git-rebase: document suppression of duplicate commits
[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
13 #ifdef NO_FAST_WORKING_DIRECTORY
14 #define FAST_WORKING_DIRECTORY 0
15 #else
16 #define FAST_WORKING_DIRECTORY 1
17 #endif
18
19 static int diff_detect_rename_default;
20 static int diff_rename_limit_default = 100;
21 static int diff_use_color_default;
22 int diff_auto_refresh_index = 1;
23
24 static char diff_colors[][COLOR_MAXLEN] = {
25         "\033[m",       /* reset */
26         "",             /* PLAIN (normal) */
27         "\033[1m",      /* METAINFO (bold) */
28         "\033[36m",     /* FRAGINFO (cyan) */
29         "\033[31m",     /* OLD (red) */
30         "\033[32m",     /* NEW (green) */
31         "\033[33m",     /* COMMIT (yellow) */
32         "\033[41m",     /* WHITESPACE (red background) */
33 };
34
35 static int parse_diff_color_slot(const char *var, int ofs)
36 {
37         if (!strcasecmp(var+ofs, "plain"))
38                 return DIFF_PLAIN;
39         if (!strcasecmp(var+ofs, "meta"))
40                 return DIFF_METAINFO;
41         if (!strcasecmp(var+ofs, "frag"))
42                 return DIFF_FRAGINFO;
43         if (!strcasecmp(var+ofs, "old"))
44                 return DIFF_FILE_OLD;
45         if (!strcasecmp(var+ofs, "new"))
46                 return DIFF_FILE_NEW;
47         if (!strcasecmp(var+ofs, "commit"))
48                 return DIFF_COMMIT;
49         if (!strcasecmp(var+ofs, "whitespace"))
50                 return DIFF_WHITESPACE;
51         die("bad config variable '%s'", var);
52 }
53
54 static struct ll_diff_driver {
55         const char *name;
56         struct ll_diff_driver *next;
57         char *cmd;
58 } *user_diff, **user_diff_tail;
59
60 static void read_config_if_needed(void)
61 {
62         if (!user_diff_tail) {
63                 user_diff_tail = &user_diff;
64                 git_config(git_diff_ui_config);
65         }
66 }
67
68 /*
69  * Currently there is only "diff.<drivername>.command" variable;
70  * because there are "diff.color.<slot>" variables, we are parsing
71  * this in a bit convoluted way to allow low level diff driver
72  * called "color".
73  */
74 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
75 {
76         const char *name;
77         int namelen;
78         struct ll_diff_driver *drv;
79
80         name = var + 5;
81         namelen = ep - name;
82         for (drv = user_diff; drv; drv = drv->next)
83                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
84                         break;
85         if (!drv) {
86                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
87                 drv->name = xmemdupz(name, namelen);
88                 if (!user_diff_tail)
89                         user_diff_tail = &user_diff;
90                 *user_diff_tail = drv;
91                 user_diff_tail = &(drv->next);
92         }
93
94         if (!value)
95                 return error("%s: lacks value", var);
96         drv->cmd = strdup(value);
97         return 0;
98 }
99
100 /*
101  * 'diff.<what>.funcname' attribute can be specified in the configuration
102  * to define a customized regexp to find the beginning of a function to
103  * be used for hunk header lines of "diff -p" style output.
104  */
105 static struct funcname_pattern {
106         char *name;
107         char *pattern;
108         struct funcname_pattern *next;
109 } *funcname_pattern_list;
110
111 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
112 {
113         const char *name;
114         int namelen;
115         struct funcname_pattern *pp;
116
117         name = var + 5; /* "diff." */
118         namelen = ep - name;
119
120         for (pp = funcname_pattern_list; pp; pp = pp->next)
121                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
122                         break;
123         if (!pp) {
124                 pp = xcalloc(1, sizeof(*pp));
125                 pp->name = xmemdupz(name, namelen);
126                 pp->next = funcname_pattern_list;
127                 funcname_pattern_list = pp;
128         }
129         if (pp->pattern)
130                 free(pp->pattern);
131         pp->pattern = xstrdup(value);
132         return 0;
133 }
134
135 /*
136  * These are to give UI layer defaults.
137  * The core-level commands such as git-diff-files should
138  * never be affected by the setting of diff.renames
139  * the user happens to have in the configuration file.
140  */
141 int git_diff_ui_config(const char *var, const char *value)
142 {
143         if (!strcmp(var, "diff.renamelimit")) {
144                 diff_rename_limit_default = git_config_int(var, value);
145                 return 0;
146         }
147         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
148                 diff_use_color_default = git_config_colorbool(var, value);
149                 return 0;
150         }
151         if (!strcmp(var, "diff.renames")) {
152                 if (!value)
153                         diff_detect_rename_default = DIFF_DETECT_RENAME;
154                 else if (!strcasecmp(value, "copies") ||
155                          !strcasecmp(value, "copy"))
156                         diff_detect_rename_default = DIFF_DETECT_COPY;
157                 else if (git_config_bool(var,value))
158                         diff_detect_rename_default = DIFF_DETECT_RENAME;
159                 return 0;
160         }
161         if (!strcmp(var, "diff.autorefreshindex")) {
162                 diff_auto_refresh_index = git_config_bool(var, value);
163                 return 0;
164         }
165         if (!prefixcmp(var, "diff.")) {
166                 const char *ep = strrchr(var, '.');
167
168                 if (ep != var + 4) {
169                         if (!strcmp(ep, ".command"))
170                                 return parse_lldiff_command(var, ep, value);
171                         if (!strcmp(ep, ".funcname"))
172                                 return parse_funcname_pattern(var, ep, value);
173                 }
174         }
175         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
176                 int slot = parse_diff_color_slot(var, 11);
177                 color_parse(value, var, diff_colors[slot]);
178                 return 0;
179         }
180
181         return git_default_config(var, value);
182 }
183
184 static char *quote_two(const char *one, const char *two)
185 {
186         int need_one = quote_c_style(one, NULL, NULL, 1);
187         int need_two = quote_c_style(two, NULL, NULL, 1);
188         struct strbuf res;
189
190         strbuf_init(&res, 0);
191         if (need_one + need_two) {
192                 strbuf_addch(&res, '"');
193                 quote_c_style(one, &res, NULL, 1);
194                 quote_c_style(two, &res, NULL, 1);
195                 strbuf_addch(&res, '"');
196         } else {
197                 strbuf_addstr(&res, one);
198                 strbuf_addstr(&res, two);
199         }
200         return strbuf_detach(&res, NULL);
201 }
202
203 static const char *external_diff(void)
204 {
205         static const char *external_diff_cmd = NULL;
206         static int done_preparing = 0;
207
208         if (done_preparing)
209                 return external_diff_cmd;
210         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
211         done_preparing = 1;
212         return external_diff_cmd;
213 }
214
215 static struct diff_tempfile {
216         const char *name; /* filename external diff should read from */
217         char hex[41];
218         char mode[10];
219         char tmp_path[PATH_MAX];
220 } diff_temp[2];
221
222 static int count_lines(const char *data, int size)
223 {
224         int count, ch, completely_empty = 1, nl_just_seen = 0;
225         count = 0;
226         while (0 < size--) {
227                 ch = *data++;
228                 if (ch == '\n') {
229                         count++;
230                         nl_just_seen = 1;
231                         completely_empty = 0;
232                 }
233                 else {
234                         nl_just_seen = 0;
235                         completely_empty = 0;
236                 }
237         }
238         if (completely_empty)
239                 return 0;
240         if (!nl_just_seen)
241                 count++; /* no trailing newline */
242         return count;
243 }
244
245 static void print_line_count(int count)
246 {
247         switch (count) {
248         case 0:
249                 printf("0,0");
250                 break;
251         case 1:
252                 printf("1");
253                 break;
254         default:
255                 printf("1,%d", count);
256                 break;
257         }
258 }
259
260 static void copy_file(int prefix, const char *data, int size,
261                 const char *set, const char *reset)
262 {
263         int ch, nl_just_seen = 1;
264         while (0 < size--) {
265                 ch = *data++;
266                 if (nl_just_seen) {
267                         fputs(set, stdout);
268                         putchar(prefix);
269                 }
270                 if (ch == '\n') {
271                         nl_just_seen = 1;
272                         fputs(reset, stdout);
273                 } else
274                         nl_just_seen = 0;
275                 putchar(ch);
276         }
277         if (!nl_just_seen)
278                 printf("%s\n\\ No newline at end of file\n", reset);
279 }
280
281 static void emit_rewrite_diff(const char *name_a,
282                               const char *name_b,
283                               struct diff_filespec *one,
284                               struct diff_filespec *two,
285                               int color_diff)
286 {
287         int lc_a, lc_b;
288         const char *name_a_tab, *name_b_tab;
289         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
290         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
291         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
292         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
293         const char *reset = diff_get_color(color_diff, DIFF_RESET);
294
295         name_a += (*name_a == '/');
296         name_b += (*name_b == '/');
297         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
298         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
299
300         diff_populate_filespec(one, 0);
301         diff_populate_filespec(two, 0);
302         lc_a = count_lines(one->data, one->size);
303         lc_b = count_lines(two->data, two->size);
304         printf("%s--- a/%s%s%s\n%s+++ b/%s%s%s\n%s@@ -",
305                metainfo, name_a, name_a_tab, reset,
306                metainfo, name_b, name_b_tab, reset, fraginfo);
307         print_line_count(lc_a);
308         printf(" +");
309         print_line_count(lc_b);
310         printf(" @@%s\n", reset);
311         if (lc_a)
312                 copy_file('-', one->data, one->size, old, reset);
313         if (lc_b)
314                 copy_file('+', two->data, two->size, new, reset);
315 }
316
317 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
318 {
319         if (!DIFF_FILE_VALID(one)) {
320                 mf->ptr = (char *)""; /* does not matter */
321                 mf->size = 0;
322                 return 0;
323         }
324         else if (diff_populate_filespec(one, 0))
325                 return -1;
326         mf->ptr = one->data;
327         mf->size = one->size;
328         return 0;
329 }
330
331 struct diff_words_buffer {
332         mmfile_t text;
333         long alloc;
334         long current; /* output pointer */
335         int suppressed_newline;
336 };
337
338 static void diff_words_append(char *line, unsigned long len,
339                 struct diff_words_buffer *buffer)
340 {
341         if (buffer->text.size + len > buffer->alloc) {
342                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
343                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
344         }
345         line++;
346         len--;
347         memcpy(buffer->text.ptr + buffer->text.size, line, len);
348         buffer->text.size += len;
349 }
350
351 struct diff_words_data {
352         struct xdiff_emit_state xm;
353         struct diff_words_buffer minus, plus;
354 };
355
356 static void print_word(struct diff_words_buffer *buffer, int len, int color,
357                 int suppress_newline)
358 {
359         const char *ptr;
360         int eol = 0;
361
362         if (len == 0)
363                 return;
364
365         ptr  = buffer->text.ptr + buffer->current;
366         buffer->current += len;
367
368         if (ptr[len - 1] == '\n') {
369                 eol = 1;
370                 len--;
371         }
372
373         fputs(diff_get_color(1, color), stdout);
374         fwrite(ptr, len, 1, stdout);
375         fputs(diff_get_color(1, DIFF_RESET), stdout);
376
377         if (eol) {
378                 if (suppress_newline)
379                         buffer->suppressed_newline = 1;
380                 else
381                         putchar('\n');
382         }
383 }
384
385 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
386 {
387         struct diff_words_data *diff_words = priv;
388
389         if (diff_words->minus.suppressed_newline) {
390                 if (line[0] != '+')
391                         putchar('\n');
392                 diff_words->minus.suppressed_newline = 0;
393         }
394
395         len--;
396         switch (line[0]) {
397                 case '-':
398                         print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
399                         break;
400                 case '+':
401                         print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
402                         break;
403                 case ' ':
404                         print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
405                         diff_words->minus.current += len;
406                         break;
407         }
408 }
409
410 /* this executes the word diff on the accumulated buffers */
411 static void diff_words_show(struct diff_words_data *diff_words)
412 {
413         xpparam_t xpp;
414         xdemitconf_t xecfg;
415         xdemitcb_t ecb;
416         mmfile_t minus, plus;
417         int i;
418
419         memset(&xecfg, 0, sizeof(xecfg));
420         minus.size = diff_words->minus.text.size;
421         minus.ptr = xmalloc(minus.size);
422         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
423         for (i = 0; i < minus.size; i++)
424                 if (isspace(minus.ptr[i]))
425                         minus.ptr[i] = '\n';
426         diff_words->minus.current = 0;
427
428         plus.size = diff_words->plus.text.size;
429         plus.ptr = xmalloc(plus.size);
430         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
431         for (i = 0; i < plus.size; i++)
432                 if (isspace(plus.ptr[i]))
433                         plus.ptr[i] = '\n';
434         diff_words->plus.current = 0;
435
436         xpp.flags = XDF_NEED_MINIMAL;
437         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
438         ecb.outf = xdiff_outf;
439         ecb.priv = diff_words;
440         diff_words->xm.consume = fn_out_diff_words_aux;
441         xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
442
443         free(minus.ptr);
444         free(plus.ptr);
445         diff_words->minus.text.size = diff_words->plus.text.size = 0;
446
447         if (diff_words->minus.suppressed_newline) {
448                 putchar('\n');
449                 diff_words->minus.suppressed_newline = 0;
450         }
451 }
452
453 struct emit_callback {
454         struct xdiff_emit_state xm;
455         int nparents, color_diff;
456         const char **label_path;
457         struct diff_words_data *diff_words;
458         int *found_changesp;
459 };
460
461 static void free_diff_words_data(struct emit_callback *ecbdata)
462 {
463         if (ecbdata->diff_words) {
464                 /* flush buffers */
465                 if (ecbdata->diff_words->minus.text.size ||
466                                 ecbdata->diff_words->plus.text.size)
467                         diff_words_show(ecbdata->diff_words);
468
469                 if (ecbdata->diff_words->minus.text.ptr)
470                         free (ecbdata->diff_words->minus.text.ptr);
471                 if (ecbdata->diff_words->plus.text.ptr)
472                         free (ecbdata->diff_words->plus.text.ptr);
473                 free(ecbdata->diff_words);
474                 ecbdata->diff_words = NULL;
475         }
476 }
477
478 const char *diff_get_color(int diff_use_color, enum color_diff ix)
479 {
480         if (diff_use_color)
481                 return diff_colors[ix];
482         return "";
483 }
484
485 static void emit_line(const char *set, const char *reset, const char *line, int len)
486 {
487         if (len > 0 && line[len-1] == '\n')
488                 len--;
489         fputs(set, stdout);
490         fwrite(line, len, 1, stdout);
491         puts(reset);
492 }
493
494 static void emit_line_with_ws(int nparents,
495                 const char *set, const char *reset, const char *ws,
496                 const char *line, int len)
497 {
498         int col0 = nparents;
499         int last_tab_in_indent = -1;
500         int last_space_in_indent = -1;
501         int i;
502         int tail = len;
503         int need_highlight_leading_space = 0;
504         /* The line is a newly added line.  Does it have funny leading
505          * whitespaces?  In indent, SP should never precede a TAB.
506          */
507         for (i = col0; i < len; i++) {
508                 if (line[i] == '\t') {
509                         last_tab_in_indent = i;
510                         if (0 <= last_space_in_indent)
511                                 need_highlight_leading_space = 1;
512                 }
513                 else if (line[i] == ' ')
514                         last_space_in_indent = i;
515                 else
516                         break;
517         }
518         fputs(set, stdout);
519         fwrite(line, col0, 1, stdout);
520         fputs(reset, stdout);
521         if (((i == len) || line[i] == '\n') && i != col0) {
522                 /* The whole line was indent */
523                 emit_line(ws, reset, line + col0, len - col0);
524                 return;
525         }
526         i = col0;
527         if (need_highlight_leading_space) {
528                 while (i < last_tab_in_indent) {
529                         if (line[i] == ' ') {
530                                 fputs(ws, stdout);
531                                 putchar(' ');
532                                 fputs(reset, stdout);
533                         }
534                         else
535                                 putchar(line[i]);
536                         i++;
537                 }
538         }
539         tail = len - 1;
540         if (line[tail] == '\n' && i < tail)
541                 tail--;
542         while (i < tail) {
543                 if (!isspace(line[tail]))
544                         break;
545                 tail--;
546         }
547         if ((i < tail && line[tail + 1] != '\n')) {
548                 /* This has whitespace between tail+1..len */
549                 fputs(set, stdout);
550                 fwrite(line + i, tail - i + 1, 1, stdout);
551                 fputs(reset, stdout);
552                 emit_line(ws, reset, line + tail + 1, len - tail - 1);
553         }
554         else
555                 emit_line(set, reset, line + i, len - i);
556 }
557
558 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
559 {
560         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
561         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
562
563         if (!*ws)
564                 emit_line(set, reset, line, len);
565         else
566                 emit_line_with_ws(ecbdata->nparents, set, reset, ws,
567                                 line, len);
568 }
569
570 static void fn_out_consume(void *priv, char *line, unsigned long len)
571 {
572         int i;
573         int color;
574         struct emit_callback *ecbdata = priv;
575         const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
576         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
577
578         *(ecbdata->found_changesp) = 1;
579
580         if (ecbdata->label_path[0]) {
581                 const char *name_a_tab, *name_b_tab;
582
583                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
584                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
585
586                 printf("%s--- %s%s%s\n",
587                        set, ecbdata->label_path[0], reset, name_a_tab);
588                 printf("%s+++ %s%s%s\n",
589                        set, ecbdata->label_path[1], reset, name_b_tab);
590                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
591         }
592
593         /* This is not really necessary for now because
594          * this codepath only deals with two-way diffs.
595          */
596         for (i = 0; i < len && line[i] == '@'; i++)
597                 ;
598         if (2 <= i && i < len && line[i] == ' ') {
599                 ecbdata->nparents = i - 1;
600                 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
601                           reset, line, len);
602                 return;
603         }
604
605         if (len < ecbdata->nparents) {
606                 set = reset;
607                 emit_line(reset, reset, line, len);
608                 return;
609         }
610
611         color = DIFF_PLAIN;
612         if (ecbdata->diff_words && ecbdata->nparents != 1)
613                 /* fall back to normal diff */
614                 free_diff_words_data(ecbdata);
615         if (ecbdata->diff_words) {
616                 if (line[0] == '-') {
617                         diff_words_append(line, len,
618                                           &ecbdata->diff_words->minus);
619                         return;
620                 } else if (line[0] == '+') {
621                         diff_words_append(line, len,
622                                           &ecbdata->diff_words->plus);
623                         return;
624                 }
625                 if (ecbdata->diff_words->minus.text.size ||
626                     ecbdata->diff_words->plus.text.size)
627                         diff_words_show(ecbdata->diff_words);
628                 line++;
629                 len--;
630                 emit_line(set, reset, line, len);
631                 return;
632         }
633         for (i = 0; i < ecbdata->nparents && len; i++) {
634                 if (line[i] == '-')
635                         color = DIFF_FILE_OLD;
636                 else if (line[i] == '+')
637                         color = DIFF_FILE_NEW;
638         }
639
640         if (color != DIFF_FILE_NEW) {
641                 emit_line(diff_get_color(ecbdata->color_diff, color),
642                           reset, line, len);
643                 return;
644         }
645         emit_add_line(reset, ecbdata, line, len);
646 }
647
648 static char *pprint_rename(const char *a, const char *b)
649 {
650         const char *old = a;
651         const char *new = b;
652         struct strbuf name;
653         int pfx_length, sfx_length;
654         int len_a = strlen(a);
655         int len_b = strlen(b);
656         int a_midlen, b_midlen;
657         int qlen_a = quote_c_style(a, NULL, NULL, 0);
658         int qlen_b = quote_c_style(b, NULL, NULL, 0);
659
660         strbuf_init(&name, 0);
661         if (qlen_a || qlen_b) {
662                 quote_c_style(a, &name, NULL, 0);
663                 strbuf_addstr(&name, " => ");
664                 quote_c_style(b, &name, NULL, 0);
665                 return strbuf_detach(&name, NULL);
666         }
667
668         /* Find common prefix */
669         pfx_length = 0;
670         while (*old && *new && *old == *new) {
671                 if (*old == '/')
672                         pfx_length = old - a + 1;
673                 old++;
674                 new++;
675         }
676
677         /* Find common suffix */
678         old = a + len_a;
679         new = b + len_b;
680         sfx_length = 0;
681         while (a <= old && b <= new && *old == *new) {
682                 if (*old == '/')
683                         sfx_length = len_a - (old - a);
684                 old--;
685                 new--;
686         }
687
688         /*
689          * pfx{mid-a => mid-b}sfx
690          * {pfx-a => pfx-b}sfx
691          * pfx{sfx-a => sfx-b}
692          * name-a => name-b
693          */
694         a_midlen = len_a - pfx_length - sfx_length;
695         b_midlen = len_b - pfx_length - sfx_length;
696         if (a_midlen < 0)
697                 a_midlen = 0;
698         if (b_midlen < 0)
699                 b_midlen = 0;
700
701         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
702         if (pfx_length + sfx_length) {
703                 strbuf_add(&name, a, pfx_length);
704                 strbuf_addch(&name, '{');
705         }
706         strbuf_add(&name, a + pfx_length, a_midlen);
707         strbuf_addstr(&name, " => ");
708         strbuf_add(&name, b + pfx_length, b_midlen);
709         if (pfx_length + sfx_length) {
710                 strbuf_addch(&name, '}');
711                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
712         }
713         return strbuf_detach(&name, NULL);
714 }
715
716 struct diffstat_t {
717         struct xdiff_emit_state xm;
718
719         int nr;
720         int alloc;
721         struct diffstat_file {
722                 char *name;
723                 unsigned is_unmerged:1;
724                 unsigned is_binary:1;
725                 unsigned is_renamed:1;
726                 unsigned int added, deleted;
727         } **files;
728 };
729
730 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
731                                           const char *name_a,
732                                           const char *name_b)
733 {
734         struct diffstat_file *x;
735         x = xcalloc(sizeof (*x), 1);
736         if (diffstat->nr == diffstat->alloc) {
737                 diffstat->alloc = alloc_nr(diffstat->alloc);
738                 diffstat->files = xrealloc(diffstat->files,
739                                 diffstat->alloc * sizeof(x));
740         }
741         diffstat->files[diffstat->nr++] = x;
742         if (name_b) {
743                 x->name = pprint_rename(name_a, name_b);
744                 x->is_renamed = 1;
745         }
746         else
747                 x->name = xstrdup(name_a);
748         return x;
749 }
750
751 static void diffstat_consume(void *priv, char *line, unsigned long len)
752 {
753         struct diffstat_t *diffstat = priv;
754         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
755
756         if (line[0] == '+')
757                 x->added++;
758         else if (line[0] == '-')
759                 x->deleted++;
760 }
761
762 const char mime_boundary_leader[] = "------------";
763
764 static int scale_linear(int it, int width, int max_change)
765 {
766         /*
767          * make sure that at least one '-' is printed if there were deletions,
768          * and likewise for '+'.
769          */
770         if (max_change < 2)
771                 return it;
772         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
773 }
774
775 static void show_name(const char *prefix, const char *name, int len,
776                       const char *reset, const char *set)
777 {
778         printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
779 }
780
781 static void show_graph(char ch, int cnt, const char *set, const char *reset)
782 {
783         if (cnt <= 0)
784                 return;
785         printf("%s", set);
786         while (cnt--)
787                 putchar(ch);
788         printf("%s", reset);
789 }
790
791 static void show_stats(struct diffstat_t* data, struct diff_options *options)
792 {
793         int i, len, add, del, total, adds = 0, dels = 0;
794         int max_change = 0, max_len = 0;
795         int total_files = data->nr;
796         int width, name_width;
797         const char *reset, *set, *add_c, *del_c;
798
799         if (data->nr == 0)
800                 return;
801
802         width = options->stat_width ? options->stat_width : 80;
803         name_width = options->stat_name_width ? options->stat_name_width : 50;
804
805         /* Sanity: give at least 5 columns to the graph,
806          * but leave at least 10 columns for the name.
807          */
808         if (width < name_width + 15) {
809                 if (name_width <= 25)
810                         width = name_width + 15;
811                 else
812                         name_width = width - 15;
813         }
814
815         /* Find the longest filename and max number of changes */
816         reset = diff_get_color(options->color_diff, DIFF_RESET);
817         set = diff_get_color(options->color_diff, DIFF_PLAIN);
818         add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
819         del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
820
821         for (i = 0; i < data->nr; i++) {
822                 struct diffstat_file *file = data->files[i];
823                 int change = file->added + file->deleted;
824
825                 if (!file->is_renamed) {  /* renames are already quoted by pprint_rename */
826                         struct strbuf buf;
827                         strbuf_init(&buf, 0);
828                         if (quote_c_style(file->name, &buf, NULL, 0)) {
829                                 free(file->name);
830                                 file->name = strbuf_detach(&buf, NULL);
831                         } else {
832                                 strbuf_release(&buf);
833                         }
834                 }
835
836                 len = strlen(file->name);
837                 if (max_len < len)
838                         max_len = len;
839
840                 if (file->is_binary || file->is_unmerged)
841                         continue;
842                 if (max_change < change)
843                         max_change = change;
844         }
845
846         /* Compute the width of the graph part;
847          * 10 is for one blank at the beginning of the line plus
848          * " | count " between the name and the graph.
849          *
850          * From here on, name_width is the width of the name area,
851          * and width is the width of the graph area.
852          */
853         name_width = (name_width < max_len) ? name_width : max_len;
854         if (width < (name_width + 10) + max_change)
855                 width = width - (name_width + 10);
856         else
857                 width = max_change;
858
859         for (i = 0; i < data->nr; i++) {
860                 const char *prefix = "";
861                 char *name = data->files[i]->name;
862                 int added = data->files[i]->added;
863                 int deleted = data->files[i]->deleted;
864                 int name_len;
865
866                 /*
867                  * "scale" the filename
868                  */
869                 len = name_width;
870                 name_len = strlen(name);
871                 if (name_width < name_len) {
872                         char *slash;
873                         prefix = "...";
874                         len -= 3;
875                         name += name_len - len;
876                         slash = strchr(name, '/');
877                         if (slash)
878                                 name = slash;
879                 }
880
881                 if (data->files[i]->is_binary) {
882                         show_name(prefix, name, len, reset, set);
883                         printf("  Bin ");
884                         printf("%s%d%s", del_c, deleted, reset);
885                         printf(" -> ");
886                         printf("%s%d%s", add_c, added, reset);
887                         printf(" bytes");
888                         printf("\n");
889                         goto free_diffstat_file;
890                 }
891                 else if (data->files[i]->is_unmerged) {
892                         show_name(prefix, name, len, reset, set);
893                         printf("  Unmerged\n");
894                         goto free_diffstat_file;
895                 }
896                 else if (!data->files[i]->is_renamed &&
897                          (added + deleted == 0)) {
898                         total_files--;
899                         goto free_diffstat_file;
900                 }
901
902                 /*
903                  * scale the add/delete
904                  */
905                 add = added;
906                 del = deleted;
907                 total = add + del;
908                 adds += add;
909                 dels += del;
910
911                 if (width <= max_change) {
912                         add = scale_linear(add, width, max_change);
913                         del = scale_linear(del, width, max_change);
914                         total = add + del;
915                 }
916                 show_name(prefix, name, len, reset, set);
917                 printf("%5d ", added + deleted);
918                 show_graph('+', add, add_c, reset);
919                 show_graph('-', del, del_c, reset);
920                 putchar('\n');
921         free_diffstat_file:
922                 free(data->files[i]->name);
923                 free(data->files[i]);
924         }
925         free(data->files);
926         printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
927                set, total_files, adds, dels, reset);
928 }
929
930 static void show_shortstats(struct diffstat_t* data)
931 {
932         int i, adds = 0, dels = 0, total_files = data->nr;
933
934         if (data->nr == 0)
935                 return;
936
937         for (i = 0; i < data->nr; i++) {
938                 if (!data->files[i]->is_binary &&
939                     !data->files[i]->is_unmerged) {
940                         int added = data->files[i]->added;
941                         int deleted= data->files[i]->deleted;
942                         if (!data->files[i]->is_renamed &&
943                             (added + deleted == 0)) {
944                                 total_files--;
945                         } else {
946                                 adds += added;
947                                 dels += deleted;
948                         }
949                 }
950                 free(data->files[i]->name);
951                 free(data->files[i]);
952         }
953         free(data->files);
954
955         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
956                total_files, adds, dels);
957 }
958
959 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
960 {
961         int i;
962
963         for (i = 0; i < data->nr; i++) {
964                 struct diffstat_file *file = data->files[i];
965
966                 if (file->is_binary)
967                         printf("-\t-\t");
968                 else
969                         printf("%d\t%d\t", file->added, file->deleted);
970                 if (!file->is_renamed) {
971                         write_name_quoted(file->name, stdout, options->line_termination);
972                 } else {
973                         fputs(file->name, stdout);
974                         putchar(options->line_termination);
975                 }
976         }
977 }
978
979 struct checkdiff_t {
980         struct xdiff_emit_state xm;
981         const char *filename;
982         int lineno, color_diff;
983 };
984
985 static void checkdiff_consume(void *priv, char *line, unsigned long len)
986 {
987         struct checkdiff_t *data = priv;
988         const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
989         const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
990         const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
991
992         if (line[0] == '+') {
993                 int i, spaces = 0, space_before_tab = 0, white_space_at_end = 0;
994
995                 /* check space before tab */
996                 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
997                         if (line[i] == ' ')
998                                 spaces++;
999                 if (line[i - 1] == '\t' && spaces)
1000                         space_before_tab = 1;
1001
1002                 /* check white space at line end */
1003                 if (line[len - 1] == '\n')
1004                         len--;
1005                 if (isspace(line[len - 1]))
1006                         white_space_at_end = 1;
1007
1008                 if (space_before_tab || white_space_at_end) {
1009                         printf("%s:%d: %s", data->filename, data->lineno, ws);
1010                         if (space_before_tab) {
1011                                 printf("space before tab");
1012                                 if (white_space_at_end)
1013                                         putchar(',');
1014                         }
1015                         if (white_space_at_end)
1016                                 printf("white space at end");
1017                         printf(":%s ", reset);
1018                         emit_line_with_ws(1, set, reset, ws, line, len);
1019                 }
1020
1021                 data->lineno++;
1022         } else if (line[0] == ' ')
1023                 data->lineno++;
1024         else if (line[0] == '@') {
1025                 char *plus = strchr(line, '+');
1026                 if (plus)
1027                         data->lineno = strtol(plus, NULL, 10);
1028                 else
1029                         die("invalid diff");
1030         }
1031 }
1032
1033 static unsigned char *deflate_it(char *data,
1034                                  unsigned long size,
1035                                  unsigned long *result_size)
1036 {
1037         int bound;
1038         unsigned char *deflated;
1039         z_stream stream;
1040
1041         memset(&stream, 0, sizeof(stream));
1042         deflateInit(&stream, zlib_compression_level);
1043         bound = deflateBound(&stream, size);
1044         deflated = xmalloc(bound);
1045         stream.next_out = deflated;
1046         stream.avail_out = bound;
1047
1048         stream.next_in = (unsigned char *)data;
1049         stream.avail_in = size;
1050         while (deflate(&stream, Z_FINISH) == Z_OK)
1051                 ; /* nothing */
1052         deflateEnd(&stream);
1053         *result_size = stream.total_out;
1054         return deflated;
1055 }
1056
1057 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
1058 {
1059         void *cp;
1060         void *delta;
1061         void *deflated;
1062         void *data;
1063         unsigned long orig_size;
1064         unsigned long delta_size;
1065         unsigned long deflate_size;
1066         unsigned long data_size;
1067
1068         /* We could do deflated delta, or we could do just deflated two,
1069          * whichever is smaller.
1070          */
1071         delta = NULL;
1072         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1073         if (one->size && two->size) {
1074                 delta = diff_delta(one->ptr, one->size,
1075                                    two->ptr, two->size,
1076                                    &delta_size, deflate_size);
1077                 if (delta) {
1078                         void *to_free = delta;
1079                         orig_size = delta_size;
1080                         delta = deflate_it(delta, delta_size, &delta_size);
1081                         free(to_free);
1082                 }
1083         }
1084
1085         if (delta && delta_size < deflate_size) {
1086                 printf("delta %lu\n", orig_size);
1087                 free(deflated);
1088                 data = delta;
1089                 data_size = delta_size;
1090         }
1091         else {
1092                 printf("literal %lu\n", two->size);
1093                 free(delta);
1094                 data = deflated;
1095                 data_size = deflate_size;
1096         }
1097
1098         /* emit data encoded in base85 */
1099         cp = data;
1100         while (data_size) {
1101                 int bytes = (52 < data_size) ? 52 : data_size;
1102                 char line[70];
1103                 data_size -= bytes;
1104                 if (bytes <= 26)
1105                         line[0] = bytes + 'A' - 1;
1106                 else
1107                         line[0] = bytes - 26 + 'a' - 1;
1108                 encode_85(line + 1, cp, bytes);
1109                 cp = (char *) cp + bytes;
1110                 puts(line);
1111         }
1112         printf("\n");
1113         free(data);
1114 }
1115
1116 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1117 {
1118         printf("GIT binary patch\n");
1119         emit_binary_diff_body(one, two);
1120         emit_binary_diff_body(two, one);
1121 }
1122
1123 static void setup_diff_attr_check(struct git_attr_check *check)
1124 {
1125         static struct git_attr *attr_diff;
1126
1127         if (!attr_diff) {
1128                 attr_diff = git_attr("diff", 4);
1129         }
1130         check[0].attr = attr_diff;
1131 }
1132
1133 static void diff_filespec_check_attr(struct diff_filespec *one)
1134 {
1135         struct git_attr_check attr_diff_check;
1136         int check_from_data = 0;
1137
1138         if (one->checked_attr)
1139                 return;
1140
1141         setup_diff_attr_check(&attr_diff_check);
1142         one->is_binary = 0;
1143         one->funcname_pattern_ident = NULL;
1144
1145         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1146                 const char *value;
1147
1148                 /* binaryness */
1149                 value = attr_diff_check.value;
1150                 if (ATTR_TRUE(value))
1151                         ;
1152                 else if (ATTR_FALSE(value))
1153                         one->is_binary = 1;
1154                 else
1155                         check_from_data = 1;
1156
1157                 /* funcname pattern ident */
1158                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1159                         ;
1160                 else
1161                         one->funcname_pattern_ident = value;
1162         }
1163
1164         if (check_from_data) {
1165                 if (!one->data && DIFF_FILE_VALID(one))
1166                         diff_populate_filespec(one, 0);
1167
1168                 if (one->data)
1169                         one->is_binary = buffer_is_binary(one->data, one->size);
1170         }
1171 }
1172
1173 int diff_filespec_is_binary(struct diff_filespec *one)
1174 {
1175         diff_filespec_check_attr(one);
1176         return one->is_binary;
1177 }
1178
1179 static const char *funcname_pattern(const char *ident)
1180 {
1181         struct funcname_pattern *pp;
1182
1183         read_config_if_needed();
1184         for (pp = funcname_pattern_list; pp; pp = pp->next)
1185                 if (!strcmp(ident, pp->name))
1186                         return pp->pattern;
1187         return NULL;
1188 }
1189
1190 static struct builtin_funcname_pattern {
1191         const char *name;
1192         const char *pattern;
1193 } builtin_funcname_pattern[] = {
1194         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1195                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1196                         "^[     ]*\\(\\([       ]*"
1197                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1198                         "[      ]*([^;]*$\\)" },
1199         { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1200 };
1201
1202 static const char *diff_funcname_pattern(struct diff_filespec *one)
1203 {
1204         const char *ident, *pattern;
1205         int i;
1206
1207         diff_filespec_check_attr(one);
1208         ident = one->funcname_pattern_ident;
1209
1210         if (!ident)
1211                 /*
1212                  * If the config file has "funcname.default" defined, that
1213                  * regexp is used; otherwise NULL is returned and xemit uses
1214                  * the built-in default.
1215                  */
1216                 return funcname_pattern("default");
1217
1218         /* Look up custom "funcname.$ident" regexp from config. */
1219         pattern = funcname_pattern(ident);
1220         if (pattern)
1221                 return pattern;
1222
1223         /*
1224          * And define built-in fallback patterns here.  Note that
1225          * these can be overriden by the user's config settings.
1226          */
1227         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1228                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1229                         return builtin_funcname_pattern[i].pattern;
1230
1231         return NULL;
1232 }
1233
1234 static void builtin_diff(const char *name_a,
1235                          const char *name_b,
1236                          struct diff_filespec *one,
1237                          struct diff_filespec *two,
1238                          const char *xfrm_msg,
1239                          struct diff_options *o,
1240                          int complete_rewrite)
1241 {
1242         mmfile_t mf1, mf2;
1243         const char *lbl[2];
1244         char *a_one, *b_two;
1245         const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
1246         const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
1247
1248         a_one = quote_two("a/", name_a + (*name_a == '/'));
1249         b_two = quote_two("b/", name_b + (*name_b == '/'));
1250         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1251         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1252         printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1253         if (lbl[0][0] == '/') {
1254                 /* /dev/null */
1255                 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1256                 if (xfrm_msg && xfrm_msg[0])
1257                         printf("%s%s%s\n", set, xfrm_msg, reset);
1258         }
1259         else if (lbl[1][0] == '/') {
1260                 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1261                 if (xfrm_msg && xfrm_msg[0])
1262                         printf("%s%s%s\n", set, xfrm_msg, reset);
1263         }
1264         else {
1265                 if (one->mode != two->mode) {
1266                         printf("%sold mode %06o%s\n", set, one->mode, reset);
1267                         printf("%snew mode %06o%s\n", set, two->mode, reset);
1268                 }
1269                 if (xfrm_msg && xfrm_msg[0])
1270                         printf("%s%s%s\n", set, xfrm_msg, reset);
1271                 /*
1272                  * we do not run diff between different kind
1273                  * of objects.
1274                  */
1275                 if ((one->mode ^ two->mode) & S_IFMT)
1276                         goto free_ab_and_return;
1277                 if (complete_rewrite) {
1278                         emit_rewrite_diff(name_a, name_b, one, two,
1279                                         o->color_diff);
1280                         o->found_changes = 1;
1281                         goto free_ab_and_return;
1282                 }
1283         }
1284
1285         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1286                 die("unable to read files to diff");
1287
1288         if (!o->text &&
1289             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1290                 /* Quite common confusing case */
1291                 if (mf1.size == mf2.size &&
1292                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1293                         goto free_ab_and_return;
1294                 if (o->binary)
1295                         emit_binary_diff(&mf1, &mf2);
1296                 else
1297                         printf("Binary files %s and %s differ\n",
1298                                lbl[0], lbl[1]);
1299                 o->found_changes = 1;
1300         }
1301         else {
1302                 /* Crazy xdl interfaces.. */
1303                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1304                 xpparam_t xpp;
1305                 xdemitconf_t xecfg;
1306                 xdemitcb_t ecb;
1307                 struct emit_callback ecbdata;
1308                 const char *funcname_pattern;
1309
1310                 funcname_pattern = diff_funcname_pattern(one);
1311                 if (!funcname_pattern)
1312                         funcname_pattern = diff_funcname_pattern(two);
1313
1314                 memset(&xecfg, 0, sizeof(xecfg));
1315                 memset(&ecbdata, 0, sizeof(ecbdata));
1316                 ecbdata.label_path = lbl;
1317                 ecbdata.color_diff = o->color_diff;
1318                 ecbdata.found_changesp = &o->found_changes;
1319                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1320                 xecfg.ctxlen = o->context;
1321                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1322                 if (funcname_pattern)
1323                         xdiff_set_find_func(&xecfg, funcname_pattern);
1324                 if (!diffopts)
1325                         ;
1326                 else if (!prefixcmp(diffopts, "--unified="))
1327                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1328                 else if (!prefixcmp(diffopts, "-u"))
1329                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1330                 ecb.outf = xdiff_outf;
1331                 ecb.priv = &ecbdata;
1332                 ecbdata.xm.consume = fn_out_consume;
1333                 if (o->color_diff_words)
1334                         ecbdata.diff_words =
1335                                 xcalloc(1, sizeof(struct diff_words_data));
1336                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1337                 if (o->color_diff_words)
1338                         free_diff_words_data(&ecbdata);
1339         }
1340
1341  free_ab_and_return:
1342         diff_free_filespec_data(one);
1343         diff_free_filespec_data(two);
1344         free(a_one);
1345         free(b_two);
1346         return;
1347 }
1348
1349 static void builtin_diffstat(const char *name_a, const char *name_b,
1350                              struct diff_filespec *one,
1351                              struct diff_filespec *two,
1352                              struct diffstat_t *diffstat,
1353                              struct diff_options *o,
1354                              int complete_rewrite)
1355 {
1356         mmfile_t mf1, mf2;
1357         struct diffstat_file *data;
1358
1359         data = diffstat_add(diffstat, name_a, name_b);
1360
1361         if (!one || !two) {
1362                 data->is_unmerged = 1;
1363                 return;
1364         }
1365         if (complete_rewrite) {
1366                 diff_populate_filespec(one, 0);
1367                 diff_populate_filespec(two, 0);
1368                 data->deleted = count_lines(one->data, one->size);
1369                 data->added = count_lines(two->data, two->size);
1370                 goto free_and_return;
1371         }
1372         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1373                 die("unable to read files to diff");
1374
1375         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1376                 data->is_binary = 1;
1377                 data->added = mf2.size;
1378                 data->deleted = mf1.size;
1379         } else {
1380                 /* Crazy xdl interfaces.. */
1381                 xpparam_t xpp;
1382                 xdemitconf_t xecfg;
1383                 xdemitcb_t ecb;
1384
1385                 memset(&xecfg, 0, sizeof(xecfg));
1386                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1387                 ecb.outf = xdiff_outf;
1388                 ecb.priv = diffstat;
1389                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1390         }
1391
1392  free_and_return:
1393         diff_free_filespec_data(one);
1394         diff_free_filespec_data(two);
1395 }
1396
1397 static void builtin_checkdiff(const char *name_a, const char *name_b,
1398                              struct diff_filespec *one,
1399                              struct diff_filespec *two, struct diff_options *o)
1400 {
1401         mmfile_t mf1, mf2;
1402         struct checkdiff_t data;
1403
1404         if (!two)
1405                 return;
1406
1407         memset(&data, 0, sizeof(data));
1408         data.xm.consume = checkdiff_consume;
1409         data.filename = name_b ? name_b : name_a;
1410         data.lineno = 0;
1411         data.color_diff = o->color_diff;
1412
1413         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1414                 die("unable to read files to diff");
1415
1416         if (diff_filespec_is_binary(two))
1417                 goto free_and_return;
1418         else {
1419                 /* Crazy xdl interfaces.. */
1420                 xpparam_t xpp;
1421                 xdemitconf_t xecfg;
1422                 xdemitcb_t ecb;
1423
1424                 memset(&xecfg, 0, sizeof(xecfg));
1425                 xpp.flags = XDF_NEED_MINIMAL;
1426                 ecb.outf = xdiff_outf;
1427                 ecb.priv = &data;
1428                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1429         }
1430  free_and_return:
1431         diff_free_filespec_data(one);
1432         diff_free_filespec_data(two);
1433 }
1434
1435 struct diff_filespec *alloc_filespec(const char *path)
1436 {
1437         int namelen = strlen(path);
1438         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1439
1440         memset(spec, 0, sizeof(*spec));
1441         spec->path = (char *)(spec + 1);
1442         memcpy(spec->path, path, namelen+1);
1443         return spec;
1444 }
1445
1446 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1447                    unsigned short mode)
1448 {
1449         if (mode) {
1450                 spec->mode = canon_mode(mode);
1451                 hashcpy(spec->sha1, sha1);
1452                 spec->sha1_valid = !is_null_sha1(sha1);
1453         }
1454 }
1455
1456 /*
1457  * Given a name and sha1 pair, if the index tells us the file in
1458  * the work tree has that object contents, return true, so that
1459  * prepare_temp_file() does not have to inflate and extract.
1460  */
1461 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1462 {
1463         struct cache_entry *ce;
1464         struct stat st;
1465         int pos, len;
1466
1467         /* We do not read the cache ourselves here, because the
1468          * benchmark with my previous version that always reads cache
1469          * shows that it makes things worse for diff-tree comparing
1470          * two linux-2.6 kernel trees in an already checked out work
1471          * tree.  This is because most diff-tree comparisons deal with
1472          * only a small number of files, while reading the cache is
1473          * expensive for a large project, and its cost outweighs the
1474          * savings we get by not inflating the object to a temporary
1475          * file.  Practically, this code only helps when we are used
1476          * by diff-cache --cached, which does read the cache before
1477          * calling us.
1478          */
1479         if (!active_cache)
1480                 return 0;
1481
1482         /* We want to avoid the working directory if our caller
1483          * doesn't need the data in a normal file, this system
1484          * is rather slow with its stat/open/mmap/close syscalls,
1485          * and the object is contained in a pack file.  The pack
1486          * is probably already open and will be faster to obtain
1487          * the data through than the working directory.  Loose
1488          * objects however would tend to be slower as they need
1489          * to be individually opened and inflated.
1490          */
1491         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1492                 return 0;
1493
1494         len = strlen(name);
1495         pos = cache_name_pos(name, len);
1496         if (pos < 0)
1497                 return 0;
1498         ce = active_cache[pos];
1499         if ((lstat(name, &st) < 0) ||
1500             !S_ISREG(st.st_mode) || /* careful! */
1501             ce_match_stat(ce, &st, 0) ||
1502             hashcmp(sha1, ce->sha1))
1503                 return 0;
1504         /* we return 1 only when we can stat, it is a regular file,
1505          * stat information matches, and sha1 recorded in the cache
1506          * matches.  I.e. we know the file in the work tree really is
1507          * the same as the <name, sha1> pair.
1508          */
1509         return 1;
1510 }
1511
1512 static int populate_from_stdin(struct diff_filespec *s)
1513 {
1514         struct strbuf buf;
1515
1516         strbuf_init(&buf, 0);
1517         if (strbuf_read(&buf, 0, 0) < 0)
1518                 return error("error while reading from stdin %s",
1519                                      strerror(errno));
1520
1521         s->should_munmap = 0;
1522         s->data = strbuf_detach(&buf, &s->size);
1523         s->should_free = 1;
1524         return 0;
1525 }
1526
1527 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1528 {
1529         int len;
1530         char *data = xmalloc(100);
1531         len = snprintf(data, 100,
1532                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1533         s->data = data;
1534         s->size = len;
1535         s->should_free = 1;
1536         if (size_only) {
1537                 s->data = NULL;
1538                 free(data);
1539         }
1540         return 0;
1541 }
1542
1543 /*
1544  * While doing rename detection and pickaxe operation, we may need to
1545  * grab the data for the blob (or file) for our own in-core comparison.
1546  * diff_filespec has data and size fields for this purpose.
1547  */
1548 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1549 {
1550         int err = 0;
1551         if (!DIFF_FILE_VALID(s))
1552                 die("internal error: asking to populate invalid file.");
1553         if (S_ISDIR(s->mode))
1554                 return -1;
1555
1556         if (s->data)
1557                 return 0;
1558
1559         if (size_only && 0 < s->size)
1560                 return 0;
1561
1562         if (S_ISGITLINK(s->mode))
1563                 return diff_populate_gitlink(s, size_only);
1564
1565         if (!s->sha1_valid ||
1566             reuse_worktree_file(s->path, s->sha1, 0)) {
1567                 struct strbuf buf;
1568                 struct stat st;
1569                 int fd;
1570
1571                 if (!strcmp(s->path, "-"))
1572                         return populate_from_stdin(s);
1573
1574                 if (lstat(s->path, &st) < 0) {
1575                         if (errno == ENOENT) {
1576                         err_empty:
1577                                 err = -1;
1578                         empty:
1579                                 s->data = (char *)"";
1580                                 s->size = 0;
1581                                 return err;
1582                         }
1583                 }
1584                 s->size = xsize_t(st.st_size);
1585                 if (!s->size)
1586                         goto empty;
1587                 if (size_only)
1588                         return 0;
1589                 if (S_ISLNK(st.st_mode)) {
1590                         int ret;
1591                         s->data = xmalloc(s->size);
1592                         s->should_free = 1;
1593                         ret = readlink(s->path, s->data, s->size);
1594                         if (ret < 0) {
1595                                 free(s->data);
1596                                 goto err_empty;
1597                         }
1598                         return 0;
1599                 }
1600                 fd = open(s->path, O_RDONLY);
1601                 if (fd < 0)
1602                         goto err_empty;
1603                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1604                 close(fd);
1605                 s->should_munmap = 1;
1606
1607                 /*
1608                  * Convert from working tree format to canonical git format
1609                  */
1610                 strbuf_init(&buf, 0);
1611                 if (convert_to_git(s->path, s->data, s->size, &buf)) {
1612                         munmap(s->data, s->size);
1613                         s->should_munmap = 0;
1614                         s->data = strbuf_detach(&buf, &s->size);
1615                         s->should_free = 1;
1616                 }
1617         }
1618         else {
1619                 enum object_type type;
1620                 if (size_only)
1621                         type = sha1_object_info(s->sha1, &s->size);
1622                 else {
1623                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1624                         s->should_free = 1;
1625                 }
1626         }
1627         return 0;
1628 }
1629
1630 void diff_free_filespec_blob(struct diff_filespec *s)
1631 {
1632         if (s->should_free)
1633                 free(s->data);
1634         else if (s->should_munmap)
1635                 munmap(s->data, s->size);
1636
1637         if (s->should_free || s->should_munmap) {
1638                 s->should_free = s->should_munmap = 0;
1639                 s->data = NULL;
1640         }
1641 }
1642
1643 void diff_free_filespec_data(struct diff_filespec *s)
1644 {
1645         diff_free_filespec_blob(s);
1646         free(s->cnt_data);
1647         s->cnt_data = NULL;
1648 }
1649
1650 static void prep_temp_blob(struct diff_tempfile *temp,
1651                            void *blob,
1652                            unsigned long size,
1653                            const unsigned char *sha1,
1654                            int mode)
1655 {
1656         int fd;
1657
1658         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1659         if (fd < 0)
1660                 die("unable to create temp-file: %s", strerror(errno));
1661         if (write_in_full(fd, blob, size) != size)
1662                 die("unable to write temp-file");
1663         close(fd);
1664         temp->name = temp->tmp_path;
1665         strcpy(temp->hex, sha1_to_hex(sha1));
1666         temp->hex[40] = 0;
1667         sprintf(temp->mode, "%06o", mode);
1668 }
1669
1670 static void prepare_temp_file(const char *name,
1671                               struct diff_tempfile *temp,
1672                               struct diff_filespec *one)
1673 {
1674         if (!DIFF_FILE_VALID(one)) {
1675         not_a_valid_file:
1676                 /* A '-' entry produces this for file-2, and
1677                  * a '+' entry produces this for file-1.
1678                  */
1679                 temp->name = "/dev/null";
1680                 strcpy(temp->hex, ".");
1681                 strcpy(temp->mode, ".");
1682                 return;
1683         }
1684
1685         if (!one->sha1_valid ||
1686             reuse_worktree_file(name, one->sha1, 1)) {
1687                 struct stat st;
1688                 if (lstat(name, &st) < 0) {
1689                         if (errno == ENOENT)
1690                                 goto not_a_valid_file;
1691                         die("stat(%s): %s", name, strerror(errno));
1692                 }
1693                 if (S_ISLNK(st.st_mode)) {
1694                         int ret;
1695                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1696                         size_t sz = xsize_t(st.st_size);
1697                         if (sizeof(buf) <= st.st_size)
1698                                 die("symlink too long: %s", name);
1699                         ret = readlink(name, buf, sz);
1700                         if (ret < 0)
1701                                 die("readlink(%s)", name);
1702                         prep_temp_blob(temp, buf, sz,
1703                                        (one->sha1_valid ?
1704                                         one->sha1 : null_sha1),
1705                                        (one->sha1_valid ?
1706                                         one->mode : S_IFLNK));
1707                 }
1708                 else {
1709                         /* we can borrow from the file in the work tree */
1710                         temp->name = name;
1711                         if (!one->sha1_valid)
1712                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1713                         else
1714                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1715                         /* Even though we may sometimes borrow the
1716                          * contents from the work tree, we always want
1717                          * one->mode.  mode is trustworthy even when
1718                          * !(one->sha1_valid), as long as
1719                          * DIFF_FILE_VALID(one).
1720                          */
1721                         sprintf(temp->mode, "%06o", one->mode);
1722                 }
1723                 return;
1724         }
1725         else {
1726                 if (diff_populate_filespec(one, 0))
1727                         die("cannot read data blob for %s", one->path);
1728                 prep_temp_blob(temp, one->data, one->size,
1729                                one->sha1, one->mode);
1730         }
1731 }
1732
1733 static void remove_tempfile(void)
1734 {
1735         int i;
1736
1737         for (i = 0; i < 2; i++)
1738                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1739                         unlink(diff_temp[i].name);
1740                         diff_temp[i].name = NULL;
1741                 }
1742 }
1743
1744 static void remove_tempfile_on_signal(int signo)
1745 {
1746         remove_tempfile();
1747         signal(SIGINT, SIG_DFL);
1748         raise(signo);
1749 }
1750
1751 static int spawn_prog(const char *pgm, const char **arg)
1752 {
1753         pid_t pid;
1754         int status;
1755
1756         fflush(NULL);
1757         pid = fork();
1758         if (pid < 0)
1759                 die("unable to fork");
1760         if (!pid) {
1761                 execvp(pgm, (char *const*) arg);
1762                 exit(255);
1763         }
1764
1765         while (waitpid(pid, &status, 0) < 0) {
1766                 if (errno == EINTR)
1767                         continue;
1768                 return -1;
1769         }
1770
1771         /* Earlier we did not check the exit status because
1772          * diff exits non-zero if files are different, and
1773          * we are not interested in knowing that.  It was a
1774          * mistake which made it harder to quit a diff-*
1775          * session that uses the git-apply-patch-script as
1776          * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
1777          * should also exit non-zero only when it wants to
1778          * abort the entire diff-* session.
1779          */
1780         if (WIFEXITED(status) && !WEXITSTATUS(status))
1781                 return 0;
1782         return -1;
1783 }
1784
1785 /* An external diff command takes:
1786  *
1787  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1788  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1789  *
1790  */
1791 static void run_external_diff(const char *pgm,
1792                               const char *name,
1793                               const char *other,
1794                               struct diff_filespec *one,
1795                               struct diff_filespec *two,
1796                               const char *xfrm_msg,
1797                               int complete_rewrite)
1798 {
1799         const char *spawn_arg[10];
1800         struct diff_tempfile *temp = diff_temp;
1801         int retval;
1802         static int atexit_asked = 0;
1803         const char *othername;
1804         const char **arg = &spawn_arg[0];
1805
1806         othername = (other? other : name);
1807         if (one && two) {
1808                 prepare_temp_file(name, &temp[0], one);
1809                 prepare_temp_file(othername, &temp[1], two);
1810                 if (! atexit_asked &&
1811                     (temp[0].name == temp[0].tmp_path ||
1812                      temp[1].name == temp[1].tmp_path)) {
1813                         atexit_asked = 1;
1814                         atexit(remove_tempfile);
1815                 }
1816                 signal(SIGINT, remove_tempfile_on_signal);
1817         }
1818
1819         if (one && two) {
1820                 *arg++ = pgm;
1821                 *arg++ = name;
1822                 *arg++ = temp[0].name;
1823                 *arg++ = temp[0].hex;
1824                 *arg++ = temp[0].mode;
1825                 *arg++ = temp[1].name;
1826                 *arg++ = temp[1].hex;
1827                 *arg++ = temp[1].mode;
1828                 if (other) {
1829                         *arg++ = other;
1830                         *arg++ = xfrm_msg;
1831                 }
1832         } else {
1833                 *arg++ = pgm;
1834                 *arg++ = name;
1835         }
1836         *arg = NULL;
1837         retval = spawn_prog(pgm, spawn_arg);
1838         remove_tempfile();
1839         if (retval) {
1840                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1841                 exit(1);
1842         }
1843 }
1844
1845 static const char *external_diff_attr(const char *name)
1846 {
1847         struct git_attr_check attr_diff_check;
1848
1849         setup_diff_attr_check(&attr_diff_check);
1850         if (!git_checkattr(name, 1, &attr_diff_check)) {
1851                 const char *value = attr_diff_check.value;
1852                 if (!ATTR_TRUE(value) &&
1853                     !ATTR_FALSE(value) &&
1854                     !ATTR_UNSET(value)) {
1855                         struct ll_diff_driver *drv;
1856
1857                         read_config_if_needed();
1858                         for (drv = user_diff; drv; drv = drv->next)
1859                                 if (!strcmp(drv->name, value))
1860                                         return drv->cmd;
1861                 }
1862         }
1863         return NULL;
1864 }
1865
1866 static void run_diff_cmd(const char *pgm,
1867                          const char *name,
1868                          const char *other,
1869                          struct diff_filespec *one,
1870                          struct diff_filespec *two,
1871                          const char *xfrm_msg,
1872                          struct diff_options *o,
1873                          int complete_rewrite)
1874 {
1875         if (!o->allow_external)
1876                 pgm = NULL;
1877         else {
1878                 const char *cmd = external_diff_attr(name);
1879                 if (cmd)
1880                         pgm = cmd;
1881         }
1882
1883         if (pgm) {
1884                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1885                                   complete_rewrite);
1886                 return;
1887         }
1888         if (one && two)
1889                 builtin_diff(name, other ? other : name,
1890                              one, two, xfrm_msg, o, complete_rewrite);
1891         else
1892                 printf("* Unmerged path %s\n", name);
1893 }
1894
1895 static void diff_fill_sha1_info(struct diff_filespec *one)
1896 {
1897         if (DIFF_FILE_VALID(one)) {
1898                 if (!one->sha1_valid) {
1899                         struct stat st;
1900                         if (!strcmp(one->path, "-")) {
1901                                 hashcpy(one->sha1, null_sha1);
1902                                 return;
1903                         }
1904                         if (lstat(one->path, &st) < 0)
1905                                 die("stat %s", one->path);
1906                         if (index_path(one->sha1, one->path, &st, 0))
1907                                 die("cannot hash %s\n", one->path);
1908                 }
1909         }
1910         else
1911                 hashclr(one->sha1);
1912 }
1913
1914 static int similarity_index(struct diff_filepair *p)
1915 {
1916         return p->score * 100 / MAX_SCORE;
1917 }
1918
1919 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1920 {
1921         const char *pgm = external_diff();
1922         struct strbuf msg;
1923         char *xfrm_msg;
1924         struct diff_filespec *one = p->one;
1925         struct diff_filespec *two = p->two;
1926         const char *name;
1927         const char *other;
1928         int complete_rewrite = 0;
1929
1930
1931         if (DIFF_PAIR_UNMERGED(p)) {
1932                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1933                 return;
1934         }
1935
1936         name  = p->one->path;
1937         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1938         diff_fill_sha1_info(one);
1939         diff_fill_sha1_info(two);
1940
1941         strbuf_init(&msg, PATH_MAX * 2 + 300);
1942         switch (p->status) {
1943         case DIFF_STATUS_COPIED:
1944                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1945                 strbuf_addstr(&msg, "\ncopy from ");
1946                 quote_c_style(name, &msg, NULL, 0);
1947                 strbuf_addstr(&msg, "\ncopy to ");
1948                 quote_c_style(other, &msg, NULL, 0);
1949                 strbuf_addch(&msg, '\n');
1950                 break;
1951         case DIFF_STATUS_RENAMED:
1952                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1953                 strbuf_addstr(&msg, "\nrename from ");
1954                 quote_c_style(name, &msg, NULL, 0);
1955                 strbuf_addstr(&msg, "\nrename to ");
1956                 quote_c_style(other, &msg, NULL, 0);
1957                 strbuf_addch(&msg, '\n');
1958                 break;
1959         case DIFF_STATUS_MODIFIED:
1960                 if (p->score) {
1961                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
1962                                         similarity_index(p));
1963                         complete_rewrite = 1;
1964                         break;
1965                 }
1966                 /* fallthru */
1967         default:
1968                 /* nothing */
1969                 ;
1970         }
1971
1972         if (hashcmp(one->sha1, two->sha1)) {
1973                 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1974
1975                 if (o->binary) {
1976                         mmfile_t mf;
1977                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
1978                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
1979                                 abbrev = 40;
1980                 }
1981                 strbuf_addf(&msg, "index %.*s..%.*s",
1982                                 abbrev, sha1_to_hex(one->sha1),
1983                                 abbrev, sha1_to_hex(two->sha1));
1984                 if (one->mode == two->mode)
1985                         strbuf_addf(&msg, " %06o", one->mode);
1986                 strbuf_addch(&msg, '\n');
1987         }
1988
1989         if (msg.len)
1990                 strbuf_setlen(&msg, msg.len - 1);
1991         xfrm_msg = msg.len ? msg.buf : NULL;
1992
1993         if (!pgm &&
1994             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1995             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1996                 /* a filepair that changes between file and symlink
1997                  * needs to be split into deletion and creation.
1998                  */
1999                 struct diff_filespec *null = alloc_filespec(two->path);
2000                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
2001                 free(null);
2002                 null = alloc_filespec(one->path);
2003                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
2004                 free(null);
2005         }
2006         else
2007                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
2008                              complete_rewrite);
2009
2010         strbuf_release(&msg);
2011 }
2012
2013 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2014                          struct diffstat_t *diffstat)
2015 {
2016         const char *name;
2017         const char *other;
2018         int complete_rewrite = 0;
2019
2020         if (DIFF_PAIR_UNMERGED(p)) {
2021                 /* unmerged */
2022                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2023                 return;
2024         }
2025
2026         name = p->one->path;
2027         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2028
2029         diff_fill_sha1_info(p->one);
2030         diff_fill_sha1_info(p->two);
2031
2032         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2033                 complete_rewrite = 1;
2034         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2035 }
2036
2037 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2038 {
2039         const char *name;
2040         const char *other;
2041
2042         if (DIFF_PAIR_UNMERGED(p)) {
2043                 /* unmerged */
2044                 return;
2045         }
2046
2047         name = p->one->path;
2048         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2049
2050         diff_fill_sha1_info(p->one);
2051         diff_fill_sha1_info(p->two);
2052
2053         builtin_checkdiff(name, other, p->one, p->two, o);
2054 }
2055
2056 void diff_setup(struct diff_options *options)
2057 {
2058         memset(options, 0, sizeof(*options));
2059         options->line_termination = '\n';
2060         options->break_opt = -1;
2061         options->rename_limit = -1;
2062         options->context = 3;
2063         options->msg_sep = "";
2064
2065         options->change = diff_change;
2066         options->add_remove = diff_addremove;
2067         options->color_diff = diff_use_color_default;
2068         options->detect_rename = diff_detect_rename_default;
2069 }
2070
2071 int diff_setup_done(struct diff_options *options)
2072 {
2073         int count = 0;
2074
2075         if (options->output_format & DIFF_FORMAT_NAME)
2076                 count++;
2077         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2078                 count++;
2079         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2080                 count++;
2081         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2082                 count++;
2083         if (count > 1)
2084                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2085
2086         if (options->find_copies_harder)
2087                 options->detect_rename = DIFF_DETECT_COPY;
2088
2089         if (options->output_format & (DIFF_FORMAT_NAME |
2090                                       DIFF_FORMAT_NAME_STATUS |
2091                                       DIFF_FORMAT_CHECKDIFF |
2092                                       DIFF_FORMAT_NO_OUTPUT))
2093                 options->output_format &= ~(DIFF_FORMAT_RAW |
2094                                             DIFF_FORMAT_NUMSTAT |
2095                                             DIFF_FORMAT_DIFFSTAT |
2096                                             DIFF_FORMAT_SHORTSTAT |
2097                                             DIFF_FORMAT_SUMMARY |
2098                                             DIFF_FORMAT_PATCH);
2099
2100         /*
2101          * These cases always need recursive; we do not drop caller-supplied
2102          * recursive bits for other formats here.
2103          */
2104         if (options->output_format & (DIFF_FORMAT_PATCH |
2105                                       DIFF_FORMAT_NUMSTAT |
2106                                       DIFF_FORMAT_DIFFSTAT |
2107                                       DIFF_FORMAT_SHORTSTAT |
2108                                       DIFF_FORMAT_SUMMARY |
2109                                       DIFF_FORMAT_CHECKDIFF))
2110                 options->recursive = 1;
2111         /*
2112          * Also pickaxe would not work very well if you do not say recursive
2113          */
2114         if (options->pickaxe)
2115                 options->recursive = 1;
2116
2117         if (options->detect_rename && options->rename_limit < 0)
2118                 options->rename_limit = diff_rename_limit_default;
2119         if (options->setup & DIFF_SETUP_USE_CACHE) {
2120                 if (!active_cache)
2121                         /* read-cache does not die even when it fails
2122                          * so it is safe for us to do this here.  Also
2123                          * it does not smudge active_cache or active_nr
2124                          * when it fails, so we do not have to worry about
2125                          * cleaning it up ourselves either.
2126                          */
2127                         read_cache();
2128         }
2129         if (options->abbrev <= 0 || 40 < options->abbrev)
2130                 options->abbrev = 40; /* full */
2131
2132         /*
2133          * It does not make sense to show the first hit we happened
2134          * to have found.  It does not make sense not to return with
2135          * exit code in such a case either.
2136          */
2137         if (options->quiet) {
2138                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2139                 options->exit_with_status = 1;
2140         }
2141
2142         /*
2143          * If we postprocess in diffcore, we cannot simply return
2144          * upon the first hit.  We need to run diff as usual.
2145          */
2146         if (options->pickaxe || options->filter)
2147                 options->quiet = 0;
2148
2149         return 0;
2150 }
2151
2152 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2153 {
2154         char c, *eq;
2155         int len;
2156
2157         if (*arg != '-')
2158                 return 0;
2159         c = *++arg;
2160         if (!c)
2161                 return 0;
2162         if (c == arg_short) {
2163                 c = *++arg;
2164                 if (!c)
2165                         return 1;
2166                 if (val && isdigit(c)) {
2167                         char *end;
2168                         int n = strtoul(arg, &end, 10);
2169                         if (*end)
2170                                 return 0;
2171                         *val = n;
2172                         return 1;
2173                 }
2174                 return 0;
2175         }
2176         if (c != '-')
2177                 return 0;
2178         arg++;
2179         eq = strchr(arg, '=');
2180         if (eq)
2181                 len = eq - arg;
2182         else
2183                 len = strlen(arg);
2184         if (!len || strncmp(arg, arg_long, len))
2185                 return 0;
2186         if (eq) {
2187                 int n;
2188                 char *end;
2189                 if (!isdigit(*++eq))
2190                         return 0;
2191                 n = strtoul(eq, &end, 10);
2192                 if (*end)
2193                         return 0;
2194                 *val = n;
2195         }
2196         return 1;
2197 }
2198
2199 static int diff_scoreopt_parse(const char *opt);
2200
2201 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2202 {
2203         const char *arg = av[0];
2204         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2205                 options->output_format |= DIFF_FORMAT_PATCH;
2206         else if (opt_arg(arg, 'U', "unified", &options->context))
2207                 options->output_format |= DIFF_FORMAT_PATCH;
2208         else if (!strcmp(arg, "--raw"))
2209                 options->output_format |= DIFF_FORMAT_RAW;
2210         else if (!strcmp(arg, "--patch-with-raw")) {
2211                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2212         }
2213         else if (!strcmp(arg, "--numstat")) {
2214                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2215         }
2216         else if (!strcmp(arg, "--shortstat")) {
2217                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2218         }
2219         else if (!prefixcmp(arg, "--stat")) {
2220                 char *end;
2221                 int width = options->stat_width;
2222                 int name_width = options->stat_name_width;
2223                 arg += 6;
2224                 end = (char *)arg;
2225
2226                 switch (*arg) {
2227                 case '-':
2228                         if (!prefixcmp(arg, "-width="))
2229                                 width = strtoul(arg + 7, &end, 10);
2230                         else if (!prefixcmp(arg, "-name-width="))
2231                                 name_width = strtoul(arg + 12, &end, 10);
2232                         break;
2233                 case '=':
2234                         width = strtoul(arg+1, &end, 10);
2235                         if (*end == ',')
2236                                 name_width = strtoul(end+1, &end, 10);
2237                 }
2238
2239                 /* Important! This checks all the error cases! */
2240                 if (*end)
2241                         return 0;
2242                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2243                 options->stat_name_width = name_width;
2244                 options->stat_width = width;
2245         }
2246         else if (!strcmp(arg, "--check"))
2247                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2248         else if (!strcmp(arg, "--summary"))
2249                 options->output_format |= DIFF_FORMAT_SUMMARY;
2250         else if (!strcmp(arg, "--patch-with-stat")) {
2251                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2252         }
2253         else if (!strcmp(arg, "-z"))
2254                 options->line_termination = 0;
2255         else if (!prefixcmp(arg, "-l"))
2256                 options->rename_limit = strtoul(arg+2, NULL, 10);
2257         else if (!strcmp(arg, "--full-index"))
2258                 options->full_index = 1;
2259         else if (!strcmp(arg, "--binary")) {
2260                 options->output_format |= DIFF_FORMAT_PATCH;
2261                 options->binary = 1;
2262         }
2263         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
2264                 options->text = 1;
2265         }
2266         else if (!strcmp(arg, "--name-only"))
2267                 options->output_format |= DIFF_FORMAT_NAME;
2268         else if (!strcmp(arg, "--name-status"))
2269                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2270         else if (!strcmp(arg, "-R"))
2271                 options->reverse_diff = 1;
2272         else if (!prefixcmp(arg, "-S"))
2273                 options->pickaxe = arg + 2;
2274         else if (!strcmp(arg, "-s")) {
2275                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2276         }
2277         else if (!prefixcmp(arg, "-O"))
2278                 options->orderfile = arg + 2;
2279         else if (!prefixcmp(arg, "--diff-filter="))
2280                 options->filter = arg + 14;
2281         else if (!strcmp(arg, "--pickaxe-all"))
2282                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2283         else if (!strcmp(arg, "--pickaxe-regex"))
2284                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2285         else if (!prefixcmp(arg, "-B")) {
2286                 if ((options->break_opt =
2287                      diff_scoreopt_parse(arg)) == -1)
2288                         return -1;
2289         }
2290         else if (!prefixcmp(arg, "-M")) {
2291                 if ((options->rename_score =
2292                      diff_scoreopt_parse(arg)) == -1)
2293                         return -1;
2294                 options->detect_rename = DIFF_DETECT_RENAME;
2295         }
2296         else if (!prefixcmp(arg, "-C")) {
2297                 if (options->detect_rename == DIFF_DETECT_COPY)
2298                         options->find_copies_harder = 1;
2299                 if ((options->rename_score =
2300                      diff_scoreopt_parse(arg)) == -1)
2301                         return -1;
2302                 options->detect_rename = DIFF_DETECT_COPY;
2303         }
2304         else if (!strcmp(arg, "--find-copies-harder"))
2305                 options->find_copies_harder = 1;
2306         else if (!strcmp(arg, "--follow"))
2307                 options->follow_renames = 1;
2308         else if (!strcmp(arg, "--abbrev"))
2309                 options->abbrev = DEFAULT_ABBREV;
2310         else if (!prefixcmp(arg, "--abbrev=")) {
2311                 options->abbrev = strtoul(arg + 9, NULL, 10);
2312                 if (options->abbrev < MINIMUM_ABBREV)
2313                         options->abbrev = MINIMUM_ABBREV;
2314                 else if (40 < options->abbrev)
2315                         options->abbrev = 40;
2316         }
2317         else if (!strcmp(arg, "--color"))
2318                 options->color_diff = 1;
2319         else if (!strcmp(arg, "--no-color"))
2320                 options->color_diff = 0;
2321         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2322                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2323         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2324                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2325         else if (!strcmp(arg, "--ignore-space-at-eol"))
2326                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2327         else if (!strcmp(arg, "--color-words"))
2328                 options->color_diff = options->color_diff_words = 1;
2329         else if (!strcmp(arg, "--no-renames"))
2330                 options->detect_rename = 0;
2331         else if (!strcmp(arg, "--exit-code"))
2332                 options->exit_with_status = 1;
2333         else if (!strcmp(arg, "--quiet"))
2334                 options->quiet = 1;
2335         else if (!strcmp(arg, "--ext-diff"))
2336                 options->allow_external = 1;
2337         else if (!strcmp(arg, "--no-ext-diff"))
2338                 options->allow_external = 0;
2339         else
2340                 return 0;
2341         return 1;
2342 }
2343
2344 static int parse_num(const char **cp_p)
2345 {
2346         unsigned long num, scale;
2347         int ch, dot;
2348         const char *cp = *cp_p;
2349
2350         num = 0;
2351         scale = 1;
2352         dot = 0;
2353         for(;;) {
2354                 ch = *cp;
2355                 if ( !dot && ch == '.' ) {
2356                         scale = 1;
2357                         dot = 1;
2358                 } else if ( ch == '%' ) {
2359                         scale = dot ? scale*100 : 100;
2360                         cp++;   /* % is always at the end */
2361                         break;
2362                 } else if ( ch >= '0' && ch <= '9' ) {
2363                         if ( scale < 100000 ) {
2364                                 scale *= 10;
2365                                 num = (num*10) + (ch-'0');
2366                         }
2367                 } else {
2368                         break;
2369                 }
2370                 cp++;
2371         }
2372         *cp_p = cp;
2373
2374         /* user says num divided by scale and we say internally that
2375          * is MAX_SCORE * num / scale.
2376          */
2377         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2378 }
2379
2380 static int diff_scoreopt_parse(const char *opt)
2381 {
2382         int opt1, opt2, cmd;
2383
2384         if (*opt++ != '-')
2385                 return -1;
2386         cmd = *opt++;
2387         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2388                 return -1; /* that is not a -M, -C nor -B option */
2389
2390         opt1 = parse_num(&opt);
2391         if (cmd != 'B')
2392                 opt2 = 0;
2393         else {
2394                 if (*opt == 0)
2395                         opt2 = 0;
2396                 else if (*opt != '/')
2397                         return -1; /* we expect -B80/99 or -B80 */
2398                 else {
2399                         opt++;
2400                         opt2 = parse_num(&opt);
2401                 }
2402         }
2403         if (*opt != 0)
2404                 return -1;
2405         return opt1 | (opt2 << 16);
2406 }
2407
2408 struct diff_queue_struct diff_queued_diff;
2409
2410 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2411 {
2412         if (queue->alloc <= queue->nr) {
2413                 queue->alloc = alloc_nr(queue->alloc);
2414                 queue->queue = xrealloc(queue->queue,
2415                                         sizeof(dp) * queue->alloc);
2416         }
2417         queue->queue[queue->nr++] = dp;
2418 }
2419
2420 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2421                                  struct diff_filespec *one,
2422                                  struct diff_filespec *two)
2423 {
2424         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2425         dp->one = one;
2426         dp->two = two;
2427         if (queue)
2428                 diff_q(queue, dp);
2429         return dp;
2430 }
2431
2432 void diff_free_filepair(struct diff_filepair *p)
2433 {
2434         diff_free_filespec_data(p->one);
2435         diff_free_filespec_data(p->two);
2436         free(p->one);
2437         free(p->two);
2438         free(p);
2439 }
2440
2441 /* This is different from find_unique_abbrev() in that
2442  * it stuffs the result with dots for alignment.
2443  */
2444 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2445 {
2446         int abblen;
2447         const char *abbrev;
2448         if (len == 40)
2449                 return sha1_to_hex(sha1);
2450
2451         abbrev = find_unique_abbrev(sha1, len);
2452         if (!abbrev)
2453                 return sha1_to_hex(sha1);
2454         abblen = strlen(abbrev);
2455         if (abblen < 37) {
2456                 static char hex[41];
2457                 if (len < abblen && abblen <= len + 2)
2458                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2459                 else
2460                         sprintf(hex, "%s...", abbrev);
2461                 return hex;
2462         }
2463         return sha1_to_hex(sha1);
2464 }
2465
2466 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2467 {
2468         int line_termination = opt->line_termination;
2469         int inter_name_termination = line_termination ? '\t' : '\0';
2470
2471         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2472                 printf(":%06o %06o %s ", p->one->mode, p->two->mode,
2473                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
2474                 printf("%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2475         }
2476         if (p->score) {
2477                 printf("%c%03d%c", p->status, similarity_index(p),
2478                            inter_name_termination);
2479         } else {
2480                 printf("%c%c", p->status, inter_name_termination);
2481         }
2482
2483         if (p->status == DIFF_STATUS_COPIED || p->status == DIFF_STATUS_RENAMED) {
2484                 write_name_quoted(p->one->path, stdout, inter_name_termination);
2485                 write_name_quoted(p->two->path, stdout, line_termination);
2486         } else {
2487                 const char *path = p->one->mode ? p->one->path : p->two->path;
2488                 write_name_quoted(path, stdout, line_termination);
2489         }
2490 }
2491
2492 int diff_unmodified_pair(struct diff_filepair *p)
2493 {
2494         /* This function is written stricter than necessary to support
2495          * the currently implemented transformers, but the idea is to
2496          * let transformers to produce diff_filepairs any way they want,
2497          * and filter and clean them up here before producing the output.
2498          */
2499         struct diff_filespec *one = p->one, *two = p->two;
2500
2501         if (DIFF_PAIR_UNMERGED(p))
2502                 return 0; /* unmerged is interesting */
2503
2504         /* deletion, addition, mode or type change
2505          * and rename are all interesting.
2506          */
2507         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2508             DIFF_PAIR_MODE_CHANGED(p) ||
2509             strcmp(one->path, two->path))
2510                 return 0;
2511
2512         /* both are valid and point at the same path.  that is, we are
2513          * dealing with a change.
2514          */
2515         if (one->sha1_valid && two->sha1_valid &&
2516             !hashcmp(one->sha1, two->sha1))
2517                 return 1; /* no change */
2518         if (!one->sha1_valid && !two->sha1_valid)
2519                 return 1; /* both look at the same file on the filesystem. */
2520         return 0;
2521 }
2522
2523 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2524 {
2525         if (diff_unmodified_pair(p))
2526                 return;
2527
2528         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2529             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2530                 return; /* no tree diffs in patch format */
2531
2532         run_diff(p, o);
2533 }
2534
2535 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2536                             struct diffstat_t *diffstat)
2537 {
2538         if (diff_unmodified_pair(p))
2539                 return;
2540
2541         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2542             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2543                 return; /* no tree diffs in patch format */
2544
2545         run_diffstat(p, o, diffstat);
2546 }
2547
2548 static void diff_flush_checkdiff(struct diff_filepair *p,
2549                 struct diff_options *o)
2550 {
2551         if (diff_unmodified_pair(p))
2552                 return;
2553
2554         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2555             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2556                 return; /* no tree diffs in patch format */
2557
2558         run_checkdiff(p, o);
2559 }
2560
2561 int diff_queue_is_empty(void)
2562 {
2563         struct diff_queue_struct *q = &diff_queued_diff;
2564         int i;
2565         for (i = 0; i < q->nr; i++)
2566                 if (!diff_unmodified_pair(q->queue[i]))
2567                         return 0;
2568         return 1;
2569 }
2570
2571 #if DIFF_DEBUG
2572 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2573 {
2574         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2575                 x, one ? one : "",
2576                 s->path,
2577                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2578                 s->mode,
2579                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2580         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2581                 x, one ? one : "",
2582                 s->size, s->xfrm_flags);
2583 }
2584
2585 void diff_debug_filepair(const struct diff_filepair *p, int i)
2586 {
2587         diff_debug_filespec(p->one, i, "one");
2588         diff_debug_filespec(p->two, i, "two");
2589         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
2590                 p->score, p->status ? p->status : '?',
2591                 p->source_stays, p->broken_pair);
2592 }
2593
2594 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2595 {
2596         int i;
2597         if (msg)
2598                 fprintf(stderr, "%s\n", msg);
2599         fprintf(stderr, "q->nr = %d\n", q->nr);
2600         for (i = 0; i < q->nr; i++) {
2601                 struct diff_filepair *p = q->queue[i];
2602                 diff_debug_filepair(p, i);
2603         }
2604 }
2605 #endif
2606
2607 static void diff_resolve_rename_copy(void)
2608 {
2609         int i, j;
2610         struct diff_filepair *p, *pp;
2611         struct diff_queue_struct *q = &diff_queued_diff;
2612
2613         diff_debug_queue("resolve-rename-copy", q);
2614
2615         for (i = 0; i < q->nr; i++) {
2616                 p = q->queue[i];
2617                 p->status = 0; /* undecided */
2618                 if (DIFF_PAIR_UNMERGED(p))
2619                         p->status = DIFF_STATUS_UNMERGED;
2620                 else if (!DIFF_FILE_VALID(p->one))
2621                         p->status = DIFF_STATUS_ADDED;
2622                 else if (!DIFF_FILE_VALID(p->two))
2623                         p->status = DIFF_STATUS_DELETED;
2624                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2625                         p->status = DIFF_STATUS_TYPE_CHANGED;
2626
2627                 /* from this point on, we are dealing with a pair
2628                  * whose both sides are valid and of the same type, i.e.
2629                  * either in-place edit or rename/copy edit.
2630                  */
2631                 else if (DIFF_PAIR_RENAME(p)) {
2632                         if (p->source_stays) {
2633                                 p->status = DIFF_STATUS_COPIED;
2634                                 continue;
2635                         }
2636                         /* See if there is some other filepair that
2637                          * copies from the same source as us.  If so
2638                          * we are a copy.  Otherwise we are either a
2639                          * copy if the path stays, or a rename if it
2640                          * does not, but we already handled "stays" case.
2641                          */
2642                         for (j = i + 1; j < q->nr; j++) {
2643                                 pp = q->queue[j];
2644                                 if (strcmp(pp->one->path, p->one->path))
2645                                         continue; /* not us */
2646                                 if (!DIFF_PAIR_RENAME(pp))
2647                                         continue; /* not a rename/copy */
2648                                 /* pp is a rename/copy from the same source */
2649                                 p->status = DIFF_STATUS_COPIED;
2650                                 break;
2651                         }
2652                         if (!p->status)
2653                                 p->status = DIFF_STATUS_RENAMED;
2654                 }
2655                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2656                          p->one->mode != p->two->mode ||
2657                          is_null_sha1(p->one->sha1))
2658                         p->status = DIFF_STATUS_MODIFIED;
2659                 else {
2660                         /* This is a "no-change" entry and should not
2661                          * happen anymore, but prepare for broken callers.
2662                          */
2663                         error("feeding unmodified %s to diffcore",
2664                               p->one->path);
2665                         p->status = DIFF_STATUS_UNKNOWN;
2666                 }
2667         }
2668         diff_debug_queue("resolve-rename-copy done", q);
2669 }
2670
2671 static int check_pair_status(struct diff_filepair *p)
2672 {
2673         switch (p->status) {
2674         case DIFF_STATUS_UNKNOWN:
2675                 return 0;
2676         case 0:
2677                 die("internal error in diff-resolve-rename-copy");
2678         default:
2679                 return 1;
2680         }
2681 }
2682
2683 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2684 {
2685         int fmt = opt->output_format;
2686
2687         if (fmt & DIFF_FORMAT_CHECKDIFF)
2688                 diff_flush_checkdiff(p, opt);
2689         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2690                 diff_flush_raw(p, opt);
2691         else if (fmt & DIFF_FORMAT_NAME)
2692                 write_name_quoted(p->two->path, stdout, opt->line_termination);
2693 }
2694
2695 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2696 {
2697         if (fs->mode)
2698                 printf(" %s mode %06o ", newdelete, fs->mode);
2699         else
2700                 printf(" %s ", newdelete);
2701         write_name_quoted(fs->path, stdout, '\n');
2702 }
2703
2704
2705 static void show_mode_change(struct diff_filepair *p, int show_name)
2706 {
2707         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2708                 printf(" mode change %06o => %06o%c", p->one->mode, p->two->mode,
2709                         show_name ? ' ' : '\n');
2710                 if (show_name) {
2711                         write_name_quoted(p->two->path, stdout, '\n');
2712                 }
2713         }
2714 }
2715
2716 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2717 {
2718         char *names = pprint_rename(p->one->path, p->two->path);
2719
2720         printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2721         free(names);
2722         show_mode_change(p, 0);
2723 }
2724
2725 static void diff_summary(struct diff_filepair *p)
2726 {
2727         switch(p->status) {
2728         case DIFF_STATUS_DELETED:
2729                 show_file_mode_name("delete", p->one);
2730                 break;
2731         case DIFF_STATUS_ADDED:
2732                 show_file_mode_name("create", p->two);
2733                 break;
2734         case DIFF_STATUS_COPIED:
2735                 show_rename_copy("copy", p);
2736                 break;
2737         case DIFF_STATUS_RENAMED:
2738                 show_rename_copy("rename", p);
2739                 break;
2740         default:
2741                 if (p->score) {
2742                         puts(" rewrite ");
2743                         write_name_quoted(p->two->path, stdout, ' ');
2744                         printf("(%d%%)\n", similarity_index(p));
2745                 }
2746                 show_mode_change(p, !p->score);
2747                 break;
2748         }
2749 }
2750
2751 struct patch_id_t {
2752         struct xdiff_emit_state xm;
2753         SHA_CTX *ctx;
2754         int patchlen;
2755 };
2756
2757 static int remove_space(char *line, int len)
2758 {
2759         int i;
2760         char *dst = line;
2761         unsigned char c;
2762
2763         for (i = 0; i < len; i++)
2764                 if (!isspace((c = line[i])))
2765                         *dst++ = c;
2766
2767         return dst - line;
2768 }
2769
2770 static void patch_id_consume(void *priv, char *line, unsigned long len)
2771 {
2772         struct patch_id_t *data = priv;
2773         int new_len;
2774
2775         /* Ignore line numbers when computing the SHA1 of the patch */
2776         if (!prefixcmp(line, "@@ -"))
2777                 return;
2778
2779         new_len = remove_space(line, len);
2780
2781         SHA1_Update(data->ctx, line, new_len);
2782         data->patchlen += new_len;
2783 }
2784
2785 /* returns 0 upon success, and writes result into sha1 */
2786 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2787 {
2788         struct diff_queue_struct *q = &diff_queued_diff;
2789         int i;
2790         SHA_CTX ctx;
2791         struct patch_id_t data;
2792         char buffer[PATH_MAX * 4 + 20];
2793
2794         SHA1_Init(&ctx);
2795         memset(&data, 0, sizeof(struct patch_id_t));
2796         data.ctx = &ctx;
2797         data.xm.consume = patch_id_consume;
2798
2799         for (i = 0; i < q->nr; i++) {
2800                 xpparam_t xpp;
2801                 xdemitconf_t xecfg;
2802                 xdemitcb_t ecb;
2803                 mmfile_t mf1, mf2;
2804                 struct diff_filepair *p = q->queue[i];
2805                 int len1, len2;
2806
2807                 memset(&xecfg, 0, sizeof(xecfg));
2808                 if (p->status == 0)
2809                         return error("internal diff status error");
2810                 if (p->status == DIFF_STATUS_UNKNOWN)
2811                         continue;
2812                 if (diff_unmodified_pair(p))
2813                         continue;
2814                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2815                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2816                         continue;
2817                 if (DIFF_PAIR_UNMERGED(p))
2818                         continue;
2819
2820                 diff_fill_sha1_info(p->one);
2821                 diff_fill_sha1_info(p->two);
2822                 if (fill_mmfile(&mf1, p->one) < 0 ||
2823                                 fill_mmfile(&mf2, p->two) < 0)
2824                         return error("unable to read files to diff");
2825
2826                 len1 = remove_space(p->one->path, strlen(p->one->path));
2827                 len2 = remove_space(p->two->path, strlen(p->two->path));
2828                 if (p->one->mode == 0)
2829                         len1 = snprintf(buffer, sizeof(buffer),
2830                                         "diff--gita/%.*sb/%.*s"
2831                                         "newfilemode%06o"
2832                                         "---/dev/null"
2833                                         "+++b/%.*s",
2834                                         len1, p->one->path,
2835                                         len2, p->two->path,
2836                                         p->two->mode,
2837                                         len2, p->two->path);
2838                 else if (p->two->mode == 0)
2839                         len1 = snprintf(buffer, sizeof(buffer),
2840                                         "diff--gita/%.*sb/%.*s"
2841                                         "deletedfilemode%06o"
2842                                         "---a/%.*s"
2843                                         "+++/dev/null",
2844                                         len1, p->one->path,
2845                                         len2, p->two->path,
2846                                         p->one->mode,
2847                                         len1, p->one->path);
2848                 else
2849                         len1 = snprintf(buffer, sizeof(buffer),
2850                                         "diff--gita/%.*sb/%.*s"
2851                                         "---a/%.*s"
2852                                         "+++b/%.*s",
2853                                         len1, p->one->path,
2854                                         len2, p->two->path,
2855                                         len1, p->one->path,
2856                                         len2, p->two->path);
2857                 SHA1_Update(&ctx, buffer, len1);
2858
2859                 xpp.flags = XDF_NEED_MINIMAL;
2860                 xecfg.ctxlen = 3;
2861                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2862                 ecb.outf = xdiff_outf;
2863                 ecb.priv = &data;
2864                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2865         }
2866
2867         SHA1_Final(sha1, &ctx);
2868         return 0;
2869 }
2870
2871 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2872 {
2873         struct diff_queue_struct *q = &diff_queued_diff;
2874         int i;
2875         int result = diff_get_patch_id(options, sha1);
2876
2877         for (i = 0; i < q->nr; i++)
2878                 diff_free_filepair(q->queue[i]);
2879
2880         free(q->queue);
2881         q->queue = NULL;
2882         q->nr = q->alloc = 0;
2883
2884         return result;
2885 }
2886
2887 static int is_summary_empty(const struct diff_queue_struct *q)
2888 {
2889         int i;
2890
2891         for (i = 0; i < q->nr; i++) {
2892                 const struct diff_filepair *p = q->queue[i];
2893
2894                 switch (p->status) {
2895                 case DIFF_STATUS_DELETED:
2896                 case DIFF_STATUS_ADDED:
2897                 case DIFF_STATUS_COPIED:
2898                 case DIFF_STATUS_RENAMED:
2899                         return 0;
2900                 default:
2901                         if (p->score)
2902                                 return 0;
2903                         if (p->one->mode && p->two->mode &&
2904                             p->one->mode != p->two->mode)
2905                                 return 0;
2906                         break;
2907                 }
2908         }
2909         return 1;
2910 }
2911
2912 void diff_flush(struct diff_options *options)
2913 {
2914         struct diff_queue_struct *q = &diff_queued_diff;
2915         int i, output_format = options->output_format;
2916         int separator = 0;
2917
2918         /*
2919          * Order: raw, stat, summary, patch
2920          * or:    name/name-status/checkdiff (other bits clear)
2921          */
2922         if (!q->nr)
2923                 goto free_queue;
2924
2925         if (output_format & (DIFF_FORMAT_RAW |
2926                              DIFF_FORMAT_NAME |
2927                              DIFF_FORMAT_NAME_STATUS |
2928                              DIFF_FORMAT_CHECKDIFF)) {
2929                 for (i = 0; i < q->nr; i++) {
2930                         struct diff_filepair *p = q->queue[i];
2931                         if (check_pair_status(p))
2932                                 flush_one_pair(p, options);
2933                 }
2934                 separator++;
2935         }
2936
2937         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2938                 struct diffstat_t diffstat;
2939
2940                 memset(&diffstat, 0, sizeof(struct diffstat_t));
2941                 diffstat.xm.consume = diffstat_consume;
2942                 for (i = 0; i < q->nr; i++) {
2943                         struct diff_filepair *p = q->queue[i];
2944                         if (check_pair_status(p))
2945                                 diff_flush_stat(p, options, &diffstat);
2946                 }
2947                 if (output_format & DIFF_FORMAT_NUMSTAT)
2948                         show_numstat(&diffstat, options);
2949                 if (output_format & DIFF_FORMAT_DIFFSTAT)
2950                         show_stats(&diffstat, options);
2951                 else if (output_format & DIFF_FORMAT_SHORTSTAT)
2952                         show_shortstats(&diffstat);
2953                 separator++;
2954         }
2955
2956         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2957                 for (i = 0; i < q->nr; i++)
2958                         diff_summary(q->queue[i]);
2959                 separator++;
2960         }
2961
2962         if (output_format & DIFF_FORMAT_PATCH) {
2963                 if (separator) {
2964                         if (options->stat_sep) {
2965                                 /* attach patch instead of inline */
2966                                 fputs(options->stat_sep, stdout);
2967                         } else {
2968                                 putchar(options->line_termination);
2969                         }
2970                 }
2971
2972                 for (i = 0; i < q->nr; i++) {
2973                         struct diff_filepair *p = q->queue[i];
2974                         if (check_pair_status(p))
2975                                 diff_flush_patch(p, options);
2976                 }
2977         }
2978
2979         if (output_format & DIFF_FORMAT_CALLBACK)
2980                 options->format_callback(q, options, options->format_callback_data);
2981
2982         for (i = 0; i < q->nr; i++)
2983                 diff_free_filepair(q->queue[i]);
2984 free_queue:
2985         free(q->queue);
2986         q->queue = NULL;
2987         q->nr = q->alloc = 0;
2988 }
2989
2990 static void diffcore_apply_filter(const char *filter)
2991 {
2992         int i;
2993         struct diff_queue_struct *q = &diff_queued_diff;
2994         struct diff_queue_struct outq;
2995         outq.queue = NULL;
2996         outq.nr = outq.alloc = 0;
2997
2998         if (!filter)
2999                 return;
3000
3001         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3002                 int found;
3003                 for (i = found = 0; !found && i < q->nr; i++) {
3004                         struct diff_filepair *p = q->queue[i];
3005                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3006                              ((p->score &&
3007                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3008                               (!p->score &&
3009                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3010                             ((p->status != DIFF_STATUS_MODIFIED) &&
3011                              strchr(filter, p->status)))
3012                                 found++;
3013                 }
3014                 if (found)
3015                         return;
3016
3017                 /* otherwise we will clear the whole queue
3018                  * by copying the empty outq at the end of this
3019                  * function, but first clear the current entries
3020                  * in the queue.
3021                  */
3022                 for (i = 0; i < q->nr; i++)
3023                         diff_free_filepair(q->queue[i]);
3024         }
3025         else {
3026                 /* Only the matching ones */
3027                 for (i = 0; i < q->nr; i++) {
3028                         struct diff_filepair *p = q->queue[i];
3029
3030                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3031                              ((p->score &&
3032                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3033                               (!p->score &&
3034                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3035                             ((p->status != DIFF_STATUS_MODIFIED) &&
3036                              strchr(filter, p->status)))
3037                                 diff_q(&outq, p);
3038                         else
3039                                 diff_free_filepair(p);
3040                 }
3041         }
3042         free(q->queue);
3043         *q = outq;
3044 }
3045
3046 /* Check whether two filespecs with the same mode and size are identical */
3047 static int diff_filespec_is_identical(struct diff_filespec *one,
3048                                       struct diff_filespec *two)
3049 {
3050         if (S_ISGITLINK(one->mode)) {
3051                 diff_fill_sha1_info(one);
3052                 diff_fill_sha1_info(two);
3053                 return !hashcmp(one->sha1, two->sha1);
3054         }
3055         if (diff_populate_filespec(one, 0))
3056                 return 0;
3057         if (diff_populate_filespec(two, 0))
3058                 return 0;
3059         return !memcmp(one->data, two->data, one->size);
3060 }
3061
3062 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3063 {
3064         int i;
3065         struct diff_queue_struct *q = &diff_queued_diff;
3066         struct diff_queue_struct outq;
3067         outq.queue = NULL;
3068         outq.nr = outq.alloc = 0;
3069
3070         for (i = 0; i < q->nr; i++) {
3071                 struct diff_filepair *p = q->queue[i];
3072
3073                 /*
3074                  * 1. Entries that come from stat info dirtyness
3075                  *    always have both sides (iow, not create/delete),
3076                  *    one side of the object name is unknown, with
3077                  *    the same mode and size.  Keep the ones that
3078                  *    do not match these criteria.  They have real
3079                  *    differences.
3080                  *
3081                  * 2. At this point, the file is known to be modified,
3082                  *    with the same mode and size, and the object
3083                  *    name of one side is unknown.  Need to inspect
3084                  *    the identical contents.
3085                  */
3086                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3087                     !DIFF_FILE_VALID(p->two) ||
3088                     (p->one->sha1_valid && p->two->sha1_valid) ||
3089                     (p->one->mode != p->two->mode) ||
3090                     diff_populate_filespec(p->one, 1) ||
3091                     diff_populate_filespec(p->two, 1) ||
3092                     (p->one->size != p->two->size) ||
3093                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3094                         diff_q(&outq, p);
3095                 else {
3096                         /*
3097                          * The caller can subtract 1 from skip_stat_unmatch
3098                          * to determine how many paths were dirty only
3099                          * due to stat info mismatch.
3100                          */
3101                         if (!diffopt->no_index)
3102                                 diffopt->skip_stat_unmatch++;
3103                         diff_free_filepair(p);
3104                 }
3105         }
3106         free(q->queue);
3107         *q = outq;
3108 }
3109
3110 void diffcore_std(struct diff_options *options)
3111 {
3112         if (options->quiet)
3113                 return;
3114
3115         if (options->skip_stat_unmatch && !options->find_copies_harder)
3116                 diffcore_skip_stat_unmatch(options);
3117         if (options->break_opt != -1)
3118                 diffcore_break(options->break_opt);
3119         if (options->detect_rename)
3120                 diffcore_rename(options);
3121         if (options->break_opt != -1)
3122                 diffcore_merge_broken();
3123         if (options->pickaxe)
3124                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3125         if (options->orderfile)
3126                 diffcore_order(options->orderfile);
3127         diff_resolve_rename_copy();
3128         diffcore_apply_filter(options->filter);
3129
3130         options->has_changes = !!diff_queued_diff.nr;
3131 }
3132
3133
3134 void diff_addremove(struct diff_options *options,
3135                     int addremove, unsigned mode,
3136                     const unsigned char *sha1,
3137                     const char *base, const char *path)
3138 {
3139         char concatpath[PATH_MAX];
3140         struct diff_filespec *one, *two;
3141
3142         /* This may look odd, but it is a preparation for
3143          * feeding "there are unchanged files which should
3144          * not produce diffs, but when you are doing copy
3145          * detection you would need them, so here they are"
3146          * entries to the diff-core.  They will be prefixed
3147          * with something like '=' or '*' (I haven't decided
3148          * which but should not make any difference).
3149          * Feeding the same new and old to diff_change()
3150          * also has the same effect.
3151          * Before the final output happens, they are pruned after
3152          * merged into rename/copy pairs as appropriate.
3153          */
3154         if (options->reverse_diff)
3155                 addremove = (addremove == '+' ? '-' :
3156                              addremove == '-' ? '+' : addremove);
3157
3158         if (!path) path = "";
3159         sprintf(concatpath, "%s%s", base, path);
3160         one = alloc_filespec(concatpath);
3161         two = alloc_filespec(concatpath);
3162
3163         if (addremove != '+')
3164                 fill_filespec(one, sha1, mode);
3165         if (addremove != '-')
3166                 fill_filespec(two, sha1, mode);
3167
3168         diff_queue(&diff_queued_diff, one, two);
3169         options->has_changes = 1;
3170 }
3171
3172 void diff_change(struct diff_options *options,
3173                  unsigned old_mode, unsigned new_mode,
3174                  const unsigned char *old_sha1,
3175                  const unsigned char *new_sha1,
3176                  const char *base, const char *path)
3177 {
3178         char concatpath[PATH_MAX];
3179         struct diff_filespec *one, *two;
3180
3181         if (options->reverse_diff) {
3182                 unsigned tmp;
3183                 const unsigned char *tmp_c;
3184                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3185                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3186         }
3187         if (!path) path = "";
3188         sprintf(concatpath, "%s%s", base, path);
3189         one = alloc_filespec(concatpath);
3190         two = alloc_filespec(concatpath);
3191         fill_filespec(one, old_sha1, old_mode);
3192         fill_filespec(two, new_sha1, new_mode);
3193
3194         diff_queue(&diff_queued_diff, one, two);
3195         options->has_changes = 1;
3196 }
3197
3198 void diff_unmerge(struct diff_options *options,
3199                   const char *path,
3200                   unsigned mode, const unsigned char *sha1)
3201 {
3202         struct diff_filespec *one, *two;
3203         one = alloc_filespec(path);
3204         two = alloc_filespec(path);
3205         fill_filespec(one, sha1, mode);
3206         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3207 }