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