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