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