diff: add filter for converting binary to text
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
14 #include "userdiff.h"
15
16 #ifdef NO_FAST_WORKING_DIRECTORY
17 #define FAST_WORKING_DIRECTORY 0
18 #else
19 #define FAST_WORKING_DIRECTORY 1
20 #endif
21
22 static int diff_detect_rename_default;
23 static int diff_rename_limit_default = 200;
24 static int diff_suppress_blank_empty;
25 int diff_use_color_default = -1;
26 static const char *external_diff_cmd_cfg;
27 int diff_auto_refresh_index = 1;
28 static int diff_mnemonic_prefix;
29
30 static char diff_colors[][COLOR_MAXLEN] = {
31         "\033[m",       /* reset */
32         "",             /* PLAIN (normal) */
33         "\033[1m",      /* METAINFO (bold) */
34         "\033[36m",     /* FRAGINFO (cyan) */
35         "\033[31m",     /* OLD (red) */
36         "\033[32m",     /* NEW (green) */
37         "\033[33m",     /* COMMIT (yellow) */
38         "\033[41m",     /* WHITESPACE (red background) */
39 };
40
41 static void diff_filespec_load_driver(struct diff_filespec *one);
42 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
43
44 static int parse_diff_color_slot(const char *var, int ofs)
45 {
46         if (!strcasecmp(var+ofs, "plain"))
47                 return DIFF_PLAIN;
48         if (!strcasecmp(var+ofs, "meta"))
49                 return DIFF_METAINFO;
50         if (!strcasecmp(var+ofs, "frag"))
51                 return DIFF_FRAGINFO;
52         if (!strcasecmp(var+ofs, "old"))
53                 return DIFF_FILE_OLD;
54         if (!strcasecmp(var+ofs, "new"))
55                 return DIFF_FILE_NEW;
56         if (!strcasecmp(var+ofs, "commit"))
57                 return DIFF_COMMIT;
58         if (!strcasecmp(var+ofs, "whitespace"))
59                 return DIFF_WHITESPACE;
60         die("bad config variable '%s'", var);
61 }
62
63 /*
64  * These are to give UI layer defaults.
65  * The core-level commands such as git-diff-files should
66  * never be affected by the setting of diff.renames
67  * the user happens to have in the configuration file.
68  */
69 int git_diff_ui_config(const char *var, const char *value, void *cb)
70 {
71         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
72                 diff_use_color_default = git_config_colorbool(var, value, -1);
73                 return 0;
74         }
75         if (!strcmp(var, "diff.renames")) {
76                 if (!value)
77                         diff_detect_rename_default = DIFF_DETECT_RENAME;
78                 else if (!strcasecmp(value, "copies") ||
79                          !strcasecmp(value, "copy"))
80                         diff_detect_rename_default = DIFF_DETECT_COPY;
81                 else if (git_config_bool(var,value))
82                         diff_detect_rename_default = DIFF_DETECT_RENAME;
83                 return 0;
84         }
85         if (!strcmp(var, "diff.autorefreshindex")) {
86                 diff_auto_refresh_index = git_config_bool(var, value);
87                 return 0;
88         }
89         if (!strcmp(var, "diff.mnemonicprefix")) {
90                 diff_mnemonic_prefix = git_config_bool(var, value);
91                 return 0;
92         }
93         if (!strcmp(var, "diff.external"))
94                 return git_config_string(&external_diff_cmd_cfg, var, value);
95
96         switch (userdiff_config_porcelain(var, value)) {
97                 case 0: break;
98                 case -1: return -1;
99                 default: return 0;
100         }
101
102         return git_diff_basic_config(var, value, cb);
103 }
104
105 int git_diff_basic_config(const char *var, const char *value, void *cb)
106 {
107         if (!strcmp(var, "diff.renamelimit")) {
108                 diff_rename_limit_default = git_config_int(var, value);
109                 return 0;
110         }
111
112         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
113                 int slot = parse_diff_color_slot(var, 11);
114                 if (!value)
115                         return config_error_nonbool(var);
116                 color_parse(value, var, diff_colors[slot]);
117                 return 0;
118         }
119
120         /* like GNU diff's --suppress-blank-empty option  */
121         if (!strcmp(var, "diff.suppress-blank-empty")) {
122                 diff_suppress_blank_empty = git_config_bool(var, value);
123                 return 0;
124         }
125
126         switch (userdiff_config_basic(var, value)) {
127                 case 0: break;
128                 case -1: return -1;
129                 default: return 0;
130         }
131
132         return git_color_default_config(var, value, cb);
133 }
134
135 static char *quote_two(const char *one, const char *two)
136 {
137         int need_one = quote_c_style(one, NULL, NULL, 1);
138         int need_two = quote_c_style(two, NULL, NULL, 1);
139         struct strbuf res = STRBUF_INIT;
140
141         if (need_one + need_two) {
142                 strbuf_addch(&res, '"');
143                 quote_c_style(one, &res, NULL, 1);
144                 quote_c_style(two, &res, NULL, 1);
145                 strbuf_addch(&res, '"');
146         } else {
147                 strbuf_addstr(&res, one);
148                 strbuf_addstr(&res, two);
149         }
150         return strbuf_detach(&res, NULL);
151 }
152
153 static const char *external_diff(void)
154 {
155         static const char *external_diff_cmd = NULL;
156         static int done_preparing = 0;
157
158         if (done_preparing)
159                 return external_diff_cmd;
160         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
161         if (!external_diff_cmd)
162                 external_diff_cmd = external_diff_cmd_cfg;
163         done_preparing = 1;
164         return external_diff_cmd;
165 }
166
167 static struct diff_tempfile {
168         const char *name; /* filename external diff should read from */
169         char hex[41];
170         char mode[10];
171         char tmp_path[PATH_MAX];
172 } diff_temp[2];
173
174 static int count_lines(const char *data, int size)
175 {
176         int count, ch, completely_empty = 1, nl_just_seen = 0;
177         count = 0;
178         while (0 < size--) {
179                 ch = *data++;
180                 if (ch == '\n') {
181                         count++;
182                         nl_just_seen = 1;
183                         completely_empty = 0;
184                 }
185                 else {
186                         nl_just_seen = 0;
187                         completely_empty = 0;
188                 }
189         }
190         if (completely_empty)
191                 return 0;
192         if (!nl_just_seen)
193                 count++; /* no trailing newline */
194         return count;
195 }
196
197 static void print_line_count(FILE *file, int count)
198 {
199         switch (count) {
200         case 0:
201                 fprintf(file, "0,0");
202                 break;
203         case 1:
204                 fprintf(file, "1");
205                 break;
206         default:
207                 fprintf(file, "1,%d", count);
208                 break;
209         }
210 }
211
212 static void copy_file_with_prefix(FILE *file,
213                                   int prefix, const char *data, int size,
214                                   const char *set, const char *reset)
215 {
216         int ch, nl_just_seen = 1;
217         while (0 < size--) {
218                 ch = *data++;
219                 if (nl_just_seen) {
220                         fputs(set, file);
221                         putc(prefix, file);
222                 }
223                 if (ch == '\n') {
224                         nl_just_seen = 1;
225                         fputs(reset, file);
226                 } else
227                         nl_just_seen = 0;
228                 putc(ch, file);
229         }
230         if (!nl_just_seen)
231                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
232 }
233
234 static void emit_rewrite_diff(const char *name_a,
235                               const char *name_b,
236                               struct diff_filespec *one,
237                               struct diff_filespec *two,
238                               struct diff_options *o)
239 {
240         int lc_a, lc_b;
241         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
242         const char *name_a_tab, *name_b_tab;
243         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
244         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
245         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
246         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
247         const char *reset = diff_get_color(color_diff, DIFF_RESET);
248         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
249         const char *a_prefix, *b_prefix;
250
251         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
252                 a_prefix = o->b_prefix;
253                 b_prefix = o->a_prefix;
254         } else {
255                 a_prefix = o->a_prefix;
256                 b_prefix = o->b_prefix;
257         }
258
259         name_a += (*name_a == '/');
260         name_b += (*name_b == '/');
261         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
262         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
263
264         strbuf_reset(&a_name);
265         strbuf_reset(&b_name);
266         quote_two_c_style(&a_name, a_prefix, name_a, 0);
267         quote_two_c_style(&b_name, b_prefix, name_b, 0);
268
269         diff_populate_filespec(one, 0);
270         diff_populate_filespec(two, 0);
271         lc_a = count_lines(one->data, one->size);
272         lc_b = count_lines(two->data, two->size);
273         fprintf(o->file,
274                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
275                 metainfo, a_name.buf, name_a_tab, reset,
276                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
277         print_line_count(o->file, lc_a);
278         fprintf(o->file, " +");
279         print_line_count(o->file, lc_b);
280         fprintf(o->file, " @@%s\n", reset);
281         if (lc_a)
282                 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
283         if (lc_b)
284                 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
285 }
286
287 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
288 {
289         if (!DIFF_FILE_VALID(one)) {
290                 mf->ptr = (char *)""; /* does not matter */
291                 mf->size = 0;
292                 return 0;
293         }
294         else if (diff_populate_filespec(one, 0))
295                 return -1;
296
297         diff_filespec_load_driver(one);
298         if (one->driver->textconv) {
299                 size_t size;
300                 mf->ptr = run_textconv(one->driver->textconv, one, &size);
301                 if (!mf->ptr)
302                         return -1;
303                 mf->size = size;
304         }
305         else {
306                 mf->ptr = one->data;
307                 mf->size = one->size;
308         }
309         return 0;
310 }
311
312 struct diff_words_buffer {
313         mmfile_t text;
314         long alloc;
315         long current; /* output pointer */
316         int suppressed_newline;
317 };
318
319 static void diff_words_append(char *line, unsigned long len,
320                 struct diff_words_buffer *buffer)
321 {
322         if (buffer->text.size + len > buffer->alloc) {
323                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
324                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
325         }
326         line++;
327         len--;
328         memcpy(buffer->text.ptr + buffer->text.size, line, len);
329         buffer->text.size += len;
330 }
331
332 struct diff_words_data {
333         struct diff_words_buffer minus, plus;
334         FILE *file;
335 };
336
337 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
338                 int suppress_newline)
339 {
340         const char *ptr;
341         int eol = 0;
342
343         if (len == 0)
344                 return;
345
346         ptr  = buffer->text.ptr + buffer->current;
347         buffer->current += len;
348
349         if (ptr[len - 1] == '\n') {
350                 eol = 1;
351                 len--;
352         }
353
354         fputs(diff_get_color(1, color), file);
355         fwrite(ptr, len, 1, file);
356         fputs(diff_get_color(1, DIFF_RESET), file);
357
358         if (eol) {
359                 if (suppress_newline)
360                         buffer->suppressed_newline = 1;
361                 else
362                         putc('\n', file);
363         }
364 }
365
366 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
367 {
368         struct diff_words_data *diff_words = priv;
369
370         if (diff_words->minus.suppressed_newline) {
371                 if (line[0] != '+')
372                         putc('\n', diff_words->file);
373                 diff_words->minus.suppressed_newline = 0;
374         }
375
376         len--;
377         switch (line[0]) {
378                 case '-':
379                         print_word(diff_words->file,
380                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
381                         break;
382                 case '+':
383                         print_word(diff_words->file,
384                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
385                         break;
386                 case ' ':
387                         print_word(diff_words->file,
388                                    &diff_words->plus, len, DIFF_PLAIN, 0);
389                         diff_words->minus.current += len;
390                         break;
391         }
392 }
393
394 /* this executes the word diff on the accumulated buffers */
395 static void diff_words_show(struct diff_words_data *diff_words)
396 {
397         xpparam_t xpp;
398         xdemitconf_t xecfg;
399         xdemitcb_t ecb;
400         mmfile_t minus, plus;
401         int i;
402
403         memset(&xecfg, 0, sizeof(xecfg));
404         minus.size = diff_words->minus.text.size;
405         minus.ptr = xmalloc(minus.size);
406         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
407         for (i = 0; i < minus.size; i++)
408                 if (isspace(minus.ptr[i]))
409                         minus.ptr[i] = '\n';
410         diff_words->minus.current = 0;
411
412         plus.size = diff_words->plus.text.size;
413         plus.ptr = xmalloc(plus.size);
414         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
415         for (i = 0; i < plus.size; i++)
416                 if (isspace(plus.ptr[i]))
417                         plus.ptr[i] = '\n';
418         diff_words->plus.current = 0;
419
420         xpp.flags = XDF_NEED_MINIMAL;
421         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
422         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
423                       &xpp, &xecfg, &ecb);
424         free(minus.ptr);
425         free(plus.ptr);
426         diff_words->minus.text.size = diff_words->plus.text.size = 0;
427
428         if (diff_words->minus.suppressed_newline) {
429                 putc('\n', diff_words->file);
430                 diff_words->minus.suppressed_newline = 0;
431         }
432 }
433
434 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
435
436 struct emit_callback {
437         int nparents, color_diff;
438         unsigned ws_rule;
439         sane_truncate_fn truncate;
440         const char **label_path;
441         struct diff_words_data *diff_words;
442         int *found_changesp;
443         FILE *file;
444 };
445
446 static void free_diff_words_data(struct emit_callback *ecbdata)
447 {
448         if (ecbdata->diff_words) {
449                 /* flush buffers */
450                 if (ecbdata->diff_words->minus.text.size ||
451                                 ecbdata->diff_words->plus.text.size)
452                         diff_words_show(ecbdata->diff_words);
453
454                 free (ecbdata->diff_words->minus.text.ptr);
455                 free (ecbdata->diff_words->plus.text.ptr);
456                 free(ecbdata->diff_words);
457                 ecbdata->diff_words = NULL;
458         }
459 }
460
461 const char *diff_get_color(int diff_use_color, enum color_diff ix)
462 {
463         if (diff_use_color)
464                 return diff_colors[ix];
465         return "";
466 }
467
468 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
469 {
470         int has_trailing_newline, has_trailing_carriage_return;
471
472         has_trailing_newline = (len > 0 && line[len-1] == '\n');
473         if (has_trailing_newline)
474                 len--;
475         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
476         if (has_trailing_carriage_return)
477                 len--;
478
479         fputs(set, file);
480         fwrite(line, len, 1, file);
481         fputs(reset, file);
482         if (has_trailing_carriage_return)
483                 fputc('\r', file);
484         if (has_trailing_newline)
485                 fputc('\n', file);
486 }
487
488 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
489 {
490         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
491         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
492
493         if (!*ws)
494                 emit_line(ecbdata->file, set, reset, line, len);
495         else {
496                 /* Emit just the prefix, then the rest. */
497                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
498                 ws_check_emit(line + ecbdata->nparents,
499                               len - ecbdata->nparents, ecbdata->ws_rule,
500                               ecbdata->file, set, reset, ws);
501         }
502 }
503
504 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
505 {
506         const char *cp;
507         unsigned long allot;
508         size_t l = len;
509
510         if (ecb->truncate)
511                 return ecb->truncate(line, len);
512         cp = line;
513         allot = l;
514         while (0 < l) {
515                 (void) utf8_width(&cp, &l);
516                 if (!cp)
517                         break; /* truncated in the middle? */
518         }
519         return allot - l;
520 }
521
522 static void fn_out_consume(void *priv, char *line, unsigned long len)
523 {
524         int i;
525         int color;
526         struct emit_callback *ecbdata = priv;
527         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
528         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
529         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
530
531         *(ecbdata->found_changesp) = 1;
532
533         if (ecbdata->label_path[0]) {
534                 const char *name_a_tab, *name_b_tab;
535
536                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
537                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
538
539                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
540                         meta, ecbdata->label_path[0], reset, name_a_tab);
541                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
542                         meta, ecbdata->label_path[1], reset, name_b_tab);
543                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
544         }
545
546         if (diff_suppress_blank_empty
547             && len == 2 && line[0] == ' ' && line[1] == '\n') {
548                 line[0] = '\n';
549                 len = 1;
550         }
551
552         /* This is not really necessary for now because
553          * this codepath only deals with two-way diffs.
554          */
555         for (i = 0; i < len && line[i] == '@'; i++)
556                 ;
557         if (2 <= i && i < len && line[i] == ' ') {
558                 ecbdata->nparents = i - 1;
559                 len = sane_truncate_line(ecbdata, line, len);
560                 emit_line(ecbdata->file,
561                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
562                           reset, line, len);
563                 if (line[len-1] != '\n')
564                         putc('\n', ecbdata->file);
565                 return;
566         }
567
568         if (len < ecbdata->nparents) {
569                 emit_line(ecbdata->file, reset, reset, line, len);
570                 return;
571         }
572
573         color = DIFF_PLAIN;
574         if (ecbdata->diff_words && ecbdata->nparents != 1)
575                 /* fall back to normal diff */
576                 free_diff_words_data(ecbdata);
577         if (ecbdata->diff_words) {
578                 if (line[0] == '-') {
579                         diff_words_append(line, len,
580                                           &ecbdata->diff_words->minus);
581                         return;
582                 } else if (line[0] == '+') {
583                         diff_words_append(line, len,
584                                           &ecbdata->diff_words->plus);
585                         return;
586                 }
587                 if (ecbdata->diff_words->minus.text.size ||
588                     ecbdata->diff_words->plus.text.size)
589                         diff_words_show(ecbdata->diff_words);
590                 line++;
591                 len--;
592                 emit_line(ecbdata->file, plain, reset, line, len);
593                 return;
594         }
595         for (i = 0; i < ecbdata->nparents && len; i++) {
596                 if (line[i] == '-')
597                         color = DIFF_FILE_OLD;
598                 else if (line[i] == '+')
599                         color = DIFF_FILE_NEW;
600         }
601
602         if (color != DIFF_FILE_NEW) {
603                 emit_line(ecbdata->file,
604                           diff_get_color(ecbdata->color_diff, color),
605                           reset, line, len);
606                 return;
607         }
608         emit_add_line(reset, ecbdata, line, len);
609 }
610
611 static char *pprint_rename(const char *a, const char *b)
612 {
613         const char *old = a;
614         const char *new = b;
615         struct strbuf name = STRBUF_INIT;
616         int pfx_length, sfx_length;
617         int len_a = strlen(a);
618         int len_b = strlen(b);
619         int a_midlen, b_midlen;
620         int qlen_a = quote_c_style(a, NULL, NULL, 0);
621         int qlen_b = quote_c_style(b, NULL, NULL, 0);
622
623         if (qlen_a || qlen_b) {
624                 quote_c_style(a, &name, NULL, 0);
625                 strbuf_addstr(&name, " => ");
626                 quote_c_style(b, &name, NULL, 0);
627                 return strbuf_detach(&name, NULL);
628         }
629
630         /* Find common prefix */
631         pfx_length = 0;
632         while (*old && *new && *old == *new) {
633                 if (*old == '/')
634                         pfx_length = old - a + 1;
635                 old++;
636                 new++;
637         }
638
639         /* Find common suffix */
640         old = a + len_a;
641         new = b + len_b;
642         sfx_length = 0;
643         while (a <= old && b <= new && *old == *new) {
644                 if (*old == '/')
645                         sfx_length = len_a - (old - a);
646                 old--;
647                 new--;
648         }
649
650         /*
651          * pfx{mid-a => mid-b}sfx
652          * {pfx-a => pfx-b}sfx
653          * pfx{sfx-a => sfx-b}
654          * name-a => name-b
655          */
656         a_midlen = len_a - pfx_length - sfx_length;
657         b_midlen = len_b - pfx_length - sfx_length;
658         if (a_midlen < 0)
659                 a_midlen = 0;
660         if (b_midlen < 0)
661                 b_midlen = 0;
662
663         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
664         if (pfx_length + sfx_length) {
665                 strbuf_add(&name, a, pfx_length);
666                 strbuf_addch(&name, '{');
667         }
668         strbuf_add(&name, a + pfx_length, a_midlen);
669         strbuf_addstr(&name, " => ");
670         strbuf_add(&name, b + pfx_length, b_midlen);
671         if (pfx_length + sfx_length) {
672                 strbuf_addch(&name, '}');
673                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
674         }
675         return strbuf_detach(&name, NULL);
676 }
677
678 struct diffstat_t {
679         int nr;
680         int alloc;
681         struct diffstat_file {
682                 char *from_name;
683                 char *name;
684                 char *print_name;
685                 unsigned is_unmerged:1;
686                 unsigned is_binary:1;
687                 unsigned is_renamed:1;
688                 unsigned int added, deleted;
689         } **files;
690 };
691
692 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
693                                           const char *name_a,
694                                           const char *name_b)
695 {
696         struct diffstat_file *x;
697         x = xcalloc(sizeof (*x), 1);
698         if (diffstat->nr == diffstat->alloc) {
699                 diffstat->alloc = alloc_nr(diffstat->alloc);
700                 diffstat->files = xrealloc(diffstat->files,
701                                 diffstat->alloc * sizeof(x));
702         }
703         diffstat->files[diffstat->nr++] = x;
704         if (name_b) {
705                 x->from_name = xstrdup(name_a);
706                 x->name = xstrdup(name_b);
707                 x->is_renamed = 1;
708         }
709         else {
710                 x->from_name = NULL;
711                 x->name = xstrdup(name_a);
712         }
713         return x;
714 }
715
716 static void diffstat_consume(void *priv, char *line, unsigned long len)
717 {
718         struct diffstat_t *diffstat = priv;
719         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
720
721         if (line[0] == '+')
722                 x->added++;
723         else if (line[0] == '-')
724                 x->deleted++;
725 }
726
727 const char mime_boundary_leader[] = "------------";
728
729 static int scale_linear(int it, int width, int max_change)
730 {
731         /*
732          * make sure that at least one '-' is printed if there were deletions,
733          * and likewise for '+'.
734          */
735         if (max_change < 2)
736                 return it;
737         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
738 }
739
740 static void show_name(FILE *file,
741                       const char *prefix, const char *name, int len,
742                       const char *reset, const char *set)
743 {
744         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
745 }
746
747 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
748 {
749         if (cnt <= 0)
750                 return;
751         fprintf(file, "%s", set);
752         while (cnt--)
753                 putc(ch, file);
754         fprintf(file, "%s", reset);
755 }
756
757 static void fill_print_name(struct diffstat_file *file)
758 {
759         char *pname;
760
761         if (file->print_name)
762                 return;
763
764         if (!file->is_renamed) {
765                 struct strbuf buf = STRBUF_INIT;
766                 if (quote_c_style(file->name, &buf, NULL, 0)) {
767                         pname = strbuf_detach(&buf, NULL);
768                 } else {
769                         pname = file->name;
770                         strbuf_release(&buf);
771                 }
772         } else {
773                 pname = pprint_rename(file->from_name, file->name);
774         }
775         file->print_name = pname;
776 }
777
778 static void show_stats(struct diffstat_t* data, struct diff_options *options)
779 {
780         int i, len, add, del, total, adds = 0, dels = 0;
781         int max_change = 0, max_len = 0;
782         int total_files = data->nr;
783         int width, name_width;
784         const char *reset, *set, *add_c, *del_c;
785
786         if (data->nr == 0)
787                 return;
788
789         width = options->stat_width ? options->stat_width : 80;
790         name_width = options->stat_name_width ? options->stat_name_width : 50;
791
792         /* Sanity: give at least 5 columns to the graph,
793          * but leave at least 10 columns for the name.
794          */
795         if (width < 25)
796                 width = 25;
797         if (name_width < 10)
798                 name_width = 10;
799         else if (width < name_width + 15)
800                 name_width = width - 15;
801
802         /* Find the longest filename and max number of changes */
803         reset = diff_get_color_opt(options, DIFF_RESET);
804         set   = diff_get_color_opt(options, DIFF_PLAIN);
805         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
806         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
807
808         for (i = 0; i < data->nr; i++) {
809                 struct diffstat_file *file = data->files[i];
810                 int change = file->added + file->deleted;
811                 fill_print_name(file);
812                 len = strlen(file->print_name);
813                 if (max_len < len)
814                         max_len = len;
815
816                 if (file->is_binary || file->is_unmerged)
817                         continue;
818                 if (max_change < change)
819                         max_change = change;
820         }
821
822         /* Compute the width of the graph part;
823          * 10 is for one blank at the beginning of the line plus
824          * " | count " between the name and the graph.
825          *
826          * From here on, name_width is the width of the name area,
827          * and width is the width of the graph area.
828          */
829         name_width = (name_width < max_len) ? name_width : max_len;
830         if (width < (name_width + 10) + max_change)
831                 width = width - (name_width + 10);
832         else
833                 width = max_change;
834
835         for (i = 0; i < data->nr; i++) {
836                 const char *prefix = "";
837                 char *name = data->files[i]->print_name;
838                 int added = data->files[i]->added;
839                 int deleted = data->files[i]->deleted;
840                 int name_len;
841
842                 /*
843                  * "scale" the filename
844                  */
845                 len = name_width;
846                 name_len = strlen(name);
847                 if (name_width < name_len) {
848                         char *slash;
849                         prefix = "...";
850                         len -= 3;
851                         name += name_len - len;
852                         slash = strchr(name, '/');
853                         if (slash)
854                                 name = slash;
855                 }
856
857                 if (data->files[i]->is_binary) {
858                         show_name(options->file, prefix, name, len, reset, set);
859                         fprintf(options->file, "  Bin ");
860                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
861                         fprintf(options->file, " -> ");
862                         fprintf(options->file, "%s%d%s", add_c, added, reset);
863                         fprintf(options->file, " bytes");
864                         fprintf(options->file, "\n");
865                         continue;
866                 }
867                 else if (data->files[i]->is_unmerged) {
868                         show_name(options->file, prefix, name, len, reset, set);
869                         fprintf(options->file, "  Unmerged\n");
870                         continue;
871                 }
872                 else if (!data->files[i]->is_renamed &&
873                          (added + deleted == 0)) {
874                         total_files--;
875                         continue;
876                 }
877
878                 /*
879                  * scale the add/delete
880                  */
881                 add = added;
882                 del = deleted;
883                 total = add + del;
884                 adds += add;
885                 dels += del;
886
887                 if (width <= max_change) {
888                         add = scale_linear(add, width, max_change);
889                         del = scale_linear(del, width, max_change);
890                         total = add + del;
891                 }
892                 show_name(options->file, prefix, name, len, reset, set);
893                 fprintf(options->file, "%5d%s", added + deleted,
894                                 added + deleted ? " " : "");
895                 show_graph(options->file, '+', add, add_c, reset);
896                 show_graph(options->file, '-', del, del_c, reset);
897                 fprintf(options->file, "\n");
898         }
899         fprintf(options->file,
900                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
901                set, total_files, adds, dels, reset);
902 }
903
904 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
905 {
906         int i, adds = 0, dels = 0, total_files = data->nr;
907
908         if (data->nr == 0)
909                 return;
910
911         for (i = 0; i < data->nr; i++) {
912                 if (!data->files[i]->is_binary &&
913                     !data->files[i]->is_unmerged) {
914                         int added = data->files[i]->added;
915                         int deleted= data->files[i]->deleted;
916                         if (!data->files[i]->is_renamed &&
917                             (added + deleted == 0)) {
918                                 total_files--;
919                         } else {
920                                 adds += added;
921                                 dels += deleted;
922                         }
923                 }
924         }
925         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
926                total_files, adds, dels);
927 }
928
929 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
930 {
931         int i;
932
933         if (data->nr == 0)
934                 return;
935
936         for (i = 0; i < data->nr; i++) {
937                 struct diffstat_file *file = data->files[i];
938
939                 if (file->is_binary)
940                         fprintf(options->file, "-\t-\t");
941                 else
942                         fprintf(options->file,
943                                 "%d\t%d\t", file->added, file->deleted);
944                 if (options->line_termination) {
945                         fill_print_name(file);
946                         if (!file->is_renamed)
947                                 write_name_quoted(file->name, options->file,
948                                                   options->line_termination);
949                         else {
950                                 fputs(file->print_name, options->file);
951                                 putc(options->line_termination, options->file);
952                         }
953                 } else {
954                         if (file->is_renamed) {
955                                 putc('\0', options->file);
956                                 write_name_quoted(file->from_name, options->file, '\0');
957                         }
958                         write_name_quoted(file->name, options->file, '\0');
959                 }
960         }
961 }
962
963 struct dirstat_file {
964         const char *name;
965         unsigned long changed;
966 };
967
968 struct dirstat_dir {
969         struct dirstat_file *files;
970         int alloc, nr, percent, cumulative;
971 };
972
973 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
974 {
975         unsigned long this_dir = 0;
976         unsigned int sources = 0;
977
978         while (dir->nr) {
979                 struct dirstat_file *f = dir->files;
980                 int namelen = strlen(f->name);
981                 unsigned long this;
982                 char *slash;
983
984                 if (namelen < baselen)
985                         break;
986                 if (memcmp(f->name, base, baselen))
987                         break;
988                 slash = strchr(f->name + baselen, '/');
989                 if (slash) {
990                         int newbaselen = slash + 1 - f->name;
991                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
992                         sources++;
993                 } else {
994                         this = f->changed;
995                         dir->files++;
996                         dir->nr--;
997                         sources += 2;
998                 }
999                 this_dir += this;
1000         }
1001
1002         /*
1003          * We don't report dirstat's for
1004          *  - the top level
1005          *  - or cases where everything came from a single directory
1006          *    under this directory (sources == 1).
1007          */
1008         if (baselen && sources != 1) {
1009                 int permille = this_dir * 1000 / changed;
1010                 if (permille) {
1011                         int percent = permille / 10;
1012                         if (percent >= dir->percent) {
1013                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1014                                 if (!dir->cumulative)
1015                                         return 0;
1016                         }
1017                 }
1018         }
1019         return this_dir;
1020 }
1021
1022 static int dirstat_compare(const void *_a, const void *_b)
1023 {
1024         const struct dirstat_file *a = _a;
1025         const struct dirstat_file *b = _b;
1026         return strcmp(a->name, b->name);
1027 }
1028
1029 static void show_dirstat(struct diff_options *options)
1030 {
1031         int i;
1032         unsigned long changed;
1033         struct dirstat_dir dir;
1034         struct diff_queue_struct *q = &diff_queued_diff;
1035
1036         dir.files = NULL;
1037         dir.alloc = 0;
1038         dir.nr = 0;
1039         dir.percent = options->dirstat_percent;
1040         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1041
1042         changed = 0;
1043         for (i = 0; i < q->nr; i++) {
1044                 struct diff_filepair *p = q->queue[i];
1045                 const char *name;
1046                 unsigned long copied, added, damage;
1047
1048                 name = p->one->path ? p->one->path : p->two->path;
1049
1050                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1051                         diff_populate_filespec(p->one, 0);
1052                         diff_populate_filespec(p->two, 0);
1053                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1054                                                &copied, &added);
1055                         diff_free_filespec_data(p->one);
1056                         diff_free_filespec_data(p->two);
1057                 } else if (DIFF_FILE_VALID(p->one)) {
1058                         diff_populate_filespec(p->one, 1);
1059                         copied = added = 0;
1060                         diff_free_filespec_data(p->one);
1061                 } else if (DIFF_FILE_VALID(p->two)) {
1062                         diff_populate_filespec(p->two, 1);
1063                         copied = 0;
1064                         added = p->two->size;
1065                         diff_free_filespec_data(p->two);
1066                 } else
1067                         continue;
1068
1069                 /*
1070                  * Original minus copied is the removed material,
1071                  * added is the new material.  They are both damages
1072                  * made to the preimage. In --dirstat-by-file mode, count
1073                  * damaged files, not damaged lines. This is done by
1074                  * counting only a single damaged line per file.
1075                  */
1076                 damage = (p->one->size - copied) + added;
1077                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1078                         damage = 1;
1079
1080                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1081                 dir.files[dir.nr].name = name;
1082                 dir.files[dir.nr].changed = damage;
1083                 changed += damage;
1084                 dir.nr++;
1085         }
1086
1087         /* This can happen even with many files, if everything was renames */
1088         if (!changed)
1089                 return;
1090
1091         /* Show all directories with more than x% of the changes */
1092         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1093         gather_dirstat(options->file, &dir, changed, "", 0);
1094 }
1095
1096 static void free_diffstat_info(struct diffstat_t *diffstat)
1097 {
1098         int i;
1099         for (i = 0; i < diffstat->nr; i++) {
1100                 struct diffstat_file *f = diffstat->files[i];
1101                 if (f->name != f->print_name)
1102                         free(f->print_name);
1103                 free(f->name);
1104                 free(f->from_name);
1105                 free(f);
1106         }
1107         free(diffstat->files);
1108 }
1109
1110 struct checkdiff_t {
1111         const char *filename;
1112         int lineno;
1113         struct diff_options *o;
1114         unsigned ws_rule;
1115         unsigned status;
1116         int trailing_blanks_start;
1117 };
1118
1119 static int is_conflict_marker(const char *line, unsigned long len)
1120 {
1121         char firstchar;
1122         int cnt;
1123
1124         if (len < 8)
1125                 return 0;
1126         firstchar = line[0];
1127         switch (firstchar) {
1128         case '=': case '>': case '<':
1129                 break;
1130         default:
1131                 return 0;
1132         }
1133         for (cnt = 1; cnt < 7; cnt++)
1134                 if (line[cnt] != firstchar)
1135                         return 0;
1136         /* line[0] thru line[6] are same as firstchar */
1137         if (firstchar == '=') {
1138                 /* divider between ours and theirs? */
1139                 if (len != 8 || line[7] != '\n')
1140                         return 0;
1141         } else if (len < 8 || !isspace(line[7])) {
1142                 /* not divider before ours nor after theirs */
1143                 return 0;
1144         }
1145         return 1;
1146 }
1147
1148 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1149 {
1150         struct checkdiff_t *data = priv;
1151         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1152         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1153         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1154         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1155         char *err;
1156
1157         if (line[0] == '+') {
1158                 unsigned bad;
1159                 data->lineno++;
1160                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1161                         data->trailing_blanks_start = 0;
1162                 else if (!data->trailing_blanks_start)
1163                         data->trailing_blanks_start = data->lineno;
1164                 if (is_conflict_marker(line + 1, len - 1)) {
1165                         data->status |= 1;
1166                         fprintf(data->o->file,
1167                                 "%s:%d: leftover conflict marker\n",
1168                                 data->filename, data->lineno);
1169                 }
1170                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1171                 if (!bad)
1172                         return;
1173                 data->status |= bad;
1174                 err = whitespace_error_string(bad);
1175                 fprintf(data->o->file, "%s:%d: %s.\n",
1176                         data->filename, data->lineno, err);
1177                 free(err);
1178                 emit_line(data->o->file, set, reset, line, 1);
1179                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1180                               data->o->file, set, reset, ws);
1181         } else if (line[0] == ' ') {
1182                 data->lineno++;
1183                 data->trailing_blanks_start = 0;
1184         } else if (line[0] == '@') {
1185                 char *plus = strchr(line, '+');
1186                 if (plus)
1187                         data->lineno = strtol(plus, NULL, 10) - 1;
1188                 else
1189                         die("invalid diff");
1190                 data->trailing_blanks_start = 0;
1191         }
1192 }
1193
1194 static unsigned char *deflate_it(char *data,
1195                                  unsigned long size,
1196                                  unsigned long *result_size)
1197 {
1198         int bound;
1199         unsigned char *deflated;
1200         z_stream stream;
1201
1202         memset(&stream, 0, sizeof(stream));
1203         deflateInit(&stream, zlib_compression_level);
1204         bound = deflateBound(&stream, size);
1205         deflated = xmalloc(bound);
1206         stream.next_out = deflated;
1207         stream.avail_out = bound;
1208
1209         stream.next_in = (unsigned char *)data;
1210         stream.avail_in = size;
1211         while (deflate(&stream, Z_FINISH) == Z_OK)
1212                 ; /* nothing */
1213         deflateEnd(&stream);
1214         *result_size = stream.total_out;
1215         return deflated;
1216 }
1217
1218 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1219 {
1220         void *cp;
1221         void *delta;
1222         void *deflated;
1223         void *data;
1224         unsigned long orig_size;
1225         unsigned long delta_size;
1226         unsigned long deflate_size;
1227         unsigned long data_size;
1228
1229         /* We could do deflated delta, or we could do just deflated two,
1230          * whichever is smaller.
1231          */
1232         delta = NULL;
1233         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1234         if (one->size && two->size) {
1235                 delta = diff_delta(one->ptr, one->size,
1236                                    two->ptr, two->size,
1237                                    &delta_size, deflate_size);
1238                 if (delta) {
1239                         void *to_free = delta;
1240                         orig_size = delta_size;
1241                         delta = deflate_it(delta, delta_size, &delta_size);
1242                         free(to_free);
1243                 }
1244         }
1245
1246         if (delta && delta_size < deflate_size) {
1247                 fprintf(file, "delta %lu\n", orig_size);
1248                 free(deflated);
1249                 data = delta;
1250                 data_size = delta_size;
1251         }
1252         else {
1253                 fprintf(file, "literal %lu\n", two->size);
1254                 free(delta);
1255                 data = deflated;
1256                 data_size = deflate_size;
1257         }
1258
1259         /* emit data encoded in base85 */
1260         cp = data;
1261         while (data_size) {
1262                 int bytes = (52 < data_size) ? 52 : data_size;
1263                 char line[70];
1264                 data_size -= bytes;
1265                 if (bytes <= 26)
1266                         line[0] = bytes + 'A' - 1;
1267                 else
1268                         line[0] = bytes - 26 + 'a' - 1;
1269                 encode_85(line + 1, cp, bytes);
1270                 cp = (char *) cp + bytes;
1271                 fputs(line, file);
1272                 fputc('\n', file);
1273         }
1274         fprintf(file, "\n");
1275         free(data);
1276 }
1277
1278 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1279 {
1280         fprintf(file, "GIT binary patch\n");
1281         emit_binary_diff_body(file, one, two);
1282         emit_binary_diff_body(file, two, one);
1283 }
1284
1285 void diff_filespec_load_driver(struct diff_filespec *one)
1286 {
1287         if (!one->driver)
1288                 one->driver = userdiff_find_by_path(one->path);
1289         if (!one->driver)
1290                 one->driver = userdiff_find_by_name("default");
1291 }
1292
1293 int diff_filespec_is_binary(struct diff_filespec *one)
1294 {
1295         if (one->is_binary == -1) {
1296                 diff_filespec_load_driver(one);
1297                 if (one->driver->binary != -1)
1298                         one->is_binary = one->driver->binary;
1299                 else {
1300                         if (!one->data && DIFF_FILE_VALID(one))
1301                                 diff_populate_filespec(one, 0);
1302                         if (one->data)
1303                                 one->is_binary = buffer_is_binary(one->data,
1304                                                 one->size);
1305                         if (one->is_binary == -1)
1306                                 one->is_binary = 0;
1307                 }
1308         }
1309         return one->is_binary;
1310 }
1311
1312 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1313 {
1314         diff_filespec_load_driver(one);
1315         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1316 }
1317
1318 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1319 {
1320         if (!options->a_prefix)
1321                 options->a_prefix = a;
1322         if (!options->b_prefix)
1323                 options->b_prefix = b;
1324 }
1325
1326 static void builtin_diff(const char *name_a,
1327                          const char *name_b,
1328                          struct diff_filespec *one,
1329                          struct diff_filespec *two,
1330                          const char *xfrm_msg,
1331                          struct diff_options *o,
1332                          int complete_rewrite)
1333 {
1334         mmfile_t mf1, mf2;
1335         const char *lbl[2];
1336         char *a_one, *b_two;
1337         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1338         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1339         const char *a_prefix, *b_prefix;
1340
1341         diff_set_mnemonic_prefix(o, "a/", "b/");
1342         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1343                 a_prefix = o->b_prefix;
1344                 b_prefix = o->a_prefix;
1345         } else {
1346                 a_prefix = o->a_prefix;
1347                 b_prefix = o->b_prefix;
1348         }
1349
1350         /* Never use a non-valid filename anywhere if at all possible */
1351         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1352         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1353
1354         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1355         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1356         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1357         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1358         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1359         if (lbl[0][0] == '/') {
1360                 /* /dev/null */
1361                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1362                 if (xfrm_msg && xfrm_msg[0])
1363                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1364         }
1365         else if (lbl[1][0] == '/') {
1366                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1367                 if (xfrm_msg && xfrm_msg[0])
1368                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1369         }
1370         else {
1371                 if (one->mode != two->mode) {
1372                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1373                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1374                 }
1375                 if (xfrm_msg && xfrm_msg[0])
1376                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1377                 /*
1378                  * we do not run diff between different kind
1379                  * of objects.
1380                  */
1381                 if ((one->mode ^ two->mode) & S_IFMT)
1382                         goto free_ab_and_return;
1383                 if (complete_rewrite) {
1384                         emit_rewrite_diff(name_a, name_b, one, two, o);
1385                         o->found_changes = 1;
1386                         goto free_ab_and_return;
1387                 }
1388         }
1389
1390         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1391                 die("unable to read files to diff");
1392
1393         if (!DIFF_OPT_TST(o, TEXT) &&
1394             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1395                 /* Quite common confusing case */
1396                 if (mf1.size == mf2.size &&
1397                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1398                         goto free_ab_and_return;
1399                 if (DIFF_OPT_TST(o, BINARY))
1400                         emit_binary_diff(o->file, &mf1, &mf2);
1401                 else
1402                         fprintf(o->file, "Binary files %s and %s differ\n",
1403                                 lbl[0], lbl[1]);
1404                 o->found_changes = 1;
1405         }
1406         else {
1407                 /* Crazy xdl interfaces.. */
1408                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1409                 xpparam_t xpp;
1410                 xdemitconf_t xecfg;
1411                 xdemitcb_t ecb;
1412                 struct emit_callback ecbdata;
1413                 const struct userdiff_funcname *pe;
1414
1415                 pe = diff_funcname_pattern(one);
1416                 if (!pe)
1417                         pe = diff_funcname_pattern(two);
1418
1419                 memset(&xecfg, 0, sizeof(xecfg));
1420                 memset(&ecbdata, 0, sizeof(ecbdata));
1421                 ecbdata.label_path = lbl;
1422                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1423                 ecbdata.found_changesp = &o->found_changes;
1424                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1425                 ecbdata.file = o->file;
1426                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1427                 xecfg.ctxlen = o->context;
1428                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1429                 if (pe)
1430                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1431                 if (!diffopts)
1432                         ;
1433                 else if (!prefixcmp(diffopts, "--unified="))
1434                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1435                 else if (!prefixcmp(diffopts, "-u"))
1436                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1437                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1438                         ecbdata.diff_words =
1439                                 xcalloc(1, sizeof(struct diff_words_data));
1440                         ecbdata.diff_words->file = o->file;
1441                 }
1442                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1443                               &xpp, &xecfg, &ecb);
1444                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1445                         free_diff_words_data(&ecbdata);
1446         }
1447
1448  free_ab_and_return:
1449         diff_free_filespec_data(one);
1450         diff_free_filespec_data(two);
1451         free(a_one);
1452         free(b_two);
1453         return;
1454 }
1455
1456 static void builtin_diffstat(const char *name_a, const char *name_b,
1457                              struct diff_filespec *one,
1458                              struct diff_filespec *two,
1459                              struct diffstat_t *diffstat,
1460                              struct diff_options *o,
1461                              int complete_rewrite)
1462 {
1463         mmfile_t mf1, mf2;
1464         struct diffstat_file *data;
1465
1466         data = diffstat_add(diffstat, name_a, name_b);
1467
1468         if (!one || !two) {
1469                 data->is_unmerged = 1;
1470                 return;
1471         }
1472         if (complete_rewrite) {
1473                 diff_populate_filespec(one, 0);
1474                 diff_populate_filespec(two, 0);
1475                 data->deleted = count_lines(one->data, one->size);
1476                 data->added = count_lines(two->data, two->size);
1477                 goto free_and_return;
1478         }
1479         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1480                 die("unable to read files to diff");
1481
1482         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1483                 data->is_binary = 1;
1484                 data->added = mf2.size;
1485                 data->deleted = mf1.size;
1486         } else {
1487                 /* Crazy xdl interfaces.. */
1488                 xpparam_t xpp;
1489                 xdemitconf_t xecfg;
1490                 xdemitcb_t ecb;
1491
1492                 memset(&xecfg, 0, sizeof(xecfg));
1493                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1494                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1495                               &xpp, &xecfg, &ecb);
1496         }
1497
1498  free_and_return:
1499         diff_free_filespec_data(one);
1500         diff_free_filespec_data(two);
1501 }
1502
1503 static void builtin_checkdiff(const char *name_a, const char *name_b,
1504                               const char *attr_path,
1505                               struct diff_filespec *one,
1506                               struct diff_filespec *two,
1507                               struct diff_options *o)
1508 {
1509         mmfile_t mf1, mf2;
1510         struct checkdiff_t data;
1511
1512         if (!two)
1513                 return;
1514
1515         memset(&data, 0, sizeof(data));
1516         data.filename = name_b ? name_b : name_a;
1517         data.lineno = 0;
1518         data.o = o;
1519         data.ws_rule = whitespace_rule(attr_path);
1520
1521         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1522                 die("unable to read files to diff");
1523
1524         /*
1525          * All the other codepaths check both sides, but not checking
1526          * the "old" side here is deliberate.  We are checking the newly
1527          * introduced changes, and as long as the "new" side is text, we
1528          * can and should check what it introduces.
1529          */
1530         if (diff_filespec_is_binary(two))
1531                 goto free_and_return;
1532         else {
1533                 /* Crazy xdl interfaces.. */
1534                 xpparam_t xpp;
1535                 xdemitconf_t xecfg;
1536                 xdemitcb_t ecb;
1537
1538                 memset(&xecfg, 0, sizeof(xecfg));
1539                 xecfg.ctxlen = 1; /* at least one context line */
1540                 xpp.flags = XDF_NEED_MINIMAL;
1541                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1542                               &xpp, &xecfg, &ecb);
1543
1544                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1545                     data.trailing_blanks_start) {
1546                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1547                                 data.filename, data.trailing_blanks_start);
1548                         data.status = 1; /* report errors */
1549                 }
1550         }
1551  free_and_return:
1552         diff_free_filespec_data(one);
1553         diff_free_filespec_data(two);
1554         if (data.status)
1555                 DIFF_OPT_SET(o, CHECK_FAILED);
1556 }
1557
1558 struct diff_filespec *alloc_filespec(const char *path)
1559 {
1560         int namelen = strlen(path);
1561         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1562
1563         memset(spec, 0, sizeof(*spec));
1564         spec->path = (char *)(spec + 1);
1565         memcpy(spec->path, path, namelen+1);
1566         spec->count = 1;
1567         spec->is_binary = -1;
1568         return spec;
1569 }
1570
1571 void free_filespec(struct diff_filespec *spec)
1572 {
1573         if (!--spec->count) {
1574                 diff_free_filespec_data(spec);
1575                 free(spec);
1576         }
1577 }
1578
1579 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1580                    unsigned short mode)
1581 {
1582         if (mode) {
1583                 spec->mode = canon_mode(mode);
1584                 hashcpy(spec->sha1, sha1);
1585                 spec->sha1_valid = !is_null_sha1(sha1);
1586         }
1587 }
1588
1589 /*
1590  * Given a name and sha1 pair, if the index tells us the file in
1591  * the work tree has that object contents, return true, so that
1592  * prepare_temp_file() does not have to inflate and extract.
1593  */
1594 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1595 {
1596         struct cache_entry *ce;
1597         struct stat st;
1598         int pos, len;
1599
1600         /* We do not read the cache ourselves here, because the
1601          * benchmark with my previous version that always reads cache
1602          * shows that it makes things worse for diff-tree comparing
1603          * two linux-2.6 kernel trees in an already checked out work
1604          * tree.  This is because most diff-tree comparisons deal with
1605          * only a small number of files, while reading the cache is
1606          * expensive for a large project, and its cost outweighs the
1607          * savings we get by not inflating the object to a temporary
1608          * file.  Practically, this code only helps when we are used
1609          * by diff-cache --cached, which does read the cache before
1610          * calling us.
1611          */
1612         if (!active_cache)
1613                 return 0;
1614
1615         /* We want to avoid the working directory if our caller
1616          * doesn't need the data in a normal file, this system
1617          * is rather slow with its stat/open/mmap/close syscalls,
1618          * and the object is contained in a pack file.  The pack
1619          * is probably already open and will be faster to obtain
1620          * the data through than the working directory.  Loose
1621          * objects however would tend to be slower as they need
1622          * to be individually opened and inflated.
1623          */
1624         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1625                 return 0;
1626
1627         len = strlen(name);
1628         pos = cache_name_pos(name, len);
1629         if (pos < 0)
1630                 return 0;
1631         ce = active_cache[pos];
1632
1633         /*
1634          * This is not the sha1 we are looking for, or
1635          * unreusable because it is not a regular file.
1636          */
1637         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1638                 return 0;
1639
1640         /*
1641          * If ce matches the file in the work tree, we can reuse it.
1642          */
1643         if (ce_uptodate(ce) ||
1644             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1645                 return 1;
1646
1647         return 0;
1648 }
1649
1650 static int populate_from_stdin(struct diff_filespec *s)
1651 {
1652         struct strbuf buf = STRBUF_INIT;
1653         size_t size = 0;
1654
1655         if (strbuf_read(&buf, 0, 0) < 0)
1656                 return error("error while reading from stdin %s",
1657                                      strerror(errno));
1658
1659         s->should_munmap = 0;
1660         s->data = strbuf_detach(&buf, &size);
1661         s->size = size;
1662         s->should_free = 1;
1663         return 0;
1664 }
1665
1666 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1667 {
1668         int len;
1669         char *data = xmalloc(100);
1670         len = snprintf(data, 100,
1671                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1672         s->data = data;
1673         s->size = len;
1674         s->should_free = 1;
1675         if (size_only) {
1676                 s->data = NULL;
1677                 free(data);
1678         }
1679         return 0;
1680 }
1681
1682 /*
1683  * While doing rename detection and pickaxe operation, we may need to
1684  * grab the data for the blob (or file) for our own in-core comparison.
1685  * diff_filespec has data and size fields for this purpose.
1686  */
1687 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1688 {
1689         int err = 0;
1690         if (!DIFF_FILE_VALID(s))
1691                 die("internal error: asking to populate invalid file.");
1692         if (S_ISDIR(s->mode))
1693                 return -1;
1694
1695         if (s->data)
1696                 return 0;
1697
1698         if (size_only && 0 < s->size)
1699                 return 0;
1700
1701         if (S_ISGITLINK(s->mode))
1702                 return diff_populate_gitlink(s, size_only);
1703
1704         if (!s->sha1_valid ||
1705             reuse_worktree_file(s->path, s->sha1, 0)) {
1706                 struct strbuf buf = STRBUF_INIT;
1707                 struct stat st;
1708                 int fd;
1709
1710                 if (!strcmp(s->path, "-"))
1711                         return populate_from_stdin(s);
1712
1713                 if (lstat(s->path, &st) < 0) {
1714                         if (errno == ENOENT) {
1715                         err_empty:
1716                                 err = -1;
1717                         empty:
1718                                 s->data = (char *)"";
1719                                 s->size = 0;
1720                                 return err;
1721                         }
1722                 }
1723                 s->size = xsize_t(st.st_size);
1724                 if (!s->size)
1725                         goto empty;
1726                 if (size_only)
1727                         return 0;
1728                 if (S_ISLNK(st.st_mode)) {
1729                         int ret;
1730                         s->data = xmalloc(s->size);
1731                         s->should_free = 1;
1732                         ret = readlink(s->path, s->data, s->size);
1733                         if (ret < 0) {
1734                                 free(s->data);
1735                                 goto err_empty;
1736                         }
1737                         return 0;
1738                 }
1739                 fd = open(s->path, O_RDONLY);
1740                 if (fd < 0)
1741                         goto err_empty;
1742                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1743                 close(fd);
1744                 s->should_munmap = 1;
1745
1746                 /*
1747                  * Convert from working tree format to canonical git format
1748                  */
1749                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1750                         size_t size = 0;
1751                         munmap(s->data, s->size);
1752                         s->should_munmap = 0;
1753                         s->data = strbuf_detach(&buf, &size);
1754                         s->size = size;
1755                         s->should_free = 1;
1756                 }
1757         }
1758         else {
1759                 enum object_type type;
1760                 if (size_only)
1761                         type = sha1_object_info(s->sha1, &s->size);
1762                 else {
1763                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1764                         s->should_free = 1;
1765                 }
1766         }
1767         return 0;
1768 }
1769
1770 void diff_free_filespec_blob(struct diff_filespec *s)
1771 {
1772         if (s->should_free)
1773                 free(s->data);
1774         else if (s->should_munmap)
1775                 munmap(s->data, s->size);
1776
1777         if (s->should_free || s->should_munmap) {
1778                 s->should_free = s->should_munmap = 0;
1779                 s->data = NULL;
1780         }
1781 }
1782
1783 void diff_free_filespec_data(struct diff_filespec *s)
1784 {
1785         diff_free_filespec_blob(s);
1786         free(s->cnt_data);
1787         s->cnt_data = NULL;
1788 }
1789
1790 static void prep_temp_blob(struct diff_tempfile *temp,
1791                            void *blob,
1792                            unsigned long size,
1793                            const unsigned char *sha1,
1794                            int mode)
1795 {
1796         int fd;
1797
1798         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1799         if (fd < 0)
1800                 die("unable to create temp-file: %s", strerror(errno));
1801         if (write_in_full(fd, blob, size) != size)
1802                 die("unable to write temp-file");
1803         close(fd);
1804         temp->name = temp->tmp_path;
1805         strcpy(temp->hex, sha1_to_hex(sha1));
1806         temp->hex[40] = 0;
1807         sprintf(temp->mode, "%06o", mode);
1808 }
1809
1810 static void prepare_temp_file(const char *name,
1811                               struct diff_tempfile *temp,
1812                               struct diff_filespec *one)
1813 {
1814         if (!DIFF_FILE_VALID(one)) {
1815         not_a_valid_file:
1816                 /* A '-' entry produces this for file-2, and
1817                  * a '+' entry produces this for file-1.
1818                  */
1819                 temp->name = "/dev/null";
1820                 strcpy(temp->hex, ".");
1821                 strcpy(temp->mode, ".");
1822                 return;
1823         }
1824
1825         if (!one->sha1_valid ||
1826             reuse_worktree_file(name, one->sha1, 1)) {
1827                 struct stat st;
1828                 if (lstat(name, &st) < 0) {
1829                         if (errno == ENOENT)
1830                                 goto not_a_valid_file;
1831                         die("stat(%s): %s", name, strerror(errno));
1832                 }
1833                 if (S_ISLNK(st.st_mode)) {
1834                         int ret;
1835                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1836                         size_t sz = xsize_t(st.st_size);
1837                         if (sizeof(buf) <= st.st_size)
1838                                 die("symlink too long: %s", name);
1839                         ret = readlink(name, buf, sz);
1840                         if (ret < 0)
1841                                 die("readlink(%s)", name);
1842                         prep_temp_blob(temp, buf, sz,
1843                                        (one->sha1_valid ?
1844                                         one->sha1 : null_sha1),
1845                                        (one->sha1_valid ?
1846                                         one->mode : S_IFLNK));
1847                 }
1848                 else {
1849                         /* we can borrow from the file in the work tree */
1850                         temp->name = name;
1851                         if (!one->sha1_valid)
1852                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1853                         else
1854                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1855                         /* Even though we may sometimes borrow the
1856                          * contents from the work tree, we always want
1857                          * one->mode.  mode is trustworthy even when
1858                          * !(one->sha1_valid), as long as
1859                          * DIFF_FILE_VALID(one).
1860                          */
1861                         sprintf(temp->mode, "%06o", one->mode);
1862                 }
1863                 return;
1864         }
1865         else {
1866                 if (diff_populate_filespec(one, 0))
1867                         die("cannot read data blob for %s", one->path);
1868                 prep_temp_blob(temp, one->data, one->size,
1869                                one->sha1, one->mode);
1870         }
1871 }
1872
1873 static void remove_tempfile(void)
1874 {
1875         int i;
1876
1877         for (i = 0; i < 2; i++)
1878                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1879                         unlink(diff_temp[i].name);
1880                         diff_temp[i].name = NULL;
1881                 }
1882 }
1883
1884 static void remove_tempfile_on_signal(int signo)
1885 {
1886         remove_tempfile();
1887         signal(SIGINT, SIG_DFL);
1888         raise(signo);
1889 }
1890
1891 /* An external diff command takes:
1892  *
1893  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1894  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1895  *
1896  */
1897 static void run_external_diff(const char *pgm,
1898                               const char *name,
1899                               const char *other,
1900                               struct diff_filespec *one,
1901                               struct diff_filespec *two,
1902                               const char *xfrm_msg,
1903                               int complete_rewrite)
1904 {
1905         const char *spawn_arg[10];
1906         struct diff_tempfile *temp = diff_temp;
1907         int retval;
1908         static int atexit_asked = 0;
1909         const char *othername;
1910         const char **arg = &spawn_arg[0];
1911
1912         othername = (other? other : name);
1913         if (one && two) {
1914                 prepare_temp_file(name, &temp[0], one);
1915                 prepare_temp_file(othername, &temp[1], two);
1916                 if (! atexit_asked &&
1917                     (temp[0].name == temp[0].tmp_path ||
1918                      temp[1].name == temp[1].tmp_path)) {
1919                         atexit_asked = 1;
1920                         atexit(remove_tempfile);
1921                 }
1922                 signal(SIGINT, remove_tempfile_on_signal);
1923         }
1924
1925         if (one && two) {
1926                 *arg++ = pgm;
1927                 *arg++ = name;
1928                 *arg++ = temp[0].name;
1929                 *arg++ = temp[0].hex;
1930                 *arg++ = temp[0].mode;
1931                 *arg++ = temp[1].name;
1932                 *arg++ = temp[1].hex;
1933                 *arg++ = temp[1].mode;
1934                 if (other) {
1935                         *arg++ = other;
1936                         *arg++ = xfrm_msg;
1937                 }
1938         } else {
1939                 *arg++ = pgm;
1940                 *arg++ = name;
1941         }
1942         *arg = NULL;
1943         fflush(NULL);
1944         retval = run_command_v_opt(spawn_arg, 0);
1945         remove_tempfile();
1946         if (retval) {
1947                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1948                 exit(1);
1949         }
1950 }
1951
1952 static void run_diff_cmd(const char *pgm,
1953                          const char *name,
1954                          const char *other,
1955                          const char *attr_path,
1956                          struct diff_filespec *one,
1957                          struct diff_filespec *two,
1958                          const char *xfrm_msg,
1959                          struct diff_options *o,
1960                          int complete_rewrite)
1961 {
1962         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
1963                 pgm = NULL;
1964         else {
1965                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
1966                 if (drv && drv->external)
1967                         pgm = drv->external;
1968         }
1969
1970         if (pgm) {
1971                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1972                                   complete_rewrite);
1973                 return;
1974         }
1975         if (one && two)
1976                 builtin_diff(name, other ? other : name,
1977                              one, two, xfrm_msg, o, complete_rewrite);
1978         else
1979                 fprintf(o->file, "* Unmerged path %s\n", name);
1980 }
1981
1982 static void diff_fill_sha1_info(struct diff_filespec *one)
1983 {
1984         if (DIFF_FILE_VALID(one)) {
1985                 if (!one->sha1_valid) {
1986                         struct stat st;
1987                         if (!strcmp(one->path, "-")) {
1988                                 hashcpy(one->sha1, null_sha1);
1989                                 return;
1990                         }
1991                         if (lstat(one->path, &st) < 0)
1992                                 die("stat %s", one->path);
1993                         if (index_path(one->sha1, one->path, &st, 0))
1994                                 die("cannot hash %s\n", one->path);
1995                 }
1996         }
1997         else
1998                 hashclr(one->sha1);
1999 }
2000
2001 static int similarity_index(struct diff_filepair *p)
2002 {
2003         return p->score * 100 / MAX_SCORE;
2004 }
2005
2006 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2007 {
2008         /* Strip the prefix but do not molest /dev/null and absolute paths */
2009         if (*namep && **namep != '/')
2010                 *namep += prefix_length;
2011         if (*otherp && **otherp != '/')
2012                 *otherp += prefix_length;
2013 }
2014
2015 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2016 {
2017         const char *pgm = external_diff();
2018         struct strbuf msg;
2019         char *xfrm_msg;
2020         struct diff_filespec *one = p->one;
2021         struct diff_filespec *two = p->two;
2022         const char *name;
2023         const char *other;
2024         const char *attr_path;
2025         int complete_rewrite = 0;
2026
2027         name  = p->one->path;
2028         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2029         attr_path = name;
2030         if (o->prefix_length)
2031                 strip_prefix(o->prefix_length, &name, &other);
2032
2033         if (DIFF_PAIR_UNMERGED(p)) {
2034                 run_diff_cmd(pgm, name, NULL, attr_path,
2035                              NULL, NULL, NULL, o, 0);
2036                 return;
2037         }
2038
2039         diff_fill_sha1_info(one);
2040         diff_fill_sha1_info(two);
2041
2042         strbuf_init(&msg, PATH_MAX * 2 + 300);
2043         switch (p->status) {
2044         case DIFF_STATUS_COPIED:
2045                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2046                 strbuf_addstr(&msg, "\ncopy from ");
2047                 quote_c_style(name, &msg, NULL, 0);
2048                 strbuf_addstr(&msg, "\ncopy to ");
2049                 quote_c_style(other, &msg, NULL, 0);
2050                 strbuf_addch(&msg, '\n');
2051                 break;
2052         case DIFF_STATUS_RENAMED:
2053                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2054                 strbuf_addstr(&msg, "\nrename from ");
2055                 quote_c_style(name, &msg, NULL, 0);
2056                 strbuf_addstr(&msg, "\nrename to ");
2057                 quote_c_style(other, &msg, NULL, 0);
2058                 strbuf_addch(&msg, '\n');
2059                 break;
2060         case DIFF_STATUS_MODIFIED:
2061                 if (p->score) {
2062                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2063                                         similarity_index(p));
2064                         complete_rewrite = 1;
2065                         break;
2066                 }
2067                 /* fallthru */
2068         default:
2069                 /* nothing */
2070                 ;
2071         }
2072
2073         if (hashcmp(one->sha1, two->sha1)) {
2074                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2075
2076                 if (DIFF_OPT_TST(o, BINARY)) {
2077                         mmfile_t mf;
2078                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2079                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2080                                 abbrev = 40;
2081                 }
2082                 strbuf_addf(&msg, "index %.*s..%.*s",
2083                                 abbrev, sha1_to_hex(one->sha1),
2084                                 abbrev, sha1_to_hex(two->sha1));
2085                 if (one->mode == two->mode)
2086                         strbuf_addf(&msg, " %06o", one->mode);
2087                 strbuf_addch(&msg, '\n');
2088         }
2089
2090         if (msg.len)
2091                 strbuf_setlen(&msg, msg.len - 1);
2092         xfrm_msg = msg.len ? msg.buf : NULL;
2093
2094         if (!pgm &&
2095             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2096             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2097                 /* a filepair that changes between file and symlink
2098                  * needs to be split into deletion and creation.
2099                  */
2100                 struct diff_filespec *null = alloc_filespec(two->path);
2101                 run_diff_cmd(NULL, name, other, attr_path,
2102                              one, null, xfrm_msg, o, 0);
2103                 free(null);
2104                 null = alloc_filespec(one->path);
2105                 run_diff_cmd(NULL, name, other, attr_path,
2106                              null, two, xfrm_msg, o, 0);
2107                 free(null);
2108         }
2109         else
2110                 run_diff_cmd(pgm, name, other, attr_path,
2111                              one, two, xfrm_msg, o, complete_rewrite);
2112
2113         strbuf_release(&msg);
2114 }
2115
2116 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2117                          struct diffstat_t *diffstat)
2118 {
2119         const char *name;
2120         const char *other;
2121         int complete_rewrite = 0;
2122
2123         if (DIFF_PAIR_UNMERGED(p)) {
2124                 /* unmerged */
2125                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2126                 return;
2127         }
2128
2129         name = p->one->path;
2130         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2131
2132         if (o->prefix_length)
2133                 strip_prefix(o->prefix_length, &name, &other);
2134
2135         diff_fill_sha1_info(p->one);
2136         diff_fill_sha1_info(p->two);
2137
2138         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2139                 complete_rewrite = 1;
2140         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2141 }
2142
2143 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2144 {
2145         const char *name;
2146         const char *other;
2147         const char *attr_path;
2148
2149         if (DIFF_PAIR_UNMERGED(p)) {
2150                 /* unmerged */
2151                 return;
2152         }
2153
2154         name = p->one->path;
2155         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2156         attr_path = other ? other : name;
2157
2158         if (o->prefix_length)
2159                 strip_prefix(o->prefix_length, &name, &other);
2160
2161         diff_fill_sha1_info(p->one);
2162         diff_fill_sha1_info(p->two);
2163
2164         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2165 }
2166
2167 void diff_setup(struct diff_options *options)
2168 {
2169         memset(options, 0, sizeof(*options));
2170
2171         options->file = stdout;
2172
2173         options->line_termination = '\n';
2174         options->break_opt = -1;
2175         options->rename_limit = -1;
2176         options->dirstat_percent = 3;
2177         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2178         options->context = 3;
2179
2180         options->change = diff_change;
2181         options->add_remove = diff_addremove;
2182         if (diff_use_color_default > 0)
2183                 DIFF_OPT_SET(options, COLOR_DIFF);
2184         else
2185                 DIFF_OPT_CLR(options, COLOR_DIFF);
2186         options->detect_rename = diff_detect_rename_default;
2187
2188         if (!diff_mnemonic_prefix) {
2189                 options->a_prefix = "a/";
2190                 options->b_prefix = "b/";
2191         }
2192 }
2193
2194 int diff_setup_done(struct diff_options *options)
2195 {
2196         int count = 0;
2197
2198         if (options->output_format & DIFF_FORMAT_NAME)
2199                 count++;
2200         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2201                 count++;
2202         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2203                 count++;
2204         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2205                 count++;
2206         if (count > 1)
2207                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2208
2209         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2210                 options->detect_rename = DIFF_DETECT_COPY;
2211
2212         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2213                 options->prefix = NULL;
2214         if (options->prefix)
2215                 options->prefix_length = strlen(options->prefix);
2216         else
2217                 options->prefix_length = 0;
2218
2219         if (options->output_format & (DIFF_FORMAT_NAME |
2220                                       DIFF_FORMAT_NAME_STATUS |
2221                                       DIFF_FORMAT_CHECKDIFF |
2222                                       DIFF_FORMAT_NO_OUTPUT))
2223                 options->output_format &= ~(DIFF_FORMAT_RAW |
2224                                             DIFF_FORMAT_NUMSTAT |
2225                                             DIFF_FORMAT_DIFFSTAT |
2226                                             DIFF_FORMAT_SHORTSTAT |
2227                                             DIFF_FORMAT_DIRSTAT |
2228                                             DIFF_FORMAT_SUMMARY |
2229                                             DIFF_FORMAT_PATCH);
2230
2231         /*
2232          * These cases always need recursive; we do not drop caller-supplied
2233          * recursive bits for other formats here.
2234          */
2235         if (options->output_format & (DIFF_FORMAT_PATCH |
2236                                       DIFF_FORMAT_NUMSTAT |
2237                                       DIFF_FORMAT_DIFFSTAT |
2238                                       DIFF_FORMAT_SHORTSTAT |
2239                                       DIFF_FORMAT_DIRSTAT |
2240                                       DIFF_FORMAT_SUMMARY |
2241                                       DIFF_FORMAT_CHECKDIFF))
2242                 DIFF_OPT_SET(options, RECURSIVE);
2243         /*
2244          * Also pickaxe would not work very well if you do not say recursive
2245          */
2246         if (options->pickaxe)
2247                 DIFF_OPT_SET(options, RECURSIVE);
2248
2249         if (options->detect_rename && options->rename_limit < 0)
2250                 options->rename_limit = diff_rename_limit_default;
2251         if (options->setup & DIFF_SETUP_USE_CACHE) {
2252                 if (!active_cache)
2253                         /* read-cache does not die even when it fails
2254                          * so it is safe for us to do this here.  Also
2255                          * it does not smudge active_cache or active_nr
2256                          * when it fails, so we do not have to worry about
2257                          * cleaning it up ourselves either.
2258                          */
2259                         read_cache();
2260         }
2261         if (options->abbrev <= 0 || 40 < options->abbrev)
2262                 options->abbrev = 40; /* full */
2263
2264         /*
2265          * It does not make sense to show the first hit we happened
2266          * to have found.  It does not make sense not to return with
2267          * exit code in such a case either.
2268          */
2269         if (DIFF_OPT_TST(options, QUIET)) {
2270                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2271                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2272         }
2273
2274         return 0;
2275 }
2276
2277 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2278 {
2279         char c, *eq;
2280         int len;
2281
2282         if (*arg != '-')
2283                 return 0;
2284         c = *++arg;
2285         if (!c)
2286                 return 0;
2287         if (c == arg_short) {
2288                 c = *++arg;
2289                 if (!c)
2290                         return 1;
2291                 if (val && isdigit(c)) {
2292                         char *end;
2293                         int n = strtoul(arg, &end, 10);
2294                         if (*end)
2295                                 return 0;
2296                         *val = n;
2297                         return 1;
2298                 }
2299                 return 0;
2300         }
2301         if (c != '-')
2302                 return 0;
2303         arg++;
2304         eq = strchr(arg, '=');
2305         if (eq)
2306                 len = eq - arg;
2307         else
2308                 len = strlen(arg);
2309         if (!len || strncmp(arg, arg_long, len))
2310                 return 0;
2311         if (eq) {
2312                 int n;
2313                 char *end;
2314                 if (!isdigit(*++eq))
2315                         return 0;
2316                 n = strtoul(eq, &end, 10);
2317                 if (*end)
2318                         return 0;
2319                 *val = n;
2320         }
2321         return 1;
2322 }
2323
2324 static int diff_scoreopt_parse(const char *opt);
2325
2326 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2327 {
2328         const char *arg = av[0];
2329
2330         /* Output format options */
2331         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2332                 options->output_format |= DIFF_FORMAT_PATCH;
2333         else if (opt_arg(arg, 'U', "unified", &options->context))
2334                 options->output_format |= DIFF_FORMAT_PATCH;
2335         else if (!strcmp(arg, "--raw"))
2336                 options->output_format |= DIFF_FORMAT_RAW;
2337         else if (!strcmp(arg, "--patch-with-raw"))
2338                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2339         else if (!strcmp(arg, "--numstat"))
2340                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2341         else if (!strcmp(arg, "--shortstat"))
2342                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2343         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2344                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2345         else if (!strcmp(arg, "--cumulative")) {
2346                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2347                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2348         } else if (opt_arg(arg, 0, "dirstat-by-file",
2349                            &options->dirstat_percent)) {
2350                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2351                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2352         }
2353         else if (!strcmp(arg, "--check"))
2354                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2355         else if (!strcmp(arg, "--summary"))
2356                 options->output_format |= DIFF_FORMAT_SUMMARY;
2357         else if (!strcmp(arg, "--patch-with-stat"))
2358                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2359         else if (!strcmp(arg, "--name-only"))
2360                 options->output_format |= DIFF_FORMAT_NAME;
2361         else if (!strcmp(arg, "--name-status"))
2362                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2363         else if (!strcmp(arg, "-s"))
2364                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2365         else if (!prefixcmp(arg, "--stat")) {
2366                 char *end;
2367                 int width = options->stat_width;
2368                 int name_width = options->stat_name_width;
2369                 arg += 6;
2370                 end = (char *)arg;
2371
2372                 switch (*arg) {
2373                 case '-':
2374                         if (!prefixcmp(arg, "-width="))
2375                                 width = strtoul(arg + 7, &end, 10);
2376                         else if (!prefixcmp(arg, "-name-width="))
2377                                 name_width = strtoul(arg + 12, &end, 10);
2378                         break;
2379                 case '=':
2380                         width = strtoul(arg+1, &end, 10);
2381                         if (*end == ',')
2382                                 name_width = strtoul(end+1, &end, 10);
2383                 }
2384
2385                 /* Important! This checks all the error cases! */
2386                 if (*end)
2387                         return 0;
2388                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2389                 options->stat_name_width = name_width;
2390                 options->stat_width = width;
2391         }
2392
2393         /* renames options */
2394         else if (!prefixcmp(arg, "-B")) {
2395                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2396                         return -1;
2397         }
2398         else if (!prefixcmp(arg, "-M")) {
2399                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2400                         return -1;
2401                 options->detect_rename = DIFF_DETECT_RENAME;
2402         }
2403         else if (!prefixcmp(arg, "-C")) {
2404                 if (options->detect_rename == DIFF_DETECT_COPY)
2405                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2406                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2407                         return -1;
2408                 options->detect_rename = DIFF_DETECT_COPY;
2409         }
2410         else if (!strcmp(arg, "--no-renames"))
2411                 options->detect_rename = 0;
2412         else if (!strcmp(arg, "--relative"))
2413                 DIFF_OPT_SET(options, RELATIVE_NAME);
2414         else if (!prefixcmp(arg, "--relative=")) {
2415                 DIFF_OPT_SET(options, RELATIVE_NAME);
2416                 options->prefix = arg + 11;
2417         }
2418
2419         /* xdiff options */
2420         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2421                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2422         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2423                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2424         else if (!strcmp(arg, "--ignore-space-at-eol"))
2425                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2426
2427         /* flags options */
2428         else if (!strcmp(arg, "--binary")) {
2429                 options->output_format |= DIFF_FORMAT_PATCH;
2430                 DIFF_OPT_SET(options, BINARY);
2431         }
2432         else if (!strcmp(arg, "--full-index"))
2433                 DIFF_OPT_SET(options, FULL_INDEX);
2434         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2435                 DIFF_OPT_SET(options, TEXT);
2436         else if (!strcmp(arg, "-R"))
2437                 DIFF_OPT_SET(options, REVERSE_DIFF);
2438         else if (!strcmp(arg, "--find-copies-harder"))
2439                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2440         else if (!strcmp(arg, "--follow"))
2441                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2442         else if (!strcmp(arg, "--color"))
2443                 DIFF_OPT_SET(options, COLOR_DIFF);
2444         else if (!strcmp(arg, "--no-color"))
2445                 DIFF_OPT_CLR(options, COLOR_DIFF);
2446         else if (!strcmp(arg, "--color-words"))
2447                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2448         else if (!strcmp(arg, "--exit-code"))
2449                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2450         else if (!strcmp(arg, "--quiet"))
2451                 DIFF_OPT_SET(options, QUIET);
2452         else if (!strcmp(arg, "--ext-diff"))
2453                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2454         else if (!strcmp(arg, "--no-ext-diff"))
2455                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2456         else if (!strcmp(arg, "--ignore-submodules"))
2457                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2458
2459         /* misc options */
2460         else if (!strcmp(arg, "-z"))
2461                 options->line_termination = 0;
2462         else if (!prefixcmp(arg, "-l"))
2463                 options->rename_limit = strtoul(arg+2, NULL, 10);
2464         else if (!prefixcmp(arg, "-S"))
2465                 options->pickaxe = arg + 2;
2466         else if (!strcmp(arg, "--pickaxe-all"))
2467                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2468         else if (!strcmp(arg, "--pickaxe-regex"))
2469                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2470         else if (!prefixcmp(arg, "-O"))
2471                 options->orderfile = arg + 2;
2472         else if (!prefixcmp(arg, "--diff-filter="))
2473                 options->filter = arg + 14;
2474         else if (!strcmp(arg, "--abbrev"))
2475                 options->abbrev = DEFAULT_ABBREV;
2476         else if (!prefixcmp(arg, "--abbrev=")) {
2477                 options->abbrev = strtoul(arg + 9, NULL, 10);
2478                 if (options->abbrev < MINIMUM_ABBREV)
2479                         options->abbrev = MINIMUM_ABBREV;
2480                 else if (40 < options->abbrev)
2481                         options->abbrev = 40;
2482         }
2483         else if (!prefixcmp(arg, "--src-prefix="))
2484                 options->a_prefix = arg + 13;
2485         else if (!prefixcmp(arg, "--dst-prefix="))
2486                 options->b_prefix = arg + 13;
2487         else if (!strcmp(arg, "--no-prefix"))
2488                 options->a_prefix = options->b_prefix = "";
2489         else if (!prefixcmp(arg, "--output=")) {
2490                 options->file = fopen(arg + strlen("--output="), "w");
2491                 options->close_file = 1;
2492         } else
2493                 return 0;
2494         return 1;
2495 }
2496
2497 static int parse_num(const char **cp_p)
2498 {
2499         unsigned long num, scale;
2500         int ch, dot;
2501         const char *cp = *cp_p;
2502
2503         num = 0;
2504         scale = 1;
2505         dot = 0;
2506         for(;;) {
2507                 ch = *cp;
2508                 if ( !dot && ch == '.' ) {
2509                         scale = 1;
2510                         dot = 1;
2511                 } else if ( ch == '%' ) {
2512                         scale = dot ? scale*100 : 100;
2513                         cp++;   /* % is always at the end */
2514                         break;
2515                 } else if ( ch >= '0' && ch <= '9' ) {
2516                         if ( scale < 100000 ) {
2517                                 scale *= 10;
2518                                 num = (num*10) + (ch-'0');
2519                         }
2520                 } else {
2521                         break;
2522                 }
2523                 cp++;
2524         }
2525         *cp_p = cp;
2526
2527         /* user says num divided by scale and we say internally that
2528          * is MAX_SCORE * num / scale.
2529          */
2530         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2531 }
2532
2533 static int diff_scoreopt_parse(const char *opt)
2534 {
2535         int opt1, opt2, cmd;
2536
2537         if (*opt++ != '-')
2538                 return -1;
2539         cmd = *opt++;
2540         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2541                 return -1; /* that is not a -M, -C nor -B option */
2542
2543         opt1 = parse_num(&opt);
2544         if (cmd != 'B')
2545                 opt2 = 0;
2546         else {
2547                 if (*opt == 0)
2548                         opt2 = 0;
2549                 else if (*opt != '/')
2550                         return -1; /* we expect -B80/99 or -B80 */
2551                 else {
2552                         opt++;
2553                         opt2 = parse_num(&opt);
2554                 }
2555         }
2556         if (*opt != 0)
2557                 return -1;
2558         return opt1 | (opt2 << 16);
2559 }
2560
2561 struct diff_queue_struct diff_queued_diff;
2562
2563 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2564 {
2565         if (queue->alloc <= queue->nr) {
2566                 queue->alloc = alloc_nr(queue->alloc);
2567                 queue->queue = xrealloc(queue->queue,
2568                                         sizeof(dp) * queue->alloc);
2569         }
2570         queue->queue[queue->nr++] = dp;
2571 }
2572
2573 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2574                                  struct diff_filespec *one,
2575                                  struct diff_filespec *two)
2576 {
2577         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2578         dp->one = one;
2579         dp->two = two;
2580         if (queue)
2581                 diff_q(queue, dp);
2582         return dp;
2583 }
2584
2585 void diff_free_filepair(struct diff_filepair *p)
2586 {
2587         free_filespec(p->one);
2588         free_filespec(p->two);
2589         free(p);
2590 }
2591
2592 /* This is different from find_unique_abbrev() in that
2593  * it stuffs the result with dots for alignment.
2594  */
2595 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2596 {
2597         int abblen;
2598         const char *abbrev;
2599         if (len == 40)
2600                 return sha1_to_hex(sha1);
2601
2602         abbrev = find_unique_abbrev(sha1, len);
2603         abblen = strlen(abbrev);
2604         if (abblen < 37) {
2605                 static char hex[41];
2606                 if (len < abblen && abblen <= len + 2)
2607                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2608                 else
2609                         sprintf(hex, "%s...", abbrev);
2610                 return hex;
2611         }
2612         return sha1_to_hex(sha1);
2613 }
2614
2615 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2616 {
2617         int line_termination = opt->line_termination;
2618         int inter_name_termination = line_termination ? '\t' : '\0';
2619
2620         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2621                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2622                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2623                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2624         }
2625         if (p->score) {
2626                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2627                         inter_name_termination);
2628         } else {
2629                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2630         }
2631
2632         if (p->status == DIFF_STATUS_COPIED ||
2633             p->status == DIFF_STATUS_RENAMED) {
2634                 const char *name_a, *name_b;
2635                 name_a = p->one->path;
2636                 name_b = p->two->path;
2637                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2638                 write_name_quoted(name_a, opt->file, inter_name_termination);
2639                 write_name_quoted(name_b, opt->file, line_termination);
2640         } else {
2641                 const char *name_a, *name_b;
2642                 name_a = p->one->mode ? p->one->path : p->two->path;
2643                 name_b = NULL;
2644                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2645                 write_name_quoted(name_a, opt->file, line_termination);
2646         }
2647 }
2648
2649 int diff_unmodified_pair(struct diff_filepair *p)
2650 {
2651         /* This function is written stricter than necessary to support
2652          * the currently implemented transformers, but the idea is to
2653          * let transformers to produce diff_filepairs any way they want,
2654          * and filter and clean them up here before producing the output.
2655          */
2656         struct diff_filespec *one = p->one, *two = p->two;
2657
2658         if (DIFF_PAIR_UNMERGED(p))
2659                 return 0; /* unmerged is interesting */
2660
2661         /* deletion, addition, mode or type change
2662          * and rename are all interesting.
2663          */
2664         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2665             DIFF_PAIR_MODE_CHANGED(p) ||
2666             strcmp(one->path, two->path))
2667                 return 0;
2668
2669         /* both are valid and point at the same path.  that is, we are
2670          * dealing with a change.
2671          */
2672         if (one->sha1_valid && two->sha1_valid &&
2673             !hashcmp(one->sha1, two->sha1))
2674                 return 1; /* no change */
2675         if (!one->sha1_valid && !two->sha1_valid)
2676                 return 1; /* both look at the same file on the filesystem. */
2677         return 0;
2678 }
2679
2680 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2681 {
2682         if (diff_unmodified_pair(p))
2683                 return;
2684
2685         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2686             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2687                 return; /* no tree diffs in patch format */
2688
2689         run_diff(p, o);
2690 }
2691
2692 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2693                             struct diffstat_t *diffstat)
2694 {
2695         if (diff_unmodified_pair(p))
2696                 return;
2697
2698         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2699             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2700                 return; /* no tree diffs in patch format */
2701
2702         run_diffstat(p, o, diffstat);
2703 }
2704
2705 static void diff_flush_checkdiff(struct diff_filepair *p,
2706                 struct diff_options *o)
2707 {
2708         if (diff_unmodified_pair(p))
2709                 return;
2710
2711         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2712             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2713                 return; /* no tree diffs in patch format */
2714
2715         run_checkdiff(p, o);
2716 }
2717
2718 int diff_queue_is_empty(void)
2719 {
2720         struct diff_queue_struct *q = &diff_queued_diff;
2721         int i;
2722         for (i = 0; i < q->nr; i++)
2723                 if (!diff_unmodified_pair(q->queue[i]))
2724                         return 0;
2725         return 1;
2726 }
2727
2728 #if DIFF_DEBUG
2729 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2730 {
2731         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2732                 x, one ? one : "",
2733                 s->path,
2734                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2735                 s->mode,
2736                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2737         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2738                 x, one ? one : "",
2739                 s->size, s->xfrm_flags);
2740 }
2741
2742 void diff_debug_filepair(const struct diff_filepair *p, int i)
2743 {
2744         diff_debug_filespec(p->one, i, "one");
2745         diff_debug_filespec(p->two, i, "two");
2746         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2747                 p->score, p->status ? p->status : '?',
2748                 p->one->rename_used, p->broken_pair);
2749 }
2750
2751 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2752 {
2753         int i;
2754         if (msg)
2755                 fprintf(stderr, "%s\n", msg);
2756         fprintf(stderr, "q->nr = %d\n", q->nr);
2757         for (i = 0; i < q->nr; i++) {
2758                 struct diff_filepair *p = q->queue[i];
2759                 diff_debug_filepair(p, i);
2760         }
2761 }
2762 #endif
2763
2764 static void diff_resolve_rename_copy(void)
2765 {
2766         int i;
2767         struct diff_filepair *p;
2768         struct diff_queue_struct *q = &diff_queued_diff;
2769
2770         diff_debug_queue("resolve-rename-copy", q);
2771
2772         for (i = 0; i < q->nr; i++) {
2773                 p = q->queue[i];
2774                 p->status = 0; /* undecided */
2775                 if (DIFF_PAIR_UNMERGED(p))
2776                         p->status = DIFF_STATUS_UNMERGED;
2777                 else if (!DIFF_FILE_VALID(p->one))
2778                         p->status = DIFF_STATUS_ADDED;
2779                 else if (!DIFF_FILE_VALID(p->two))
2780                         p->status = DIFF_STATUS_DELETED;
2781                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2782                         p->status = DIFF_STATUS_TYPE_CHANGED;
2783
2784                 /* from this point on, we are dealing with a pair
2785                  * whose both sides are valid and of the same type, i.e.
2786                  * either in-place edit or rename/copy edit.
2787                  */
2788                 else if (DIFF_PAIR_RENAME(p)) {
2789                         /*
2790                          * A rename might have re-connected a broken
2791                          * pair up, causing the pathnames to be the
2792                          * same again. If so, that's not a rename at
2793                          * all, just a modification..
2794                          *
2795                          * Otherwise, see if this source was used for
2796                          * multiple renames, in which case we decrement
2797                          * the count, and call it a copy.
2798                          */
2799                         if (!strcmp(p->one->path, p->two->path))
2800                                 p->status = DIFF_STATUS_MODIFIED;
2801                         else if (--p->one->rename_used > 0)
2802                                 p->status = DIFF_STATUS_COPIED;
2803                         else
2804                                 p->status = DIFF_STATUS_RENAMED;
2805                 }
2806                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2807                          p->one->mode != p->two->mode ||
2808                          is_null_sha1(p->one->sha1))
2809                         p->status = DIFF_STATUS_MODIFIED;
2810                 else {
2811                         /* This is a "no-change" entry and should not
2812                          * happen anymore, but prepare for broken callers.
2813                          */
2814                         error("feeding unmodified %s to diffcore",
2815                               p->one->path);
2816                         p->status = DIFF_STATUS_UNKNOWN;
2817                 }
2818         }
2819         diff_debug_queue("resolve-rename-copy done", q);
2820 }
2821
2822 static int check_pair_status(struct diff_filepair *p)
2823 {
2824         switch (p->status) {
2825         case DIFF_STATUS_UNKNOWN:
2826                 return 0;
2827         case 0:
2828                 die("internal error in diff-resolve-rename-copy");
2829         default:
2830                 return 1;
2831         }
2832 }
2833
2834 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2835 {
2836         int fmt = opt->output_format;
2837
2838         if (fmt & DIFF_FORMAT_CHECKDIFF)
2839                 diff_flush_checkdiff(p, opt);
2840         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2841                 diff_flush_raw(p, opt);
2842         else if (fmt & DIFF_FORMAT_NAME) {
2843                 const char *name_a, *name_b;
2844                 name_a = p->two->path;
2845                 name_b = NULL;
2846                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2847                 write_name_quoted(name_a, opt->file, opt->line_termination);
2848         }
2849 }
2850
2851 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2852 {
2853         if (fs->mode)
2854                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2855         else
2856                 fprintf(file, " %s ", newdelete);
2857         write_name_quoted(fs->path, file, '\n');
2858 }
2859
2860
2861 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2862 {
2863         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2864                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2865                         show_name ? ' ' : '\n');
2866                 if (show_name) {
2867                         write_name_quoted(p->two->path, file, '\n');
2868                 }
2869         }
2870 }
2871
2872 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2873 {
2874         char *names = pprint_rename(p->one->path, p->two->path);
2875
2876         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2877         free(names);
2878         show_mode_change(file, p, 0);
2879 }
2880
2881 static void diff_summary(FILE *file, struct diff_filepair *p)
2882 {
2883         switch(p->status) {
2884         case DIFF_STATUS_DELETED:
2885                 show_file_mode_name(file, "delete", p->one);
2886                 break;
2887         case DIFF_STATUS_ADDED:
2888                 show_file_mode_name(file, "create", p->two);
2889                 break;
2890         case DIFF_STATUS_COPIED:
2891                 show_rename_copy(file, "copy", p);
2892                 break;
2893         case DIFF_STATUS_RENAMED:
2894                 show_rename_copy(file, "rename", p);
2895                 break;
2896         default:
2897                 if (p->score) {
2898                         fputs(" rewrite ", file);
2899                         write_name_quoted(p->two->path, file, ' ');
2900                         fprintf(file, "(%d%%)\n", similarity_index(p));
2901                 }
2902                 show_mode_change(file, p, !p->score);
2903                 break;
2904         }
2905 }
2906
2907 struct patch_id_t {
2908         git_SHA_CTX *ctx;
2909         int patchlen;
2910 };
2911
2912 static int remove_space(char *line, int len)
2913 {
2914         int i;
2915         char *dst = line;
2916         unsigned char c;
2917
2918         for (i = 0; i < len; i++)
2919                 if (!isspace((c = line[i])))
2920                         *dst++ = c;
2921
2922         return dst - line;
2923 }
2924
2925 static void patch_id_consume(void *priv, char *line, unsigned long len)
2926 {
2927         struct patch_id_t *data = priv;
2928         int new_len;
2929
2930         /* Ignore line numbers when computing the SHA1 of the patch */
2931         if (!prefixcmp(line, "@@ -"))
2932                 return;
2933
2934         new_len = remove_space(line, len);
2935
2936         git_SHA1_Update(data->ctx, line, new_len);
2937         data->patchlen += new_len;
2938 }
2939
2940 /* returns 0 upon success, and writes result into sha1 */
2941 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2942 {
2943         struct diff_queue_struct *q = &diff_queued_diff;
2944         int i;
2945         git_SHA_CTX ctx;
2946         struct patch_id_t data;
2947         char buffer[PATH_MAX * 4 + 20];
2948
2949         git_SHA1_Init(&ctx);
2950         memset(&data, 0, sizeof(struct patch_id_t));
2951         data.ctx = &ctx;
2952
2953         for (i = 0; i < q->nr; i++) {
2954                 xpparam_t xpp;
2955                 xdemitconf_t xecfg;
2956                 xdemitcb_t ecb;
2957                 mmfile_t mf1, mf2;
2958                 struct diff_filepair *p = q->queue[i];
2959                 int len1, len2;
2960
2961                 memset(&xecfg, 0, sizeof(xecfg));
2962                 if (p->status == 0)
2963                         return error("internal diff status error");
2964                 if (p->status == DIFF_STATUS_UNKNOWN)
2965                         continue;
2966                 if (diff_unmodified_pair(p))
2967                         continue;
2968                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2969                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2970                         continue;
2971                 if (DIFF_PAIR_UNMERGED(p))
2972                         continue;
2973
2974                 diff_fill_sha1_info(p->one);
2975                 diff_fill_sha1_info(p->two);
2976                 if (fill_mmfile(&mf1, p->one) < 0 ||
2977                                 fill_mmfile(&mf2, p->two) < 0)
2978                         return error("unable to read files to diff");
2979
2980                 len1 = remove_space(p->one->path, strlen(p->one->path));
2981                 len2 = remove_space(p->two->path, strlen(p->two->path));
2982                 if (p->one->mode == 0)
2983                         len1 = snprintf(buffer, sizeof(buffer),
2984                                         "diff--gita/%.*sb/%.*s"
2985                                         "newfilemode%06o"
2986                                         "---/dev/null"
2987                                         "+++b/%.*s",
2988                                         len1, p->one->path,
2989                                         len2, p->two->path,
2990                                         p->two->mode,
2991                                         len2, p->two->path);
2992                 else if (p->two->mode == 0)
2993                         len1 = snprintf(buffer, sizeof(buffer),
2994                                         "diff--gita/%.*sb/%.*s"
2995                                         "deletedfilemode%06o"
2996                                         "---a/%.*s"
2997                                         "+++/dev/null",
2998                                         len1, p->one->path,
2999                                         len2, p->two->path,
3000                                         p->one->mode,
3001                                         len1, p->one->path);
3002                 else
3003                         len1 = snprintf(buffer, sizeof(buffer),
3004                                         "diff--gita/%.*sb/%.*s"
3005                                         "---a/%.*s"
3006                                         "+++b/%.*s",
3007                                         len1, p->one->path,
3008                                         len2, p->two->path,
3009                                         len1, p->one->path,
3010                                         len2, p->two->path);
3011                 git_SHA1_Update(&ctx, buffer, len1);
3012
3013                 xpp.flags = XDF_NEED_MINIMAL;
3014                 xecfg.ctxlen = 3;
3015                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3016                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3017                               &xpp, &xecfg, &ecb);
3018         }
3019
3020         git_SHA1_Final(sha1, &ctx);
3021         return 0;
3022 }
3023
3024 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3025 {
3026         struct diff_queue_struct *q = &diff_queued_diff;
3027         int i;
3028         int result = diff_get_patch_id(options, sha1);
3029
3030         for (i = 0; i < q->nr; i++)
3031                 diff_free_filepair(q->queue[i]);
3032
3033         free(q->queue);
3034         q->queue = NULL;
3035         q->nr = q->alloc = 0;
3036
3037         return result;
3038 }
3039
3040 static int is_summary_empty(const struct diff_queue_struct *q)
3041 {
3042         int i;
3043
3044         for (i = 0; i < q->nr; i++) {
3045                 const struct diff_filepair *p = q->queue[i];
3046
3047                 switch (p->status) {
3048                 case DIFF_STATUS_DELETED:
3049                 case DIFF_STATUS_ADDED:
3050                 case DIFF_STATUS_COPIED:
3051                 case DIFF_STATUS_RENAMED:
3052                         return 0;
3053                 default:
3054                         if (p->score)
3055                                 return 0;
3056                         if (p->one->mode && p->two->mode &&
3057                             p->one->mode != p->two->mode)
3058                                 return 0;
3059                         break;
3060                 }
3061         }
3062         return 1;
3063 }
3064
3065 void diff_flush(struct diff_options *options)
3066 {
3067         struct diff_queue_struct *q = &diff_queued_diff;
3068         int i, output_format = options->output_format;
3069         int separator = 0;
3070
3071         /*
3072          * Order: raw, stat, summary, patch
3073          * or:    name/name-status/checkdiff (other bits clear)
3074          */
3075         if (!q->nr)
3076                 goto free_queue;
3077
3078         if (output_format & (DIFF_FORMAT_RAW |
3079                              DIFF_FORMAT_NAME |
3080                              DIFF_FORMAT_NAME_STATUS |
3081                              DIFF_FORMAT_CHECKDIFF)) {
3082                 for (i = 0; i < q->nr; i++) {
3083                         struct diff_filepair *p = q->queue[i];
3084                         if (check_pair_status(p))
3085                                 flush_one_pair(p, options);
3086                 }
3087                 separator++;
3088         }
3089
3090         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3091                 struct diffstat_t diffstat;
3092
3093                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3094                 for (i = 0; i < q->nr; i++) {
3095                         struct diff_filepair *p = q->queue[i];
3096                         if (check_pair_status(p))
3097                                 diff_flush_stat(p, options, &diffstat);
3098                 }
3099                 if (output_format & DIFF_FORMAT_NUMSTAT)
3100                         show_numstat(&diffstat, options);
3101                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3102                         show_stats(&diffstat, options);
3103                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3104                         show_shortstats(&diffstat, options);
3105                 free_diffstat_info(&diffstat);
3106                 separator++;
3107         }
3108         if (output_format & DIFF_FORMAT_DIRSTAT)
3109                 show_dirstat(options);
3110
3111         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3112                 for (i = 0; i < q->nr; i++)
3113                         diff_summary(options->file, q->queue[i]);
3114                 separator++;
3115         }
3116
3117         if (output_format & DIFF_FORMAT_PATCH) {
3118                 if (separator) {
3119                         putc(options->line_termination, options->file);
3120                         if (options->stat_sep) {
3121                                 /* attach patch instead of inline */
3122                                 fputs(options->stat_sep, options->file);
3123                         }
3124                 }
3125
3126                 for (i = 0; i < q->nr; i++) {
3127                         struct diff_filepair *p = q->queue[i];
3128                         if (check_pair_status(p))
3129                                 diff_flush_patch(p, options);
3130                 }
3131         }
3132
3133         if (output_format & DIFF_FORMAT_CALLBACK)
3134                 options->format_callback(q, options, options->format_callback_data);
3135
3136         for (i = 0; i < q->nr; i++)
3137                 diff_free_filepair(q->queue[i]);
3138 free_queue:
3139         free(q->queue);
3140         q->queue = NULL;
3141         q->nr = q->alloc = 0;
3142         if (options->close_file)
3143                 fclose(options->file);
3144 }
3145
3146 static void diffcore_apply_filter(const char *filter)
3147 {
3148         int i;
3149         struct diff_queue_struct *q = &diff_queued_diff;
3150         struct diff_queue_struct outq;
3151         outq.queue = NULL;
3152         outq.nr = outq.alloc = 0;
3153
3154         if (!filter)
3155                 return;
3156
3157         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3158                 int found;
3159                 for (i = found = 0; !found && i < q->nr; i++) {
3160                         struct diff_filepair *p = q->queue[i];
3161                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3162                              ((p->score &&
3163                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3164                               (!p->score &&
3165                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3166                             ((p->status != DIFF_STATUS_MODIFIED) &&
3167                              strchr(filter, p->status)))
3168                                 found++;
3169                 }
3170                 if (found)
3171                         return;
3172
3173                 /* otherwise we will clear the whole queue
3174                  * by copying the empty outq at the end of this
3175                  * function, but first clear the current entries
3176                  * in the queue.
3177                  */
3178                 for (i = 0; i < q->nr; i++)
3179                         diff_free_filepair(q->queue[i]);
3180         }
3181         else {
3182                 /* Only the matching ones */
3183                 for (i = 0; i < q->nr; i++) {
3184                         struct diff_filepair *p = q->queue[i];
3185
3186                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3187                              ((p->score &&
3188                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3189                               (!p->score &&
3190                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3191                             ((p->status != DIFF_STATUS_MODIFIED) &&
3192                              strchr(filter, p->status)))
3193                                 diff_q(&outq, p);
3194                         else
3195                                 diff_free_filepair(p);
3196                 }
3197         }
3198         free(q->queue);
3199         *q = outq;
3200 }
3201
3202 /* Check whether two filespecs with the same mode and size are identical */
3203 static int diff_filespec_is_identical(struct diff_filespec *one,
3204                                       struct diff_filespec *two)
3205 {
3206         if (S_ISGITLINK(one->mode))
3207                 return 0;
3208         if (diff_populate_filespec(one, 0))
3209                 return 0;
3210         if (diff_populate_filespec(two, 0))
3211                 return 0;
3212         return !memcmp(one->data, two->data, one->size);
3213 }
3214
3215 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3216 {
3217         int i;
3218         struct diff_queue_struct *q = &diff_queued_diff;
3219         struct diff_queue_struct outq;
3220         outq.queue = NULL;
3221         outq.nr = outq.alloc = 0;
3222
3223         for (i = 0; i < q->nr; i++) {
3224                 struct diff_filepair *p = q->queue[i];
3225
3226                 /*
3227                  * 1. Entries that come from stat info dirtyness
3228                  *    always have both sides (iow, not create/delete),
3229                  *    one side of the object name is unknown, with
3230                  *    the same mode and size.  Keep the ones that
3231                  *    do not match these criteria.  They have real
3232                  *    differences.
3233                  *
3234                  * 2. At this point, the file is known to be modified,
3235                  *    with the same mode and size, and the object
3236                  *    name of one side is unknown.  Need to inspect
3237                  *    the identical contents.
3238                  */
3239                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3240                     !DIFF_FILE_VALID(p->two) ||
3241                     (p->one->sha1_valid && p->two->sha1_valid) ||
3242                     (p->one->mode != p->two->mode) ||
3243                     diff_populate_filespec(p->one, 1) ||
3244                     diff_populate_filespec(p->two, 1) ||
3245                     (p->one->size != p->two->size) ||
3246                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3247                         diff_q(&outq, p);
3248                 else {
3249                         /*
3250                          * The caller can subtract 1 from skip_stat_unmatch
3251                          * to determine how many paths were dirty only
3252                          * due to stat info mismatch.
3253                          */
3254                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3255                                 diffopt->skip_stat_unmatch++;
3256                         diff_free_filepair(p);
3257                 }
3258         }
3259         free(q->queue);
3260         *q = outq;
3261 }
3262
3263 void diffcore_std(struct diff_options *options)
3264 {
3265         if (options->skip_stat_unmatch)
3266                 diffcore_skip_stat_unmatch(options);
3267         if (options->break_opt != -1)
3268                 diffcore_break(options->break_opt);
3269         if (options->detect_rename)
3270                 diffcore_rename(options);
3271         if (options->break_opt != -1)
3272                 diffcore_merge_broken();
3273         if (options->pickaxe)
3274                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3275         if (options->orderfile)
3276                 diffcore_order(options->orderfile);
3277         diff_resolve_rename_copy();
3278         diffcore_apply_filter(options->filter);
3279
3280         if (diff_queued_diff.nr)
3281                 DIFF_OPT_SET(options, HAS_CHANGES);
3282         else
3283                 DIFF_OPT_CLR(options, HAS_CHANGES);
3284 }
3285
3286 int diff_result_code(struct diff_options *opt, int status)
3287 {
3288         int result = 0;
3289         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3290             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3291                 return status;
3292         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3293             DIFF_OPT_TST(opt, HAS_CHANGES))
3294                 result |= 01;
3295         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3296             DIFF_OPT_TST(opt, CHECK_FAILED))
3297                 result |= 02;
3298         return result;
3299 }
3300
3301 void diff_addremove(struct diff_options *options,
3302                     int addremove, unsigned mode,
3303                     const unsigned char *sha1,
3304                     const char *concatpath)
3305 {
3306         struct diff_filespec *one, *two;
3307
3308         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3309                 return;
3310
3311         /* This may look odd, but it is a preparation for
3312          * feeding "there are unchanged files which should
3313          * not produce diffs, but when you are doing copy
3314          * detection you would need them, so here they are"
3315          * entries to the diff-core.  They will be prefixed
3316          * with something like '=' or '*' (I haven't decided
3317          * which but should not make any difference).
3318          * Feeding the same new and old to diff_change()
3319          * also has the same effect.
3320          * Before the final output happens, they are pruned after
3321          * merged into rename/copy pairs as appropriate.
3322          */
3323         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3324                 addremove = (addremove == '+' ? '-' :
3325                              addremove == '-' ? '+' : addremove);
3326
3327         if (options->prefix &&
3328             strncmp(concatpath, options->prefix, options->prefix_length))
3329                 return;
3330
3331         one = alloc_filespec(concatpath);
3332         two = alloc_filespec(concatpath);
3333
3334         if (addremove != '+')
3335                 fill_filespec(one, sha1, mode);
3336         if (addremove != '-')
3337                 fill_filespec(two, sha1, mode);
3338
3339         diff_queue(&diff_queued_diff, one, two);
3340         DIFF_OPT_SET(options, HAS_CHANGES);
3341 }
3342
3343 void diff_change(struct diff_options *options,
3344                  unsigned old_mode, unsigned new_mode,
3345                  const unsigned char *old_sha1,
3346                  const unsigned char *new_sha1,
3347                  const char *concatpath)
3348 {
3349         struct diff_filespec *one, *two;
3350
3351         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3352                         && S_ISGITLINK(new_mode))
3353                 return;
3354
3355         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3356                 unsigned tmp;
3357                 const unsigned char *tmp_c;
3358                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3359                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3360         }
3361
3362         if (options->prefix &&
3363             strncmp(concatpath, options->prefix, options->prefix_length))
3364                 return;
3365
3366         one = alloc_filespec(concatpath);
3367         two = alloc_filespec(concatpath);
3368         fill_filespec(one, old_sha1, old_mode);
3369         fill_filespec(two, new_sha1, new_mode);
3370
3371         diff_queue(&diff_queued_diff, one, two);
3372         DIFF_OPT_SET(options, HAS_CHANGES);
3373 }
3374
3375 void diff_unmerge(struct diff_options *options,
3376                   const char *path,
3377                   unsigned mode, const unsigned char *sha1)
3378 {
3379         struct diff_filespec *one, *two;
3380
3381         if (options->prefix &&
3382             strncmp(path, options->prefix, options->prefix_length))
3383                 return;
3384
3385         one = alloc_filespec(path);
3386         two = alloc_filespec(path);
3387         fill_filespec(one, sha1, mode);
3388         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3389 }
3390
3391 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3392                 size_t *outsize)
3393 {
3394         struct diff_tempfile temp;
3395         const char *argv[3];
3396         const char **arg = argv;
3397         struct child_process child;
3398         struct strbuf buf = STRBUF_INIT;
3399
3400         prepare_temp_file(spec->path, &temp, spec);
3401         *arg++ = pgm;
3402         *arg++ = temp.name;
3403         *arg = NULL;
3404
3405         memset(&child, 0, sizeof(child));
3406         child.argv = argv;
3407         child.out = -1;
3408         if (start_command(&child) != 0 ||
3409             strbuf_read(&buf, child.out, 0) < 0 ||
3410             finish_command(&child) != 0) {
3411                 if (temp.name == temp.tmp_path)
3412                         unlink(temp.name);
3413                 error("error running textconv command '%s'", pgm);
3414                 return NULL;
3415         }
3416         if (temp.name == temp.tmp_path)
3417                 unlink(temp.name);
3418
3419         return strbuf_detach(&buf, outsize);
3420 }