sizing of default
4208          * abbreviation length is used--we would be fed -1 in "len" in
4209          * that case, and will end up always appending three-dots, but
4210          * the automatic sizing is supposed to give abblen that ensures
4211          * uniqueness across all objects (statistically speaking).
4212          */
4213         if (abblen < GIT_SHA1_HEXSZ - 3) {
4214                 static char hex[GIT_SHA1_HEXSZ + 1];
4215                 if (len < abblen && abblen <= len + 2)
4216                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4217                 else
4218                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4219                 return hex;
4220         }
4221
4222         return oid_to_hex(oid);
4223 }
4224
4225 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4226 {
4227         int line_termination = opt->line_termination;
4228         int inter_name_termination = line_termination ? '\t' : '\0';
4229
4230         fprintf(opt->file, "%s", diff_line_prefix(opt));
4231         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4232                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4233                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
4234                 fprintf(opt->file, "%s ",
4235                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
4236         }
4237         if (p->score) {
4238                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4239                         inter_name_termination);
4240         } else {
4241                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4242         }
4243
4244         if (p->status == DIFF_STATUS_COPIED ||
4245             p->status == DIFF_STATUS_RENAMED) {
4246                 const char *name_a, *name_b;
4247                 name_a = p->one->path;
4248                 name_b = p->two->path;
4249                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4250                 write_name_quoted(name_a, opt->file, inter_name_termination);
4251                 write_name_quoted(name_b, opt->file, line_termination);
4252         } else {
4253                 const char *name_a, *name_b;
4254                 name_a = p->one->mode ? p->one->path : p->two->path;
4255                 name_b = NULL;
4256                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4257                 write_name_quoted(name_a, opt->file, line_termination);
4258         }
4259 }
4260
4261 int diff_unmodified_pair(struct diff_filepair *p)
4262 {
4263         /* This function is written stricter than necessary to support
4264          * the currently implemented transformers, but the idea is to
4265          * let transformers to produce diff_filepairs any way they want,
4266          * and filter and clean them up here before producing the output.
4267          */
4268         struct diff_filespec *one = p->one, *two = p->two;
4269
4270         if (DIFF_PAIR_UNMERGED(p))
4271                 return 0; /* unmerged is interesting */
4272
4273         /* deletion, addition, mode or type change
4274          * and rename are all interesting.
4275          */
4276         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4277             DIFF_PAIR_MODE_CHANGED(p) ||
4278             strcmp(one->path, two->path))
4279                 return 0;
4280
4281         /* both are valid and point at the same path.  that is, we are
4282          * dealing with a change.
4283          */
4284         if (one->oid_valid && two->oid_valid &&
4285             !oidcmp(&one->oid, &two->oid) &&
4286             !one->dirty_submodule && !two->dirty_submodule)
4287                 return 1; /* no change */
4288         if (!one->oid_valid && !two->oid_valid)
4289                 return 1; /* both look at the same file on the filesystem. */
4290         return 0;
4291 }
4292
4293 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4294 {
4295         if (diff_unmodified_pair(p))
4296                 return;
4297
4298         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4299             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4300                 return; /* no tree diffs in patch format */
4301
4302         run_diff(p, o);
4303 }
4304
4305 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4306                             struct diffstat_t *diffstat)
4307 {
4308         if (diff_unmodified_pair(p))
4309                 return;
4310
4311         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4312             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4313                 return; /* no useful stat for tree diffs */
4314
4315         run_diffstat(p, o, diffstat);
4316 }
4317
4318 static void diff_flush_checkdiff(struct diff_filepair *p,
4319                 struct diff_options *o)
4320 {
4321         if (diff_unmodified_pair(p))
4322                 return;
4323
4324         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4325             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4326                 return; /* nothing to check in tree diffs */
4327
4328         run_checkdiff(p, o);
4329 }
4330
4331 int diff_queue_is_empty(void)
4332 {
4333         struct diff_queue_struct *q = &diff_queued_diff;
4334         int i;
4335         for (i = 0; i < q->nr; i++)
4336                 if (!diff_unmodified_pair(q->queue[i]))
4337                         return 0;
4338         return 1;
4339 }
4340
4341 #if DIFF_DEBUG
4342 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4343 {
4344         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4345                 x, one ? one : "",
4346                 s->path,
4347                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4348                 s->mode,
4349                 s->oid_valid ? oid_to_hex(&s->oid) : "");
4350         fprintf(stderr, "queue[%d] %s size %lu\n",
4351                 x, one ? one : "",
4352                 s->size);
4353 }
4354
4355 void diff_debug_filepair(const struct diff_filepair *p, int i)
4356 {
4357         diff_debug_filespec(p->one, i, "one");
4358         diff_debug_filespec(p->two, i, "two");
4359         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4360                 p->score, p->status ? p->status : '?',
4361                 p->one->rename_used, p->broken_pair);
4362 }
4363
4364 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4365 {
4366         int i;
4367         if (msg)
4368                 fprintf(stderr, "%s\n", msg);
4369         fprintf(stderr, "q->nr = %d\n", q->nr);
4370         for (i = 0; i < q->nr; i++) {
4371                 struct diff_filepair *p = q->queue[i];
4372                 diff_debug_filepair(p, i);
4373         }
4374 }
4375 #endif
4376
4377 static void diff_resolve_rename_copy(void)
4378 {
4379         int i;
4380         struct diff_filepair *p;
4381         struct diff_queue_struct *q = &diff_queued_diff;
4382
4383         diff_debug_queue("resolve-rename-copy", q);
4384
4385         for (i = 0; i < q->nr; i++) {
4386                 p = q->queue[i];
4387                 p->status = 0; /* undecided */
4388                 if (DIFF_PAIR_UNMERGED(p))
4389                         p->status = DIFF_STATUS_UNMERGED;
4390                 else if (!DIFF_FILE_VALID(p->one))
4391                         p->status = DIFF_STATUS_ADDED;
4392                 else if (!DIFF_FILE_VALID(p->two))
4393                         p->status = DIFF_STATUS_DELETED;
4394                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4395                         p->status = DIFF_STATUS_TYPE_CHANGED;
4396
4397                 /* from this point on, we are dealing with a pair
4398                  * whose both sides are valid and of the same type, i.e.
4399                  * either in-place edit or rename/copy edit.
4400                  */
4401                 else if (DIFF_PAIR_RENAME(p)) {
4402                         /*
4403                          * A rename might have re-connected a broken
4404                          * pair up, causing the pathnames to be the
4405                          * same again. If so, that's not a rename at
4406                          * all, just a modification..
4407                          *
4408                          * Otherwise, see if this source was used for
4409                          * multiple renames, in which case we decrement
4410                          * the count, and call it a copy.
4411                          */
4412                         if (!strcmp(p->one->path, p->two->path))
4413                                 p->status = DIFF_STATUS_MODIFIED;
4414                         else if (--p->one->rename_used > 0)
4415                                 p->status = DIFF_STATUS_COPIED;
4416                         else
4417                                 p->status = DIFF_STATUS_RENAMED;
4418                 }
4419                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
4420                          p->one->mode != p->two->mode ||
4421                          p->one->dirty_submodule ||
4422                          p->two->dirty_submodule ||
4423                          is_null_oid(&p->one->oid))
4424                         p->status = DIFF_STATUS_MODIFIED;
4425                 else {
4426                         /* This is a "no-change" entry and should not
4427                          * happen anymore, but prepare for broken callers.
4428                          */
4429                         error("feeding unmodified %s to diffcore",
4430                               p->one->path);
4431                         p->status = DIFF_STATUS_UNKNOWN;
4432                 }
4433         }
4434         diff_debug_queue("resolve-rename-copy done", q);
4435 }
4436
4437 static int check_pair_status(struct diff_filepair *p)
4438 {
4439         switch (p->status) {
4440         case DIFF_STATUS_UNKNOWN:
4441                 return 0;
4442         case 0:
4443                 die("internal error in diff-resolve-rename-copy");
4444         default:
4445                 return 1;
4446         }
4447 }
4448
4449 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4450 {
4451         int fmt = opt->output_format;
4452
4453         if (fmt & DIFF_FORMAT_CHECKDIFF)
4454                 diff_flush_checkdiff(p, opt);
4455         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4456                 diff_flush_raw(p, opt);
4457         else if (fmt & DIFF_FORMAT_NAME) {
4458                 const char *name_a, *name_b;
4459                 name_a = p->two->path;
4460                 name_b = NULL;
4461                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4462                 write_name_quoted(name_a, opt->file, opt->line_termination);
4463         }
4464 }
4465
4466 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4467 {
4468         if (fs->mode)
4469                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4470         else
4471                 fprintf(file, " %s ", newdelete);
4472         write_name_quoted(fs->path, file, '\n');
4473 }
4474
4475
4476 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4477                 const char *line_prefix)
4478 {
4479         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4480                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4481                         p->two->mode, show_name ? ' ' : '\n');
4482                 if (show_name) {
4483                         write_name_quoted(p->two->path, file, '\n');
4484                 }
4485         }
4486 }
4487
4488 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4489                         const char *line_prefix)
4490 {
4491         char *names = pprint_rename(p->one->path, p->two->path);
4492
4493         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4494         free(names);
4495         show_mode_change(file, p, 0, line_prefix);
4496 }
4497
4498 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4499 {
4500         FILE *file = opt->file;
4501         const char *line_prefix = diff_line_prefix(opt);
4502
4503         switch(p->status) {
4504         case DIFF_STATUS_DELETED:
4505                 fputs(line_prefix, file);
4506                 show_file_mode_name(file, "delete", p->one);
4507                 break;
4508         case DIFF_STATUS_ADDED:
4509                 fputs(line_prefix, file);
4510                 show_file_mode_name(file, "create", p->two);
4511                 break;
4512         case DIFF_STATUS_COPIED:
4513                 fputs(line_prefix, file);
4514                 show_rename_copy(file, "copy", p, line_prefix);
4515                 break;
4516         case DIFF_STATUS_RENAMED:
4517                 fputs(line_prefix, file);
4518                 show_rename_copy(file, "rename", p, line_prefix);
4519                 break;
4520         default:
4521                 if (p->score) {
4522                         fprintf(file, "%s rewrite ", line_prefix);
4523                         write_name_quoted(p->two->path, file, ' ');
4524                         fprintf(file, "(%d%%)\n", similarity_index(p));
4525                 }
4526                 show_mode_change(file, p, !p->score, line_prefix);
4527                 break;
4528         }
4529 }
4530
4531 struct patch_id_t {
4532         git_SHA_CTX *ctx;
4533         int patchlen;
4534 };
4535
4536 static int remove_space(char *line, int len)
4537 {
4538         int i;
4539         char *dst = line;
4540         unsigned char c;
4541
4542         for (i = 0; i < len; i++)
4543                 if (!isspace((c = line[i])))
4544                         *dst++ = c;
4545
4546         return dst - line;
4547 }
4548
4549 static void patch_id_consume(void *priv, char *line, unsigned long len)
4550 {
4551         struct patch_id_t *data = priv;
4552         int new_len;
4553
4554         /* Ignore line numbers when computing the SHA1 of the patch */
4555         if (starts_with(line, "@@ -"))
4556                 return;
4557
4558         new_len = remove_space(line, len);
4559
4560         git_SHA1_Update(data->ctx, line, new_len);
4561         data->patchlen += new_len;
4562 }
4563
4564 /* returns 0 upon success, and writes result into sha1 */
4565 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4566 {
4567         struct diff_queue_struct *q = &diff_queued_diff;
4568         int i;
4569         git_SHA_CTX ctx;
4570         struct patch_id_t data;
4571         char buffer[PATH_MAX * 4 + 20];
4572
4573         git_SHA1_Init(&ctx);
4574         memset(&data, 0, sizeof(struct patch_id_t));
4575         data.ctx = &ctx;
4576
4577         for (i = 0; i < q->nr; i++) {
4578                 xpparam_t xpp;
4579                 xdemitconf_t xecfg;
4580                 mmfile_t mf1, mf2;
4581                 struct diff_filepair *p = q->queue[i];
4582                 int len1, len2;
4583
4584                 memset(&xpp, 0, sizeof(xpp));
4585                 memset(&xecfg, 0, sizeof(xecfg));
4586                 if (p->status == 0)
4587                         return error("internal diff status error");
4588                 if (p->status == DIFF_STATUS_UNKNOWN)
4589                         continue;
4590                 if (diff_unmodified_pair(p))
4591                         continue;
4592                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4593                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4594                         continue;
4595                 if (DIFF_PAIR_UNMERGED(p))
4596                         continue;
4597
4598                 diff_fill_sha1_info(p->one);
4599                 diff_fill_sha1_info(p->two);
4600
4601                 len1 = remove_space(p->one->path, strlen(p->one->path));
4602                 len2 = remove_space(p->two->path, strlen(p->two->path));
4603                 if (p->one->mode == 0)
4604                         len1 = snprintf(buffer, sizeof(buffer),
4605                                         "diff--gita/%.*sb/%.*s"
4606                                         "newfilemode%06o"
4607                                         "---/dev/null"
4608                                         "+++b/%.*s",
4609                                         len1, p->one->path,
4610                                         len2, p->two->path,
4611                                         p->two->mode,
4612                                         len2, p->two->path);
4613                 else if (p->two->mode == 0)
4614                         len1 = snprintf(buffer, sizeof(buffer),
4615                                         "diff--gita/%.*sb/%.*s"
4616                                         "deletedfilemode%06o"
4617                                         "---a/%.*s"
4618                                         "+++/dev/null",
4619                                         len1, p->one->path,
4620                                         len2, p->two->path,
4621                                         p->one->mode,
4622                                         len1, p->one->path);
4623                 else
4624                         len1 = snprintf(buffer, sizeof(buffer),
4625                                         "diff--gita/%.*sb/%.*s"
4626                                         "---a/%.*s"
4627                                         "+++b/%.*s",
4628                                         len1, p->one->path,
4629                                         len2, p->two->path,
4630                                         len1, p->one->path,
4631                                         len2, p->two->path);
4632                 git_SHA1_Update(&ctx, buffer, len1);
4633
4634                 if (diff_header_only)
4635                         continue;
4636
4637                 if (fill_mmfile(&mf1, p->one) < 0 ||
4638                     fill_mmfile(&mf2, p->two) < 0)
4639                         return error("unable to read files to diff");
4640
4641                 if (diff_filespec_is_binary(p->one) ||
4642                     diff_filespec_is_binary(p->two)) {
4643                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
4644                                         40);
4645                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
4646                                         40);
4647                         continue;
4648                 }
4649
4650                 xpp.flags = 0;
4651                 xecfg.ctxlen = 3;
4652                 xecfg.flags = 0;
4653                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4654                                   &xpp, &xecfg))
4655                         return error("unable to generate patch-id diff for %s",
4656                                      p->one->path);
4657         }
4658
4659         git_SHA1_Final(sha1, &ctx);
4660         return 0;
4661 }
4662
4663 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4664 {
4665         struct diff_queue_struct *q = &diff_queued_diff;
4666         int i;
4667         int result = diff_get_patch_id(options, sha1, diff_header_only);
4668
4669         for (i = 0; i < q->nr; i++)
4670                 diff_free_filepair(q->queue[i]);
4671
4672         free(q->queue);
4673         DIFF_QUEUE_CLEAR(q);
4674
4675         return result;
4676 }
4677
4678 static int is_summary_empty(const struct diff_queue_struct *q)
4679 {
4680         int i;
4681
4682         for (i = 0; i < q->nr; i++) {
4683                 const struct diff_filepair *p = q->queue[i];
4684
4685                 switch (p->status) {
4686                 case DIFF_STATUS_DELETED:
4687                 case DIFF_STATUS_ADDED:
4688                 case DIFF_STATUS_COPIED:
4689                 case DIFF_STATUS_RENAMED:
4690                         return 0;
4691                 default:
4692                         if (p->score)
4693                                 return 0;
4694                         if (p->one->mode && p->two->mode &&
4695                             p->one->mode != p->two->mode)
4696                                 return 0;
4697                         break;
4698                 }
4699         }
4700         return 1;
4701 }
4702
4703 static const char rename_limit_warning[] =
4704 N_("inexact rename detection was skipped due to too many files.");
4705
4706 static const char degrade_cc_to_c_warning[] =
4707 N_("only found copies from modified paths due to too many files.");
4708
4709 static const char rename_limit_advice[] =
4710 N_("you may want to set your %s variable to at least "
4711    "%d and retry the command.");
4712
4713 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4714 {
4715         if (degraded_cc)
4716                 warning(_(degrade_cc_to_c_warning));
4717         else if (needed)
4718                 warning(_(rename_limit_warning));
4719         else
4720                 return;
4721         if (0 < needed && needed < 32767)
4722                 warning(_(rename_limit_advice), varname, needed);
4723 }
4724
4725 void diff_flush(struct diff_options *options)
4726 {
4727         struct diff_queue_struct *q = &diff_queued_diff;
4728         int i, output_format = options->output_format;
4729         int separator = 0;
4730         int dirstat_by_line = 0;
4731
4732         /*
4733          * Order: raw, stat, summary, patch
4734          * or:    name/name-status/checkdiff (other bits clear)
4735          */
4736         if (!q->nr)
4737                 goto free_queue;
4738
4739         if (output_format & (DIFF_FORMAT_RAW |
4740                              DIFF_FORMAT_NAME |
4741                              DIFF_FORMAT_NAME_STATUS |
4742                              DIFF_FORMAT_CHECKDIFF)) {
4743                 for (i = 0; i < q->nr; i++) {
4744                         struct diff_filepair *p = q->queue[i];
4745                         if (check_pair_status(p))
4746                                 flush_one_pair(p, options);
4747                 }
4748                 separator++;
4749         }
4750
4751         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4752                 dirstat_by_line = 1;
4753
4754         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4755             dirstat_by_line) {
4756                 struct diffstat_t diffstat;
4757
4758                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4759                 for (i = 0; i < q->nr; i++) {
4760                         struct diff_filepair *p = q->queue[i];
4761                         if (check_pair_status(p))
4762                                 diff_flush_stat(p, options, &diffstat);
4763                 }
4764                 if (output_format & DIFF_FORMAT_NUMSTAT)
4765                         show_numstat(&diffstat, options);
4766                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4767                         show_stats(&diffstat, options);
4768                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4769                         show_shortstats(&diffstat, options);
4770                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4771                         show_dirstat_by_line(&diffstat, options);
4772                 free_diffstat_info(&diffstat);
4773                 separator++;
4774         }
4775         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4776                 show_dirstat(options);
4777
4778         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4779                 for (i = 0; i < q->nr; i++) {
4780                         diff_summary(options, q->queue[i]);
4781                 }
4782                 separator++;
4783         }
4784
4785         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4786             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4787             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4788                 /*
4789                  * run diff_flush_patch for the exit status. setting
4790                  * options->file to /dev/null should be safe, because we
4791                  * aren't supposed to produce any output anyway.
4792                  */
4793                 if (options->close_file)
4794                         fclose(options->file);
4795                 options->file = fopen("/dev/null", "w");
4796                 if (!options->file)
4797                         die_errno("Could not open /dev/null");
4798                 options->close_file = 1;
4799                 for (i = 0; i < q->nr; i++) {
4800                         struct diff_filepair *p = q->queue[i];
4801                         if (check_pair_status(p))
4802                                 diff_flush_patch(p, options);
4803                         if (options->found_changes)
4804                                 break;
4805                 }
4806         }
4807
4808         if (output_format & DIFF_FORMAT_PATCH) {
4809                 if (separator) {
4810                         fprintf(options->file, "%s%c",
4811                                 diff_line_prefix(options),
4812                                 options->line_termination);
4813                         if (options->stat_sep) {
4814                                 /* attach patch instead of inline */
4815                                 fputs(options->stat_sep, options->file);
4816                         }
4817                 }
4818
4819                 for (i = 0; i < q->nr; i++) {
4820                         struct diff_filepair *p = q->queue[i];
4821                         if (check_pair_status(p))
4822                                 diff_flush_patch(p, options);
4823                 }
4824         }
4825
4826         if (output_format & DIFF_FORMAT_CALLBACK)
4827                 options->format_callback(q, options, options->format_callback_data);
4828
4829         for (i = 0; i < q->nr; i++)
4830                 diff_free_filepair(q->queue[i]);
4831 free_queue:
4832         free(q->queue);
4833         DIFF_QUEUE_CLEAR(q);
4834         if (options->close_file)
4835                 fclose(options->file);
4836
4837         /*
4838          * Report the content-level differences with HAS_CHANGES;
4839          * diff_addremove/diff_change does not set the bit when
4840          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4841          */
4842         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4843                 if (options->found_changes)
4844                         DIFF_OPT_SET(options, HAS_CHANGES);
4845                 else
4846                         DIFF_OPT_CLR(options, HAS_CHANGES);
4847         }
4848 }
4849
4850 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4851 {
4852         return (((p->status == DIFF_STATUS_MODIFIED) &&
4853                  ((p->score &&
4854                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4855                   (!p->score &&
4856                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4857                 ((p->status != DIFF_STATUS_MODIFIED) &&
4858                  filter_bit_tst(p->status, options)));
4859 }
4860
4861 static void diffcore_apply_filter(struct diff_options *options)
4862 {
4863         int i;
4864         struct diff_queue_struct *q = &diff_queued_diff;
4865         struct diff_queue_struct outq;
4866
4867         DIFF_QUEUE_CLEAR(&outq);
4868
4869         if (!options->filter)
4870                 return;
4871
4872         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4873                 int found;
4874                 for (i = found = 0; !found && i < q->nr; i++) {
4875                         if (match_filter(options, q->queue[i]))
4876                                 found++;
4877                 }
4878                 if (found)
4879                         return;
4880
4881                 /* otherwise we will clear the whole queue
4882                  * by copying the empty outq at the end of this
4883                  * function, but first clear the current entries
4884                  * in the queue.
4885                  */
4886                 for (i = 0; i < q->nr; i++)
4887                         diff_free_filepair(q->queue[i]);
4888         }
4889         else {
4890                 /* Only the matching ones */
4891                 for (i = 0; i < q->nr; i++) {
4892                         struct diff_filepair *p = q->queue[i];
4893                         if (match_filter(options, p))
4894                                 diff_q(&outq, p);
4895                         else
4896                                 diff_free_filepair(p);
4897                 }
4898         }
4899         free(q->queue);
4900         *q = outq;
4901 }
4902
4903 /* Check whether two filespecs with the same mode and size are identical */
4904 static int diff_filespec_is_identical(struct diff_filespec *one,
4905                                       struct diff_filespec *two)
4906 {
4907         if (S_ISGITLINK(one->mode))
4908                 return 0;
4909         if (diff_populate_filespec(one, 0))
4910                 return 0;
4911         if (diff_populate_filespec(two, 0))
4912                 return 0;
4913         return !memcmp(one->data, two->data, one->size);
4914 }
4915
4916 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4917 {
4918         if (p->done_skip_stat_unmatch)
4919                 return p->skip_stat_unmatch_result;
4920
4921         p->done_skip_stat_unmatch = 1;
4922         p->skip_stat_unmatch_result = 0;
4923         /*
4924          * 1. Entries that come from stat info dirtiness
4925          *    always have both sides (iow, not create/delete),
4926          *    one side of the object name is unknown, with
4927          *    the same mode and size.  Keep the ones that
4928          *    do not match these criteria.  They have real
4929          *    differences.
4930          *
4931          * 2. At this point, the file is known to be modified,
4932          *    with the same mode and size, and the object
4933          *    name of one side is unknown.  Need to inspect
4934          *    the identical contents.
4935          */
4936         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4937             !DIFF_FILE_VALID(p->two) ||
4938             (p->one->oid_valid && p->two->oid_valid) ||
4939             (p->one->mode != p->two->mode) ||
4940             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4941             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4942             (p->one->size != p->two->size) ||
4943             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4944                 p->skip_stat_unmatch_result = 1;
4945         return p->skip_stat_unmatch_result;
4946 }
4947
4948 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4949 {
4950         int i;
4951         struct diff_queue_struct *q = &diff_queued_diff;
4952         struct diff_queue_struct outq;
4953         DIFF_QUEUE_CLEAR(&outq);
4954
4955         for (i = 0; i < q->nr; i++) {
4956                 struct diff_filepair *p = q->queue[i];
4957
4958                 if (diff_filespec_check_stat_unmatch(p))
4959                         diff_q(&outq, p);
4960                 else {
4961                         /*
4962                          * The caller can subtract 1 from skip_stat_unmatch
4963                          * to determine how many paths were dirty only
4964                          * due to stat info mismatch.
4965                          */
4966                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4967                                 diffopt->skip_stat_unmatch++;
4968                         diff_free_filepair(p);
4969                 }
4970         }
4971         free(q->queue);
4972         *q = outq;
4973 }
4974
4975 static int diffnamecmp(const void *a_, const void *b_)
4976 {
4977         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4978         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4979         const char *name_a, *name_b;
4980
4981         name_a = a->one ? a->one->path : a->two->path;
4982         name_b = b->one ? b->one->path : b->two->path;
4983         return strcmp(name_a, name_b);
4984 }
4985
4986 void diffcore_fix_diff_index(struct diff_options *options)
4987 {
4988         struct diff_queue_struct *q = &diff_queued_diff;
4989         QSORT(q->queue, q->nr, diffnamecmp);
4990 }
4991
4992 void diffcore_std(struct diff_options *options)
4993 {
4994         /* NOTE please keep the following in sync with diff_tree_combined() */
4995         if (options->skip_stat_unmatch)
4996                 diffcore_skip_stat_unmatch(options);
4997         if (!options->found_follow) {
4998                 /* See try_to_follow_renames() in tree-diff.c */
4999                 if (options->break_opt != -1)
5000                         diffcore_break(options->break_opt);
5001                 if (options->detect_rename)
5002                         diffcore_rename(options);
5003                 if (options->break_opt != -1)
5004                         diffcore_merge_broken();
5005         }
5006         if (options->pickaxe)
5007                 diffcore_pickaxe(options);
5008         if (options->orderfile)
5009                 diffcore_order(options->orderfile);
5010         if (!options->found_follow)
5011                 /* See try_to_follow_renames() in tree-diff.c */
5012                 diff_resolve_rename_copy();
5013         diffcore_apply_filter(options);
5014
5015         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5016                 DIFF_OPT_SET(options, HAS_CHANGES);
5017         else
5018                 DIFF_OPT_CLR(options, HAS_CHANGES);
5019
5020         options->found_follow = 0;
5021 }
5022
5023 int diff_result_code(struct diff_options *opt, int status)
5024 {
5025         int result = 0;
5026
5027         diff_warn_rename_limit("diff.renameLimit",
5028                                opt->needed_rename_limit,
5029                                opt->degraded_cc_to_c);
5030         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5031             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5032                 return status;
5033         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5034             DIFF_OPT_TST(opt, HAS_CHANGES))
5035                 result |= 01;
5036         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5037             DIFF_OPT_TST(opt, CHECK_FAILED))
5038                 result |= 02;
5039         return result;
5040 }
5041
5042 int diff_can_quit_early(struct diff_options *opt)
5043 {
5044         return (DIFF_OPT_TST(opt, QUICK) &&
5045                 !opt->filter &&
5046                 DIFF_OPT_TST(opt, HAS_CHANGES));
5047 }
5048
5049 /*
5050  * Shall changes to this submodule be ignored?
5051  *
5052  * Submodule changes can be configured to be ignored separately for each path,
5053  * but that configuration can be overridden from the command line.
5054  */
5055 static int is_submodule_ignored(const char *path, struct diff_options *options)
5056 {
5057         int ignored = 0;
5058         unsigned orig_flags = options->flags;
5059         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
5060                 set_diffopt_flags_from_submodule_config(options, path);
5061         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
5062                 ignored = 1;
5063         options->flags = orig_flags;
5064         return ignored;
5065 }
5066
5067 void diff_addremove(struct diff_options *options,
5068                     int addremove, unsigned mode,
5069                     const unsigned char *sha1,
5070                     int sha1_valid,
5071                     const char *concatpath, unsigned dirty_submodule)
5072 {
5073         struct diff_filespec *one, *two;
5074
5075         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5076                 return;
5077
5078         /* This may look odd, but it is a preparation for
5079          * feeding "there are unchanged files which should
5080          * not produce diffs, but when you are doing copy
5081          * detection you would need them, so here they are"
5082          * entries to the diff-core.  They will be prefixed
5083          * with something like '=' or '*' (I haven't decided
5084          * which but should not make any difference).
5085          * Feeding the same new and old to diff_change()
5086          * also has the same effect.
5087          * Before the final output happens, they are pruned after
5088          * merged into rename/copy pairs as appropriate.
5089          */
5090         if (DIFF_OPT_TST(options, REVERSE_DIFF))
5091                 addremove = (addremove == '+' ? '-' :
5092                              addremove == '-' ? '+' : addremove);
5093
5094         if (options->prefix &&
5095             strncmp(concatpath, options->prefix, options->prefix_length))
5096                 return;
5097
5098         one = alloc_filespec(concatpath);
5099         two = alloc_filespec(concatpath);
5100
5101         if (addremove != '+')
5102                 fill_filespec(one, sha1, sha1_valid, mode);
5103         if (addremove != '-') {
5104                 fill_filespec(two, sha1, sha1_valid, mode);
5105                 two->dirty_submodule = dirty_submodule;
5106         }
5107
5108         diff_queue(&diff_queued_diff, one, two);
5109         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5110                 DIFF_OPT_SET(options, HAS_CHANGES);
5111 }
5112
5113 void diff_change(struct diff_options *options,
5114                  unsigned old_mode, unsigned new_mode,
5115                  const unsigned char *old_sha1,
5116                  const unsigned char *new_sha1,
5117                  int old_sha1_valid, int new_sha1_valid,
5118                  const char *concatpath,
5119                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5120 {
5121         struct diff_filespec *one, *two;
5122         struct diff_filepair *p;
5123
5124         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5125             is_submodule_ignored(concatpath, options))
5126                 return;
5127
5128         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5129                 unsigned tmp;
5130                 const unsigned char *tmp_c;
5131                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
5132                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
5133                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
5134                         new_sha1_valid = tmp;
5135                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
5136                         new_dirty_submodule = tmp;
5137         }
5138
5139         if (options->prefix &&
5140             strncmp(concatpath, options->prefix, options->prefix_length))
5141                 return;
5142
5143         one = alloc_filespec(concatpath);
5144         two = alloc_filespec(concatpath);
5145         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5146         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5147         one->dirty_submodule = old_dirty_submodule;
5148         two->dirty_submodule = new_dirty_submodule;
5149         p = diff_queue(&diff_queued_diff, one, two);
5150
5151         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5152                 return;
5153
5154         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5155             !diff_filespec_check_stat_unmatch(p))
5156                 return;
5157
5158         DIFF_OPT_SET(options, HAS_CHANGES);
5159 }
5160
5161 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5162 {
5163         struct diff_filepair *pair;
5164         struct diff_filespec *one, *two;
5165
5166         if (options->prefix &&
5167             strncmp(path, options->prefix, options->prefix_length))
5168                 return NULL;
5169
5170         one = alloc_filespec(path);
5171         two = alloc_filespec(path);
5172         pair = diff_queue(&diff_queued_diff, one, two);
5173         pair->is_unmerged = 1;
5174         return pair;
5175 }
5176
5177 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5178                 size_t *outsize)
5179 {
5180         struct diff_tempfile *temp;
5181         const char *argv[3];
5182         const char **arg = argv;
5183         struct child_process child = CHILD_PROCESS_INIT;
5184         struct strbuf buf = STRBUF_INIT;
5185         int err = 0;
5186
5187         temp = prepare_temp_file(spec->path, spec);
5188         *arg++ = pgm;
5189         *arg++ = temp->name;
5190         *arg = NULL;
5191
5192         child.use_shell = 1;
5193         child.argv = argv;
5194         child.out = -1;
5195         if (start_command(&child)) {
5196                 remove_tempfile();
5197                 return NULL;
5198         }
5199
5200         if (strbuf_read(&buf, child.out, 0) < 0)
5201                 err = error("error reading from textconv command '%s'", pgm);
5202         close(child.out);
5203
5204         if (finish_command(&child) || err) {
5205                 strbuf_release(&buf);
5206                 remove_tempfile();
5207                 return NULL;
5208         }
5209         remove_tempfile();
5210
5211         return strbuf_detach(&buf, outsize);
5212 }
5213
5214 size_t fill_textconv(struct userdiff_driver *driver,
5215                      struct diff_filespec *df,
5216                      char **outbuf)
5217 {
5218         size_t size;
5219
5220         if (!driver) {
5221                 if (!DIFF_FILE_VALID(df)) {
5222                         *outbuf = "";
5223                         return 0;
5224                 }
5225                 if (diff_populate_filespec(df, 0))
5226                         die("unable to read files to diff");
5227                 *outbuf = df->data;
5228                 return df->size;
5229         }
5230
5231         if (!driver->textconv)
5232                 die("BUG: fill_textconv called with non-textconv driver");
5233
5234         if (driver->textconv_cache && df->oid_valid) {
5235                 *outbuf = notes_cache_get(driver->textconv_cache,
5236                                           df->oid.hash,
5237                                           &size);
5238                 if (*outbuf)
5239                         return size;
5240         }
5241
5242         *outbuf = run_textconv(driver->textconv, df, &size);
5243         if (!*outbuf)
5244                 die("unable to read files to diff");
5245
5246         if (driver->textconv_cache && df->oid_valid) {
5247                 /* ignore errors, as we might be in a readonly repository */
5248                 notes_cache_put(driver->textconv_cache, df->oid.hash, *outbuf,
5249                                 size);
5250                 /*
5251                  * we could save up changes and flush them all at the end,
5252                  * but we would need an extra call after all diffing is done.
5253                  * Since generating a cache entry is the slow path anyway,
5254                  * this extra overhead probably isn't a big deal.
5255                  */
5256                 notes_cache_write(driver->textconv_cache);
5257         }
5258
5259         return size;
5260 }
5261
5262 void setup_diff_pager(struct diff_options *opt)
5263 {
5264         /*
5265          * If the user asked for our exit code, then either they want --quiet
5266          * or --exit-code. We should definitely not bother with a pager in the
5267          * former case, as we will generate no output. Since we still properly
5268          * report our exit code even when a pager is run, we _could_ run a
5269          * pager with --exit-code. But since we have not done so historically,
5270          * and because it is easy to find people oneline advising "git diff
5271          * --exit-code" in hooks and other scripts, we do not do so.
5272          */
5273         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5274             check_pager_config("diff") != 0)
5275                 setup_pager();
5276 }