git-svn: add ability to specify --commit-url for dcommit
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
14
15 #ifdef NO_FAST_WORKING_DIRECTORY
16 #define FAST_WORKING_DIRECTORY 0
17 #else
18 #define FAST_WORKING_DIRECTORY 1
19 #endif
20
21 static int diff_detect_rename_default;
22 static int diff_rename_limit_default = 200;
23 int diff_use_color_default = -1;
24 static const char *external_diff_cmd_cfg;
25 int diff_auto_refresh_index = 1;
26
27 static char diff_colors[][COLOR_MAXLEN] = {
28         "\033[m",       /* reset */
29         "",             /* PLAIN (normal) */
30         "\033[1m",      /* METAINFO (bold) */
31         "\033[36m",     /* FRAGINFO (cyan) */
32         "\033[31m",     /* OLD (red) */
33         "\033[32m",     /* NEW (green) */
34         "\033[33m",     /* COMMIT (yellow) */
35         "\033[41m",     /* WHITESPACE (red background) */
36 };
37
38 static int parse_diff_color_slot(const char *var, int ofs)
39 {
40         if (!strcasecmp(var+ofs, "plain"))
41                 return DIFF_PLAIN;
42         if (!strcasecmp(var+ofs, "meta"))
43                 return DIFF_METAINFO;
44         if (!strcasecmp(var+ofs, "frag"))
45                 return DIFF_FRAGINFO;
46         if (!strcasecmp(var+ofs, "old"))
47                 return DIFF_FILE_OLD;
48         if (!strcasecmp(var+ofs, "new"))
49                 return DIFF_FILE_NEW;
50         if (!strcasecmp(var+ofs, "commit"))
51                 return DIFF_COMMIT;
52         if (!strcasecmp(var+ofs, "whitespace"))
53                 return DIFF_WHITESPACE;
54         die("bad config variable '%s'", var);
55 }
56
57 static struct ll_diff_driver {
58         const char *name;
59         struct ll_diff_driver *next;
60         const char *cmd;
61 } *user_diff, **user_diff_tail;
62
63 /*
64  * Currently there is only "diff.<drivername>.command" variable;
65  * because there are "diff.color.<slot>" variables, we are parsing
66  * this in a bit convoluted way to allow low level diff driver
67  * called "color".
68  */
69 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
70 {
71         const char *name;
72         int namelen;
73         struct ll_diff_driver *drv;
74
75         name = var + 5;
76         namelen = ep - name;
77         for (drv = user_diff; drv; drv = drv->next)
78                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
79                         break;
80         if (!drv) {
81                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
82                 drv->name = xmemdupz(name, namelen);
83                 if (!user_diff_tail)
84                         user_diff_tail = &user_diff;
85                 *user_diff_tail = drv;
86                 user_diff_tail = &(drv->next);
87         }
88
89         return git_config_string(&(drv->cmd), var, value);
90 }
91
92 /*
93  * 'diff.<what>.funcname' attribute can be specified in the configuration
94  * to define a customized regexp to find the beginning of a function to
95  * be used for hunk header lines of "diff -p" style output.
96  */
97 static struct funcname_pattern {
98         char *name;
99         char *pattern;
100         struct funcname_pattern *next;
101 } *funcname_pattern_list;
102
103 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
104 {
105         const char *name;
106         int namelen;
107         struct funcname_pattern *pp;
108
109         name = var + 5; /* "diff." */
110         namelen = ep - name;
111
112         for (pp = funcname_pattern_list; pp; pp = pp->next)
113                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
114                         break;
115         if (!pp) {
116                 pp = xcalloc(1, sizeof(*pp));
117                 pp->name = xmemdupz(name, namelen);
118                 pp->next = funcname_pattern_list;
119                 funcname_pattern_list = pp;
120         }
121         free(pp->pattern);
122         pp->pattern = xstrdup(value);
123         return 0;
124 }
125
126 /*
127  * These are to give UI layer defaults.
128  * The core-level commands such as git-diff-files should
129  * never be affected by the setting of diff.renames
130  * the user happens to have in the configuration file.
131  */
132 int git_diff_ui_config(const char *var, const char *value, void *cb)
133 {
134         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
135                 diff_use_color_default = git_config_colorbool(var, value, -1);
136                 return 0;
137         }
138         if (!strcmp(var, "diff.renames")) {
139                 if (!value)
140                         diff_detect_rename_default = DIFF_DETECT_RENAME;
141                 else if (!strcasecmp(value, "copies") ||
142                          !strcasecmp(value, "copy"))
143                         diff_detect_rename_default = DIFF_DETECT_COPY;
144                 else if (git_config_bool(var,value))
145                         diff_detect_rename_default = DIFF_DETECT_RENAME;
146                 return 0;
147         }
148         if (!strcmp(var, "diff.autorefreshindex")) {
149                 diff_auto_refresh_index = git_config_bool(var, value);
150                 return 0;
151         }
152         if (!strcmp(var, "diff.external"))
153                 return git_config_string(&external_diff_cmd_cfg, var, value);
154         if (!prefixcmp(var, "diff.")) {
155                 const char *ep = strrchr(var, '.');
156
157                 if (ep != var + 4 && !strcmp(ep, ".command"))
158                         return parse_lldiff_command(var, ep, value);
159         }
160
161         return git_diff_basic_config(var, value, cb);
162 }
163
164 int git_diff_basic_config(const char *var, const char *value, void *cb)
165 {
166         if (!strcmp(var, "diff.renamelimit")) {
167                 diff_rename_limit_default = git_config_int(var, value);
168                 return 0;
169         }
170
171         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
172                 int slot = parse_diff_color_slot(var, 11);
173                 if (!value)
174                         return config_error_nonbool(var);
175                 color_parse(value, var, diff_colors[slot]);
176                 return 0;
177         }
178
179         if (!prefixcmp(var, "diff.")) {
180                 const char *ep = strrchr(var, '.');
181                 if (ep != var + 4) {
182                         if (!strcmp(ep, ".funcname")) {
183                                 if (!value)
184                                         return config_error_nonbool(var);
185                                 return parse_funcname_pattern(var, ep, value);
186                         }
187                 }
188         }
189
190         return git_color_default_config(var, value, cb);
191 }
192
193 static char *quote_two(const char *one, const char *two)
194 {
195         int need_one = quote_c_style(one, NULL, NULL, 1);
196         int need_two = quote_c_style(two, NULL, NULL, 1);
197         struct strbuf res;
198
199         strbuf_init(&res, 0);
200         if (need_one + need_two) {
201                 strbuf_addch(&res, '"');
202                 quote_c_style(one, &res, NULL, 1);
203                 quote_c_style(two, &res, NULL, 1);
204                 strbuf_addch(&res, '"');
205         } else {
206                 strbuf_addstr(&res, one);
207                 strbuf_addstr(&res, two);
208         }
209         return strbuf_detach(&res, NULL);
210 }
211
212 static const char *external_diff(void)
213 {
214         static const char *external_diff_cmd = NULL;
215         static int done_preparing = 0;
216
217         if (done_preparing)
218                 return external_diff_cmd;
219         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
220         if (!external_diff_cmd)
221                 external_diff_cmd = external_diff_cmd_cfg;
222         done_preparing = 1;
223         return external_diff_cmd;
224 }
225
226 static struct diff_tempfile {
227         const char *name; /* filename external diff should read from */
228         char hex[41];
229         char mode[10];
230         char tmp_path[PATH_MAX];
231 } diff_temp[2];
232
233 static int count_lines(const char *data, int size)
234 {
235         int count, ch, completely_empty = 1, nl_just_seen = 0;
236         count = 0;
237         while (0 < size--) {
238                 ch = *data++;
239                 if (ch == '\n') {
240                         count++;
241                         nl_just_seen = 1;
242                         completely_empty = 0;
243                 }
244                 else {
245                         nl_just_seen = 0;
246                         completely_empty = 0;
247                 }
248         }
249         if (completely_empty)
250                 return 0;
251         if (!nl_just_seen)
252                 count++; /* no trailing newline */
253         return count;
254 }
255
256 static void print_line_count(FILE *file, int count)
257 {
258         switch (count) {
259         case 0:
260                 fprintf(file, "0,0");
261                 break;
262         case 1:
263                 fprintf(file, "1");
264                 break;
265         default:
266                 fprintf(file, "1,%d", count);
267                 break;
268         }
269 }
270
271 static void copy_file_with_prefix(FILE *file,
272                                   int prefix, const char *data, int size,
273                                   const char *set, const char *reset)
274 {
275         int ch, nl_just_seen = 1;
276         while (0 < size--) {
277                 ch = *data++;
278                 if (nl_just_seen) {
279                         fputs(set, file);
280                         putc(prefix, file);
281                 }
282                 if (ch == '\n') {
283                         nl_just_seen = 1;
284                         fputs(reset, file);
285                 } else
286                         nl_just_seen = 0;
287                 putc(ch, file);
288         }
289         if (!nl_just_seen)
290                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
291 }
292
293 static void emit_rewrite_diff(const char *name_a,
294                               const char *name_b,
295                               struct diff_filespec *one,
296                               struct diff_filespec *two,
297                               struct diff_options *o)
298 {
299         int lc_a, lc_b;
300         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
301         const char *name_a_tab, *name_b_tab;
302         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
303         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
304         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
305         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
306         const char *reset = diff_get_color(color_diff, DIFF_RESET);
307         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
308
309         name_a += (*name_a == '/');
310         name_b += (*name_b == '/');
311         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
312         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
313
314         strbuf_reset(&a_name);
315         strbuf_reset(&b_name);
316         quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
317         quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
318
319         diff_populate_filespec(one, 0);
320         diff_populate_filespec(two, 0);
321         lc_a = count_lines(one->data, one->size);
322         lc_b = count_lines(two->data, two->size);
323         fprintf(o->file,
324                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
325                 metainfo, a_name.buf, name_a_tab, reset,
326                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
327         print_line_count(o->file, lc_a);
328         fprintf(o->file, " +");
329         print_line_count(o->file, lc_b);
330         fprintf(o->file, " @@%s\n", reset);
331         if (lc_a)
332                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
333         if (lc_b)
334                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
335 }
336
337 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
338 {
339         if (!DIFF_FILE_VALID(one)) {
340                 mf->ptr = (char *)""; /* does not matter */
341                 mf->size = 0;
342                 return 0;
343         }
344         else if (diff_populate_filespec(one, 0))
345                 return -1;
346         mf->ptr = one->data;
347         mf->size = one->size;
348         return 0;
349 }
350
351 struct diff_words_buffer {
352         mmfile_t text;
353         long alloc;
354         long current; /* output pointer */
355         int suppressed_newline;
356 };
357
358 static void diff_words_append(char *line, unsigned long len,
359                 struct diff_words_buffer *buffer)
360 {
361         if (buffer->text.size + len > buffer->alloc) {
362                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
363                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
364         }
365         line++;
366         len--;
367         memcpy(buffer->text.ptr + buffer->text.size, line, len);
368         buffer->text.size += len;
369 }
370
371 struct diff_words_data {
372         struct xdiff_emit_state xm;
373         struct diff_words_buffer minus, plus;
374         FILE *file;
375 };
376
377 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
378                 int suppress_newline)
379 {
380         const char *ptr;
381         int eol = 0;
382
383         if (len == 0)
384                 return;
385
386         ptr  = buffer->text.ptr + buffer->current;
387         buffer->current += len;
388
389         if (ptr[len - 1] == '\n') {
390                 eol = 1;
391                 len--;
392         }
393
394         fputs(diff_get_color(1, color), file);
395         fwrite(ptr, len, 1, file);
396         fputs(diff_get_color(1, DIFF_RESET), file);
397
398         if (eol) {
399                 if (suppress_newline)
400                         buffer->suppressed_newline = 1;
401                 else
402                         putc('\n', file);
403         }
404 }
405
406 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
407 {
408         struct diff_words_data *diff_words = priv;
409
410         if (diff_words->minus.suppressed_newline) {
411                 if (line[0] != '+')
412                         putc('\n', diff_words->file);
413                 diff_words->minus.suppressed_newline = 0;
414         }
415
416         len--;
417         switch (line[0]) {
418                 case '-':
419                         print_word(diff_words->file,
420                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
421                         break;
422                 case '+':
423                         print_word(diff_words->file,
424                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
425                         break;
426                 case ' ':
427                         print_word(diff_words->file,
428                                    &diff_words->plus, len, DIFF_PLAIN, 0);
429                         diff_words->minus.current += len;
430                         break;
431         }
432 }
433
434 /* this executes the word diff on the accumulated buffers */
435 static void diff_words_show(struct diff_words_data *diff_words)
436 {
437         xpparam_t xpp;
438         xdemitconf_t xecfg;
439         xdemitcb_t ecb;
440         mmfile_t minus, plus;
441         int i;
442
443         memset(&xecfg, 0, sizeof(xecfg));
444         minus.size = diff_words->minus.text.size;
445         minus.ptr = xmalloc(minus.size);
446         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
447         for (i = 0; i < minus.size; i++)
448                 if (isspace(minus.ptr[i]))
449                         minus.ptr[i] = '\n';
450         diff_words->minus.current = 0;
451
452         plus.size = diff_words->plus.text.size;
453         plus.ptr = xmalloc(plus.size);
454         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
455         for (i = 0; i < plus.size; i++)
456                 if (isspace(plus.ptr[i]))
457                         plus.ptr[i] = '\n';
458         diff_words->plus.current = 0;
459
460         xpp.flags = XDF_NEED_MINIMAL;
461         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
462         ecb.outf = xdiff_outf;
463         ecb.priv = diff_words;
464         diff_words->xm.consume = fn_out_diff_words_aux;
465         xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
466
467         free(minus.ptr);
468         free(plus.ptr);
469         diff_words->minus.text.size = diff_words->plus.text.size = 0;
470
471         if (diff_words->minus.suppressed_newline) {
472                 putc('\n', diff_words->file);
473                 diff_words->minus.suppressed_newline = 0;
474         }
475 }
476
477 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
478
479 struct emit_callback {
480         struct xdiff_emit_state xm;
481         int nparents, color_diff;
482         unsigned ws_rule;
483         sane_truncate_fn truncate;
484         const char **label_path;
485         struct diff_words_data *diff_words;
486         int *found_changesp;
487         FILE *file;
488 };
489
490 static void free_diff_words_data(struct emit_callback *ecbdata)
491 {
492         if (ecbdata->diff_words) {
493                 /* flush buffers */
494                 if (ecbdata->diff_words->minus.text.size ||
495                                 ecbdata->diff_words->plus.text.size)
496                         diff_words_show(ecbdata->diff_words);
497
498                 free (ecbdata->diff_words->minus.text.ptr);
499                 free (ecbdata->diff_words->plus.text.ptr);
500                 free(ecbdata->diff_words);
501                 ecbdata->diff_words = NULL;
502         }
503 }
504
505 const char *diff_get_color(int diff_use_color, enum color_diff ix)
506 {
507         if (diff_use_color)
508                 return diff_colors[ix];
509         return "";
510 }
511
512 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
513 {
514         int has_trailing_newline = (len > 0 && line[len-1] == '\n');
515         if (has_trailing_newline)
516                 len--;
517
518         fputs(set, file);
519         fwrite(line, len, 1, file);
520         fputs(reset, file);
521         if (has_trailing_newline)
522                 fputc('\n', file);
523 }
524
525 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
526 {
527         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
528         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
529
530         if (!*ws)
531                 emit_line(ecbdata->file, set, reset, line, len);
532         else {
533                 /* Emit just the prefix, then the rest. */
534                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
535                 ws_check_emit(line + ecbdata->nparents,
536                               len - ecbdata->nparents, ecbdata->ws_rule,
537                               ecbdata->file, set, reset, ws);
538         }
539 }
540
541 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
542 {
543         const char *cp;
544         unsigned long allot;
545         size_t l = len;
546
547         if (ecb->truncate)
548                 return ecb->truncate(line, len);
549         cp = line;
550         allot = l;
551         while (0 < l) {
552                 (void) utf8_width(&cp, &l);
553                 if (!cp)
554                         break; /* truncated in the middle? */
555         }
556         return allot - l;
557 }
558
559 static void fn_out_consume(void *priv, char *line, unsigned long len)
560 {
561         int i;
562         int color;
563         struct emit_callback *ecbdata = priv;
564         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
565         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
566         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
567
568         *(ecbdata->found_changesp) = 1;
569
570         if (ecbdata->label_path[0]) {
571                 const char *name_a_tab, *name_b_tab;
572
573                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
574                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
575
576                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
577                         meta, ecbdata->label_path[0], reset, name_a_tab);
578                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
579                         meta, ecbdata->label_path[1], reset, name_b_tab);
580                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
581         }
582
583         /* This is not really necessary for now because
584          * this codepath only deals with two-way diffs.
585          */
586         for (i = 0; i < len && line[i] == '@'; i++)
587                 ;
588         if (2 <= i && i < len && line[i] == ' ') {
589                 ecbdata->nparents = i - 1;
590                 len = sane_truncate_line(ecbdata, line, len);
591                 emit_line(ecbdata->file,
592                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
593                           reset, line, len);
594                 if (line[len-1] != '\n')
595                         putc('\n', ecbdata->file);
596                 return;
597         }
598
599         if (len < ecbdata->nparents) {
600                 emit_line(ecbdata->file, reset, reset, line, len);
601                 return;
602         }
603
604         color = DIFF_PLAIN;
605         if (ecbdata->diff_words && ecbdata->nparents != 1)
606                 /* fall back to normal diff */
607                 free_diff_words_data(ecbdata);
608         if (ecbdata->diff_words) {
609                 if (line[0] == '-') {
610                         diff_words_append(line, len,
611                                           &ecbdata->diff_words->minus);
612                         return;
613                 } else if (line[0] == '+') {
614                         diff_words_append(line, len,
615                                           &ecbdata->diff_words->plus);
616                         return;
617                 }
618                 if (ecbdata->diff_words->minus.text.size ||
619                     ecbdata->diff_words->plus.text.size)
620                         diff_words_show(ecbdata->diff_words);
621                 line++;
622                 len--;
623                 emit_line(ecbdata->file, plain, reset, line, len);
624                 return;
625         }
626         for (i = 0; i < ecbdata->nparents && len; i++) {
627                 if (line[i] == '-')
628                         color = DIFF_FILE_OLD;
629                 else if (line[i] == '+')
630                         color = DIFF_FILE_NEW;
631         }
632
633         if (color != DIFF_FILE_NEW) {
634                 emit_line(ecbdata->file,
635                           diff_get_color(ecbdata->color_diff, color),
636                           reset, line, len);
637                 return;
638         }
639         emit_add_line(reset, ecbdata, line, len);
640 }
641
642 static char *pprint_rename(const char *a, const char *b)
643 {
644         const char *old = a;
645         const char *new = b;
646         struct strbuf name;
647         int pfx_length, sfx_length;
648         int len_a = strlen(a);
649         int len_b = strlen(b);
650         int a_midlen, b_midlen;
651         int qlen_a = quote_c_style(a, NULL, NULL, 0);
652         int qlen_b = quote_c_style(b, NULL, NULL, 0);
653
654         strbuf_init(&name, 0);
655         if (qlen_a || qlen_b) {
656                 quote_c_style(a, &name, NULL, 0);
657                 strbuf_addstr(&name, " => ");
658                 quote_c_style(b, &name, NULL, 0);
659                 return strbuf_detach(&name, NULL);
660         }
661
662         /* Find common prefix */
663         pfx_length = 0;
664         while (*old && *new && *old == *new) {
665                 if (*old == '/')
666                         pfx_length = old - a + 1;
667                 old++;
668                 new++;
669         }
670
671         /* Find common suffix */
672         old = a + len_a;
673         new = b + len_b;
674         sfx_length = 0;
675         while (a <= old && b <= new && *old == *new) {
676                 if (*old == '/')
677                         sfx_length = len_a - (old - a);
678                 old--;
679                 new--;
680         }
681
682         /*
683          * pfx{mid-a => mid-b}sfx
684          * {pfx-a => pfx-b}sfx
685          * pfx{sfx-a => sfx-b}
686          * name-a => name-b
687          */
688         a_midlen = len_a - pfx_length - sfx_length;
689         b_midlen = len_b - pfx_length - sfx_length;
690         if (a_midlen < 0)
691                 a_midlen = 0;
692         if (b_midlen < 0)
693                 b_midlen = 0;
694
695         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
696         if (pfx_length + sfx_length) {
697                 strbuf_add(&name, a, pfx_length);
698                 strbuf_addch(&name, '{');
699         }
700         strbuf_add(&name, a + pfx_length, a_midlen);
701         strbuf_addstr(&name, " => ");
702         strbuf_add(&name, b + pfx_length, b_midlen);
703         if (pfx_length + sfx_length) {
704                 strbuf_addch(&name, '}');
705                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
706         }
707         return strbuf_detach(&name, NULL);
708 }
709
710 struct diffstat_t {
711         struct xdiff_emit_state xm;
712
713         int nr;
714         int alloc;
715         struct diffstat_file {
716                 char *from_name;
717                 char *name;
718                 char *print_name;
719                 unsigned is_unmerged:1;
720                 unsigned is_binary:1;
721                 unsigned is_renamed:1;
722                 unsigned int added, deleted;
723         } **files;
724 };
725
726 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
727                                           const char *name_a,
728                                           const char *name_b)
729 {
730         struct diffstat_file *x;
731         x = xcalloc(sizeof (*x), 1);
732         if (diffstat->nr == diffstat->alloc) {
733                 diffstat->alloc = alloc_nr(diffstat->alloc);
734                 diffstat->files = xrealloc(diffstat->files,
735                                 diffstat->alloc * sizeof(x));
736         }
737         diffstat->files[diffstat->nr++] = x;
738         if (name_b) {
739                 x->from_name = xstrdup(name_a);
740                 x->name = xstrdup(name_b);
741                 x->is_renamed = 1;
742         }
743         else {
744                 x->from_name = NULL;
745                 x->name = xstrdup(name_a);
746         }
747         return x;
748 }
749
750 static void diffstat_consume(void *priv, char *line, unsigned long len)
751 {
752         struct diffstat_t *diffstat = priv;
753         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
754
755         if (line[0] == '+')
756                 x->added++;
757         else if (line[0] == '-')
758                 x->deleted++;
759 }
760
761 const char mime_boundary_leader[] = "------------";
762
763 static int scale_linear(int it, int width, int max_change)
764 {
765         /*
766          * make sure that at least one '-' is printed if there were deletions,
767          * and likewise for '+'.
768          */
769         if (max_change < 2)
770                 return it;
771         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
772 }
773
774 static void show_name(FILE *file,
775                       const char *prefix, const char *name, int len,
776                       const char *reset, const char *set)
777 {
778         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
779 }
780
781 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
782 {
783         if (cnt <= 0)
784                 return;
785         fprintf(file, "%s", set);
786         while (cnt--)
787                 putc(ch, file);
788         fprintf(file, "%s", reset);
789 }
790
791 static void fill_print_name(struct diffstat_file *file)
792 {
793         char *pname;
794
795         if (file->print_name)
796                 return;
797
798         if (!file->is_renamed) {
799                 struct strbuf buf;
800                 strbuf_init(&buf, 0);
801                 if (quote_c_style(file->name, &buf, NULL, 0)) {
802                         pname = strbuf_detach(&buf, NULL);
803                 } else {
804                         pname = file->name;
805                         strbuf_release(&buf);
806                 }
807         } else {
808                 pname = pprint_rename(file->from_name, file->name);
809         }
810         file->print_name = pname;
811 }
812
813 static void show_stats(struct diffstat_t* data, struct diff_options *options)
814 {
815         int i, len, add, del, total, adds = 0, dels = 0;
816         int max_change = 0, max_len = 0;
817         int total_files = data->nr;
818         int width, name_width;
819         const char *reset, *set, *add_c, *del_c;
820
821         if (data->nr == 0)
822                 return;
823
824         width = options->stat_width ? options->stat_width : 80;
825         name_width = options->stat_name_width ? options->stat_name_width : 50;
826
827         /* Sanity: give at least 5 columns to the graph,
828          * but leave at least 10 columns for the name.
829          */
830         if (width < 25)
831                 width = 25;
832         if (name_width < 10)
833                 name_width = 10;
834         else if (width < name_width + 15)
835                 name_width = width - 15;
836
837         /* Find the longest filename and max number of changes */
838         reset = diff_get_color_opt(options, DIFF_RESET);
839         set   = diff_get_color_opt(options, DIFF_PLAIN);
840         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
841         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
842
843         for (i = 0; i < data->nr; i++) {
844                 struct diffstat_file *file = data->files[i];
845                 int change = file->added + file->deleted;
846                 fill_print_name(file);
847                 len = strlen(file->print_name);
848                 if (max_len < len)
849                         max_len = len;
850
851                 if (file->is_binary || file->is_unmerged)
852                         continue;
853                 if (max_change < change)
854                         max_change = change;
855         }
856
857         /* Compute the width of the graph part;
858          * 10 is for one blank at the beginning of the line plus
859          * " | count " between the name and the graph.
860          *
861          * From here on, name_width is the width of the name area,
862          * and width is the width of the graph area.
863          */
864         name_width = (name_width < max_len) ? name_width : max_len;
865         if (width < (name_width + 10) + max_change)
866                 width = width - (name_width + 10);
867         else
868                 width = max_change;
869
870         for (i = 0; i < data->nr; i++) {
871                 const char *prefix = "";
872                 char *name = data->files[i]->print_name;
873                 int added = data->files[i]->added;
874                 int deleted = data->files[i]->deleted;
875                 int name_len;
876
877                 /*
878                  * "scale" the filename
879                  */
880                 len = name_width;
881                 name_len = strlen(name);
882                 if (name_width < name_len) {
883                         char *slash;
884                         prefix = "...";
885                         len -= 3;
886                         name += name_len - len;
887                         slash = strchr(name, '/');
888                         if (slash)
889                                 name = slash;
890                 }
891
892                 if (data->files[i]->is_binary) {
893                         show_name(options->file, prefix, name, len, reset, set);
894                         fprintf(options->file, "  Bin ");
895                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
896                         fprintf(options->file, " -> ");
897                         fprintf(options->file, "%s%d%s", add_c, added, reset);
898                         fprintf(options->file, " bytes");
899                         fprintf(options->file, "\n");
900                         continue;
901                 }
902                 else if (data->files[i]->is_unmerged) {
903                         show_name(options->file, prefix, name, len, reset, set);
904                         fprintf(options->file, "  Unmerged\n");
905                         continue;
906                 }
907                 else if (!data->files[i]->is_renamed &&
908                          (added + deleted == 0)) {
909                         total_files--;
910                         continue;
911                 }
912
913                 /*
914                  * scale the add/delete
915                  */
916                 add = added;
917                 del = deleted;
918                 total = add + del;
919                 adds += add;
920                 dels += del;
921
922                 if (width <= max_change) {
923                         add = scale_linear(add, width, max_change);
924                         del = scale_linear(del, width, max_change);
925                         total = add + del;
926                 }
927                 show_name(options->file, prefix, name, len, reset, set);
928                 fprintf(options->file, "%5d%s", added + deleted,
929                                 added + deleted ? " " : "");
930                 show_graph(options->file, '+', add, add_c, reset);
931                 show_graph(options->file, '-', del, del_c, reset);
932                 fprintf(options->file, "\n");
933         }
934         fprintf(options->file,
935                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
936                set, total_files, adds, dels, reset);
937 }
938
939 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
940 {
941         int i, adds = 0, dels = 0, total_files = data->nr;
942
943         if (data->nr == 0)
944                 return;
945
946         for (i = 0; i < data->nr; i++) {
947                 if (!data->files[i]->is_binary &&
948                     !data->files[i]->is_unmerged) {
949                         int added = data->files[i]->added;
950                         int deleted= data->files[i]->deleted;
951                         if (!data->files[i]->is_renamed &&
952                             (added + deleted == 0)) {
953                                 total_files--;
954                         } else {
955                                 adds += added;
956                                 dels += deleted;
957                         }
958                 }
959         }
960         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
961                total_files, adds, dels);
962 }
963
964 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
965 {
966         int i;
967
968         if (data->nr == 0)
969                 return;
970
971         for (i = 0; i < data->nr; i++) {
972                 struct diffstat_file *file = data->files[i];
973
974                 if (file->is_binary)
975                         fprintf(options->file, "-\t-\t");
976                 else
977                         fprintf(options->file,
978                                 "%d\t%d\t", file->added, file->deleted);
979                 if (options->line_termination) {
980                         fill_print_name(file);
981                         if (!file->is_renamed)
982                                 write_name_quoted(file->name, options->file,
983                                                   options->line_termination);
984                         else {
985                                 fputs(file->print_name, options->file);
986                                 putc(options->line_termination, options->file);
987                         }
988                 } else {
989                         if (file->is_renamed) {
990                                 putc('\0', options->file);
991                                 write_name_quoted(file->from_name, options->file, '\0');
992                         }
993                         write_name_quoted(file->name, options->file, '\0');
994                 }
995         }
996 }
997
998 struct dirstat_file {
999         const char *name;
1000         unsigned long changed;
1001 };
1002
1003 struct dirstat_dir {
1004         struct dirstat_file *files;
1005         int alloc, nr, percent, cumulative;
1006 };
1007
1008 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1009 {
1010         unsigned long this_dir = 0;
1011         unsigned int sources = 0;
1012
1013         while (dir->nr) {
1014                 struct dirstat_file *f = dir->files;
1015                 int namelen = strlen(f->name);
1016                 unsigned long this;
1017                 char *slash;
1018
1019                 if (namelen < baselen)
1020                         break;
1021                 if (memcmp(f->name, base, baselen))
1022                         break;
1023                 slash = strchr(f->name + baselen, '/');
1024                 if (slash) {
1025                         int newbaselen = slash + 1 - f->name;
1026                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1027                         sources++;
1028                 } else {
1029                         this = f->changed;
1030                         dir->files++;
1031                         dir->nr--;
1032                         sources += 2;
1033                 }
1034                 this_dir += this;
1035         }
1036
1037         /*
1038          * We don't report dirstat's for
1039          *  - the top level
1040          *  - or cases where everything came from a single directory
1041          *    under this directory (sources == 1).
1042          */
1043         if (baselen && sources != 1) {
1044                 int permille = this_dir * 1000 / changed;
1045                 if (permille) {
1046                         int percent = permille / 10;
1047                         if (percent >= dir->percent) {
1048                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1049                                 if (!dir->cumulative)
1050                                         return 0;
1051                         }
1052                 }
1053         }
1054         return this_dir;
1055 }
1056
1057 static void show_dirstat(struct diff_options *options)
1058 {
1059         int i;
1060         unsigned long changed;
1061         struct dirstat_dir dir;
1062         struct diff_queue_struct *q = &diff_queued_diff;
1063
1064         dir.files = NULL;
1065         dir.alloc = 0;
1066         dir.nr = 0;
1067         dir.percent = options->dirstat_percent;
1068         dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1069
1070         changed = 0;
1071         for (i = 0; i < q->nr; i++) {
1072                 struct diff_filepair *p = q->queue[i];
1073                 const char *name;
1074                 unsigned long copied, added, damage;
1075
1076                 name = p->one->path ? p->one->path : p->two->path;
1077
1078                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1079                         diff_populate_filespec(p->one, 0);
1080                         diff_populate_filespec(p->two, 0);
1081                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1082                                                &copied, &added);
1083                         diff_free_filespec_data(p->one);
1084                         diff_free_filespec_data(p->two);
1085                 } else if (DIFF_FILE_VALID(p->one)) {
1086                         diff_populate_filespec(p->one, 1);
1087                         copied = added = 0;
1088                         diff_free_filespec_data(p->one);
1089                 } else if (DIFF_FILE_VALID(p->two)) {
1090                         diff_populate_filespec(p->two, 1);
1091                         copied = 0;
1092                         added = p->two->size;
1093                         diff_free_filespec_data(p->two);
1094                 } else
1095                         continue;
1096
1097                 /*
1098                  * Original minus copied is the removed material,
1099                  * added is the new material.  They are both damages
1100                  * made to the preimage.
1101                  */
1102                 damage = (p->one->size - copied) + added;
1103
1104                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1105                 dir.files[dir.nr].name = name;
1106                 dir.files[dir.nr].changed = damage;
1107                 changed += damage;
1108                 dir.nr++;
1109         }
1110
1111         /* This can happen even with many files, if everything was renames */
1112         if (!changed)
1113                 return;
1114
1115         /* Show all directories with more than x% of the changes */
1116         gather_dirstat(options->file, &dir, changed, "", 0);
1117 }
1118
1119 static void free_diffstat_info(struct diffstat_t *diffstat)
1120 {
1121         int i;
1122         for (i = 0; i < diffstat->nr; i++) {
1123                 struct diffstat_file *f = diffstat->files[i];
1124                 if (f->name != f->print_name)
1125                         free(f->print_name);
1126                 free(f->name);
1127                 free(f->from_name);
1128                 free(f);
1129         }
1130         free(diffstat->files);
1131 }
1132
1133 struct checkdiff_t {
1134         struct xdiff_emit_state xm;
1135         const char *filename;
1136         int lineno;
1137         struct diff_options *o;
1138         unsigned ws_rule;
1139         unsigned status;
1140         int trailing_blanks_start;
1141 };
1142
1143 static int is_conflict_marker(const char *line, unsigned long len)
1144 {
1145         char firstchar;
1146         int cnt;
1147
1148         if (len < 8)
1149                 return 0;
1150         firstchar = line[0];
1151         switch (firstchar) {
1152         case '=': case '>': case '<':
1153                 break;
1154         default:
1155                 return 0;
1156         }
1157         for (cnt = 1; cnt < 7; cnt++)
1158                 if (line[cnt] != firstchar)
1159                         return 0;
1160         /* line[0] thru line[6] are same as firstchar */
1161         if (firstchar == '=') {
1162                 /* divider between ours and theirs? */
1163                 if (len != 8 || line[7] != '\n')
1164                         return 0;
1165         } else if (len < 8 || !isspace(line[7])) {
1166                 /* not divider before ours nor after theirs */
1167                 return 0;
1168         }
1169         return 1;
1170 }
1171
1172 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1173 {
1174         struct checkdiff_t *data = priv;
1175         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1176         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1177         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1178         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1179         char *err;
1180
1181         if (line[0] == '+') {
1182                 unsigned bad;
1183                 data->lineno++;
1184                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1185                         data->trailing_blanks_start = 0;
1186                 else if (!data->trailing_blanks_start)
1187                         data->trailing_blanks_start = data->lineno;
1188                 if (is_conflict_marker(line + 1, len - 1)) {
1189                         data->status |= 1;
1190                         fprintf(data->o->file,
1191                                 "%s:%d: leftover conflict marker\n",
1192                                 data->filename, data->lineno);
1193                 }
1194                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1195                 if (!bad)
1196                         return;
1197                 data->status |= bad;
1198                 err = whitespace_error_string(bad);
1199                 fprintf(data->o->file, "%s:%d: %s.\n",
1200                         data->filename, data->lineno, err);
1201                 free(err);
1202                 emit_line(data->o->file, set, reset, line, 1);
1203                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1204                               data->o->file, set, reset, ws);
1205         } else if (line[0] == ' ') {
1206                 data->lineno++;
1207                 data->trailing_blanks_start = 0;
1208         } else if (line[0] == '@') {
1209                 char *plus = strchr(line, '+');
1210                 if (plus)
1211                         data->lineno = strtol(plus, NULL, 10) - 1;
1212                 else
1213                         die("invalid diff");
1214                 data->trailing_blanks_start = 0;
1215         }
1216 }
1217
1218 static unsigned char *deflate_it(char *data,
1219                                  unsigned long size,
1220                                  unsigned long *result_size)
1221 {
1222         int bound;
1223         unsigned char *deflated;
1224         z_stream stream;
1225
1226         memset(&stream, 0, sizeof(stream));
1227         deflateInit(&stream, zlib_compression_level);
1228         bound = deflateBound(&stream, size);
1229         deflated = xmalloc(bound);
1230         stream.next_out = deflated;
1231         stream.avail_out = bound;
1232
1233         stream.next_in = (unsigned char *)data;
1234         stream.avail_in = size;
1235         while (deflate(&stream, Z_FINISH) == Z_OK)
1236                 ; /* nothing */
1237         deflateEnd(&stream);
1238         *result_size = stream.total_out;
1239         return deflated;
1240 }
1241
1242 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1243 {
1244         void *cp;
1245         void *delta;
1246         void *deflated;
1247         void *data;
1248         unsigned long orig_size;
1249         unsigned long delta_size;
1250         unsigned long deflate_size;
1251         unsigned long data_size;
1252
1253         /* We could do deflated delta, or we could do just deflated two,
1254          * whichever is smaller.
1255          */
1256         delta = NULL;
1257         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1258         if (one->size && two->size) {
1259                 delta = diff_delta(one->ptr, one->size,
1260                                    two->ptr, two->size,
1261                                    &delta_size, deflate_size);
1262                 if (delta) {
1263                         void *to_free = delta;
1264                         orig_size = delta_size;
1265                         delta = deflate_it(delta, delta_size, &delta_size);
1266                         free(to_free);
1267                 }
1268         }
1269
1270         if (delta && delta_size < deflate_size) {
1271                 fprintf(file, "delta %lu\n", orig_size);
1272                 free(deflated);
1273                 data = delta;
1274                 data_size = delta_size;
1275         }
1276         else {
1277                 fprintf(file, "literal %lu\n", two->size);
1278                 free(delta);
1279                 data = deflated;
1280                 data_size = deflate_size;
1281         }
1282
1283         /* emit data encoded in base85 */
1284         cp = data;
1285         while (data_size) {
1286                 int bytes = (52 < data_size) ? 52 : data_size;
1287                 char line[70];
1288                 data_size -= bytes;
1289                 if (bytes <= 26)
1290                         line[0] = bytes + 'A' - 1;
1291                 else
1292                         line[0] = bytes - 26 + 'a' - 1;
1293                 encode_85(line + 1, cp, bytes);
1294                 cp = (char *) cp + bytes;
1295                 fputs(line, file);
1296                 fputc('\n', file);
1297         }
1298         fprintf(file, "\n");
1299         free(data);
1300 }
1301
1302 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1303 {
1304         fprintf(file, "GIT binary patch\n");
1305         emit_binary_diff_body(file, one, two);
1306         emit_binary_diff_body(file, two, one);
1307 }
1308
1309 static void setup_diff_attr_check(struct git_attr_check *check)
1310 {
1311         static struct git_attr *attr_diff;
1312
1313         if (!attr_diff) {
1314                 attr_diff = git_attr("diff", 4);
1315         }
1316         check[0].attr = attr_diff;
1317 }
1318
1319 static void diff_filespec_check_attr(struct diff_filespec *one)
1320 {
1321         struct git_attr_check attr_diff_check;
1322         int check_from_data = 0;
1323
1324         if (one->checked_attr)
1325                 return;
1326
1327         setup_diff_attr_check(&attr_diff_check);
1328         one->is_binary = 0;
1329         one->funcname_pattern_ident = NULL;
1330
1331         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1332                 const char *value;
1333
1334                 /* binaryness */
1335                 value = attr_diff_check.value;
1336                 if (ATTR_TRUE(value))
1337                         ;
1338                 else if (ATTR_FALSE(value))
1339                         one->is_binary = 1;
1340                 else
1341                         check_from_data = 1;
1342
1343                 /* funcname pattern ident */
1344                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1345                         ;
1346                 else
1347                         one->funcname_pattern_ident = value;
1348         }
1349
1350         if (check_from_data) {
1351                 if (!one->data && DIFF_FILE_VALID(one))
1352                         diff_populate_filespec(one, 0);
1353
1354                 if (one->data)
1355                         one->is_binary = buffer_is_binary(one->data, one->size);
1356         }
1357 }
1358
1359 int diff_filespec_is_binary(struct diff_filespec *one)
1360 {
1361         diff_filespec_check_attr(one);
1362         return one->is_binary;
1363 }
1364
1365 static const char *funcname_pattern(const char *ident)
1366 {
1367         struct funcname_pattern *pp;
1368
1369         for (pp = funcname_pattern_list; pp; pp = pp->next)
1370                 if (!strcmp(ident, pp->name))
1371                         return pp->pattern;
1372         return NULL;
1373 }
1374
1375 static struct builtin_funcname_pattern {
1376         const char *name;
1377         const char *pattern;
1378 } builtin_funcname_pattern[] = {
1379         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1380                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1381                         "^[     ]*\\(\\([       ]*"
1382                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1383                         "[      ]*([^;]*\\)$" },
1384         { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1385                         "destructor\\|interface\\|implementation\\|"
1386                         "initialization\\|finalization\\)[ \t]*.*\\)$"
1387                         "\\|"
1388                         "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1389                         },
1390         { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1391         { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1392 };
1393
1394 static const char *diff_funcname_pattern(struct diff_filespec *one)
1395 {
1396         const char *ident, *pattern;
1397         int i;
1398
1399         diff_filespec_check_attr(one);
1400         ident = one->funcname_pattern_ident;
1401
1402         if (!ident)
1403                 /*
1404                  * If the config file has "funcname.default" defined, that
1405                  * regexp is used; otherwise NULL is returned and xemit uses
1406                  * the built-in default.
1407                  */
1408                 return funcname_pattern("default");
1409
1410         /* Look up custom "funcname.$ident" regexp from config. */
1411         pattern = funcname_pattern(ident);
1412         if (pattern)
1413                 return pattern;
1414
1415         /*
1416          * And define built-in fallback patterns here.  Note that
1417          * these can be overridden by the user's config settings.
1418          */
1419         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1420                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1421                         return builtin_funcname_pattern[i].pattern;
1422
1423         return NULL;
1424 }
1425
1426 static void builtin_diff(const char *name_a,
1427                          const char *name_b,
1428                          struct diff_filespec *one,
1429                          struct diff_filespec *two,
1430                          const char *xfrm_msg,
1431                          struct diff_options *o,
1432                          int complete_rewrite)
1433 {
1434         mmfile_t mf1, mf2;
1435         const char *lbl[2];
1436         char *a_one, *b_two;
1437         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1438         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1439
1440         a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1441         b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1442         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1443         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1444         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1445         if (lbl[0][0] == '/') {
1446                 /* /dev/null */
1447                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1448                 if (xfrm_msg && xfrm_msg[0])
1449                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1450         }
1451         else if (lbl[1][0] == '/') {
1452                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1453                 if (xfrm_msg && xfrm_msg[0])
1454                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1455         }
1456         else {
1457                 if (one->mode != two->mode) {
1458                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1459                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1460                 }
1461                 if (xfrm_msg && xfrm_msg[0])
1462                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1463                 /*
1464                  * we do not run diff between different kind
1465                  * of objects.
1466                  */
1467                 if ((one->mode ^ two->mode) & S_IFMT)
1468                         goto free_ab_and_return;
1469                 if (complete_rewrite) {
1470                         emit_rewrite_diff(name_a, name_b, one, two, o);
1471                         o->found_changes = 1;
1472                         goto free_ab_and_return;
1473                 }
1474         }
1475
1476         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1477                 die("unable to read files to diff");
1478
1479         if (!DIFF_OPT_TST(o, TEXT) &&
1480             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1481                 /* Quite common confusing case */
1482                 if (mf1.size == mf2.size &&
1483                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1484                         goto free_ab_and_return;
1485                 if (DIFF_OPT_TST(o, BINARY))
1486                         emit_binary_diff(o->file, &mf1, &mf2);
1487                 else
1488                         fprintf(o->file, "Binary files %s and %s differ\n",
1489                                 lbl[0], lbl[1]);
1490                 o->found_changes = 1;
1491         }
1492         else {
1493                 /* Crazy xdl interfaces.. */
1494                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1495                 xpparam_t xpp;
1496                 xdemitconf_t xecfg;
1497                 xdemitcb_t ecb;
1498                 struct emit_callback ecbdata;
1499                 const char *funcname_pattern;
1500
1501                 funcname_pattern = diff_funcname_pattern(one);
1502                 if (!funcname_pattern)
1503                         funcname_pattern = diff_funcname_pattern(two);
1504
1505                 memset(&xecfg, 0, sizeof(xecfg));
1506                 memset(&ecbdata, 0, sizeof(ecbdata));
1507                 ecbdata.label_path = lbl;
1508                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1509                 ecbdata.found_changesp = &o->found_changes;
1510                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1511                 ecbdata.file = o->file;
1512                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1513                 xecfg.ctxlen = o->context;
1514                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1515                 if (funcname_pattern)
1516                         xdiff_set_find_func(&xecfg, funcname_pattern);
1517                 if (!diffopts)
1518                         ;
1519                 else if (!prefixcmp(diffopts, "--unified="))
1520                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1521                 else if (!prefixcmp(diffopts, "-u"))
1522                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1523                 ecb.outf = xdiff_outf;
1524                 ecb.priv = &ecbdata;
1525                 ecbdata.xm.consume = fn_out_consume;
1526                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1527                         ecbdata.diff_words =
1528                                 xcalloc(1, sizeof(struct diff_words_data));
1529                         ecbdata.diff_words->file = o->file;
1530                 }
1531                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1532                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1533                         free_diff_words_data(&ecbdata);
1534         }
1535
1536  free_ab_and_return:
1537         diff_free_filespec_data(one);
1538         diff_free_filespec_data(two);
1539         free(a_one);
1540         free(b_two);
1541         return;
1542 }
1543
1544 static void builtin_diffstat(const char *name_a, const char *name_b,
1545                              struct diff_filespec *one,
1546                              struct diff_filespec *two,
1547                              struct diffstat_t *diffstat,
1548                              struct diff_options *o,
1549                              int complete_rewrite)
1550 {
1551         mmfile_t mf1, mf2;
1552         struct diffstat_file *data;
1553
1554         data = diffstat_add(diffstat, name_a, name_b);
1555
1556         if (!one || !two) {
1557                 data->is_unmerged = 1;
1558                 return;
1559         }
1560         if (complete_rewrite) {
1561                 diff_populate_filespec(one, 0);
1562                 diff_populate_filespec(two, 0);
1563                 data->deleted = count_lines(one->data, one->size);
1564                 data->added = count_lines(two->data, two->size);
1565                 goto free_and_return;
1566         }
1567         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1568                 die("unable to read files to diff");
1569
1570         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1571                 data->is_binary = 1;
1572                 data->added = mf2.size;
1573                 data->deleted = mf1.size;
1574         } else {
1575                 /* Crazy xdl interfaces.. */
1576                 xpparam_t xpp;
1577                 xdemitconf_t xecfg;
1578                 xdemitcb_t ecb;
1579
1580                 memset(&xecfg, 0, sizeof(xecfg));
1581                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1582                 ecb.outf = xdiff_outf;
1583                 ecb.priv = diffstat;
1584                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1585         }
1586
1587  free_and_return:
1588         diff_free_filespec_data(one);
1589         diff_free_filespec_data(two);
1590 }
1591
1592 static void builtin_checkdiff(const char *name_a, const char *name_b,
1593                               const char *attr_path,
1594                               struct diff_filespec *one,
1595                               struct diff_filespec *two,
1596                               struct diff_options *o)
1597 {
1598         mmfile_t mf1, mf2;
1599         struct checkdiff_t data;
1600
1601         if (!two)
1602                 return;
1603
1604         memset(&data, 0, sizeof(data));
1605         data.xm.consume = checkdiff_consume;
1606         data.filename = name_b ? name_b : name_a;
1607         data.lineno = 0;
1608         data.o = o;
1609         data.ws_rule = whitespace_rule(attr_path);
1610
1611         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1612                 die("unable to read files to diff");
1613
1614         /*
1615          * All the other codepaths check both sides, but not checking
1616          * the "old" side here is deliberate.  We are checking the newly
1617          * introduced changes, and as long as the "new" side is text, we
1618          * can and should check what it introduces.
1619          */
1620         if (diff_filespec_is_binary(two))
1621                 goto free_and_return;
1622         else {
1623                 /* Crazy xdl interfaces.. */
1624                 xpparam_t xpp;
1625                 xdemitconf_t xecfg;
1626                 xdemitcb_t ecb;
1627
1628                 memset(&xecfg, 0, sizeof(xecfg));
1629                 xpp.flags = XDF_NEED_MINIMAL;
1630                 ecb.outf = xdiff_outf;
1631                 ecb.priv = &data;
1632                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1633
1634                 if (data.trailing_blanks_start) {
1635                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1636                                 data.filename, data.trailing_blanks_start);
1637                         data.status = 1; /* report errors */
1638                 }
1639         }
1640  free_and_return:
1641         diff_free_filespec_data(one);
1642         diff_free_filespec_data(two);
1643         if (data.status)
1644                 DIFF_OPT_SET(o, CHECK_FAILED);
1645 }
1646
1647 struct diff_filespec *alloc_filespec(const char *path)
1648 {
1649         int namelen = strlen(path);
1650         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1651
1652         memset(spec, 0, sizeof(*spec));
1653         spec->path = (char *)(spec + 1);
1654         memcpy(spec->path, path, namelen+1);
1655         spec->count = 1;
1656         return spec;
1657 }
1658
1659 void free_filespec(struct diff_filespec *spec)
1660 {
1661         if (!--spec->count) {
1662                 diff_free_filespec_data(spec);
1663                 free(spec);
1664         }
1665 }
1666
1667 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1668                    unsigned short mode)
1669 {
1670         if (mode) {
1671                 spec->mode = canon_mode(mode);
1672                 hashcpy(spec->sha1, sha1);
1673                 spec->sha1_valid = !is_null_sha1(sha1);
1674         }
1675 }
1676
1677 /*
1678  * Given a name and sha1 pair, if the index tells us the file in
1679  * the work tree has that object contents, return true, so that
1680  * prepare_temp_file() does not have to inflate and extract.
1681  */
1682 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1683 {
1684         struct cache_entry *ce;
1685         struct stat st;
1686         int pos, len;
1687
1688         /* We do not read the cache ourselves here, because the
1689          * benchmark with my previous version that always reads cache
1690          * shows that it makes things worse for diff-tree comparing
1691          * two linux-2.6 kernel trees in an already checked out work
1692          * tree.  This is because most diff-tree comparisons deal with
1693          * only a small number of files, while reading the cache is
1694          * expensive for a large project, and its cost outweighs the
1695          * savings we get by not inflating the object to a temporary
1696          * file.  Practically, this code only helps when we are used
1697          * by diff-cache --cached, which does read the cache before
1698          * calling us.
1699          */
1700         if (!active_cache)
1701                 return 0;
1702
1703         /* We want to avoid the working directory if our caller
1704          * doesn't need the data in a normal file, this system
1705          * is rather slow with its stat/open/mmap/close syscalls,
1706          * and the object is contained in a pack file.  The pack
1707          * is probably already open and will be faster to obtain
1708          * the data through than the working directory.  Loose
1709          * objects however would tend to be slower as they need
1710          * to be individually opened and inflated.
1711          */
1712         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1713                 return 0;
1714
1715         len = strlen(name);
1716         pos = cache_name_pos(name, len);
1717         if (pos < 0)
1718                 return 0;
1719         ce = active_cache[pos];
1720
1721         /*
1722          * This is not the sha1 we are looking for, or
1723          * unreusable because it is not a regular file.
1724          */
1725         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1726                 return 0;
1727
1728         /*
1729          * If ce matches the file in the work tree, we can reuse it.
1730          */
1731         if (ce_uptodate(ce) ||
1732             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1733                 return 1;
1734
1735         return 0;
1736 }
1737
1738 static int populate_from_stdin(struct diff_filespec *s)
1739 {
1740         struct strbuf buf;
1741         size_t size = 0;
1742
1743         strbuf_init(&buf, 0);
1744         if (strbuf_read(&buf, 0, 0) < 0)
1745                 return error("error while reading from stdin %s",
1746                                      strerror(errno));
1747
1748         s->should_munmap = 0;
1749         s->data = strbuf_detach(&buf, &size);
1750         s->size = size;
1751         s->should_free = 1;
1752         return 0;
1753 }
1754
1755 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1756 {
1757         int len;
1758         char *data = xmalloc(100);
1759         len = snprintf(data, 100,
1760                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1761         s->data = data;
1762         s->size = len;
1763         s->should_free = 1;
1764         if (size_only) {
1765                 s->data = NULL;
1766                 free(data);
1767         }
1768         return 0;
1769 }
1770
1771 /*
1772  * While doing rename detection and pickaxe operation, we may need to
1773  * grab the data for the blob (or file) for our own in-core comparison.
1774  * diff_filespec has data and size fields for this purpose.
1775  */
1776 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1777 {
1778         int err = 0;
1779         if (!DIFF_FILE_VALID(s))
1780                 die("internal error: asking to populate invalid file.");
1781         if (S_ISDIR(s->mode))
1782                 return -1;
1783
1784         if (s->data)
1785                 return 0;
1786
1787         if (size_only && 0 < s->size)
1788                 return 0;
1789
1790         if (S_ISGITLINK(s->mode))
1791                 return diff_populate_gitlink(s, size_only);
1792
1793         if (!s->sha1_valid ||
1794             reuse_worktree_file(s->path, s->sha1, 0)) {
1795                 struct strbuf buf;
1796                 struct stat st;
1797                 int fd;
1798
1799                 if (!strcmp(s->path, "-"))
1800                         return populate_from_stdin(s);
1801
1802                 if (lstat(s->path, &st) < 0) {
1803                         if (errno == ENOENT) {
1804                         err_empty:
1805                                 err = -1;
1806                         empty:
1807                                 s->data = (char *)"";
1808                                 s->size = 0;
1809                                 return err;
1810                         }
1811                 }
1812                 s->size = xsize_t(st.st_size);
1813                 if (!s->size)
1814                         goto empty;
1815                 if (size_only)
1816                         return 0;
1817                 if (S_ISLNK(st.st_mode)) {
1818                         int ret;
1819                         s->data = xmalloc(s->size);
1820                         s->should_free = 1;
1821                         ret = readlink(s->path, s->data, s->size);
1822                         if (ret < 0) {
1823                                 free(s->data);
1824                                 goto err_empty;
1825                         }
1826                         return 0;
1827                 }
1828                 fd = open(s->path, O_RDONLY);
1829                 if (fd < 0)
1830                         goto err_empty;
1831                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1832                 close(fd);
1833                 s->should_munmap = 1;
1834
1835                 /*
1836                  * Convert from working tree format to canonical git format
1837                  */
1838                 strbuf_init(&buf, 0);
1839                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1840                         size_t size = 0;
1841                         munmap(s->data, s->size);
1842                         s->should_munmap = 0;
1843                         s->data = strbuf_detach(&buf, &size);
1844                         s->size = size;
1845                         s->should_free = 1;
1846                 }
1847         }
1848         else {
1849                 enum object_type type;
1850                 if (size_only)
1851                         type = sha1_object_info(s->sha1, &s->size);
1852                 else {
1853                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1854                         s->should_free = 1;
1855                 }
1856         }
1857         return 0;
1858 }
1859
1860 void diff_free_filespec_blob(struct diff_filespec *s)
1861 {
1862         if (s->should_free)
1863                 free(s->data);
1864         else if (s->should_munmap)
1865                 munmap(s->data, s->size);
1866
1867         if (s->should_free || s->should_munmap) {
1868                 s->should_free = s->should_munmap = 0;
1869                 s->data = NULL;
1870         }
1871 }
1872
1873 void diff_free_filespec_data(struct diff_filespec *s)
1874 {
1875         diff_free_filespec_blob(s);
1876         free(s->cnt_data);
1877         s->cnt_data = NULL;
1878 }
1879
1880 static void prep_temp_blob(struct diff_tempfile *temp,
1881                            void *blob,
1882                            unsigned long size,
1883                            const unsigned char *sha1,
1884                            int mode)
1885 {
1886         int fd;
1887
1888         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1889         if (fd < 0)
1890                 die("unable to create temp-file: %s", strerror(errno));
1891         if (write_in_full(fd, blob, size) != size)
1892                 die("unable to write temp-file");
1893         close(fd);
1894         temp->name = temp->tmp_path;
1895         strcpy(temp->hex, sha1_to_hex(sha1));
1896         temp->hex[40] = 0;
1897         sprintf(temp->mode, "%06o", mode);
1898 }
1899
1900 static void prepare_temp_file(const char *name,
1901                               struct diff_tempfile *temp,
1902                               struct diff_filespec *one)
1903 {
1904         if (!DIFF_FILE_VALID(one)) {
1905         not_a_valid_file:
1906                 /* A '-' entry produces this for file-2, and
1907                  * a '+' entry produces this for file-1.
1908                  */
1909                 temp->name = "/dev/null";
1910                 strcpy(temp->hex, ".");
1911                 strcpy(temp->mode, ".");
1912                 return;
1913         }
1914
1915         if (!one->sha1_valid ||
1916             reuse_worktree_file(name, one->sha1, 1)) {
1917                 struct stat st;
1918                 if (lstat(name, &st) < 0) {
1919                         if (errno == ENOENT)
1920                                 goto not_a_valid_file;
1921                         die("stat(%s): %s", name, strerror(errno));
1922                 }
1923                 if (S_ISLNK(st.st_mode)) {
1924                         int ret;
1925                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1926                         size_t sz = xsize_t(st.st_size);
1927                         if (sizeof(buf) <= st.st_size)
1928                                 die("symlink too long: %s", name);
1929                         ret = readlink(name, buf, sz);
1930                         if (ret < 0)
1931                                 die("readlink(%s)", name);
1932                         prep_temp_blob(temp, buf, sz,
1933                                        (one->sha1_valid ?
1934                                         one->sha1 : null_sha1),
1935                                        (one->sha1_valid ?
1936                                         one->mode : S_IFLNK));
1937                 }
1938                 else {
1939                         /* we can borrow from the file in the work tree */
1940                         temp->name = name;
1941                         if (!one->sha1_valid)
1942                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1943                         else
1944                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1945                         /* Even though we may sometimes borrow the
1946                          * contents from the work tree, we always want
1947                          * one->mode.  mode is trustworthy even when
1948                          * !(one->sha1_valid), as long as
1949                          * DIFF_FILE_VALID(one).
1950                          */
1951                         sprintf(temp->mode, "%06o", one->mode);
1952                 }
1953                 return;
1954         }
1955         else {
1956                 if (diff_populate_filespec(one, 0))
1957                         die("cannot read data blob for %s", one->path);
1958                 prep_temp_blob(temp, one->data, one->size,
1959                                one->sha1, one->mode);
1960         }
1961 }
1962
1963 static void remove_tempfile(void)
1964 {
1965         int i;
1966
1967         for (i = 0; i < 2; i++)
1968                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1969                         unlink(diff_temp[i].name);
1970                         diff_temp[i].name = NULL;
1971                 }
1972 }
1973
1974 static void remove_tempfile_on_signal(int signo)
1975 {
1976         remove_tempfile();
1977         signal(SIGINT, SIG_DFL);
1978         raise(signo);
1979 }
1980
1981 /* An external diff command takes:
1982  *
1983  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1984  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1985  *
1986  */
1987 static void run_external_diff(const char *pgm,
1988                               const char *name,
1989                               const char *other,
1990                               struct diff_filespec *one,
1991                               struct diff_filespec *two,
1992                               const char *xfrm_msg,
1993                               int complete_rewrite)
1994 {
1995         const char *spawn_arg[10];
1996         struct diff_tempfile *temp = diff_temp;
1997         int retval;
1998         static int atexit_asked = 0;
1999         const char *othername;
2000         const char **arg = &spawn_arg[0];
2001
2002         othername = (other? other : name);
2003         if (one && two) {
2004                 prepare_temp_file(name, &temp[0], one);
2005                 prepare_temp_file(othername, &temp[1], two);
2006                 if (! atexit_asked &&
2007                     (temp[0].name == temp[0].tmp_path ||
2008                      temp[1].name == temp[1].tmp_path)) {
2009                         atexit_asked = 1;
2010                         atexit(remove_tempfile);
2011                 }
2012                 signal(SIGINT, remove_tempfile_on_signal);
2013         }
2014
2015         if (one && two) {
2016                 *arg++ = pgm;
2017                 *arg++ = name;
2018                 *arg++ = temp[0].name;
2019                 *arg++ = temp[0].hex;
2020                 *arg++ = temp[0].mode;
2021                 *arg++ = temp[1].name;
2022                 *arg++ = temp[1].hex;
2023                 *arg++ = temp[1].mode;
2024                 if (other) {
2025                         *arg++ = other;
2026                         *arg++ = xfrm_msg;
2027                 }
2028         } else {
2029                 *arg++ = pgm;
2030                 *arg++ = name;
2031         }
2032         *arg = NULL;
2033         fflush(NULL);
2034         retval = run_command_v_opt(spawn_arg, 0);
2035         remove_tempfile();
2036         if (retval) {
2037                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2038                 exit(1);
2039         }
2040 }
2041
2042 static const char *external_diff_attr(const char *name)
2043 {
2044         struct git_attr_check attr_diff_check;
2045
2046         if (!name)
2047                 return NULL;
2048
2049         setup_diff_attr_check(&attr_diff_check);
2050         if (!git_checkattr(name, 1, &attr_diff_check)) {
2051                 const char *value = attr_diff_check.value;
2052                 if (!ATTR_TRUE(value) &&
2053                     !ATTR_FALSE(value) &&
2054                     !ATTR_UNSET(value)) {
2055                         struct ll_diff_driver *drv;
2056
2057                         for (drv = user_diff; drv; drv = drv->next)
2058                                 if (!strcmp(drv->name, value))
2059                                         return drv->cmd;
2060                 }
2061         }
2062         return NULL;
2063 }
2064
2065 static void run_diff_cmd(const char *pgm,
2066                          const char *name,
2067                          const char *other,
2068                          const char *attr_path,
2069                          struct diff_filespec *one,
2070                          struct diff_filespec *two,
2071                          const char *xfrm_msg,
2072                          struct diff_options *o,
2073                          int complete_rewrite)
2074 {
2075         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2076                 pgm = NULL;
2077         else {
2078                 const char *cmd = external_diff_attr(attr_path);
2079                 if (cmd)
2080                         pgm = cmd;
2081         }
2082
2083         if (pgm) {
2084                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2085                                   complete_rewrite);
2086                 return;
2087         }
2088         if (one && two)
2089                 builtin_diff(name, other ? other : name,
2090                              one, two, xfrm_msg, o, complete_rewrite);
2091         else
2092                 fprintf(o->file, "* Unmerged path %s\n", name);
2093 }
2094
2095 static void diff_fill_sha1_info(struct diff_filespec *one)
2096 {
2097         if (DIFF_FILE_VALID(one)) {
2098                 if (!one->sha1_valid) {
2099                         struct stat st;
2100                         if (!strcmp(one->path, "-")) {
2101                                 hashcpy(one->sha1, null_sha1);
2102                                 return;
2103                         }
2104                         if (lstat(one->path, &st) < 0)
2105                                 die("stat %s", one->path);
2106                         if (index_path(one->sha1, one->path, &st, 0))
2107                                 die("cannot hash %s\n", one->path);
2108                 }
2109         }
2110         else
2111                 hashclr(one->sha1);
2112 }
2113
2114 static int similarity_index(struct diff_filepair *p)
2115 {
2116         return p->score * 100 / MAX_SCORE;
2117 }
2118
2119 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2120 {
2121         /* Strip the prefix but do not molest /dev/null and absolute paths */
2122         if (*namep && **namep != '/')
2123                 *namep += prefix_length;
2124         if (*otherp && **otherp != '/')
2125                 *otherp += prefix_length;
2126 }
2127
2128 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2129 {
2130         const char *pgm = external_diff();
2131         struct strbuf msg;
2132         char *xfrm_msg;
2133         struct diff_filespec *one = p->one;
2134         struct diff_filespec *two = p->two;
2135         const char *name;
2136         const char *other;
2137         const char *attr_path;
2138         int complete_rewrite = 0;
2139
2140         name  = p->one->path;
2141         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2142         attr_path = name;
2143         if (o->prefix_length)
2144                 strip_prefix(o->prefix_length, &name, &other);
2145
2146         if (DIFF_PAIR_UNMERGED(p)) {
2147                 run_diff_cmd(pgm, name, NULL, attr_path,
2148                              NULL, NULL, NULL, o, 0);
2149                 return;
2150         }
2151
2152         diff_fill_sha1_info(one);
2153         diff_fill_sha1_info(two);
2154
2155         strbuf_init(&msg, PATH_MAX * 2 + 300);
2156         switch (p->status) {
2157         case DIFF_STATUS_COPIED:
2158                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2159                 strbuf_addstr(&msg, "\ncopy from ");
2160                 quote_c_style(name, &msg, NULL, 0);
2161                 strbuf_addstr(&msg, "\ncopy to ");
2162                 quote_c_style(other, &msg, NULL, 0);
2163                 strbuf_addch(&msg, '\n');
2164                 break;
2165         case DIFF_STATUS_RENAMED:
2166                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2167                 strbuf_addstr(&msg, "\nrename from ");
2168                 quote_c_style(name, &msg, NULL, 0);
2169                 strbuf_addstr(&msg, "\nrename to ");
2170                 quote_c_style(other, &msg, NULL, 0);
2171                 strbuf_addch(&msg, '\n');
2172                 break;
2173         case DIFF_STATUS_MODIFIED:
2174                 if (p->score) {
2175                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2176                                         similarity_index(p));
2177                         complete_rewrite = 1;
2178                         break;
2179                 }
2180                 /* fallthru */
2181         default:
2182                 /* nothing */
2183                 ;
2184         }
2185
2186         if (hashcmp(one->sha1, two->sha1)) {
2187                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2188
2189                 if (DIFF_OPT_TST(o, BINARY)) {
2190                         mmfile_t mf;
2191                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2192                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2193                                 abbrev = 40;
2194                 }
2195                 strbuf_addf(&msg, "index %.*s..%.*s",
2196                                 abbrev, sha1_to_hex(one->sha1),
2197                                 abbrev, sha1_to_hex(two->sha1));
2198                 if (one->mode == two->mode)
2199                         strbuf_addf(&msg, " %06o", one->mode);
2200                 strbuf_addch(&msg, '\n');
2201         }
2202
2203         if (msg.len)
2204                 strbuf_setlen(&msg, msg.len - 1);
2205         xfrm_msg = msg.len ? msg.buf : NULL;
2206
2207         if (!pgm &&
2208             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2209             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2210                 /* a filepair that changes between file and symlink
2211                  * needs to be split into deletion and creation.
2212                  */
2213                 struct diff_filespec *null = alloc_filespec(two->path);
2214                 run_diff_cmd(NULL, name, other, attr_path,
2215                              one, null, xfrm_msg, o, 0);
2216                 free(null);
2217                 null = alloc_filespec(one->path);
2218                 run_diff_cmd(NULL, name, other, attr_path,
2219                              null, two, xfrm_msg, o, 0);
2220                 free(null);
2221         }
2222         else
2223                 run_diff_cmd(pgm, name, other, attr_path,
2224                              one, two, xfrm_msg, o, complete_rewrite);
2225
2226         strbuf_release(&msg);
2227 }
2228
2229 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2230                          struct diffstat_t *diffstat)
2231 {
2232         const char *name;
2233         const char *other;
2234         int complete_rewrite = 0;
2235
2236         if (DIFF_PAIR_UNMERGED(p)) {
2237                 /* unmerged */
2238                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2239                 return;
2240         }
2241
2242         name = p->one->path;
2243         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2244
2245         if (o->prefix_length)
2246                 strip_prefix(o->prefix_length, &name, &other);
2247
2248         diff_fill_sha1_info(p->one);
2249         diff_fill_sha1_info(p->two);
2250
2251         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2252                 complete_rewrite = 1;
2253         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2254 }
2255
2256 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2257 {
2258         const char *name;
2259         const char *other;
2260         const char *attr_path;
2261
2262         if (DIFF_PAIR_UNMERGED(p)) {
2263                 /* unmerged */
2264                 return;
2265         }
2266
2267         name = p->one->path;
2268         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2269         attr_path = other ? other : name;
2270
2271         if (o->prefix_length)
2272                 strip_prefix(o->prefix_length, &name, &other);
2273
2274         diff_fill_sha1_info(p->one);
2275         diff_fill_sha1_info(p->two);
2276
2277         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2278 }
2279
2280 void diff_setup(struct diff_options *options)
2281 {
2282         memset(options, 0, sizeof(*options));
2283
2284         options->file = stdout;
2285
2286         options->line_termination = '\n';
2287         options->break_opt = -1;
2288         options->rename_limit = -1;
2289         options->dirstat_percent = 3;
2290         options->context = 3;
2291
2292         options->change = diff_change;
2293         options->add_remove = diff_addremove;
2294         if (diff_use_color_default > 0)
2295                 DIFF_OPT_SET(options, COLOR_DIFF);
2296         else
2297                 DIFF_OPT_CLR(options, COLOR_DIFF);
2298         options->detect_rename = diff_detect_rename_default;
2299
2300         options->a_prefix = "a/";
2301         options->b_prefix = "b/";
2302 }
2303
2304 int diff_setup_done(struct diff_options *options)
2305 {
2306         int count = 0;
2307
2308         if (options->output_format & DIFF_FORMAT_NAME)
2309                 count++;
2310         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2311                 count++;
2312         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2313                 count++;
2314         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2315                 count++;
2316         if (count > 1)
2317                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2318
2319         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2320                 options->detect_rename = DIFF_DETECT_COPY;
2321
2322         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2323                 options->prefix = NULL;
2324         if (options->prefix)
2325                 options->prefix_length = strlen(options->prefix);
2326         else
2327                 options->prefix_length = 0;
2328
2329         if (options->output_format & (DIFF_FORMAT_NAME |
2330                                       DIFF_FORMAT_NAME_STATUS |
2331                                       DIFF_FORMAT_CHECKDIFF |
2332                                       DIFF_FORMAT_NO_OUTPUT))
2333                 options->output_format &= ~(DIFF_FORMAT_RAW |
2334                                             DIFF_FORMAT_NUMSTAT |
2335                                             DIFF_FORMAT_DIFFSTAT |
2336                                             DIFF_FORMAT_SHORTSTAT |
2337                                             DIFF_FORMAT_DIRSTAT |
2338                                             DIFF_FORMAT_SUMMARY |
2339                                             DIFF_FORMAT_PATCH);
2340
2341         /*
2342          * These cases always need recursive; we do not drop caller-supplied
2343          * recursive bits for other formats here.
2344          */
2345         if (options->output_format & (DIFF_FORMAT_PATCH |
2346                                       DIFF_FORMAT_NUMSTAT |
2347                                       DIFF_FORMAT_DIFFSTAT |
2348                                       DIFF_FORMAT_SHORTSTAT |
2349                                       DIFF_FORMAT_DIRSTAT |
2350                                       DIFF_FORMAT_SUMMARY |
2351                                       DIFF_FORMAT_CHECKDIFF))
2352                 DIFF_OPT_SET(options, RECURSIVE);
2353         /*
2354          * Also pickaxe would not work very well if you do not say recursive
2355          */
2356         if (options->pickaxe)
2357                 DIFF_OPT_SET(options, RECURSIVE);
2358
2359         if (options->detect_rename && options->rename_limit < 0)
2360                 options->rename_limit = diff_rename_limit_default;
2361         if (options->setup & DIFF_SETUP_USE_CACHE) {
2362                 if (!active_cache)
2363                         /* read-cache does not die even when it fails
2364                          * so it is safe for us to do this here.  Also
2365                          * it does not smudge active_cache or active_nr
2366                          * when it fails, so we do not have to worry about
2367                          * cleaning it up ourselves either.
2368                          */
2369                         read_cache();
2370         }
2371         if (options->abbrev <= 0 || 40 < options->abbrev)
2372                 options->abbrev = 40; /* full */
2373
2374         /*
2375          * It does not make sense to show the first hit we happened
2376          * to have found.  It does not make sense not to return with
2377          * exit code in such a case either.
2378          */
2379         if (DIFF_OPT_TST(options, QUIET)) {
2380                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2381                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2382         }
2383
2384         /*
2385          * If we postprocess in diffcore, we cannot simply return
2386          * upon the first hit.  We need to run diff as usual.
2387          */
2388         if (options->pickaxe || options->filter)
2389                 DIFF_OPT_CLR(options, QUIET);
2390
2391         return 0;
2392 }
2393
2394 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2395 {
2396         char c, *eq;
2397         int len;
2398
2399         if (*arg != '-')
2400                 return 0;
2401         c = *++arg;
2402         if (!c)
2403                 return 0;
2404         if (c == arg_short) {
2405                 c = *++arg;
2406                 if (!c)
2407                         return 1;
2408                 if (val && isdigit(c)) {
2409                         char *end;
2410                         int n = strtoul(arg, &end, 10);
2411                         if (*end)
2412                                 return 0;
2413                         *val = n;
2414                         return 1;
2415                 }
2416                 return 0;
2417         }
2418         if (c != '-')
2419                 return 0;
2420         arg++;
2421         eq = strchr(arg, '=');
2422         if (eq)
2423                 len = eq - arg;
2424         else
2425                 len = strlen(arg);
2426         if (!len || strncmp(arg, arg_long, len))
2427                 return 0;
2428         if (eq) {
2429                 int n;
2430                 char *end;
2431                 if (!isdigit(*++eq))
2432                         return 0;
2433                 n = strtoul(eq, &end, 10);
2434                 if (*end)
2435                         return 0;
2436                 *val = n;
2437         }
2438         return 1;
2439 }
2440
2441 static int diff_scoreopt_parse(const char *opt);
2442
2443 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2444 {
2445         const char *arg = av[0];
2446
2447         /* Output format options */
2448         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2449                 options->output_format |= DIFF_FORMAT_PATCH;
2450         else if (opt_arg(arg, 'U', "unified", &options->context))
2451                 options->output_format |= DIFF_FORMAT_PATCH;
2452         else if (!strcmp(arg, "--raw"))
2453                 options->output_format |= DIFF_FORMAT_RAW;
2454         else if (!strcmp(arg, "--patch-with-raw"))
2455                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2456         else if (!strcmp(arg, "--numstat"))
2457                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2458         else if (!strcmp(arg, "--shortstat"))
2459                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2460         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2461                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2462         else if (!strcmp(arg, "--cumulative"))
2463                 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2464         else if (!strcmp(arg, "--check"))
2465                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2466         else if (!strcmp(arg, "--summary"))
2467                 options->output_format |= DIFF_FORMAT_SUMMARY;
2468         else if (!strcmp(arg, "--patch-with-stat"))
2469                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2470         else if (!strcmp(arg, "--name-only"))
2471                 options->output_format |= DIFF_FORMAT_NAME;
2472         else if (!strcmp(arg, "--name-status"))
2473                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2474         else if (!strcmp(arg, "-s"))
2475                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2476         else if (!prefixcmp(arg, "--stat")) {
2477                 char *end;
2478                 int width = options->stat_width;
2479                 int name_width = options->stat_name_width;
2480                 arg += 6;
2481                 end = (char *)arg;
2482
2483                 switch (*arg) {
2484                 case '-':
2485                         if (!prefixcmp(arg, "-width="))
2486                                 width = strtoul(arg + 7, &end, 10);
2487                         else if (!prefixcmp(arg, "-name-width="))
2488                                 name_width = strtoul(arg + 12, &end, 10);
2489                         break;
2490                 case '=':
2491                         width = strtoul(arg+1, &end, 10);
2492                         if (*end == ',')
2493                                 name_width = strtoul(end+1, &end, 10);
2494                 }
2495
2496                 /* Important! This checks all the error cases! */
2497                 if (*end)
2498                         return 0;
2499                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2500                 options->stat_name_width = name_width;
2501                 options->stat_width = width;
2502         }
2503
2504         /* renames options */
2505         else if (!prefixcmp(arg, "-B")) {
2506                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2507                         return -1;
2508         }
2509         else if (!prefixcmp(arg, "-M")) {
2510                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2511                         return -1;
2512                 options->detect_rename = DIFF_DETECT_RENAME;
2513         }
2514         else if (!prefixcmp(arg, "-C")) {
2515                 if (options->detect_rename == DIFF_DETECT_COPY)
2516                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2517                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2518                         return -1;
2519                 options->detect_rename = DIFF_DETECT_COPY;
2520         }
2521         else if (!strcmp(arg, "--no-renames"))
2522                 options->detect_rename = 0;
2523         else if (!strcmp(arg, "--relative"))
2524                 DIFF_OPT_SET(options, RELATIVE_NAME);
2525         else if (!prefixcmp(arg, "--relative=")) {
2526                 DIFF_OPT_SET(options, RELATIVE_NAME);
2527                 options->prefix = arg + 11;
2528         }
2529
2530         /* xdiff options */
2531         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2532                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2533         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2534                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2535         else if (!strcmp(arg, "--ignore-space-at-eol"))
2536                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2537
2538         /* flags options */
2539         else if (!strcmp(arg, "--binary")) {
2540                 options->output_format |= DIFF_FORMAT_PATCH;
2541                 DIFF_OPT_SET(options, BINARY);
2542         }
2543         else if (!strcmp(arg, "--full-index"))
2544                 DIFF_OPT_SET(options, FULL_INDEX);
2545         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2546                 DIFF_OPT_SET(options, TEXT);
2547         else if (!strcmp(arg, "-R"))
2548                 DIFF_OPT_SET(options, REVERSE_DIFF);
2549         else if (!strcmp(arg, "--find-copies-harder"))
2550                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2551         else if (!strcmp(arg, "--follow"))
2552                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2553         else if (!strcmp(arg, "--color"))
2554                 DIFF_OPT_SET(options, COLOR_DIFF);
2555         else if (!strcmp(arg, "--no-color"))
2556                 DIFF_OPT_CLR(options, COLOR_DIFF);
2557         else if (!strcmp(arg, "--color-words"))
2558                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2559         else if (!strcmp(arg, "--exit-code"))
2560                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2561         else if (!strcmp(arg, "--quiet"))
2562                 DIFF_OPT_SET(options, QUIET);
2563         else if (!strcmp(arg, "--ext-diff"))
2564                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2565         else if (!strcmp(arg, "--no-ext-diff"))
2566                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2567         else if (!strcmp(arg, "--ignore-submodules"))
2568                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2569
2570         /* misc options */
2571         else if (!strcmp(arg, "-z"))
2572                 options->line_termination = 0;
2573         else if (!prefixcmp(arg, "-l"))
2574                 options->rename_limit = strtoul(arg+2, NULL, 10);
2575         else if (!prefixcmp(arg, "-S"))
2576                 options->pickaxe = arg + 2;
2577         else if (!strcmp(arg, "--pickaxe-all"))
2578                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2579         else if (!strcmp(arg, "--pickaxe-regex"))
2580                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2581         else if (!prefixcmp(arg, "-O"))
2582                 options->orderfile = arg + 2;
2583         else if (!prefixcmp(arg, "--diff-filter="))
2584                 options->filter = arg + 14;
2585         else if (!strcmp(arg, "--abbrev"))
2586                 options->abbrev = DEFAULT_ABBREV;
2587         else if (!prefixcmp(arg, "--abbrev=")) {
2588                 options->abbrev = strtoul(arg + 9, NULL, 10);
2589                 if (options->abbrev < MINIMUM_ABBREV)
2590                         options->abbrev = MINIMUM_ABBREV;
2591                 else if (40 < options->abbrev)
2592                         options->abbrev = 40;
2593         }
2594         else if (!prefixcmp(arg, "--src-prefix="))
2595                 options->a_prefix = arg + 13;
2596         else if (!prefixcmp(arg, "--dst-prefix="))
2597                 options->b_prefix = arg + 13;
2598         else if (!strcmp(arg, "--no-prefix"))
2599                 options->a_prefix = options->b_prefix = "";
2600         else if (!prefixcmp(arg, "--output=")) {
2601                 options->file = fopen(arg + strlen("--output="), "w");
2602                 options->close_file = 1;
2603         } else
2604                 return 0;
2605         return 1;
2606 }
2607
2608 static int parse_num(const char **cp_p)
2609 {
2610         unsigned long num, scale;
2611         int ch, dot;
2612         const char *cp = *cp_p;
2613
2614         num = 0;
2615         scale = 1;
2616         dot = 0;
2617         for(;;) {
2618                 ch = *cp;
2619                 if ( !dot && ch == '.' ) {
2620                         scale = 1;
2621                         dot = 1;
2622                 } else if ( ch == '%' ) {
2623                         scale = dot ? scale*100 : 100;
2624                         cp++;   /* % is always at the end */
2625                         break;
2626                 } else if ( ch >= '0' && ch <= '9' ) {
2627                         if ( scale < 100000 ) {
2628                                 scale *= 10;
2629                                 num = (num*10) + (ch-'0');
2630                         }
2631                 } else {
2632                         break;
2633                 }
2634                 cp++;
2635         }
2636         *cp_p = cp;
2637
2638         /* user says num divided by scale and we say internally that
2639          * is MAX_SCORE * num / scale.
2640          */
2641         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2642 }
2643
2644 static int diff_scoreopt_parse(const char *opt)
2645 {
2646         int opt1, opt2, cmd;
2647
2648         if (*opt++ != '-')
2649                 return -1;
2650         cmd = *opt++;
2651         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2652                 return -1; /* that is not a -M, -C nor -B option */
2653
2654         opt1 = parse_num(&opt);
2655         if (cmd != 'B')
2656                 opt2 = 0;
2657         else {
2658                 if (*opt == 0)
2659                         opt2 = 0;
2660                 else if (*opt != '/')
2661                         return -1; /* we expect -B80/99 or -B80 */
2662                 else {
2663                         opt++;
2664                         opt2 = parse_num(&opt);
2665                 }
2666         }
2667         if (*opt != 0)
2668                 return -1;
2669         return opt1 | (opt2 << 16);
2670 }
2671
2672 struct diff_queue_struct diff_queued_diff;
2673
2674 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2675 {
2676         if (queue->alloc <= queue->nr) {
2677                 queue->alloc = alloc_nr(queue->alloc);
2678                 queue->queue = xrealloc(queue->queue,
2679                                         sizeof(dp) * queue->alloc);
2680         }
2681         queue->queue[queue->nr++] = dp;
2682 }
2683
2684 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2685                                  struct diff_filespec *one,
2686                                  struct diff_filespec *two)
2687 {
2688         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2689         dp->one = one;
2690         dp->two = two;
2691         if (queue)
2692                 diff_q(queue, dp);
2693         return dp;
2694 }
2695
2696 void diff_free_filepair(struct diff_filepair *p)
2697 {
2698         free_filespec(p->one);
2699         free_filespec(p->two);
2700         free(p);
2701 }
2702
2703 /* This is different from find_unique_abbrev() in that
2704  * it stuffs the result with dots for alignment.
2705  */
2706 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2707 {
2708         int abblen;
2709         const char *abbrev;
2710         if (len == 40)
2711                 return sha1_to_hex(sha1);
2712
2713         abbrev = find_unique_abbrev(sha1, len);
2714         abblen = strlen(abbrev);
2715         if (abblen < 37) {
2716                 static char hex[41];
2717                 if (len < abblen && abblen <= len + 2)
2718                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2719                 else
2720                         sprintf(hex, "%s...", abbrev);
2721                 return hex;
2722         }
2723         return sha1_to_hex(sha1);
2724 }
2725
2726 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2727 {
2728         int line_termination = opt->line_termination;
2729         int inter_name_termination = line_termination ? '\t' : '\0';
2730
2731         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2732                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2733                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2734                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2735         }
2736         if (p->score) {
2737                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2738                         inter_name_termination);
2739         } else {
2740                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2741         }
2742
2743         if (p->status == DIFF_STATUS_COPIED ||
2744             p->status == DIFF_STATUS_RENAMED) {
2745                 const char *name_a, *name_b;
2746                 name_a = p->one->path;
2747                 name_b = p->two->path;
2748                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2749                 write_name_quoted(name_a, opt->file, inter_name_termination);
2750                 write_name_quoted(name_b, opt->file, line_termination);
2751         } else {
2752                 const char *name_a, *name_b;
2753                 name_a = p->one->mode ? p->one->path : p->two->path;
2754                 name_b = NULL;
2755                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2756                 write_name_quoted(name_a, opt->file, line_termination);
2757         }
2758 }
2759
2760 int diff_unmodified_pair(struct diff_filepair *p)
2761 {
2762         /* This function is written stricter than necessary to support
2763          * the currently implemented transformers, but the idea is to
2764          * let transformers to produce diff_filepairs any way they want,
2765          * and filter and clean them up here before producing the output.
2766          */
2767         struct diff_filespec *one = p->one, *two = p->two;
2768
2769         if (DIFF_PAIR_UNMERGED(p))
2770                 return 0; /* unmerged is interesting */
2771
2772         /* deletion, addition, mode or type change
2773          * and rename are all interesting.
2774          */
2775         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2776             DIFF_PAIR_MODE_CHANGED(p) ||
2777             strcmp(one->path, two->path))
2778                 return 0;
2779
2780         /* both are valid and point at the same path.  that is, we are
2781          * dealing with a change.
2782          */
2783         if (one->sha1_valid && two->sha1_valid &&
2784             !hashcmp(one->sha1, two->sha1))
2785                 return 1; /* no change */
2786         if (!one->sha1_valid && !two->sha1_valid)
2787                 return 1; /* both look at the same file on the filesystem. */
2788         return 0;
2789 }
2790
2791 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2792 {
2793         if (diff_unmodified_pair(p))
2794                 return;
2795
2796         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2797             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2798                 return; /* no tree diffs in patch format */
2799
2800         run_diff(p, o);
2801 }
2802
2803 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2804                             struct diffstat_t *diffstat)
2805 {
2806         if (diff_unmodified_pair(p))
2807                 return;
2808
2809         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2810             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2811                 return; /* no tree diffs in patch format */
2812
2813         run_diffstat(p, o, diffstat);
2814 }
2815
2816 static void diff_flush_checkdiff(struct diff_filepair *p,
2817                 struct diff_options *o)
2818 {
2819         if (diff_unmodified_pair(p))
2820                 return;
2821
2822         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2823             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2824                 return; /* no tree diffs in patch format */
2825
2826         run_checkdiff(p, o);
2827 }
2828
2829 int diff_queue_is_empty(void)
2830 {
2831         struct diff_queue_struct *q = &diff_queued_diff;
2832         int i;
2833         for (i = 0; i < q->nr; i++)
2834                 if (!diff_unmodified_pair(q->queue[i]))
2835                         return 0;
2836         return 1;
2837 }
2838
2839 #if DIFF_DEBUG
2840 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2841 {
2842         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2843                 x, one ? one : "",
2844                 s->path,
2845                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2846                 s->mode,
2847                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2848         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2849                 x, one ? one : "",
2850                 s->size, s->xfrm_flags);
2851 }
2852
2853 void diff_debug_filepair(const struct diff_filepair *p, int i)
2854 {
2855         diff_debug_filespec(p->one, i, "one");
2856         diff_debug_filespec(p->two, i, "two");
2857         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2858                 p->score, p->status ? p->status : '?',
2859                 p->one->rename_used, p->broken_pair);
2860 }
2861
2862 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2863 {
2864         int i;
2865         if (msg)
2866                 fprintf(stderr, "%s\n", msg);
2867         fprintf(stderr, "q->nr = %d\n", q->nr);
2868         for (i = 0; i < q->nr; i++) {
2869                 struct diff_filepair *p = q->queue[i];
2870                 diff_debug_filepair(p, i);
2871         }
2872 }
2873 #endif
2874
2875 static void diff_resolve_rename_copy(void)
2876 {
2877         int i;
2878         struct diff_filepair *p;
2879         struct diff_queue_struct *q = &diff_queued_diff;
2880
2881         diff_debug_queue("resolve-rename-copy", q);
2882
2883         for (i = 0; i < q->nr; i++) {
2884                 p = q->queue[i];
2885                 p->status = 0; /* undecided */
2886                 if (DIFF_PAIR_UNMERGED(p))
2887                         p->status = DIFF_STATUS_UNMERGED;
2888                 else if (!DIFF_FILE_VALID(p->one))
2889                         p->status = DIFF_STATUS_ADDED;
2890                 else if (!DIFF_FILE_VALID(p->two))
2891                         p->status = DIFF_STATUS_DELETED;
2892                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2893                         p->status = DIFF_STATUS_TYPE_CHANGED;
2894
2895                 /* from this point on, we are dealing with a pair
2896                  * whose both sides are valid and of the same type, i.e.
2897                  * either in-place edit or rename/copy edit.
2898                  */
2899                 else if (DIFF_PAIR_RENAME(p)) {
2900                         /*
2901                          * A rename might have re-connected a broken
2902                          * pair up, causing the pathnames to be the
2903                          * same again. If so, that's not a rename at
2904                          * all, just a modification..
2905                          *
2906                          * Otherwise, see if this source was used for
2907                          * multiple renames, in which case we decrement
2908                          * the count, and call it a copy.
2909                          */
2910                         if (!strcmp(p->one->path, p->two->path))
2911                                 p->status = DIFF_STATUS_MODIFIED;
2912                         else if (--p->one->rename_used > 0)
2913                                 p->status = DIFF_STATUS_COPIED;
2914                         else
2915                                 p->status = DIFF_STATUS_RENAMED;
2916                 }
2917                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2918                          p->one->mode != p->two->mode ||
2919                          is_null_sha1(p->one->sha1))
2920                         p->status = DIFF_STATUS_MODIFIED;
2921                 else {
2922                         /* This is a "no-change" entry and should not
2923                          * happen anymore, but prepare for broken callers.
2924                          */
2925                         error("feeding unmodified %s to diffcore",
2926                               p->one->path);
2927                         p->status = DIFF_STATUS_UNKNOWN;
2928                 }
2929         }
2930         diff_debug_queue("resolve-rename-copy done", q);
2931 }
2932
2933 static int check_pair_status(struct diff_filepair *p)
2934 {
2935         switch (p->status) {
2936         case DIFF_STATUS_UNKNOWN:
2937                 return 0;
2938         case 0:
2939                 die("internal error in diff-resolve-rename-copy");
2940         default:
2941                 return 1;
2942         }
2943 }
2944
2945 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2946 {
2947         int fmt = opt->output_format;
2948
2949         if (fmt & DIFF_FORMAT_CHECKDIFF)
2950                 diff_flush_checkdiff(p, opt);
2951         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2952                 diff_flush_raw(p, opt);
2953         else if (fmt & DIFF_FORMAT_NAME) {
2954                 const char *name_a, *name_b;
2955                 name_a = p->two->path;
2956                 name_b = NULL;
2957                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2958                 write_name_quoted(name_a, opt->file, opt->line_termination);
2959         }
2960 }
2961
2962 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2963 {
2964         if (fs->mode)
2965                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2966         else
2967                 fprintf(file, " %s ", newdelete);
2968         write_name_quoted(fs->path, file, '\n');
2969 }
2970
2971
2972 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2973 {
2974         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2975                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2976                         show_name ? ' ' : '\n');
2977                 if (show_name) {
2978                         write_name_quoted(p->two->path, file, '\n');
2979                 }
2980         }
2981 }
2982
2983 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2984 {
2985         char *names = pprint_rename(p->one->path, p->two->path);
2986
2987         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2988         free(names);
2989         show_mode_change(file, p, 0);
2990 }
2991
2992 static void diff_summary(FILE *file, struct diff_filepair *p)
2993 {
2994         switch(p->status) {
2995         case DIFF_STATUS_DELETED:
2996                 show_file_mode_name(file, "delete", p->one);
2997                 break;
2998         case DIFF_STATUS_ADDED:
2999                 show_file_mode_name(file, "create", p->two);
3000                 break;
3001         case DIFF_STATUS_COPIED:
3002                 show_rename_copy(file, "copy", p);
3003                 break;
3004         case DIFF_STATUS_RENAMED:
3005                 show_rename_copy(file, "rename", p);
3006                 break;
3007         default:
3008                 if (p->score) {
3009                         fputs(" rewrite ", file);
3010                         write_name_quoted(p->two->path, file, ' ');
3011                         fprintf(file, "(%d%%)\n", similarity_index(p));
3012                 }
3013                 show_mode_change(file, p, !p->score);
3014                 break;
3015         }
3016 }
3017
3018 struct patch_id_t {
3019         struct xdiff_emit_state xm;
3020         SHA_CTX *ctx;
3021         int patchlen;
3022 };
3023
3024 static int remove_space(char *line, int len)
3025 {
3026         int i;
3027         char *dst = line;
3028         unsigned char c;
3029
3030         for (i = 0; i < len; i++)
3031                 if (!isspace((c = line[i])))
3032                         *dst++ = c;
3033
3034         return dst - line;
3035 }
3036
3037 static void patch_id_consume(void *priv, char *line, unsigned long len)
3038 {
3039         struct patch_id_t *data = priv;
3040         int new_len;
3041
3042         /* Ignore line numbers when computing the SHA1 of the patch */
3043         if (!prefixcmp(line, "@@ -"))
3044                 return;
3045
3046         new_len = remove_space(line, len);
3047
3048         SHA1_Update(data->ctx, line, new_len);
3049         data->patchlen += new_len;
3050 }
3051
3052 /* returns 0 upon success, and writes result into sha1 */
3053 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3054 {
3055         struct diff_queue_struct *q = &diff_queued_diff;
3056         int i;
3057         SHA_CTX ctx;
3058         struct patch_id_t data;
3059         char buffer[PATH_MAX * 4 + 20];
3060
3061         SHA1_Init(&ctx);
3062         memset(&data, 0, sizeof(struct patch_id_t));
3063         data.ctx = &ctx;
3064         data.xm.consume = patch_id_consume;
3065
3066         for (i = 0; i < q->nr; i++) {
3067                 xpparam_t xpp;
3068                 xdemitconf_t xecfg;
3069                 xdemitcb_t ecb;
3070                 mmfile_t mf1, mf2;
3071                 struct diff_filepair *p = q->queue[i];
3072                 int len1, len2;
3073
3074                 memset(&xecfg, 0, sizeof(xecfg));
3075                 if (p->status == 0)
3076                         return error("internal diff status error");
3077                 if (p->status == DIFF_STATUS_UNKNOWN)
3078                         continue;
3079                 if (diff_unmodified_pair(p))
3080                         continue;
3081                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3082                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3083                         continue;
3084                 if (DIFF_PAIR_UNMERGED(p))
3085                         continue;
3086
3087                 diff_fill_sha1_info(p->one);
3088                 diff_fill_sha1_info(p->two);
3089                 if (fill_mmfile(&mf1, p->one) < 0 ||
3090                                 fill_mmfile(&mf2, p->two) < 0)
3091                         return error("unable to read files to diff");
3092
3093                 len1 = remove_space(p->one->path, strlen(p->one->path));
3094                 len2 = remove_space(p->two->path, strlen(p->two->path));
3095                 if (p->one->mode == 0)
3096                         len1 = snprintf(buffer, sizeof(buffer),
3097                                         "diff--gita/%.*sb/%.*s"
3098                                         "newfilemode%06o"
3099                                         "---/dev/null"
3100                                         "+++b/%.*s",
3101                                         len1, p->one->path,
3102                                         len2, p->two->path,
3103                                         p->two->mode,
3104                                         len2, p->two->path);
3105                 else if (p->two->mode == 0)
3106                         len1 = snprintf(buffer, sizeof(buffer),
3107                                         "diff--gita/%.*sb/%.*s"
3108                                         "deletedfilemode%06o"
3109                                         "---a/%.*s"
3110                                         "+++/dev/null",
3111                                         len1, p->one->path,
3112                                         len2, p->two->path,
3113                                         p->one->mode,
3114                                         len1, p->one->path);
3115                 else
3116                         len1 = snprintf(buffer, sizeof(buffer),
3117                                         "diff--gita/%.*sb/%.*s"
3118                                         "---a/%.*s"
3119                                         "+++b/%.*s",
3120                                         len1, p->one->path,
3121                                         len2, p->two->path,
3122                                         len1, p->one->path,
3123                                         len2, p->two->path);
3124                 SHA1_Update(&ctx, buffer, len1);
3125
3126                 xpp.flags = XDF_NEED_MINIMAL;
3127                 xecfg.ctxlen = 3;
3128                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3129                 ecb.outf = xdiff_outf;
3130                 ecb.priv = &data;
3131                 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3132         }
3133
3134         SHA1_Final(sha1, &ctx);
3135         return 0;
3136 }
3137
3138 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3139 {
3140         struct diff_queue_struct *q = &diff_queued_diff;
3141         int i;
3142         int result = diff_get_patch_id(options, sha1);
3143
3144         for (i = 0; i < q->nr; i++)
3145                 diff_free_filepair(q->queue[i]);
3146
3147         free(q->queue);
3148         q->queue = NULL;
3149         q->nr = q->alloc = 0;
3150
3151         return result;
3152 }
3153
3154 static int is_summary_empty(const struct diff_queue_struct *q)
3155 {
3156         int i;
3157
3158         for (i = 0; i < q->nr; i++) {
3159                 const struct diff_filepair *p = q->queue[i];
3160
3161                 switch (p->status) {
3162                 case DIFF_STATUS_DELETED:
3163                 case DIFF_STATUS_ADDED:
3164                 case DIFF_STATUS_COPIED:
3165                 case DIFF_STATUS_RENAMED:
3166                         return 0;
3167                 default:
3168                         if (p->score)
3169                                 return 0;
3170                         if (p->one->mode && p->two->mode &&
3171                             p->one->mode != p->two->mode)
3172                                 return 0;
3173                         break;
3174                 }
3175         }
3176         return 1;
3177 }
3178
3179 void diff_flush(struct diff_options *options)
3180 {
3181         struct diff_queue_struct *q = &diff_queued_diff;
3182         int i, output_format = options->output_format;
3183         int separator = 0;
3184
3185         /*
3186          * Order: raw, stat, summary, patch
3187          * or:    name/name-status/checkdiff (other bits clear)
3188          */
3189         if (!q->nr)
3190                 goto free_queue;
3191
3192         if (output_format & (DIFF_FORMAT_RAW |
3193                              DIFF_FORMAT_NAME |
3194                              DIFF_FORMAT_NAME_STATUS |
3195                              DIFF_FORMAT_CHECKDIFF)) {
3196                 for (i = 0; i < q->nr; i++) {
3197                         struct diff_filepair *p = q->queue[i];
3198                         if (check_pair_status(p))
3199                                 flush_one_pair(p, options);
3200                 }
3201                 separator++;
3202         }
3203
3204         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3205                 struct diffstat_t diffstat;
3206
3207                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3208                 diffstat.xm.consume = diffstat_consume;
3209                 for (i = 0; i < q->nr; i++) {
3210                         struct diff_filepair *p = q->queue[i];
3211                         if (check_pair_status(p))
3212                                 diff_flush_stat(p, options, &diffstat);
3213                 }
3214                 if (output_format & DIFF_FORMAT_NUMSTAT)
3215                         show_numstat(&diffstat, options);
3216                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3217                         show_stats(&diffstat, options);
3218                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3219                         show_shortstats(&diffstat, options);
3220                 free_diffstat_info(&diffstat);
3221                 separator++;
3222         }
3223         if (output_format & DIFF_FORMAT_DIRSTAT)
3224                 show_dirstat(options);
3225
3226         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3227                 for (i = 0; i < q->nr; i++)
3228                         diff_summary(options->file, q->queue[i]);
3229                 separator++;
3230         }
3231
3232         if (output_format & DIFF_FORMAT_PATCH) {
3233                 if (separator) {
3234                         putc(options->line_termination, options->file);
3235                         if (options->stat_sep) {
3236                                 /* attach patch instead of inline */
3237                                 fputs(options->stat_sep, options->file);
3238                         }
3239                 }
3240
3241                 for (i = 0; i < q->nr; i++) {
3242                         struct diff_filepair *p = q->queue[i];
3243                         if (check_pair_status(p))
3244                                 diff_flush_patch(p, options);
3245                 }
3246         }
3247
3248         if (output_format & DIFF_FORMAT_CALLBACK)
3249                 options->format_callback(q, options, options->format_callback_data);
3250
3251         for (i = 0; i < q->nr; i++)
3252                 diff_free_filepair(q->queue[i]);
3253 free_queue:
3254         free(q->queue);
3255         q->queue = NULL;
3256         q->nr = q->alloc = 0;
3257         if (options->close_file)
3258                 fclose(options->file);
3259 }
3260
3261 static void diffcore_apply_filter(const char *filter)
3262 {
3263         int i;
3264         struct diff_queue_struct *q = &diff_queued_diff;
3265         struct diff_queue_struct outq;
3266         outq.queue = NULL;
3267         outq.nr = outq.alloc = 0;
3268
3269         if (!filter)
3270                 return;
3271
3272         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3273                 int found;
3274                 for (i = found = 0; !found && i < q->nr; i++) {
3275                         struct diff_filepair *p = q->queue[i];
3276                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3277                              ((p->score &&
3278                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3279                               (!p->score &&
3280                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3281                             ((p->status != DIFF_STATUS_MODIFIED) &&
3282                              strchr(filter, p->status)))
3283                                 found++;
3284                 }
3285                 if (found)
3286                         return;
3287
3288                 /* otherwise we will clear the whole queue
3289                  * by copying the empty outq at the end of this
3290                  * function, but first clear the current entries
3291                  * in the queue.
3292                  */
3293                 for (i = 0; i < q->nr; i++)
3294                         diff_free_filepair(q->queue[i]);
3295         }
3296         else {
3297                 /* Only the matching ones */
3298                 for (i = 0; i < q->nr; i++) {
3299                         struct diff_filepair *p = q->queue[i];
3300
3301                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3302                              ((p->score &&
3303                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3304                               (!p->score &&
3305                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3306                             ((p->status != DIFF_STATUS_MODIFIED) &&
3307                              strchr(filter, p->status)))
3308                                 diff_q(&outq, p);
3309                         else
3310                                 diff_free_filepair(p);
3311                 }
3312         }
3313         free(q->queue);
3314         *q = outq;
3315 }
3316
3317 /* Check whether two filespecs with the same mode and size are identical */
3318 static int diff_filespec_is_identical(struct diff_filespec *one,
3319                                       struct diff_filespec *two)
3320 {
3321         if (S_ISGITLINK(one->mode))
3322                 return 0;
3323         if (diff_populate_filespec(one, 0))
3324                 return 0;
3325         if (diff_populate_filespec(two, 0))
3326                 return 0;
3327         return !memcmp(one->data, two->data, one->size);
3328 }
3329
3330 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3331 {
3332         int i;
3333         struct diff_queue_struct *q = &diff_queued_diff;
3334         struct diff_queue_struct outq;
3335         outq.queue = NULL;
3336         outq.nr = outq.alloc = 0;
3337
3338         for (i = 0; i < q->nr; i++) {
3339                 struct diff_filepair *p = q->queue[i];
3340
3341                 /*
3342                  * 1. Entries that come from stat info dirtyness
3343                  *    always have both sides (iow, not create/delete),
3344                  *    one side of the object name is unknown, with
3345                  *    the same mode and size.  Keep the ones that
3346                  *    do not match these criteria.  They have real
3347                  *    differences.
3348                  *
3349                  * 2. At this point, the file is known to be modified,
3350                  *    with the same mode and size, and the object
3351                  *    name of one side is unknown.  Need to inspect
3352                  *    the identical contents.
3353                  */
3354                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3355                     !DIFF_FILE_VALID(p->two) ||
3356                     (p->one->sha1_valid && p->two->sha1_valid) ||
3357                     (p->one->mode != p->two->mode) ||
3358                     diff_populate_filespec(p->one, 1) ||
3359                     diff_populate_filespec(p->two, 1) ||
3360                     (p->one->size != p->two->size) ||
3361                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3362                         diff_q(&outq, p);
3363                 else {
3364                         /*
3365                          * The caller can subtract 1 from skip_stat_unmatch
3366                          * to determine how many paths were dirty only
3367                          * due to stat info mismatch.
3368                          */
3369                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3370                                 diffopt->skip_stat_unmatch++;
3371                         diff_free_filepair(p);
3372                 }
3373         }
3374         free(q->queue);
3375         *q = outq;
3376 }
3377
3378 void diffcore_std(struct diff_options *options)
3379 {
3380         if (DIFF_OPT_TST(options, QUIET))
3381                 return;
3382
3383         if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3384                 diffcore_skip_stat_unmatch(options);
3385         if (options->break_opt != -1)
3386                 diffcore_break(options->break_opt);
3387         if (options->detect_rename)
3388                 diffcore_rename(options);
3389         if (options->break_opt != -1)
3390                 diffcore_merge_broken();
3391         if (options->pickaxe)
3392                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3393         if (options->orderfile)
3394                 diffcore_order(options->orderfile);
3395         diff_resolve_rename_copy();
3396         diffcore_apply_filter(options->filter);
3397
3398         if (diff_queued_diff.nr)
3399                 DIFF_OPT_SET(options, HAS_CHANGES);
3400         else
3401                 DIFF_OPT_CLR(options, HAS_CHANGES);
3402 }
3403
3404 int diff_result_code(struct diff_options *opt, int status)
3405 {
3406         int result = 0;
3407         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3408             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3409                 return status;
3410         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3411             DIFF_OPT_TST(opt, HAS_CHANGES))
3412                 result |= 01;
3413         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3414             DIFF_OPT_TST(opt, CHECK_FAILED))
3415                 result |= 02;
3416         return result;
3417 }
3418
3419 void diff_addremove(struct diff_options *options,
3420                     int addremove, unsigned mode,
3421                     const unsigned char *sha1,
3422                     const char *concatpath)
3423 {
3424         struct diff_filespec *one, *two;
3425
3426         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3427                 return;
3428
3429         /* This may look odd, but it is a preparation for
3430          * feeding "there are unchanged files which should
3431          * not produce diffs, but when you are doing copy
3432          * detection you would need them, so here they are"
3433          * entries to the diff-core.  They will be prefixed
3434          * with something like '=' or '*' (I haven't decided
3435          * which but should not make any difference).
3436          * Feeding the same new and old to diff_change()
3437          * also has the same effect.
3438          * Before the final output happens, they are pruned after
3439          * merged into rename/copy pairs as appropriate.
3440          */
3441         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3442                 addremove = (addremove == '+' ? '-' :
3443                              addremove == '-' ? '+' : addremove);
3444
3445         if (options->prefix &&
3446             strncmp(concatpath, options->prefix, options->prefix_length))
3447                 return;
3448
3449         one = alloc_filespec(concatpath);
3450         two = alloc_filespec(concatpath);
3451
3452         if (addremove != '+')
3453                 fill_filespec(one, sha1, mode);
3454         if (addremove != '-')
3455                 fill_filespec(two, sha1, mode);
3456
3457         diff_queue(&diff_queued_diff, one, two);
3458         DIFF_OPT_SET(options, HAS_CHANGES);
3459 }
3460
3461 void diff_change(struct diff_options *options,
3462                  unsigned old_mode, unsigned new_mode,
3463                  const unsigned char *old_sha1,
3464                  const unsigned char *new_sha1,
3465                  const char *concatpath)
3466 {
3467         struct diff_filespec *one, *two;
3468
3469         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3470                         && S_ISGITLINK(new_mode))
3471                 return;
3472
3473         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3474                 unsigned tmp;
3475                 const unsigned char *tmp_c;
3476                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3477                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3478         }
3479
3480         if (options->prefix &&
3481             strncmp(concatpath, options->prefix, options->prefix_length))
3482                 return;
3483
3484         one = alloc_filespec(concatpath);
3485         two = alloc_filespec(concatpath);
3486         fill_filespec(one, old_sha1, old_mode);
3487         fill_filespec(two, new_sha1, new_mode);
3488
3489         diff_queue(&diff_queued_diff, one, two);
3490         DIFF_OPT_SET(options, HAS_CHANGES);
3491 }
3492
3493 void diff_unmerge(struct diff_options *options,
3494                   const char *path,
3495                   unsigned mode, const unsigned char *sha1)
3496 {
3497         struct diff_filespec *one, *two;
3498
3499         if (options->prefix &&
3500             strncmp(path, options->prefix, options->prefix_length))
3501                 return;
3502
3503         one = alloc_filespec(path);
3504         two = alloc_filespec(path);
3505         fill_filespec(one, sha1, mode);
3506         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3507 }