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