Merge branch 'cb/match_refs_internal_tail'
[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 #include "sigchain.h"
16
17 #ifdef NO_FAST_WORKING_DIRECTORY
18 #define FAST_WORKING_DIRECTORY 0
19 #else
20 #define FAST_WORKING_DIRECTORY 1
21 #endif
22
23 static int diff_detect_rename_default;
24 static int diff_rename_limit_default = 200;
25 static int diff_suppress_blank_empty;
26 int diff_use_color_default = -1;
27 static const char *diff_word_regex_cfg;
28 static const char *external_diff_cmd_cfg;
29 int diff_auto_refresh_index = 1;
30 static int diff_mnemonic_prefix;
31
32 static char diff_colors[][COLOR_MAXLEN] = {
33         GIT_COLOR_RESET,
34         GIT_COLOR_NORMAL,       /* PLAIN */
35         GIT_COLOR_BOLD,         /* METAINFO */
36         GIT_COLOR_CYAN,         /* FRAGINFO */
37         GIT_COLOR_RED,          /* OLD */
38         GIT_COLOR_GREEN,        /* NEW */
39         GIT_COLOR_YELLOW,       /* COMMIT */
40         GIT_COLOR_BG_RED,       /* WHITESPACE */
41 };
42
43 static void diff_filespec_load_driver(struct diff_filespec *one);
44 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
45
46 static int parse_diff_color_slot(const char *var, int ofs)
47 {
48         if (!strcasecmp(var+ofs, "plain"))
49                 return DIFF_PLAIN;
50         if (!strcasecmp(var+ofs, "meta"))
51                 return DIFF_METAINFO;
52         if (!strcasecmp(var+ofs, "frag"))
53                 return DIFF_FRAGINFO;
54         if (!strcasecmp(var+ofs, "old"))
55                 return DIFF_FILE_OLD;
56         if (!strcasecmp(var+ofs, "new"))
57                 return DIFF_FILE_NEW;
58         if (!strcasecmp(var+ofs, "commit"))
59                 return DIFF_COMMIT;
60         if (!strcasecmp(var+ofs, "whitespace"))
61                 return DIFF_WHITESPACE;
62         die("bad config variable '%s'", var);
63 }
64
65 static int git_config_rename(const char *var, const char *value)
66 {
67         if (!value)
68                 return DIFF_DETECT_RENAME;
69         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
70                 return  DIFF_DETECT_COPY;
71         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
72 }
73
74 /*
75  * These are to give UI layer defaults.
76  * The core-level commands such as git-diff-files should
77  * never be affected by the setting of diff.renames
78  * the user happens to have in the configuration file.
79  */
80 int git_diff_ui_config(const char *var, const char *value, void *cb)
81 {
82         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
83                 diff_use_color_default = git_config_colorbool(var, value, -1);
84                 return 0;
85         }
86         if (!strcmp(var, "diff.renames")) {
87                 diff_detect_rename_default = git_config_rename(var, value);
88                 return 0;
89         }
90         if (!strcmp(var, "diff.autorefreshindex")) {
91                 diff_auto_refresh_index = git_config_bool(var, value);
92                 return 0;
93         }
94         if (!strcmp(var, "diff.mnemonicprefix")) {
95                 diff_mnemonic_prefix = git_config_bool(var, value);
96                 return 0;
97         }
98         if (!strcmp(var, "diff.external"))
99                 return git_config_string(&external_diff_cmd_cfg, var, value);
100         if (!strcmp(var, "diff.wordregex"))
101                 return git_config_string(&diff_word_regex_cfg, var, value);
102
103         return git_diff_basic_config(var, value, cb);
104 }
105
106 int git_diff_basic_config(const char *var, const char *value, void *cb)
107 {
108         if (!strcmp(var, "diff.renamelimit")) {
109                 diff_rename_limit_default = git_config_int(var, value);
110                 return 0;
111         }
112
113         switch (userdiff_config(var, value)) {
114                 case 0: break;
115                 case -1: return -1;
116                 default: return 0;
117         }
118
119         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
120                 int slot = parse_diff_color_slot(var, 11);
121                 if (!value)
122                         return config_error_nonbool(var);
123                 color_parse(value, var, diff_colors[slot]);
124                 return 0;
125         }
126
127         /* like GNU diff's --suppress-blank-empty option  */
128         if (!strcmp(var, "diff.suppressblankempty") ||
129                         /* for backwards compatibility */
130                         !strcmp(var, "diff.suppress-blank-empty")) {
131                 diff_suppress_blank_empty = git_config_bool(var, value);
132                 return 0;
133         }
134
135         return git_color_default_config(var, value, cb);
136 }
137
138 static char *quote_two(const char *one, const char *two)
139 {
140         int need_one = quote_c_style(one, NULL, NULL, 1);
141         int need_two = quote_c_style(two, NULL, NULL, 1);
142         struct strbuf res = STRBUF_INIT;
143
144         if (need_one + need_two) {
145                 strbuf_addch(&res, '"');
146                 quote_c_style(one, &res, NULL, 1);
147                 quote_c_style(two, &res, NULL, 1);
148                 strbuf_addch(&res, '"');
149         } else {
150                 strbuf_addstr(&res, one);
151                 strbuf_addstr(&res, two);
152         }
153         return strbuf_detach(&res, NULL);
154 }
155
156 static const char *external_diff(void)
157 {
158         static const char *external_diff_cmd = NULL;
159         static int done_preparing = 0;
160
161         if (done_preparing)
162                 return external_diff_cmd;
163         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
164         if (!external_diff_cmd)
165                 external_diff_cmd = external_diff_cmd_cfg;
166         done_preparing = 1;
167         return external_diff_cmd;
168 }
169
170 static struct diff_tempfile {
171         const char *name; /* filename external diff should read from */
172         char hex[41];
173         char mode[10];
174         char tmp_path[PATH_MAX];
175 } diff_temp[2];
176
177 static struct diff_tempfile *claim_diff_tempfile(void) {
178         int i;
179         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
180                 if (!diff_temp[i].name)
181                         return diff_temp + i;
182         die("BUG: diff is failing to clean up its tempfiles");
183 }
184
185 static int remove_tempfile_installed;
186
187 static void remove_tempfile(void)
188 {
189         int i;
190         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
191                 if (diff_temp[i].name == diff_temp[i].tmp_path)
192                         unlink_or_warn(diff_temp[i].name);
193                 diff_temp[i].name = NULL;
194         }
195 }
196
197 static void remove_tempfile_on_signal(int signo)
198 {
199         remove_tempfile();
200         sigchain_pop(signo);
201         raise(signo);
202 }
203
204 static int count_lines(const char *data, int size)
205 {
206         int count, ch, completely_empty = 1, nl_just_seen = 0;
207         count = 0;
208         while (0 < size--) {
209                 ch = *data++;
210                 if (ch == '\n') {
211                         count++;
212                         nl_just_seen = 1;
213                         completely_empty = 0;
214                 }
215                 else {
216                         nl_just_seen = 0;
217                         completely_empty = 0;
218                 }
219         }
220         if (completely_empty)
221                 return 0;
222         if (!nl_just_seen)
223                 count++; /* no trailing newline */
224         return count;
225 }
226
227 static void print_line_count(FILE *file, int count)
228 {
229         switch (count) {
230         case 0:
231                 fprintf(file, "0,0");
232                 break;
233         case 1:
234                 fprintf(file, "1");
235                 break;
236         default:
237                 fprintf(file, "1,%d", count);
238                 break;
239         }
240 }
241
242 static void copy_file_with_prefix(FILE *file,
243                                   int prefix, const char *data, int size,
244                                   const char *set, const char *reset)
245 {
246         int ch, nl_just_seen = 1;
247         while (0 < size--) {
248                 ch = *data++;
249                 if (nl_just_seen) {
250                         fputs(set, file);
251                         putc(prefix, file);
252                 }
253                 if (ch == '\n') {
254                         nl_just_seen = 1;
255                         fputs(reset, file);
256                 } else
257                         nl_just_seen = 0;
258                 putc(ch, file);
259         }
260         if (!nl_just_seen)
261                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
262 }
263
264 static void emit_rewrite_diff(const char *name_a,
265                               const char *name_b,
266                               struct diff_filespec *one,
267                               struct diff_filespec *two,
268                               const char *textconv_one,
269                               const char *textconv_two,
270                               struct diff_options *o)
271 {
272         int lc_a, lc_b;
273         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
274         const char *name_a_tab, *name_b_tab;
275         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
276         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
277         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
278         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
279         const char *reset = diff_get_color(color_diff, DIFF_RESET);
280         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
281         const char *a_prefix, *b_prefix;
282         const char *data_one, *data_two;
283         size_t size_one, size_two;
284
285         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
286                 a_prefix = o->b_prefix;
287                 b_prefix = o->a_prefix;
288         } else {
289                 a_prefix = o->a_prefix;
290                 b_prefix = o->b_prefix;
291         }
292
293         name_a += (*name_a == '/');
294         name_b += (*name_b == '/');
295         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
296         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
297
298         strbuf_reset(&a_name);
299         strbuf_reset(&b_name);
300         quote_two_c_style(&a_name, a_prefix, name_a, 0);
301         quote_two_c_style(&b_name, b_prefix, name_b, 0);
302
303         diff_populate_filespec(one, 0);
304         diff_populate_filespec(two, 0);
305         if (textconv_one) {
306                 data_one = run_textconv(textconv_one, one, &size_one);
307                 if (!data_one)
308                         die("unable to read files to diff");
309         }
310         else {
311                 data_one = one->data;
312                 size_one = one->size;
313         }
314         if (textconv_two) {
315                 data_two = run_textconv(textconv_two, two, &size_two);
316                 if (!data_two)
317                         die("unable to read files to diff");
318         }
319         else {
320                 data_two = two->data;
321                 size_two = two->size;
322         }
323
324         lc_a = count_lines(data_one, size_one);
325         lc_b = count_lines(data_two, size_two);
326         fprintf(o->file,
327                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
328                 metainfo, a_name.buf, name_a_tab, reset,
329                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
330         print_line_count(o->file, lc_a);
331         fprintf(o->file, " +");
332         print_line_count(o->file, lc_b);
333         fprintf(o->file, " @@%s\n", reset);
334         if (lc_a)
335                 copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
336         if (lc_b)
337                 copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
338 }
339
340 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
341 {
342         if (!DIFF_FILE_VALID(one)) {
343                 mf->ptr = (char *)""; /* does not matter */
344                 mf->size = 0;
345                 return 0;
346         }
347         else if (diff_populate_filespec(one, 0))
348                 return -1;
349
350         mf->ptr = one->data;
351         mf->size = one->size;
352         return 0;
353 }
354
355 struct diff_words_buffer {
356         mmfile_t text;
357         long alloc;
358         struct diff_words_orig {
359                 const char *begin, *end;
360         } *orig;
361         int orig_nr, orig_alloc;
362 };
363
364 static void diff_words_append(char *line, unsigned long len,
365                 struct diff_words_buffer *buffer)
366 {
367         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
368         line++;
369         len--;
370         memcpy(buffer->text.ptr + buffer->text.size, line, len);
371         buffer->text.size += len;
372         buffer->text.ptr[buffer->text.size] = '\0';
373 }
374
375 struct diff_words_data {
376         struct diff_words_buffer minus, plus;
377         const char *current_plus;
378         FILE *file;
379         regex_t *word_regex;
380 };
381
382 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
383 {
384         struct diff_words_data *diff_words = priv;
385         int minus_first, minus_len, plus_first, plus_len;
386         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
387
388         if (line[0] != '@' || parse_hunk_header(line, len,
389                         &minus_first, &minus_len, &plus_first, &plus_len))
390                 return;
391
392         /* POSIX requires that first be decremented by one if len == 0... */
393         if (minus_len) {
394                 minus_begin = diff_words->minus.orig[minus_first].begin;
395                 minus_end =
396                         diff_words->minus.orig[minus_first + minus_len - 1].end;
397         } else
398                 minus_begin = minus_end =
399                         diff_words->minus.orig[minus_first].end;
400
401         if (plus_len) {
402                 plus_begin = diff_words->plus.orig[plus_first].begin;
403                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
404         } else
405                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
406
407         if (diff_words->current_plus != plus_begin)
408                 fwrite(diff_words->current_plus,
409                                 plus_begin - diff_words->current_plus, 1,
410                                 diff_words->file);
411         if (minus_begin != minus_end)
412                 color_fwrite_lines(diff_words->file,
413                                 diff_get_color(1, DIFF_FILE_OLD),
414                                 minus_end - minus_begin, minus_begin);
415         if (plus_begin != plus_end)
416                 color_fwrite_lines(diff_words->file,
417                                 diff_get_color(1, DIFF_FILE_NEW),
418                                 plus_end - plus_begin, plus_begin);
419
420         diff_words->current_plus = plus_end;
421 }
422
423 /* This function starts looking at *begin, and returns 0 iff a word was found. */
424 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
425                 int *begin, int *end)
426 {
427         if (word_regex && *begin < buffer->size) {
428                 regmatch_t match[1];
429                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
430                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
431                                         '\n', match[0].rm_eo - match[0].rm_so);
432                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
433                         *begin += match[0].rm_so;
434                         return *begin >= *end;
435                 }
436                 return -1;
437         }
438
439         /* find the next word */
440         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
441                 (*begin)++;
442         if (*begin >= buffer->size)
443                 return -1;
444
445         /* find the end of the word */
446         *end = *begin + 1;
447         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
448                 (*end)++;
449
450         return 0;
451 }
452
453 /*
454  * This function splits the words in buffer->text, stores the list with
455  * newline separator into out, and saves the offsets of the original words
456  * in buffer->orig.
457  */
458 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
459                 regex_t *word_regex)
460 {
461         int i, j;
462         long alloc = 0;
463
464         out->size = 0;
465         out->ptr = NULL;
466
467         /* fake an empty "0th" word */
468         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
469         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
470         buffer->orig_nr = 1;
471
472         for (i = 0; i < buffer->text.size; i++) {
473                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
474                         return;
475
476                 /* store original boundaries */
477                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
478                                 buffer->orig_alloc);
479                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
480                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
481                 buffer->orig_nr++;
482
483                 /* store one word */
484                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
485                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
486                 out->ptr[out->size + j - i] = '\n';
487                 out->size += j - i + 1;
488
489                 i = j - 1;
490         }
491 }
492
493 /* this executes the word diff on the accumulated buffers */
494 static void diff_words_show(struct diff_words_data *diff_words)
495 {
496         xpparam_t xpp;
497         xdemitconf_t xecfg;
498         xdemitcb_t ecb;
499         mmfile_t minus, plus;
500
501         /* special case: only removal */
502         if (!diff_words->plus.text.size) {
503                 color_fwrite_lines(diff_words->file,
504                         diff_get_color(1, DIFF_FILE_OLD),
505                         diff_words->minus.text.size, diff_words->minus.text.ptr);
506                 diff_words->minus.text.size = 0;
507                 return;
508         }
509
510         diff_words->current_plus = diff_words->plus.text.ptr;
511
512         memset(&xpp, 0, sizeof(xpp));
513         memset(&xecfg, 0, sizeof(xecfg));
514         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
515         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
516         xpp.flags = XDF_NEED_MINIMAL;
517         /* as only the hunk header will be parsed, we need a 0-context */
518         xecfg.ctxlen = 0;
519         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
520                       &xpp, &xecfg, &ecb);
521         free(minus.ptr);
522         free(plus.ptr);
523         if (diff_words->current_plus != diff_words->plus.text.ptr +
524                         diff_words->plus.text.size)
525                 fwrite(diff_words->current_plus,
526                         diff_words->plus.text.ptr + diff_words->plus.text.size
527                         - diff_words->current_plus, 1,
528                         diff_words->file);
529         diff_words->minus.text.size = diff_words->plus.text.size = 0;
530 }
531
532 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
533
534 struct emit_callback {
535         int nparents, color_diff;
536         unsigned ws_rule;
537         sane_truncate_fn truncate;
538         const char **label_path;
539         struct diff_words_data *diff_words;
540         int *found_changesp;
541         FILE *file;
542 };
543
544 static void free_diff_words_data(struct emit_callback *ecbdata)
545 {
546         if (ecbdata->diff_words) {
547                 /* flush buffers */
548                 if (ecbdata->diff_words->minus.text.size ||
549                                 ecbdata->diff_words->plus.text.size)
550                         diff_words_show(ecbdata->diff_words);
551
552                 free (ecbdata->diff_words->minus.text.ptr);
553                 free (ecbdata->diff_words->minus.orig);
554                 free (ecbdata->diff_words->plus.text.ptr);
555                 free (ecbdata->diff_words->plus.orig);
556                 free(ecbdata->diff_words->word_regex);
557                 free(ecbdata->diff_words);
558                 ecbdata->diff_words = NULL;
559         }
560 }
561
562 const char *diff_get_color(int diff_use_color, enum color_diff ix)
563 {
564         if (diff_use_color)
565                 return diff_colors[ix];
566         return "";
567 }
568
569 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
570 {
571         int has_trailing_newline, has_trailing_carriage_return;
572
573         has_trailing_newline = (len > 0 && line[len-1] == '\n');
574         if (has_trailing_newline)
575                 len--;
576         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
577         if (has_trailing_carriage_return)
578                 len--;
579
580         fputs(set, file);
581         fwrite(line, len, 1, file);
582         fputs(reset, file);
583         if (has_trailing_carriage_return)
584                 fputc('\r', file);
585         if (has_trailing_newline)
586                 fputc('\n', file);
587 }
588
589 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
590 {
591         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
592         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
593
594         if (!*ws)
595                 emit_line(ecbdata->file, set, reset, line, len);
596         else {
597                 /* Emit just the prefix, then the rest. */
598                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
599                 ws_check_emit(line + ecbdata->nparents,
600                               len - ecbdata->nparents, ecbdata->ws_rule,
601                               ecbdata->file, set, reset, ws);
602         }
603 }
604
605 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
606 {
607         const char *cp;
608         unsigned long allot;
609         size_t l = len;
610
611         if (ecb->truncate)
612                 return ecb->truncate(line, len);
613         cp = line;
614         allot = l;
615         while (0 < l) {
616                 (void) utf8_width(&cp, &l);
617                 if (!cp)
618                         break; /* truncated in the middle? */
619         }
620         return allot - l;
621 }
622
623 static void fn_out_consume(void *priv, char *line, unsigned long len)
624 {
625         int i;
626         int color;
627         struct emit_callback *ecbdata = priv;
628         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
629         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
630         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
631
632         *(ecbdata->found_changesp) = 1;
633
634         if (ecbdata->label_path[0]) {
635                 const char *name_a_tab, *name_b_tab;
636
637                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
638                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
639
640                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
641                         meta, ecbdata->label_path[0], reset, name_a_tab);
642                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
643                         meta, ecbdata->label_path[1], reset, name_b_tab);
644                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
645         }
646
647         if (diff_suppress_blank_empty
648             && len == 2 && line[0] == ' ' && line[1] == '\n') {
649                 line[0] = '\n';
650                 len = 1;
651         }
652
653         /* This is not really necessary for now because
654          * this codepath only deals with two-way diffs.
655          */
656         for (i = 0; i < len && line[i] == '@'; i++)
657                 ;
658         if (2 <= i && i < len && line[i] == ' ') {
659                 ecbdata->nparents = i - 1;
660                 len = sane_truncate_line(ecbdata, line, len);
661                 emit_line(ecbdata->file,
662                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
663                           reset, line, len);
664                 if (line[len-1] != '\n')
665                         putc('\n', ecbdata->file);
666                 return;
667         }
668
669         if (len < ecbdata->nparents) {
670                 emit_line(ecbdata->file, reset, reset, line, len);
671                 return;
672         }
673
674         color = DIFF_PLAIN;
675         if (ecbdata->diff_words && ecbdata->nparents != 1)
676                 /* fall back to normal diff */
677                 free_diff_words_data(ecbdata);
678         if (ecbdata->diff_words) {
679                 if (line[0] == '-') {
680                         diff_words_append(line, len,
681                                           &ecbdata->diff_words->minus);
682                         return;
683                 } else if (line[0] == '+') {
684                         diff_words_append(line, len,
685                                           &ecbdata->diff_words->plus);
686                         return;
687                 }
688                 if (ecbdata->diff_words->minus.text.size ||
689                     ecbdata->diff_words->plus.text.size)
690                         diff_words_show(ecbdata->diff_words);
691                 line++;
692                 len--;
693                 emit_line(ecbdata->file, plain, reset, line, len);
694                 return;
695         }
696         for (i = 0; i < ecbdata->nparents && len; i++) {
697                 if (line[i] == '-')
698                         color = DIFF_FILE_OLD;
699                 else if (line[i] == '+')
700                         color = DIFF_FILE_NEW;
701         }
702
703         if (color != DIFF_FILE_NEW) {
704                 emit_line(ecbdata->file,
705                           diff_get_color(ecbdata->color_diff, color),
706                           reset, line, len);
707                 return;
708         }
709         emit_add_line(reset, ecbdata, line, len);
710 }
711
712 static char *pprint_rename(const char *a, const char *b)
713 {
714         const char *old = a;
715         const char *new = b;
716         struct strbuf name = STRBUF_INIT;
717         int pfx_length, sfx_length;
718         int len_a = strlen(a);
719         int len_b = strlen(b);
720         int a_midlen, b_midlen;
721         int qlen_a = quote_c_style(a, NULL, NULL, 0);
722         int qlen_b = quote_c_style(b, NULL, NULL, 0);
723
724         if (qlen_a || qlen_b) {
725                 quote_c_style(a, &name, NULL, 0);
726                 strbuf_addstr(&name, " => ");
727                 quote_c_style(b, &name, NULL, 0);
728                 return strbuf_detach(&name, NULL);
729         }
730
731         /* Find common prefix */
732         pfx_length = 0;
733         while (*old && *new && *old == *new) {
734                 if (*old == '/')
735                         pfx_length = old - a + 1;
736                 old++;
737                 new++;
738         }
739
740         /* Find common suffix */
741         old = a + len_a;
742         new = b + len_b;
743         sfx_length = 0;
744         while (a <= old && b <= new && *old == *new) {
745                 if (*old == '/')
746                         sfx_length = len_a - (old - a);
747                 old--;
748                 new--;
749         }
750
751         /*
752          * pfx{mid-a => mid-b}sfx
753          * {pfx-a => pfx-b}sfx
754          * pfx{sfx-a => sfx-b}
755          * name-a => name-b
756          */
757         a_midlen = len_a - pfx_length - sfx_length;
758         b_midlen = len_b - pfx_length - sfx_length;
759         if (a_midlen < 0)
760                 a_midlen = 0;
761         if (b_midlen < 0)
762                 b_midlen = 0;
763
764         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
765         if (pfx_length + sfx_length) {
766                 strbuf_add(&name, a, pfx_length);
767                 strbuf_addch(&name, '{');
768         }
769         strbuf_add(&name, a + pfx_length, a_midlen);
770         strbuf_addstr(&name, " => ");
771         strbuf_add(&name, b + pfx_length, b_midlen);
772         if (pfx_length + sfx_length) {
773                 strbuf_addch(&name, '}');
774                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
775         }
776         return strbuf_detach(&name, NULL);
777 }
778
779 struct diffstat_t {
780         int nr;
781         int alloc;
782         struct diffstat_file {
783                 char *from_name;
784                 char *name;
785                 char *print_name;
786                 unsigned is_unmerged:1;
787                 unsigned is_binary:1;
788                 unsigned is_renamed:1;
789                 unsigned int added, deleted;
790         } **files;
791 };
792
793 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
794                                           const char *name_a,
795                                           const char *name_b)
796 {
797         struct diffstat_file *x;
798         x = xcalloc(sizeof (*x), 1);
799         if (diffstat->nr == diffstat->alloc) {
800                 diffstat->alloc = alloc_nr(diffstat->alloc);
801                 diffstat->files = xrealloc(diffstat->files,
802                                 diffstat->alloc * sizeof(x));
803         }
804         diffstat->files[diffstat->nr++] = x;
805         if (name_b) {
806                 x->from_name = xstrdup(name_a);
807                 x->name = xstrdup(name_b);
808                 x->is_renamed = 1;
809         }
810         else {
811                 x->from_name = NULL;
812                 x->name = xstrdup(name_a);
813         }
814         return x;
815 }
816
817 static void diffstat_consume(void *priv, char *line, unsigned long len)
818 {
819         struct diffstat_t *diffstat = priv;
820         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
821
822         if (line[0] == '+')
823                 x->added++;
824         else if (line[0] == '-')
825                 x->deleted++;
826 }
827
828 const char mime_boundary_leader[] = "------------";
829
830 static int scale_linear(int it, int width, int max_change)
831 {
832         /*
833          * make sure that at least one '-' is printed if there were deletions,
834          * and likewise for '+'.
835          */
836         if (max_change < 2)
837                 return it;
838         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
839 }
840
841 static void show_name(FILE *file,
842                       const char *prefix, const char *name, int len)
843 {
844         fprintf(file, " %s%-*s |", prefix, len, name);
845 }
846
847 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
848 {
849         if (cnt <= 0)
850                 return;
851         fprintf(file, "%s", set);
852         while (cnt--)
853                 putc(ch, file);
854         fprintf(file, "%s", reset);
855 }
856
857 static void fill_print_name(struct diffstat_file *file)
858 {
859         char *pname;
860
861         if (file->print_name)
862                 return;
863
864         if (!file->is_renamed) {
865                 struct strbuf buf = STRBUF_INIT;
866                 if (quote_c_style(file->name, &buf, NULL, 0)) {
867                         pname = strbuf_detach(&buf, NULL);
868                 } else {
869                         pname = file->name;
870                         strbuf_release(&buf);
871                 }
872         } else {
873                 pname = pprint_rename(file->from_name, file->name);
874         }
875         file->print_name = pname;
876 }
877
878 static void show_stats(struct diffstat_t *data, struct diff_options *options)
879 {
880         int i, len, add, del, adds = 0, dels = 0;
881         int max_change = 0, max_len = 0;
882         int total_files = data->nr;
883         int width, name_width;
884         const char *reset, *set, *add_c, *del_c;
885
886         if (data->nr == 0)
887                 return;
888
889         width = options->stat_width ? options->stat_width : 80;
890         name_width = options->stat_name_width ? options->stat_name_width : 50;
891
892         /* Sanity: give at least 5 columns to the graph,
893          * but leave at least 10 columns for the name.
894          */
895         if (width < 25)
896                 width = 25;
897         if (name_width < 10)
898                 name_width = 10;
899         else if (width < name_width + 15)
900                 name_width = width - 15;
901
902         /* Find the longest filename and max number of changes */
903         reset = diff_get_color_opt(options, DIFF_RESET);
904         set   = diff_get_color_opt(options, DIFF_PLAIN);
905         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
906         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
907
908         for (i = 0; i < data->nr; i++) {
909                 struct diffstat_file *file = data->files[i];
910                 int change = file->added + file->deleted;
911                 fill_print_name(file);
912                 len = strlen(file->print_name);
913                 if (max_len < len)
914                         max_len = len;
915
916                 if (file->is_binary || file->is_unmerged)
917                         continue;
918                 if (max_change < change)
919                         max_change = change;
920         }
921
922         /* Compute the width of the graph part;
923          * 10 is for one blank at the beginning of the line plus
924          * " | count " between the name and the graph.
925          *
926          * From here on, name_width is the width of the name area,
927          * and width is the width of the graph area.
928          */
929         name_width = (name_width < max_len) ? name_width : max_len;
930         if (width < (name_width + 10) + max_change)
931                 width = width - (name_width + 10);
932         else
933                 width = max_change;
934
935         for (i = 0; i < data->nr; i++) {
936                 const char *prefix = "";
937                 char *name = data->files[i]->print_name;
938                 int added = data->files[i]->added;
939                 int deleted = data->files[i]->deleted;
940                 int name_len;
941
942                 /*
943                  * "scale" the filename
944                  */
945                 len = name_width;
946                 name_len = strlen(name);
947                 if (name_width < name_len) {
948                         char *slash;
949                         prefix = "...";
950                         len -= 3;
951                         name += name_len - len;
952                         slash = strchr(name, '/');
953                         if (slash)
954                                 name = slash;
955                 }
956
957                 if (data->files[i]->is_binary) {
958                         show_name(options->file, prefix, name, len);
959                         fprintf(options->file, "  Bin ");
960                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
961                         fprintf(options->file, " -> ");
962                         fprintf(options->file, "%s%d%s", add_c, added, reset);
963                         fprintf(options->file, " bytes");
964                         fprintf(options->file, "\n");
965                         continue;
966                 }
967                 else if (data->files[i]->is_unmerged) {
968                         show_name(options->file, prefix, name, len);
969                         fprintf(options->file, "  Unmerged\n");
970                         continue;
971                 }
972                 else if (!data->files[i]->is_renamed &&
973                          (added + deleted == 0)) {
974                         total_files--;
975                         continue;
976                 }
977
978                 /*
979                  * scale the add/delete
980                  */
981                 add = added;
982                 del = deleted;
983                 adds += add;
984                 dels += del;
985
986                 if (width <= max_change) {
987                         add = scale_linear(add, width, max_change);
988                         del = scale_linear(del, width, max_change);
989                 }
990                 show_name(options->file, prefix, name, len);
991                 fprintf(options->file, "%5d%s", added + deleted,
992                                 added + deleted ? " " : "");
993                 show_graph(options->file, '+', add, add_c, reset);
994                 show_graph(options->file, '-', del, del_c, reset);
995                 fprintf(options->file, "\n");
996         }
997         fprintf(options->file,
998                " %d files changed, %d insertions(+), %d deletions(-)\n",
999                total_files, adds, dels);
1000 }
1001
1002 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
1003 {
1004         int i, adds = 0, dels = 0, total_files = data->nr;
1005
1006         if (data->nr == 0)
1007                 return;
1008
1009         for (i = 0; i < data->nr; i++) {
1010                 if (!data->files[i]->is_binary &&
1011                     !data->files[i]->is_unmerged) {
1012                         int added = data->files[i]->added;
1013                         int deleted= data->files[i]->deleted;
1014                         if (!data->files[i]->is_renamed &&
1015                             (added + deleted == 0)) {
1016                                 total_files--;
1017                         } else {
1018                                 adds += added;
1019                                 dels += deleted;
1020                         }
1021                 }
1022         }
1023         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1024                total_files, adds, dels);
1025 }
1026
1027 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1028 {
1029         int i;
1030
1031         if (data->nr == 0)
1032                 return;
1033
1034         for (i = 0; i < data->nr; i++) {
1035                 struct diffstat_file *file = data->files[i];
1036
1037                 if (file->is_binary)
1038                         fprintf(options->file, "-\t-\t");
1039                 else
1040                         fprintf(options->file,
1041                                 "%d\t%d\t", file->added, file->deleted);
1042                 if (options->line_termination) {
1043                         fill_print_name(file);
1044                         if (!file->is_renamed)
1045                                 write_name_quoted(file->name, options->file,
1046                                                   options->line_termination);
1047                         else {
1048                                 fputs(file->print_name, options->file);
1049                                 putc(options->line_termination, options->file);
1050                         }
1051                 } else {
1052                         if (file->is_renamed) {
1053                                 putc('\0', options->file);
1054                                 write_name_quoted(file->from_name, options->file, '\0');
1055                         }
1056                         write_name_quoted(file->name, options->file, '\0');
1057                 }
1058         }
1059 }
1060
1061 struct dirstat_file {
1062         const char *name;
1063         unsigned long changed;
1064 };
1065
1066 struct dirstat_dir {
1067         struct dirstat_file *files;
1068         int alloc, nr, percent, cumulative;
1069 };
1070
1071 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1072 {
1073         unsigned long this_dir = 0;
1074         unsigned int sources = 0;
1075
1076         while (dir->nr) {
1077                 struct dirstat_file *f = dir->files;
1078                 int namelen = strlen(f->name);
1079                 unsigned long this;
1080                 char *slash;
1081
1082                 if (namelen < baselen)
1083                         break;
1084                 if (memcmp(f->name, base, baselen))
1085                         break;
1086                 slash = strchr(f->name + baselen, '/');
1087                 if (slash) {
1088                         int newbaselen = slash + 1 - f->name;
1089                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1090                         sources++;
1091                 } else {
1092                         this = f->changed;
1093                         dir->files++;
1094                         dir->nr--;
1095                         sources += 2;
1096                 }
1097                 this_dir += this;
1098         }
1099
1100         /*
1101          * We don't report dirstat's for
1102          *  - the top level
1103          *  - or cases where everything came from a single directory
1104          *    under this directory (sources == 1).
1105          */
1106         if (baselen && sources != 1) {
1107                 int permille = this_dir * 1000 / changed;
1108                 if (permille) {
1109                         int percent = permille / 10;
1110                         if (percent >= dir->percent) {
1111                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1112                                 if (!dir->cumulative)
1113                                         return 0;
1114                         }
1115                 }
1116         }
1117         return this_dir;
1118 }
1119
1120 static int dirstat_compare(const void *_a, const void *_b)
1121 {
1122         const struct dirstat_file *a = _a;
1123         const struct dirstat_file *b = _b;
1124         return strcmp(a->name, b->name);
1125 }
1126
1127 static void show_dirstat(struct diff_options *options)
1128 {
1129         int i;
1130         unsigned long changed;
1131         struct dirstat_dir dir;
1132         struct diff_queue_struct *q = &diff_queued_diff;
1133
1134         dir.files = NULL;
1135         dir.alloc = 0;
1136         dir.nr = 0;
1137         dir.percent = options->dirstat_percent;
1138         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1139
1140         changed = 0;
1141         for (i = 0; i < q->nr; i++) {
1142                 struct diff_filepair *p = q->queue[i];
1143                 const char *name;
1144                 unsigned long copied, added, damage;
1145
1146                 name = p->one->path ? p->one->path : p->two->path;
1147
1148                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1149                         diff_populate_filespec(p->one, 0);
1150                         diff_populate_filespec(p->two, 0);
1151                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1152                                                &copied, &added);
1153                         diff_free_filespec_data(p->one);
1154                         diff_free_filespec_data(p->two);
1155                 } else if (DIFF_FILE_VALID(p->one)) {
1156                         diff_populate_filespec(p->one, 1);
1157                         copied = added = 0;
1158                         diff_free_filespec_data(p->one);
1159                 } else if (DIFF_FILE_VALID(p->two)) {
1160                         diff_populate_filespec(p->two, 1);
1161                         copied = 0;
1162                         added = p->two->size;
1163                         diff_free_filespec_data(p->two);
1164                 } else
1165                         continue;
1166
1167                 /*
1168                  * Original minus copied is the removed material,
1169                  * added is the new material.  They are both damages
1170                  * made to the preimage. In --dirstat-by-file mode, count
1171                  * damaged files, not damaged lines. This is done by
1172                  * counting only a single damaged line per file.
1173                  */
1174                 damage = (p->one->size - copied) + added;
1175                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1176                         damage = 1;
1177
1178                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1179                 dir.files[dir.nr].name = name;
1180                 dir.files[dir.nr].changed = damage;
1181                 changed += damage;
1182                 dir.nr++;
1183         }
1184
1185         /* This can happen even with many files, if everything was renames */
1186         if (!changed)
1187                 return;
1188
1189         /* Show all directories with more than x% of the changes */
1190         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1191         gather_dirstat(options->file, &dir, changed, "", 0);
1192 }
1193
1194 static void free_diffstat_info(struct diffstat_t *diffstat)
1195 {
1196         int i;
1197         for (i = 0; i < diffstat->nr; i++) {
1198                 struct diffstat_file *f = diffstat->files[i];
1199                 if (f->name != f->print_name)
1200                         free(f->print_name);
1201                 free(f->name);
1202                 free(f->from_name);
1203                 free(f);
1204         }
1205         free(diffstat->files);
1206 }
1207
1208 struct checkdiff_t {
1209         const char *filename;
1210         int lineno;
1211         struct diff_options *o;
1212         unsigned ws_rule;
1213         unsigned status;
1214         int trailing_blanks_start;
1215 };
1216
1217 static int is_conflict_marker(const char *line, unsigned long len)
1218 {
1219         char firstchar;
1220         int cnt;
1221
1222         if (len < 8)
1223                 return 0;
1224         firstchar = line[0];
1225         switch (firstchar) {
1226         case '=': case '>': case '<':
1227                 break;
1228         default:
1229                 return 0;
1230         }
1231         for (cnt = 1; cnt < 7; cnt++)
1232                 if (line[cnt] != firstchar)
1233                         return 0;
1234         /* line[0] thru line[6] are same as firstchar */
1235         if (firstchar == '=') {
1236                 /* divider between ours and theirs? */
1237                 if (len != 8 || line[7] != '\n')
1238                         return 0;
1239         } else if (len < 8 || !isspace(line[7])) {
1240                 /* not divider before ours nor after theirs */
1241                 return 0;
1242         }
1243         return 1;
1244 }
1245
1246 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1247 {
1248         struct checkdiff_t *data = priv;
1249         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1250         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1251         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1252         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1253         char *err;
1254
1255         if (line[0] == '+') {
1256                 unsigned bad;
1257                 data->lineno++;
1258                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1259                         data->trailing_blanks_start = 0;
1260                 else if (!data->trailing_blanks_start)
1261                         data->trailing_blanks_start = data->lineno;
1262                 if (is_conflict_marker(line + 1, len - 1)) {
1263                         data->status |= 1;
1264                         fprintf(data->o->file,
1265                                 "%s:%d: leftover conflict marker\n",
1266                                 data->filename, data->lineno);
1267                 }
1268                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1269                 if (!bad)
1270                         return;
1271                 data->status |= bad;
1272                 err = whitespace_error_string(bad);
1273                 fprintf(data->o->file, "%s:%d: %s.\n",
1274                         data->filename, data->lineno, err);
1275                 free(err);
1276                 emit_line(data->o->file, set, reset, line, 1);
1277                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1278                               data->o->file, set, reset, ws);
1279         } else if (line[0] == ' ') {
1280                 data->lineno++;
1281                 data->trailing_blanks_start = 0;
1282         } else if (line[0] == '@') {
1283                 char *plus = strchr(line, '+');
1284                 if (plus)
1285                         data->lineno = strtol(plus, NULL, 10) - 1;
1286                 else
1287                         die("invalid diff");
1288                 data->trailing_blanks_start = 0;
1289         }
1290 }
1291
1292 static unsigned char *deflate_it(char *data,
1293                                  unsigned long size,
1294                                  unsigned long *result_size)
1295 {
1296         int bound;
1297         unsigned char *deflated;
1298         z_stream stream;
1299
1300         memset(&stream, 0, sizeof(stream));
1301         deflateInit(&stream, zlib_compression_level);
1302         bound = deflateBound(&stream, size);
1303         deflated = xmalloc(bound);
1304         stream.next_out = deflated;
1305         stream.avail_out = bound;
1306
1307         stream.next_in = (unsigned char *)data;
1308         stream.avail_in = size;
1309         while (deflate(&stream, Z_FINISH) == Z_OK)
1310                 ; /* nothing */
1311         deflateEnd(&stream);
1312         *result_size = stream.total_out;
1313         return deflated;
1314 }
1315
1316 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1317 {
1318         void *cp;
1319         void *delta;
1320         void *deflated;
1321         void *data;
1322         unsigned long orig_size;
1323         unsigned long delta_size;
1324         unsigned long deflate_size;
1325         unsigned long data_size;
1326
1327         /* We could do deflated delta, or we could do just deflated two,
1328          * whichever is smaller.
1329          */
1330         delta = NULL;
1331         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1332         if (one->size && two->size) {
1333                 delta = diff_delta(one->ptr, one->size,
1334                                    two->ptr, two->size,
1335                                    &delta_size, deflate_size);
1336                 if (delta) {
1337                         void *to_free = delta;
1338                         orig_size = delta_size;
1339                         delta = deflate_it(delta, delta_size, &delta_size);
1340                         free(to_free);
1341                 }
1342         }
1343
1344         if (delta && delta_size < deflate_size) {
1345                 fprintf(file, "delta %lu\n", orig_size);
1346                 free(deflated);
1347                 data = delta;
1348                 data_size = delta_size;
1349         }
1350         else {
1351                 fprintf(file, "literal %lu\n", two->size);
1352                 free(delta);
1353                 data = deflated;
1354                 data_size = deflate_size;
1355         }
1356
1357         /* emit data encoded in base85 */
1358         cp = data;
1359         while (data_size) {
1360                 int bytes = (52 < data_size) ? 52 : data_size;
1361                 char line[70];
1362                 data_size -= bytes;
1363                 if (bytes <= 26)
1364                         line[0] = bytes + 'A' - 1;
1365                 else
1366                         line[0] = bytes - 26 + 'a' - 1;
1367                 encode_85(line + 1, cp, bytes);
1368                 cp = (char *) cp + bytes;
1369                 fputs(line, file);
1370                 fputc('\n', file);
1371         }
1372         fprintf(file, "\n");
1373         free(data);
1374 }
1375
1376 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1377 {
1378         fprintf(file, "GIT binary patch\n");
1379         emit_binary_diff_body(file, one, two);
1380         emit_binary_diff_body(file, two, one);
1381 }
1382
1383 static void diff_filespec_load_driver(struct diff_filespec *one)
1384 {
1385         if (!one->driver)
1386                 one->driver = userdiff_find_by_path(one->path);
1387         if (!one->driver)
1388                 one->driver = userdiff_find_by_name("default");
1389 }
1390
1391 int diff_filespec_is_binary(struct diff_filespec *one)
1392 {
1393         if (one->is_binary == -1) {
1394                 diff_filespec_load_driver(one);
1395                 if (one->driver->binary != -1)
1396                         one->is_binary = one->driver->binary;
1397                 else {
1398                         if (!one->data && DIFF_FILE_VALID(one))
1399                                 diff_populate_filespec(one, 0);
1400                         if (one->data)
1401                                 one->is_binary = buffer_is_binary(one->data,
1402                                                 one->size);
1403                         if (one->is_binary == -1)
1404                                 one->is_binary = 0;
1405                 }
1406         }
1407         return one->is_binary;
1408 }
1409
1410 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1411 {
1412         diff_filespec_load_driver(one);
1413         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1414 }
1415
1416 static const char *userdiff_word_regex(struct diff_filespec *one)
1417 {
1418         diff_filespec_load_driver(one);
1419         return one->driver->word_regex;
1420 }
1421
1422 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1423 {
1424         if (!options->a_prefix)
1425                 options->a_prefix = a;
1426         if (!options->b_prefix)
1427                 options->b_prefix = b;
1428 }
1429
1430 static const char *get_textconv(struct diff_filespec *one)
1431 {
1432         if (!DIFF_FILE_VALID(one))
1433                 return NULL;
1434         if (!S_ISREG(one->mode))
1435                 return NULL;
1436         diff_filespec_load_driver(one);
1437         return one->driver->textconv;
1438 }
1439
1440 static void builtin_diff(const char *name_a,
1441                          const char *name_b,
1442                          struct diff_filespec *one,
1443                          struct diff_filespec *two,
1444                          const char *xfrm_msg,
1445                          struct diff_options *o,
1446                          int complete_rewrite)
1447 {
1448         mmfile_t mf1, mf2;
1449         const char *lbl[2];
1450         char *a_one, *b_two;
1451         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1452         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1453         const char *a_prefix, *b_prefix;
1454         const char *textconv_one = NULL, *textconv_two = NULL;
1455
1456         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1457                 textconv_one = get_textconv(one);
1458                 textconv_two = get_textconv(two);
1459         }
1460
1461         diff_set_mnemonic_prefix(o, "a/", "b/");
1462         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1463                 a_prefix = o->b_prefix;
1464                 b_prefix = o->a_prefix;
1465         } else {
1466                 a_prefix = o->a_prefix;
1467                 b_prefix = o->b_prefix;
1468         }
1469
1470         /* Never use a non-valid filename anywhere if at all possible */
1471         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1472         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1473
1474         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1475         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1476         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1477         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1478         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1479         if (lbl[0][0] == '/') {
1480                 /* /dev/null */
1481                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1482                 if (xfrm_msg && xfrm_msg[0])
1483                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1484         }
1485         else if (lbl[1][0] == '/') {
1486                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1487                 if (xfrm_msg && xfrm_msg[0])
1488                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1489         }
1490         else {
1491                 if (one->mode != two->mode) {
1492                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1493                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1494                 }
1495                 if (xfrm_msg && xfrm_msg[0])
1496                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1497                 /*
1498                  * we do not run diff between different kind
1499                  * of objects.
1500                  */
1501                 if ((one->mode ^ two->mode) & S_IFMT)
1502                         goto free_ab_and_return;
1503                 if (complete_rewrite &&
1504                     (textconv_one || !diff_filespec_is_binary(one)) &&
1505                     (textconv_two || !diff_filespec_is_binary(two))) {
1506                         emit_rewrite_diff(name_a, name_b, one, two,
1507                                                 textconv_one, textconv_two, o);
1508                         o->found_changes = 1;
1509                         goto free_ab_and_return;
1510                 }
1511         }
1512
1513         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1514                 die("unable to read files to diff");
1515
1516         if (!DIFF_OPT_TST(o, TEXT) &&
1517             ( (diff_filespec_is_binary(one) && !textconv_one) ||
1518               (diff_filespec_is_binary(two) && !textconv_two) )) {
1519                 /* Quite common confusing case */
1520                 if (mf1.size == mf2.size &&
1521                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1522                         goto free_ab_and_return;
1523                 if (DIFF_OPT_TST(o, BINARY))
1524                         emit_binary_diff(o->file, &mf1, &mf2);
1525                 else
1526                         fprintf(o->file, "Binary files %s and %s differ\n",
1527                                 lbl[0], lbl[1]);
1528                 o->found_changes = 1;
1529         }
1530         else {
1531                 /* Crazy xdl interfaces.. */
1532                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1533                 xpparam_t xpp;
1534                 xdemitconf_t xecfg;
1535                 xdemitcb_t ecb;
1536                 struct emit_callback ecbdata;
1537                 const struct userdiff_funcname *pe;
1538
1539                 if (textconv_one) {
1540                         size_t size;
1541                         mf1.ptr = run_textconv(textconv_one, one, &size);
1542                         if (!mf1.ptr)
1543                                 die("unable to read files to diff");
1544                         mf1.size = size;
1545                 }
1546                 if (textconv_two) {
1547                         size_t size;
1548                         mf2.ptr = run_textconv(textconv_two, two, &size);
1549                         if (!mf2.ptr)
1550                                 die("unable to read files to diff");
1551                         mf2.size = size;
1552                 }
1553
1554                 pe = diff_funcname_pattern(one);
1555                 if (!pe)
1556                         pe = diff_funcname_pattern(two);
1557
1558                 memset(&xpp, 0, sizeof(xpp));
1559                 memset(&xecfg, 0, sizeof(xecfg));
1560                 memset(&ecbdata, 0, sizeof(ecbdata));
1561                 ecbdata.label_path = lbl;
1562                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1563                 ecbdata.found_changesp = &o->found_changes;
1564                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1565                 ecbdata.file = o->file;
1566                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1567                 xecfg.ctxlen = o->context;
1568                 xecfg.interhunkctxlen = o->interhunkcontext;
1569                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1570                 if (pe)
1571                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1572                 if (!diffopts)
1573                         ;
1574                 else if (!prefixcmp(diffopts, "--unified="))
1575                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1576                 else if (!prefixcmp(diffopts, "-u"))
1577                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1578                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1579                         ecbdata.diff_words =
1580                                 xcalloc(1, sizeof(struct diff_words_data));
1581                         ecbdata.diff_words->file = o->file;
1582                         if (!o->word_regex)
1583                                 o->word_regex = userdiff_word_regex(one);
1584                         if (!o->word_regex)
1585                                 o->word_regex = userdiff_word_regex(two);
1586                         if (!o->word_regex)
1587                                 o->word_regex = diff_word_regex_cfg;
1588                         if (o->word_regex) {
1589                                 ecbdata.diff_words->word_regex = (regex_t *)
1590                                         xmalloc(sizeof(regex_t));
1591                                 if (regcomp(ecbdata.diff_words->word_regex,
1592                                                 o->word_regex,
1593                                                 REG_EXTENDED | REG_NEWLINE))
1594                                         die ("Invalid regular expression: %s",
1595                                                         o->word_regex);
1596                         }
1597                 }
1598                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1599                               &xpp, &xecfg, &ecb);
1600                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1601                         free_diff_words_data(&ecbdata);
1602                 if (textconv_one)
1603                         free(mf1.ptr);
1604                 if (textconv_two)
1605                         free(mf2.ptr);
1606         }
1607
1608  free_ab_and_return:
1609         diff_free_filespec_data(one);
1610         diff_free_filespec_data(two);
1611         free(a_one);
1612         free(b_two);
1613         return;
1614 }
1615
1616 static void builtin_diffstat(const char *name_a, const char *name_b,
1617                              struct diff_filespec *one,
1618                              struct diff_filespec *two,
1619                              struct diffstat_t *diffstat,
1620                              struct diff_options *o,
1621                              int complete_rewrite)
1622 {
1623         mmfile_t mf1, mf2;
1624         struct diffstat_file *data;
1625
1626         data = diffstat_add(diffstat, name_a, name_b);
1627
1628         if (!one || !two) {
1629                 data->is_unmerged = 1;
1630                 return;
1631         }
1632         if (complete_rewrite) {
1633                 diff_populate_filespec(one, 0);
1634                 diff_populate_filespec(two, 0);
1635                 data->deleted = count_lines(one->data, one->size);
1636                 data->added = count_lines(two->data, two->size);
1637                 goto free_and_return;
1638         }
1639         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1640                 die("unable to read files to diff");
1641
1642         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1643                 data->is_binary = 1;
1644                 data->added = mf2.size;
1645                 data->deleted = mf1.size;
1646         } else {
1647                 /* Crazy xdl interfaces.. */
1648                 xpparam_t xpp;
1649                 xdemitconf_t xecfg;
1650                 xdemitcb_t ecb;
1651
1652                 memset(&xpp, 0, sizeof(xpp));
1653                 memset(&xecfg, 0, sizeof(xecfg));
1654                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1655                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1656                               &xpp, &xecfg, &ecb);
1657         }
1658
1659  free_and_return:
1660         diff_free_filespec_data(one);
1661         diff_free_filespec_data(two);
1662 }
1663
1664 static void builtin_checkdiff(const char *name_a, const char *name_b,
1665                               const char *attr_path,
1666                               struct diff_filespec *one,
1667                               struct diff_filespec *two,
1668                               struct diff_options *o)
1669 {
1670         mmfile_t mf1, mf2;
1671         struct checkdiff_t data;
1672
1673         if (!two)
1674                 return;
1675
1676         memset(&data, 0, sizeof(data));
1677         data.filename = name_b ? name_b : name_a;
1678         data.lineno = 0;
1679         data.o = o;
1680         data.ws_rule = whitespace_rule(attr_path);
1681
1682         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1683                 die("unable to read files to diff");
1684
1685         /*
1686          * All the other codepaths check both sides, but not checking
1687          * the "old" side here is deliberate.  We are checking the newly
1688          * introduced changes, and as long as the "new" side is text, we
1689          * can and should check what it introduces.
1690          */
1691         if (diff_filespec_is_binary(two))
1692                 goto free_and_return;
1693         else {
1694                 /* Crazy xdl interfaces.. */
1695                 xpparam_t xpp;
1696                 xdemitconf_t xecfg;
1697                 xdemitcb_t ecb;
1698
1699                 memset(&xpp, 0, sizeof(xpp));
1700                 memset(&xecfg, 0, sizeof(xecfg));
1701                 xecfg.ctxlen = 1; /* at least one context line */
1702                 xpp.flags = XDF_NEED_MINIMAL;
1703                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1704                               &xpp, &xecfg, &ecb);
1705
1706                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1707                     data.trailing_blanks_start) {
1708                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1709                                 data.filename, data.trailing_blanks_start);
1710                         data.status = 1; /* report errors */
1711                 }
1712         }
1713  free_and_return:
1714         diff_free_filespec_data(one);
1715         diff_free_filespec_data(two);
1716         if (data.status)
1717                 DIFF_OPT_SET(o, CHECK_FAILED);
1718 }
1719
1720 struct diff_filespec *alloc_filespec(const char *path)
1721 {
1722         int namelen = strlen(path);
1723         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1724
1725         memset(spec, 0, sizeof(*spec));
1726         spec->path = (char *)(spec + 1);
1727         memcpy(spec->path, path, namelen+1);
1728         spec->count = 1;
1729         spec->is_binary = -1;
1730         return spec;
1731 }
1732
1733 void free_filespec(struct diff_filespec *spec)
1734 {
1735         if (!--spec->count) {
1736                 diff_free_filespec_data(spec);
1737                 free(spec);
1738         }
1739 }
1740
1741 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1742                    unsigned short mode)
1743 {
1744         if (mode) {
1745                 spec->mode = canon_mode(mode);
1746                 hashcpy(spec->sha1, sha1);
1747                 spec->sha1_valid = !is_null_sha1(sha1);
1748         }
1749 }
1750
1751 /*
1752  * Given a name and sha1 pair, if the index tells us the file in
1753  * the work tree has that object contents, return true, so that
1754  * prepare_temp_file() does not have to inflate and extract.
1755  */
1756 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1757 {
1758         struct cache_entry *ce;
1759         struct stat st;
1760         int pos, len;
1761
1762         /*
1763          * We do not read the cache ourselves here, because the
1764          * benchmark with my previous version that always reads cache
1765          * shows that it makes things worse for diff-tree comparing
1766          * two linux-2.6 kernel trees in an already checked out work
1767          * tree.  This is because most diff-tree comparisons deal with
1768          * only a small number of files, while reading the cache is
1769          * expensive for a large project, and its cost outweighs the
1770          * savings we get by not inflating the object to a temporary
1771          * file.  Practically, this code only helps when we are used
1772          * by diff-cache --cached, which does read the cache before
1773          * calling us.
1774          */
1775         if (!active_cache)
1776                 return 0;
1777
1778         /* We want to avoid the working directory if our caller
1779          * doesn't need the data in a normal file, this system
1780          * is rather slow with its stat/open/mmap/close syscalls,
1781          * and the object is contained in a pack file.  The pack
1782          * is probably already open and will be faster to obtain
1783          * the data through than the working directory.  Loose
1784          * objects however would tend to be slower as they need
1785          * to be individually opened and inflated.
1786          */
1787         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1788                 return 0;
1789
1790         len = strlen(name);
1791         pos = cache_name_pos(name, len);
1792         if (pos < 0)
1793                 return 0;
1794         ce = active_cache[pos];
1795
1796         /*
1797          * This is not the sha1 we are looking for, or
1798          * unreusable because it is not a regular file.
1799          */
1800         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1801                 return 0;
1802
1803         /*
1804          * If ce is marked as "assume unchanged", there is no
1805          * guarantee that work tree matches what we are looking for.
1806          */
1807         if (ce->ce_flags & CE_VALID)
1808                 return 0;
1809
1810         /*
1811          * If ce matches the file in the work tree, we can reuse it.
1812          */
1813         if (ce_uptodate(ce) ||
1814             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1815                 return 1;
1816
1817         return 0;
1818 }
1819
1820 static int populate_from_stdin(struct diff_filespec *s)
1821 {
1822         struct strbuf buf = STRBUF_INIT;
1823         size_t size = 0;
1824
1825         if (strbuf_read(&buf, 0, 0) < 0)
1826                 return error("error while reading from stdin %s",
1827                                      strerror(errno));
1828
1829         s->should_munmap = 0;
1830         s->data = strbuf_detach(&buf, &size);
1831         s->size = size;
1832         s->should_free = 1;
1833         return 0;
1834 }
1835
1836 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1837 {
1838         int len;
1839         char *data = xmalloc(100);
1840         len = snprintf(data, 100,
1841                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1842         s->data = data;
1843         s->size = len;
1844         s->should_free = 1;
1845         if (size_only) {
1846                 s->data = NULL;
1847                 free(data);
1848         }
1849         return 0;
1850 }
1851
1852 /*
1853  * While doing rename detection and pickaxe operation, we may need to
1854  * grab the data for the blob (or file) for our own in-core comparison.
1855  * diff_filespec has data and size fields for this purpose.
1856  */
1857 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1858 {
1859         int err = 0;
1860         if (!DIFF_FILE_VALID(s))
1861                 die("internal error: asking to populate invalid file.");
1862         if (S_ISDIR(s->mode))
1863                 return -1;
1864
1865         if (s->data)
1866                 return 0;
1867
1868         if (size_only && 0 < s->size)
1869                 return 0;
1870
1871         if (S_ISGITLINK(s->mode))
1872                 return diff_populate_gitlink(s, size_only);
1873
1874         if (!s->sha1_valid ||
1875             reuse_worktree_file(s->path, s->sha1, 0)) {
1876                 struct strbuf buf = STRBUF_INIT;
1877                 struct stat st;
1878                 int fd;
1879
1880                 if (!strcmp(s->path, "-"))
1881                         return populate_from_stdin(s);
1882
1883                 if (lstat(s->path, &st) < 0) {
1884                         if (errno == ENOENT) {
1885                         err_empty:
1886                                 err = -1;
1887                         empty:
1888                                 s->data = (char *)"";
1889                                 s->size = 0;
1890                                 return err;
1891                         }
1892                 }
1893                 s->size = xsize_t(st.st_size);
1894                 if (!s->size)
1895                         goto empty;
1896                 if (S_ISLNK(st.st_mode)) {
1897                         struct strbuf sb = STRBUF_INIT;
1898
1899                         if (strbuf_readlink(&sb, s->path, s->size))
1900                                 goto err_empty;
1901                         s->size = sb.len;
1902                         s->data = strbuf_detach(&sb, NULL);
1903                         s->should_free = 1;
1904                         return 0;
1905                 }
1906                 if (size_only)
1907                         return 0;
1908                 fd = open(s->path, O_RDONLY);
1909                 if (fd < 0)
1910                         goto err_empty;
1911                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1912                 close(fd);
1913                 s->should_munmap = 1;
1914
1915                 /*
1916                  * Convert from working tree format to canonical git format
1917                  */
1918                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1919                         size_t size = 0;
1920                         munmap(s->data, s->size);
1921                         s->should_munmap = 0;
1922                         s->data = strbuf_detach(&buf, &size);
1923                         s->size = size;
1924                         s->should_free = 1;
1925                 }
1926         }
1927         else {
1928                 enum object_type type;
1929                 if (size_only)
1930                         type = sha1_object_info(s->sha1, &s->size);
1931                 else {
1932                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1933                         s->should_free = 1;
1934                 }
1935         }
1936         return 0;
1937 }
1938
1939 void diff_free_filespec_blob(struct diff_filespec *s)
1940 {
1941         if (s->should_free)
1942                 free(s->data);
1943         else if (s->should_munmap)
1944                 munmap(s->data, s->size);
1945
1946         if (s->should_free || s->should_munmap) {
1947                 s->should_free = s->should_munmap = 0;
1948                 s->data = NULL;
1949         }
1950 }
1951
1952 void diff_free_filespec_data(struct diff_filespec *s)
1953 {
1954         diff_free_filespec_blob(s);
1955         free(s->cnt_data);
1956         s->cnt_data = NULL;
1957 }
1958
1959 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
1960                            void *blob,
1961                            unsigned long size,
1962                            const unsigned char *sha1,
1963                            int mode)
1964 {
1965         int fd;
1966         struct strbuf buf = STRBUF_INIT;
1967         struct strbuf template = STRBUF_INIT;
1968         char *path_dup = xstrdup(path);
1969         const char *base = basename(path_dup);
1970
1971         /* Generate "XXXXXX_basename.ext" */
1972         strbuf_addstr(&template, "XXXXXX_");
1973         strbuf_addstr(&template, base);
1974
1975         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
1976                         strlen(base) + 1);
1977         if (fd < 0)
1978                 die("unable to create temp-file: %s", strerror(errno));
1979         if (convert_to_working_tree(path,
1980                         (const char *)blob, (size_t)size, &buf)) {
1981                 blob = buf.buf;
1982                 size = buf.len;
1983         }
1984         if (write_in_full(fd, blob, size) != size)
1985                 die("unable to write temp-file");
1986         close(fd);
1987         temp->name = temp->tmp_path;
1988         strcpy(temp->hex, sha1_to_hex(sha1));
1989         temp->hex[40] = 0;
1990         sprintf(temp->mode, "%06o", mode);
1991         strbuf_release(&buf);
1992         strbuf_release(&template);
1993         free(path_dup);
1994 }
1995
1996 static struct diff_tempfile *prepare_temp_file(const char *name,
1997                 struct diff_filespec *one)
1998 {
1999         struct diff_tempfile *temp = claim_diff_tempfile();
2000
2001         if (!DIFF_FILE_VALID(one)) {
2002         not_a_valid_file:
2003                 /* A '-' entry produces this for file-2, and
2004                  * a '+' entry produces this for file-1.
2005                  */
2006                 temp->name = "/dev/null";
2007                 strcpy(temp->hex, ".");
2008                 strcpy(temp->mode, ".");
2009                 return temp;
2010         }
2011
2012         if (!remove_tempfile_installed) {
2013                 atexit(remove_tempfile);
2014                 sigchain_push_common(remove_tempfile_on_signal);
2015                 remove_tempfile_installed = 1;
2016         }
2017
2018         if (!one->sha1_valid ||
2019             reuse_worktree_file(name, one->sha1, 1)) {
2020                 struct stat st;
2021                 if (lstat(name, &st) < 0) {
2022                         if (errno == ENOENT)
2023                                 goto not_a_valid_file;
2024                         die("stat(%s): %s", name, strerror(errno));
2025                 }
2026                 if (S_ISLNK(st.st_mode)) {
2027                         struct strbuf sb = STRBUF_INIT;
2028                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2029                                 die("readlink(%s)", name);
2030                         prep_temp_blob(name, temp, sb.buf, sb.len,
2031                                        (one->sha1_valid ?
2032                                         one->sha1 : null_sha1),
2033                                        (one->sha1_valid ?
2034                                         one->mode : S_IFLNK));
2035                         strbuf_release(&sb);
2036                 }
2037                 else {
2038                         /* we can borrow from the file in the work tree */
2039                         temp->name = name;
2040                         if (!one->sha1_valid)
2041                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2042                         else
2043                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2044                         /* Even though we may sometimes borrow the
2045                          * contents from the work tree, we always want
2046                          * one->mode.  mode is trustworthy even when
2047                          * !(one->sha1_valid), as long as
2048                          * DIFF_FILE_VALID(one).
2049                          */
2050                         sprintf(temp->mode, "%06o", one->mode);
2051                 }
2052                 return temp;
2053         }
2054         else {
2055                 if (diff_populate_filespec(one, 0))
2056                         die("cannot read data blob for %s", one->path);
2057                 prep_temp_blob(name, temp, one->data, one->size,
2058                                one->sha1, one->mode);
2059         }
2060         return temp;
2061 }
2062
2063 /* An external diff command takes:
2064  *
2065  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2066  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2067  *
2068  */
2069 static void run_external_diff(const char *pgm,
2070                               const char *name,
2071                               const char *other,
2072                               struct diff_filespec *one,
2073                               struct diff_filespec *two,
2074                               const char *xfrm_msg,
2075                               int complete_rewrite)
2076 {
2077         const char *spawn_arg[10];
2078         int retval;
2079         const char **arg = &spawn_arg[0];
2080
2081         if (one && two) {
2082                 struct diff_tempfile *temp_one, *temp_two;
2083                 const char *othername = (other ? other : name);
2084                 temp_one = prepare_temp_file(name, one);
2085                 temp_two = prepare_temp_file(othername, two);
2086                 *arg++ = pgm;
2087                 *arg++ = name;
2088                 *arg++ = temp_one->name;
2089                 *arg++ = temp_one->hex;
2090                 *arg++ = temp_one->mode;
2091                 *arg++ = temp_two->name;
2092                 *arg++ = temp_two->hex;
2093                 *arg++ = temp_two->mode;
2094                 if (other) {
2095                         *arg++ = other;
2096                         *arg++ = xfrm_msg;
2097                 }
2098         } else {
2099                 *arg++ = pgm;
2100                 *arg++ = name;
2101         }
2102         *arg = NULL;
2103         fflush(NULL);
2104         retval = run_command_v_opt(spawn_arg, 0);
2105         remove_tempfile();
2106         if (retval) {
2107                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2108                 exit(1);
2109         }
2110 }
2111
2112 static int similarity_index(struct diff_filepair *p)
2113 {
2114         return p->score * 100 / MAX_SCORE;
2115 }
2116
2117 static void fill_metainfo(struct strbuf *msg,
2118                           const char *name,
2119                           const char *other,
2120                           struct diff_filespec *one,
2121                           struct diff_filespec *two,
2122                           struct diff_options *o,
2123                           struct diff_filepair *p)
2124 {
2125         strbuf_init(msg, PATH_MAX * 2 + 300);
2126         switch (p->status) {
2127         case DIFF_STATUS_COPIED:
2128                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2129                 strbuf_addstr(msg, "\ncopy from ");
2130                 quote_c_style(name, msg, NULL, 0);
2131                 strbuf_addstr(msg, "\ncopy to ");
2132                 quote_c_style(other, msg, NULL, 0);
2133                 strbuf_addch(msg, '\n');
2134                 break;
2135         case DIFF_STATUS_RENAMED:
2136                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2137                 strbuf_addstr(msg, "\nrename from ");
2138                 quote_c_style(name, msg, NULL, 0);
2139                 strbuf_addstr(msg, "\nrename to ");
2140                 quote_c_style(other, msg, NULL, 0);
2141                 strbuf_addch(msg, '\n');
2142                 break;
2143         case DIFF_STATUS_MODIFIED:
2144                 if (p->score) {
2145                         strbuf_addf(msg, "dissimilarity index %d%%\n",
2146                                     similarity_index(p));
2147                         break;
2148                 }
2149                 /* fallthru */
2150         default:
2151                 /* nothing */
2152                 ;
2153         }
2154         if (one && two && hashcmp(one->sha1, two->sha1)) {
2155                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2156
2157                 if (DIFF_OPT_TST(o, BINARY)) {
2158                         mmfile_t mf;
2159                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2160                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2161                                 abbrev = 40;
2162                 }
2163                 strbuf_addf(msg, "index %.*s..%.*s",
2164                             abbrev, sha1_to_hex(one->sha1),
2165                             abbrev, sha1_to_hex(two->sha1));
2166                 if (one->mode == two->mode)
2167                         strbuf_addf(msg, " %06o", one->mode);
2168                 strbuf_addch(msg, '\n');
2169         }
2170         if (msg->len)
2171                 strbuf_setlen(msg, msg->len - 1);
2172 }
2173
2174 static void run_diff_cmd(const char *pgm,
2175                          const char *name,
2176                          const char *other,
2177                          const char *attr_path,
2178                          struct diff_filespec *one,
2179                          struct diff_filespec *two,
2180                          struct strbuf *msg,
2181                          struct diff_options *o,
2182                          struct diff_filepair *p)
2183 {
2184         const char *xfrm_msg = NULL;
2185         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2186
2187         if (msg) {
2188                 fill_metainfo(msg, name, other, one, two, o, p);
2189                 xfrm_msg = msg->len ? msg->buf : NULL;
2190         }
2191
2192         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2193                 pgm = NULL;
2194         else {
2195                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2196                 if (drv && drv->external)
2197                         pgm = drv->external;
2198         }
2199
2200         if (pgm) {
2201                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2202                                   complete_rewrite);
2203                 return;
2204         }
2205         if (one && two)
2206                 builtin_diff(name, other ? other : name,
2207                              one, two, xfrm_msg, o, complete_rewrite);
2208         else
2209                 fprintf(o->file, "* Unmerged path %s\n", name);
2210 }
2211
2212 static void diff_fill_sha1_info(struct diff_filespec *one)
2213 {
2214         if (DIFF_FILE_VALID(one)) {
2215                 if (!one->sha1_valid) {
2216                         struct stat st;
2217                         if (!strcmp(one->path, "-")) {
2218                                 hashcpy(one->sha1, null_sha1);
2219                                 return;
2220                         }
2221                         if (lstat(one->path, &st) < 0)
2222                                 die("stat %s", one->path);
2223                         if (index_path(one->sha1, one->path, &st, 0))
2224                                 die("cannot hash %s", one->path);
2225                 }
2226         }
2227         else
2228                 hashclr(one->sha1);
2229 }
2230
2231 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2232 {
2233         /* Strip the prefix but do not molest /dev/null and absolute paths */
2234         if (*namep && **namep != '/')
2235                 *namep += prefix_length;
2236         if (*otherp && **otherp != '/')
2237                 *otherp += prefix_length;
2238 }
2239
2240 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2241 {
2242         const char *pgm = external_diff();
2243         struct strbuf msg;
2244         struct diff_filespec *one = p->one;
2245         struct diff_filespec *two = p->two;
2246         const char *name;
2247         const char *other;
2248         const char *attr_path;
2249
2250         name  = p->one->path;
2251         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2252         attr_path = name;
2253         if (o->prefix_length)
2254                 strip_prefix(o->prefix_length, &name, &other);
2255
2256         if (DIFF_PAIR_UNMERGED(p)) {
2257                 run_diff_cmd(pgm, name, NULL, attr_path,
2258                              NULL, NULL, NULL, o, p);
2259                 return;
2260         }
2261
2262         diff_fill_sha1_info(one);
2263         diff_fill_sha1_info(two);
2264
2265         if (!pgm &&
2266             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2267             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2268                 /*
2269                  * a filepair that changes between file and symlink
2270                  * needs to be split into deletion and creation.
2271                  */
2272                 struct diff_filespec *null = alloc_filespec(two->path);
2273                 run_diff_cmd(NULL, name, other, attr_path,
2274                              one, null, &msg, o, p);
2275                 free(null);
2276                 strbuf_release(&msg);
2277
2278                 null = alloc_filespec(one->path);
2279                 run_diff_cmd(NULL, name, other, attr_path,
2280                              null, two, &msg, o, p);
2281                 free(null);
2282         }
2283         else
2284                 run_diff_cmd(pgm, name, other, attr_path,
2285                              one, two, &msg, o, p);
2286
2287         strbuf_release(&msg);
2288 }
2289
2290 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2291                          struct diffstat_t *diffstat)
2292 {
2293         const char *name;
2294         const char *other;
2295         int complete_rewrite = 0;
2296
2297         if (DIFF_PAIR_UNMERGED(p)) {
2298                 /* unmerged */
2299                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2300                 return;
2301         }
2302
2303         name = p->one->path;
2304         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2305
2306         if (o->prefix_length)
2307                 strip_prefix(o->prefix_length, &name, &other);
2308
2309         diff_fill_sha1_info(p->one);
2310         diff_fill_sha1_info(p->two);
2311
2312         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2313                 complete_rewrite = 1;
2314         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2315 }
2316
2317 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2318 {
2319         const char *name;
2320         const char *other;
2321         const char *attr_path;
2322
2323         if (DIFF_PAIR_UNMERGED(p)) {
2324                 /* unmerged */
2325                 return;
2326         }
2327
2328         name = p->one->path;
2329         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2330         attr_path = other ? other : name;
2331
2332         if (o->prefix_length)
2333                 strip_prefix(o->prefix_length, &name, &other);
2334
2335         diff_fill_sha1_info(p->one);
2336         diff_fill_sha1_info(p->two);
2337
2338         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2339 }
2340
2341 void diff_setup(struct diff_options *options)
2342 {
2343         memset(options, 0, sizeof(*options));
2344
2345         options->file = stdout;
2346
2347         options->line_termination = '\n';
2348         options->break_opt = -1;
2349         options->rename_limit = -1;
2350         options->dirstat_percent = 3;
2351         options->context = 3;
2352
2353         options->change = diff_change;
2354         options->add_remove = diff_addremove;
2355         if (diff_use_color_default > 0)
2356                 DIFF_OPT_SET(options, COLOR_DIFF);
2357         options->detect_rename = diff_detect_rename_default;
2358
2359         if (!diff_mnemonic_prefix) {
2360                 options->a_prefix = "a/";
2361                 options->b_prefix = "b/";
2362         }
2363 }
2364
2365 int diff_setup_done(struct diff_options *options)
2366 {
2367         int count = 0;
2368
2369         if (options->output_format & DIFF_FORMAT_NAME)
2370                 count++;
2371         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2372                 count++;
2373         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2374                 count++;
2375         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2376                 count++;
2377         if (count > 1)
2378                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2379
2380         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2381                 options->detect_rename = DIFF_DETECT_COPY;
2382
2383         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2384                 options->prefix = NULL;
2385         if (options->prefix)
2386                 options->prefix_length = strlen(options->prefix);
2387         else
2388                 options->prefix_length = 0;
2389
2390         if (options->output_format & (DIFF_FORMAT_NAME |
2391                                       DIFF_FORMAT_NAME_STATUS |
2392                                       DIFF_FORMAT_CHECKDIFF |
2393                                       DIFF_FORMAT_NO_OUTPUT))
2394                 options->output_format &= ~(DIFF_FORMAT_RAW |
2395                                             DIFF_FORMAT_NUMSTAT |
2396                                             DIFF_FORMAT_DIFFSTAT |
2397                                             DIFF_FORMAT_SHORTSTAT |
2398                                             DIFF_FORMAT_DIRSTAT |
2399                                             DIFF_FORMAT_SUMMARY |
2400                                             DIFF_FORMAT_PATCH);
2401
2402         /*
2403          * These cases always need recursive; we do not drop caller-supplied
2404          * recursive bits for other formats here.
2405          */
2406         if (options->output_format & (DIFF_FORMAT_PATCH |
2407                                       DIFF_FORMAT_NUMSTAT |
2408                                       DIFF_FORMAT_DIFFSTAT |
2409                                       DIFF_FORMAT_SHORTSTAT |
2410                                       DIFF_FORMAT_DIRSTAT |
2411                                       DIFF_FORMAT_SUMMARY |
2412                                       DIFF_FORMAT_CHECKDIFF))
2413                 DIFF_OPT_SET(options, RECURSIVE);
2414         /*
2415          * Also pickaxe would not work very well if you do not say recursive
2416          */
2417         if (options->pickaxe)
2418                 DIFF_OPT_SET(options, RECURSIVE);
2419
2420         if (options->detect_rename && options->rename_limit < 0)
2421                 options->rename_limit = diff_rename_limit_default;
2422         if (options->setup & DIFF_SETUP_USE_CACHE) {
2423                 if (!active_cache)
2424                         /* read-cache does not die even when it fails
2425                          * so it is safe for us to do this here.  Also
2426                          * it does not smudge active_cache or active_nr
2427                          * when it fails, so we do not have to worry about
2428                          * cleaning it up ourselves either.
2429                          */
2430                         read_cache();
2431         }
2432         if (options->abbrev <= 0 || 40 < options->abbrev)
2433                 options->abbrev = 40; /* full */
2434
2435         /*
2436          * It does not make sense to show the first hit we happened
2437          * to have found.  It does not make sense not to return with
2438          * exit code in such a case either.
2439          */
2440         if (DIFF_OPT_TST(options, QUIET)) {
2441                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2442                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2443         }
2444
2445         return 0;
2446 }
2447
2448 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2449 {
2450         char c, *eq;
2451         int len;
2452
2453         if (*arg != '-')
2454                 return 0;
2455         c = *++arg;
2456         if (!c)
2457                 return 0;
2458         if (c == arg_short) {
2459                 c = *++arg;
2460                 if (!c)
2461                         return 1;
2462                 if (val && isdigit(c)) {
2463                         char *end;
2464                         int n = strtoul(arg, &end, 10);
2465                         if (*end)
2466                                 return 0;
2467                         *val = n;
2468                         return 1;
2469                 }
2470                 return 0;
2471         }
2472         if (c != '-')
2473                 return 0;
2474         arg++;
2475         eq = strchr(arg, '=');
2476         if (eq)
2477                 len = eq - arg;
2478         else
2479                 len = strlen(arg);
2480         if (!len || strncmp(arg, arg_long, len))
2481                 return 0;
2482         if (eq) {
2483                 int n;
2484                 char *end;
2485                 if (!isdigit(*++eq))
2486                         return 0;
2487                 n = strtoul(eq, &end, 10);
2488                 if (*end)
2489                         return 0;
2490                 *val = n;
2491         }
2492         return 1;
2493 }
2494
2495 static int diff_scoreopt_parse(const char *opt);
2496
2497 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2498 {
2499         const char *arg = av[0];
2500
2501         /* Output format options */
2502         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2503                 options->output_format |= DIFF_FORMAT_PATCH;
2504         else if (opt_arg(arg, 'U', "unified", &options->context))
2505                 options->output_format |= DIFF_FORMAT_PATCH;
2506         else if (!strcmp(arg, "--raw"))
2507                 options->output_format |= DIFF_FORMAT_RAW;
2508         else if (!strcmp(arg, "--patch-with-raw"))
2509                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2510         else if (!strcmp(arg, "--numstat"))
2511                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2512         else if (!strcmp(arg, "--shortstat"))
2513                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2514         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2515                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2516         else if (!strcmp(arg, "--cumulative")) {
2517                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2518                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2519         } else if (opt_arg(arg, 0, "dirstat-by-file",
2520                            &options->dirstat_percent)) {
2521                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2522                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2523         }
2524         else if (!strcmp(arg, "--check"))
2525                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2526         else if (!strcmp(arg, "--summary"))
2527                 options->output_format |= DIFF_FORMAT_SUMMARY;
2528         else if (!strcmp(arg, "--patch-with-stat"))
2529                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2530         else if (!strcmp(arg, "--name-only"))
2531                 options->output_format |= DIFF_FORMAT_NAME;
2532         else if (!strcmp(arg, "--name-status"))
2533                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2534         else if (!strcmp(arg, "-s"))
2535                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2536         else if (!prefixcmp(arg, "--stat")) {
2537                 char *end;
2538                 int width = options->stat_width;
2539                 int name_width = options->stat_name_width;
2540                 arg += 6;
2541                 end = (char *)arg;
2542
2543                 switch (*arg) {
2544                 case '-':
2545                         if (!prefixcmp(arg, "-width="))
2546                                 width = strtoul(arg + 7, &end, 10);
2547                         else if (!prefixcmp(arg, "-name-width="))
2548                                 name_width = strtoul(arg + 12, &end, 10);
2549                         break;
2550                 case '=':
2551                         width = strtoul(arg+1, &end, 10);
2552                         if (*end == ',')
2553                                 name_width = strtoul(end+1, &end, 10);
2554                 }
2555
2556                 /* Important! This checks all the error cases! */
2557                 if (*end)
2558                         return 0;
2559                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2560                 options->stat_name_width = name_width;
2561                 options->stat_width = width;
2562         }
2563
2564         /* renames options */
2565         else if (!prefixcmp(arg, "-B")) {
2566                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2567                         return -1;
2568         }
2569         else if (!prefixcmp(arg, "-M")) {
2570                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2571                         return -1;
2572                 options->detect_rename = DIFF_DETECT_RENAME;
2573         }
2574         else if (!prefixcmp(arg, "-C")) {
2575                 if (options->detect_rename == DIFF_DETECT_COPY)
2576                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2577                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2578                         return -1;
2579                 options->detect_rename = DIFF_DETECT_COPY;
2580         }
2581         else if (!strcmp(arg, "--no-renames"))
2582                 options->detect_rename = 0;
2583         else if (!strcmp(arg, "--relative"))
2584                 DIFF_OPT_SET(options, RELATIVE_NAME);
2585         else if (!prefixcmp(arg, "--relative=")) {
2586                 DIFF_OPT_SET(options, RELATIVE_NAME);
2587                 options->prefix = arg + 11;
2588         }
2589
2590         /* xdiff options */
2591         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2592                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2593         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2594                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2595         else if (!strcmp(arg, "--ignore-space-at-eol"))
2596                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2597         else if (!strcmp(arg, "--patience"))
2598                 DIFF_XDL_SET(options, PATIENCE_DIFF);
2599
2600         /* flags options */
2601         else if (!strcmp(arg, "--binary")) {
2602                 options->output_format |= DIFF_FORMAT_PATCH;
2603                 DIFF_OPT_SET(options, BINARY);
2604         }
2605         else if (!strcmp(arg, "--full-index"))
2606                 DIFF_OPT_SET(options, FULL_INDEX);
2607         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2608                 DIFF_OPT_SET(options, TEXT);
2609         else if (!strcmp(arg, "-R"))
2610                 DIFF_OPT_SET(options, REVERSE_DIFF);
2611         else if (!strcmp(arg, "--find-copies-harder"))
2612                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2613         else if (!strcmp(arg, "--follow"))
2614                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2615         else if (!strcmp(arg, "--color"))
2616                 DIFF_OPT_SET(options, COLOR_DIFF);
2617         else if (!strcmp(arg, "--no-color"))
2618                 DIFF_OPT_CLR(options, COLOR_DIFF);
2619         else if (!strcmp(arg, "--color-words")) {
2620                 DIFF_OPT_SET(options, COLOR_DIFF);
2621                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2622         }
2623         else if (!prefixcmp(arg, "--color-words=")) {
2624                 DIFF_OPT_SET(options, COLOR_DIFF);
2625                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2626                 options->word_regex = arg + 14;
2627         }
2628         else if (!strcmp(arg, "--exit-code"))
2629                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2630         else if (!strcmp(arg, "--quiet"))
2631                 DIFF_OPT_SET(options, QUIET);
2632         else if (!strcmp(arg, "--ext-diff"))
2633                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2634         else if (!strcmp(arg, "--no-ext-diff"))
2635                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2636         else if (!strcmp(arg, "--textconv"))
2637                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2638         else if (!strcmp(arg, "--no-textconv"))
2639                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2640         else if (!strcmp(arg, "--ignore-submodules"))
2641                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2642
2643         /* misc options */
2644         else if (!strcmp(arg, "-z"))
2645                 options->line_termination = 0;
2646         else if (!prefixcmp(arg, "-l"))
2647                 options->rename_limit = strtoul(arg+2, NULL, 10);
2648         else if (!prefixcmp(arg, "-S"))
2649                 options->pickaxe = arg + 2;
2650         else if (!strcmp(arg, "--pickaxe-all"))
2651                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2652         else if (!strcmp(arg, "--pickaxe-regex"))
2653                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2654         else if (!prefixcmp(arg, "-O"))
2655                 options->orderfile = arg + 2;
2656         else if (!prefixcmp(arg, "--diff-filter="))
2657                 options->filter = arg + 14;
2658         else if (!strcmp(arg, "--abbrev"))
2659                 options->abbrev = DEFAULT_ABBREV;
2660         else if (!prefixcmp(arg, "--abbrev=")) {
2661                 options->abbrev = strtoul(arg + 9, NULL, 10);
2662                 if (options->abbrev < MINIMUM_ABBREV)
2663                         options->abbrev = MINIMUM_ABBREV;
2664                 else if (40 < options->abbrev)
2665                         options->abbrev = 40;
2666         }
2667         else if (!prefixcmp(arg, "--src-prefix="))
2668                 options->a_prefix = arg + 13;
2669         else if (!prefixcmp(arg, "--dst-prefix="))
2670                 options->b_prefix = arg + 13;
2671         else if (!strcmp(arg, "--no-prefix"))
2672                 options->a_prefix = options->b_prefix = "";
2673         else if (opt_arg(arg, '\0', "inter-hunk-context",
2674                          &options->interhunkcontext))
2675                 ;
2676         else if (!prefixcmp(arg, "--output=")) {
2677                 options->file = fopen(arg + strlen("--output="), "w");
2678                 options->close_file = 1;
2679         } else
2680                 return 0;
2681         return 1;
2682 }
2683
2684 static int parse_num(const char **cp_p)
2685 {
2686         unsigned long num, scale;
2687         int ch, dot;
2688         const char *cp = *cp_p;
2689
2690         num = 0;
2691         scale = 1;
2692         dot = 0;
2693         for(;;) {
2694                 ch = *cp;
2695                 if ( !dot && ch == '.' ) {
2696                         scale = 1;
2697                         dot = 1;
2698                 } else if ( ch == '%' ) {
2699                         scale = dot ? scale*100 : 100;
2700                         cp++;   /* % is always at the end */
2701                         break;
2702                 } else if ( ch >= '0' && ch <= '9' ) {
2703                         if ( scale < 100000 ) {
2704                                 scale *= 10;
2705                                 num = (num*10) + (ch-'0');
2706                         }
2707                 } else {
2708                         break;
2709                 }
2710                 cp++;
2711         }
2712         *cp_p = cp;
2713
2714         /* user says num divided by scale and we say internally that
2715          * is MAX_SCORE * num / scale.
2716          */
2717         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2718 }
2719
2720 static int diff_scoreopt_parse(const char *opt)
2721 {
2722         int opt1, opt2, cmd;
2723
2724         if (*opt++ != '-')
2725                 return -1;
2726         cmd = *opt++;
2727         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2728                 return -1; /* that is not a -M, -C nor -B option */
2729
2730         opt1 = parse_num(&opt);
2731         if (cmd != 'B')
2732                 opt2 = 0;
2733         else {
2734                 if (*opt == 0)
2735                         opt2 = 0;
2736                 else if (*opt != '/')
2737                         return -1; /* we expect -B80/99 or -B80 */
2738                 else {
2739                         opt++;
2740                         opt2 = parse_num(&opt);
2741                 }
2742         }
2743         if (*opt != 0)
2744                 return -1;
2745         return opt1 | (opt2 << 16);
2746 }
2747
2748 struct diff_queue_struct diff_queued_diff;
2749
2750 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2751 {
2752         if (queue->alloc <= queue->nr) {
2753                 queue->alloc = alloc_nr(queue->alloc);
2754                 queue->queue = xrealloc(queue->queue,
2755                                         sizeof(dp) * queue->alloc);
2756         }
2757         queue->queue[queue->nr++] = dp;
2758 }
2759
2760 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2761                                  struct diff_filespec *one,
2762                                  struct diff_filespec *two)
2763 {
2764         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2765         dp->one = one;
2766         dp->two = two;
2767         if (queue)
2768                 diff_q(queue, dp);
2769         return dp;
2770 }
2771
2772 void diff_free_filepair(struct diff_filepair *p)
2773 {
2774         free_filespec(p->one);
2775         free_filespec(p->two);
2776         free(p);
2777 }
2778
2779 /* This is different from find_unique_abbrev() in that
2780  * it stuffs the result with dots for alignment.
2781  */
2782 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2783 {
2784         int abblen;
2785         const char *abbrev;
2786         if (len == 40)
2787                 return sha1_to_hex(sha1);
2788
2789         abbrev = find_unique_abbrev(sha1, len);
2790         abblen = strlen(abbrev);
2791         if (abblen < 37) {
2792                 static char hex[41];
2793                 if (len < abblen && abblen <= len + 2)
2794                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2795                 else
2796                         sprintf(hex, "%s...", abbrev);
2797                 return hex;
2798         }
2799         return sha1_to_hex(sha1);
2800 }
2801
2802 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2803 {
2804         int line_termination = opt->line_termination;
2805         int inter_name_termination = line_termination ? '\t' : '\0';
2806
2807         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2808                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2809                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2810                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2811         }
2812         if (p->score) {
2813                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2814                         inter_name_termination);
2815         } else {
2816                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2817         }
2818
2819         if (p->status == DIFF_STATUS_COPIED ||
2820             p->status == DIFF_STATUS_RENAMED) {
2821                 const char *name_a, *name_b;
2822                 name_a = p->one->path;
2823                 name_b = p->two->path;
2824                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2825                 write_name_quoted(name_a, opt->file, inter_name_termination);
2826                 write_name_quoted(name_b, opt->file, line_termination);
2827         } else {
2828                 const char *name_a, *name_b;
2829                 name_a = p->one->mode ? p->one->path : p->two->path;
2830                 name_b = NULL;
2831                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2832                 write_name_quoted(name_a, opt->file, line_termination);
2833         }
2834 }
2835
2836 int diff_unmodified_pair(struct diff_filepair *p)
2837 {
2838         /* This function is written stricter than necessary to support
2839          * the currently implemented transformers, but the idea is to
2840          * let transformers to produce diff_filepairs any way they want,
2841          * and filter and clean them up here before producing the output.
2842          */
2843         struct diff_filespec *one = p->one, *two = p->two;
2844
2845         if (DIFF_PAIR_UNMERGED(p))
2846                 return 0; /* unmerged is interesting */
2847
2848         /* deletion, addition, mode or type change
2849          * and rename are all interesting.
2850          */
2851         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2852             DIFF_PAIR_MODE_CHANGED(p) ||
2853             strcmp(one->path, two->path))
2854                 return 0;
2855
2856         /* both are valid and point at the same path.  that is, we are
2857          * dealing with a change.
2858          */
2859         if (one->sha1_valid && two->sha1_valid &&
2860             !hashcmp(one->sha1, two->sha1))
2861                 return 1; /* no change */
2862         if (!one->sha1_valid && !two->sha1_valid)
2863                 return 1; /* both look at the same file on the filesystem. */
2864         return 0;
2865 }
2866
2867 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2868 {
2869         if (diff_unmodified_pair(p))
2870                 return;
2871
2872         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2873             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2874                 return; /* no tree diffs in patch format */
2875
2876         run_diff(p, o);
2877 }
2878
2879 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2880                             struct diffstat_t *diffstat)
2881 {
2882         if (diff_unmodified_pair(p))
2883                 return;
2884
2885         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2886             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2887                 return; /* no tree diffs in patch format */
2888
2889         run_diffstat(p, o, diffstat);
2890 }
2891
2892 static void diff_flush_checkdiff(struct diff_filepair *p,
2893                 struct diff_options *o)
2894 {
2895         if (diff_unmodified_pair(p))
2896                 return;
2897
2898         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2899             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2900                 return; /* no tree diffs in patch format */
2901
2902         run_checkdiff(p, o);
2903 }
2904
2905 int diff_queue_is_empty(void)
2906 {
2907         struct diff_queue_struct *q = &diff_queued_diff;
2908         int i;
2909         for (i = 0; i < q->nr; i++)
2910                 if (!diff_unmodified_pair(q->queue[i]))
2911                         return 0;
2912         return 1;
2913 }
2914
2915 #if DIFF_DEBUG
2916 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2917 {
2918         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2919                 x, one ? one : "",
2920                 s->path,
2921                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2922                 s->mode,
2923                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2924         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2925                 x, one ? one : "",
2926                 s->size, s->xfrm_flags);
2927 }
2928
2929 void diff_debug_filepair(const struct diff_filepair *p, int i)
2930 {
2931         diff_debug_filespec(p->one, i, "one");
2932         diff_debug_filespec(p->two, i, "two");
2933         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2934                 p->score, p->status ? p->status : '?',
2935                 p->one->rename_used, p->broken_pair);
2936 }
2937
2938 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2939 {
2940         int i;
2941         if (msg)
2942                 fprintf(stderr, "%s\n", msg);
2943         fprintf(stderr, "q->nr = %d\n", q->nr);
2944         for (i = 0; i < q->nr; i++) {
2945                 struct diff_filepair *p = q->queue[i];
2946                 diff_debug_filepair(p, i);
2947         }
2948 }
2949 #endif
2950
2951 static void diff_resolve_rename_copy(void)
2952 {
2953         int i;
2954         struct diff_filepair *p;
2955         struct diff_queue_struct *q = &diff_queued_diff;
2956
2957         diff_debug_queue("resolve-rename-copy", q);
2958
2959         for (i = 0; i < q->nr; i++) {
2960                 p = q->queue[i];
2961                 p->status = 0; /* undecided */
2962                 if (DIFF_PAIR_UNMERGED(p))
2963                         p->status = DIFF_STATUS_UNMERGED;
2964                 else if (!DIFF_FILE_VALID(p->one))
2965                         p->status = DIFF_STATUS_ADDED;
2966                 else if (!DIFF_FILE_VALID(p->two))
2967                         p->status = DIFF_STATUS_DELETED;
2968                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2969                         p->status = DIFF_STATUS_TYPE_CHANGED;
2970
2971                 /* from this point on, we are dealing with a pair
2972                  * whose both sides are valid and of the same type, i.e.
2973                  * either in-place edit or rename/copy edit.
2974                  */
2975                 else if (DIFF_PAIR_RENAME(p)) {
2976                         /*
2977                          * A rename might have re-connected a broken
2978                          * pair up, causing the pathnames to be the
2979                          * same again. If so, that's not a rename at
2980                          * all, just a modification..
2981                          *
2982                          * Otherwise, see if this source was used for
2983                          * multiple renames, in which case we decrement
2984                          * the count, and call it a copy.
2985                          */
2986                         if (!strcmp(p->one->path, p->two->path))
2987                                 p->status = DIFF_STATUS_MODIFIED;
2988                         else if (--p->one->rename_used > 0)
2989                                 p->status = DIFF_STATUS_COPIED;
2990                         else
2991                                 p->status = DIFF_STATUS_RENAMED;
2992                 }
2993                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2994                          p->one->mode != p->two->mode ||
2995                          is_null_sha1(p->one->sha1))
2996                         p->status = DIFF_STATUS_MODIFIED;
2997                 else {
2998                         /* This is a "no-change" entry and should not
2999                          * happen anymore, but prepare for broken callers.
3000                          */
3001                         error("feeding unmodified %s to diffcore",
3002                               p->one->path);
3003                         p->status = DIFF_STATUS_UNKNOWN;
3004                 }
3005         }
3006         diff_debug_queue("resolve-rename-copy done", q);
3007 }
3008
3009 static int check_pair_status(struct diff_filepair *p)
3010 {
3011         switch (p->status) {
3012         case DIFF_STATUS_UNKNOWN:
3013                 return 0;
3014         case 0:
3015                 die("internal error in diff-resolve-rename-copy");
3016         default:
3017                 return 1;
3018         }
3019 }
3020
3021 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3022 {
3023         int fmt = opt->output_format;
3024
3025         if (fmt & DIFF_FORMAT_CHECKDIFF)
3026                 diff_flush_checkdiff(p, opt);
3027         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3028                 diff_flush_raw(p, opt);
3029         else if (fmt & DIFF_FORMAT_NAME) {
3030                 const char *name_a, *name_b;
3031                 name_a = p->two->path;
3032                 name_b = NULL;
3033                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3034                 write_name_quoted(name_a, opt->file, opt->line_termination);
3035         }
3036 }
3037
3038 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3039 {
3040         if (fs->mode)
3041                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3042         else
3043                 fprintf(file, " %s ", newdelete);
3044         write_name_quoted(fs->path, file, '\n');
3045 }
3046
3047
3048 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3049 {
3050         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3051                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3052                         show_name ? ' ' : '\n');
3053                 if (show_name) {
3054                         write_name_quoted(p->two->path, file, '\n');
3055                 }
3056         }
3057 }
3058
3059 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3060 {
3061         char *names = pprint_rename(p->one->path, p->two->path);
3062
3063         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3064         free(names);
3065         show_mode_change(file, p, 0);
3066 }
3067
3068 static void diff_summary(FILE *file, struct diff_filepair *p)
3069 {
3070         switch(p->status) {
3071         case DIFF_STATUS_DELETED:
3072                 show_file_mode_name(file, "delete", p->one);
3073                 break;
3074         case DIFF_STATUS_ADDED:
3075                 show_file_mode_name(file, "create", p->two);
3076                 break;
3077         case DIFF_STATUS_COPIED:
3078                 show_rename_copy(file, "copy", p);
3079                 break;
3080         case DIFF_STATUS_RENAMED:
3081                 show_rename_copy(file, "rename", p);
3082                 break;
3083         default:
3084                 if (p->score) {
3085                         fputs(" rewrite ", file);
3086                         write_name_quoted(p->two->path, file, ' ');
3087                         fprintf(file, "(%d%%)\n", similarity_index(p));
3088                 }
3089                 show_mode_change(file, p, !p->score);
3090                 break;
3091         }
3092 }
3093
3094 struct patch_id_t {
3095         git_SHA_CTX *ctx;
3096         int patchlen;
3097 };
3098
3099 static int remove_space(char *line, int len)
3100 {
3101         int i;
3102         char *dst = line;
3103         unsigned char c;
3104
3105         for (i = 0; i < len; i++)
3106                 if (!isspace((c = line[i])))
3107                         *dst++ = c;
3108
3109         return dst - line;
3110 }
3111
3112 static void patch_id_consume(void *priv, char *line, unsigned long len)
3113 {
3114         struct patch_id_t *data = priv;
3115         int new_len;
3116
3117         /* Ignore line numbers when computing the SHA1 of the patch */
3118         if (!prefixcmp(line, "@@ -"))
3119                 return;
3120
3121         new_len = remove_space(line, len);
3122
3123         git_SHA1_Update(data->ctx, line, new_len);
3124         data->patchlen += new_len;
3125 }
3126
3127 /* returns 0 upon success, and writes result into sha1 */
3128 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3129 {
3130         struct diff_queue_struct *q = &diff_queued_diff;
3131         int i;
3132         git_SHA_CTX ctx;
3133         struct patch_id_t data;
3134         char buffer[PATH_MAX * 4 + 20];
3135
3136         git_SHA1_Init(&ctx);
3137         memset(&data, 0, sizeof(struct patch_id_t));
3138         data.ctx = &ctx;
3139
3140         for (i = 0; i < q->nr; i++) {
3141                 xpparam_t xpp;
3142                 xdemitconf_t xecfg;
3143                 xdemitcb_t ecb;
3144                 mmfile_t mf1, mf2;
3145                 struct diff_filepair *p = q->queue[i];
3146                 int len1, len2;
3147
3148                 memset(&xpp, 0, sizeof(xpp));
3149                 memset(&xecfg, 0, sizeof(xecfg));
3150                 if (p->status == 0)
3151                         return error("internal diff status error");
3152                 if (p->status == DIFF_STATUS_UNKNOWN)
3153                         continue;
3154                 if (diff_unmodified_pair(p))
3155                         continue;
3156                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3157                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3158                         continue;
3159                 if (DIFF_PAIR_UNMERGED(p))
3160                         continue;
3161
3162                 diff_fill_sha1_info(p->one);
3163                 diff_fill_sha1_info(p->two);
3164                 if (fill_mmfile(&mf1, p->one) < 0 ||
3165                                 fill_mmfile(&mf2, p->two) < 0)
3166                         return error("unable to read files to diff");
3167
3168                 len1 = remove_space(p->one->path, strlen(p->one->path));
3169                 len2 = remove_space(p->two->path, strlen(p->two->path));
3170                 if (p->one->mode == 0)
3171                         len1 = snprintf(buffer, sizeof(buffer),
3172                                         "diff--gita/%.*sb/%.*s"
3173                                         "newfilemode%06o"
3174                                         "---/dev/null"
3175                                         "+++b/%.*s",
3176                                         len1, p->one->path,
3177                                         len2, p->two->path,
3178                                         p->two->mode,
3179                                         len2, p->two->path);
3180                 else if (p->two->mode == 0)
3181                         len1 = snprintf(buffer, sizeof(buffer),
3182                                         "diff--gita/%.*sb/%.*s"
3183                                         "deletedfilemode%06o"
3184                                         "---a/%.*s"
3185                                         "+++/dev/null",
3186                                         len1, p->one->path,
3187                                         len2, p->two->path,
3188                                         p->one->mode,
3189                                         len1, p->one->path);
3190                 else
3191                         len1 = snprintf(buffer, sizeof(buffer),
3192                                         "diff--gita/%.*sb/%.*s"
3193                                         "---a/%.*s"
3194                                         "+++b/%.*s",
3195                                         len1, p->one->path,
3196                                         len2, p->two->path,
3197                                         len1, p->one->path,
3198                                         len2, p->two->path);
3199                 git_SHA1_Update(&ctx, buffer, len1);
3200
3201                 xpp.flags = XDF_NEED_MINIMAL;
3202                 xecfg.ctxlen = 3;
3203                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3204                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3205                               &xpp, &xecfg, &ecb);
3206         }
3207
3208         git_SHA1_Final(sha1, &ctx);
3209         return 0;
3210 }
3211
3212 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3213 {
3214         struct diff_queue_struct *q = &diff_queued_diff;
3215         int i;
3216         int result = diff_get_patch_id(options, sha1);
3217
3218         for (i = 0; i < q->nr; i++)
3219                 diff_free_filepair(q->queue[i]);
3220
3221         free(q->queue);
3222         q->queue = NULL;
3223         q->nr = q->alloc = 0;
3224
3225         return result;
3226 }
3227
3228 static int is_summary_empty(const struct diff_queue_struct *q)
3229 {
3230         int i;
3231
3232         for (i = 0; i < q->nr; i++) {
3233                 const struct diff_filepair *p = q->queue[i];
3234
3235                 switch (p->status) {
3236                 case DIFF_STATUS_DELETED:
3237                 case DIFF_STATUS_ADDED:
3238                 case DIFF_STATUS_COPIED:
3239                 case DIFF_STATUS_RENAMED:
3240                         return 0;
3241                 default:
3242                         if (p->score)
3243                                 return 0;
3244                         if (p->one->mode && p->two->mode &&
3245                             p->one->mode != p->two->mode)
3246                                 return 0;
3247                         break;
3248                 }
3249         }
3250         return 1;
3251 }
3252
3253 void diff_flush(struct diff_options *options)
3254 {
3255         struct diff_queue_struct *q = &diff_queued_diff;
3256         int i, output_format = options->output_format;
3257         int separator = 0;
3258
3259         /*
3260          * Order: raw, stat, summary, patch
3261          * or:    name/name-status/checkdiff (other bits clear)
3262          */
3263         if (!q->nr)
3264                 goto free_queue;
3265
3266         if (output_format & (DIFF_FORMAT_RAW |
3267                              DIFF_FORMAT_NAME |
3268                              DIFF_FORMAT_NAME_STATUS |
3269                              DIFF_FORMAT_CHECKDIFF)) {
3270                 for (i = 0; i < q->nr; i++) {
3271                         struct diff_filepair *p = q->queue[i];
3272                         if (check_pair_status(p))
3273                                 flush_one_pair(p, options);
3274                 }
3275                 separator++;
3276         }
3277
3278         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3279                 struct diffstat_t diffstat;
3280
3281                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3282                 for (i = 0; i < q->nr; i++) {
3283                         struct diff_filepair *p = q->queue[i];
3284                         if (check_pair_status(p))
3285                                 diff_flush_stat(p, options, &diffstat);
3286                 }
3287                 if (output_format & DIFF_FORMAT_NUMSTAT)
3288                         show_numstat(&diffstat, options);
3289                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3290                         show_stats(&diffstat, options);
3291                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3292                         show_shortstats(&diffstat, options);
3293                 free_diffstat_info(&diffstat);
3294                 separator++;
3295         }
3296         if (output_format & DIFF_FORMAT_DIRSTAT)
3297                 show_dirstat(options);
3298
3299         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3300                 for (i = 0; i < q->nr; i++)
3301                         diff_summary(options->file, q->queue[i]);
3302                 separator++;
3303         }
3304
3305         if (output_format & DIFF_FORMAT_PATCH) {
3306                 if (separator) {
3307                         putc(options->line_termination, options->file);
3308                         if (options->stat_sep) {
3309                                 /* attach patch instead of inline */
3310                                 fputs(options->stat_sep, options->file);
3311                         }
3312                 }
3313
3314                 for (i = 0; i < q->nr; i++) {
3315                         struct diff_filepair *p = q->queue[i];
3316                         if (check_pair_status(p))
3317                                 diff_flush_patch(p, options);
3318                 }
3319         }
3320
3321         if (output_format & DIFF_FORMAT_CALLBACK)
3322                 options->format_callback(q, options, options->format_callback_data);
3323
3324         for (i = 0; i < q->nr; i++)
3325                 diff_free_filepair(q->queue[i]);
3326 free_queue:
3327         free(q->queue);
3328         q->queue = NULL;
3329         q->nr = q->alloc = 0;
3330         if (options->close_file)
3331                 fclose(options->file);
3332 }
3333
3334 static void diffcore_apply_filter(const char *filter)
3335 {
3336         int i;
3337         struct diff_queue_struct *q = &diff_queued_diff;
3338         struct diff_queue_struct outq;
3339         outq.queue = NULL;
3340         outq.nr = outq.alloc = 0;
3341
3342         if (!filter)
3343                 return;
3344
3345         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3346                 int found;
3347                 for (i = found = 0; !found && i < q->nr; i++) {
3348                         struct diff_filepair *p = q->queue[i];
3349                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3350                              ((p->score &&
3351                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3352                               (!p->score &&
3353                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3354                             ((p->status != DIFF_STATUS_MODIFIED) &&
3355                              strchr(filter, p->status)))
3356                                 found++;
3357                 }
3358                 if (found)
3359                         return;
3360
3361                 /* otherwise we will clear the whole queue
3362                  * by copying the empty outq at the end of this
3363                  * function, but first clear the current entries
3364                  * in the queue.
3365                  */
3366                 for (i = 0; i < q->nr; i++)
3367                         diff_free_filepair(q->queue[i]);
3368         }
3369         else {
3370                 /* Only the matching ones */
3371                 for (i = 0; i < q->nr; i++) {
3372                         struct diff_filepair *p = q->queue[i];
3373
3374                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3375                              ((p->score &&
3376                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3377                               (!p->score &&
3378                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3379                             ((p->status != DIFF_STATUS_MODIFIED) &&
3380                              strchr(filter, p->status)))
3381                                 diff_q(&outq, p);
3382                         else
3383                                 diff_free_filepair(p);
3384                 }
3385         }
3386         free(q->queue);
3387         *q = outq;
3388 }
3389
3390 /* Check whether two filespecs with the same mode and size are identical */
3391 static int diff_filespec_is_identical(struct diff_filespec *one,
3392                                       struct diff_filespec *two)
3393 {
3394         if (S_ISGITLINK(one->mode))
3395                 return 0;
3396         if (diff_populate_filespec(one, 0))
3397                 return 0;
3398         if (diff_populate_filespec(two, 0))
3399                 return 0;
3400         return !memcmp(one->data, two->data, one->size);
3401 }
3402
3403 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3404 {
3405         int i;
3406         struct diff_queue_struct *q = &diff_queued_diff;
3407         struct diff_queue_struct outq;
3408         outq.queue = NULL;
3409         outq.nr = outq.alloc = 0;
3410
3411         for (i = 0; i < q->nr; i++) {
3412                 struct diff_filepair *p = q->queue[i];
3413
3414                 /*
3415                  * 1. Entries that come from stat info dirtyness
3416                  *    always have both sides (iow, not create/delete),
3417                  *    one side of the object name is unknown, with
3418                  *    the same mode and size.  Keep the ones that
3419                  *    do not match these criteria.  They have real
3420                  *    differences.
3421                  *
3422                  * 2. At this point, the file is known to be modified,
3423                  *    with the same mode and size, and the object
3424                  *    name of one side is unknown.  Need to inspect
3425                  *    the identical contents.
3426                  */
3427                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3428                     !DIFF_FILE_VALID(p->two) ||
3429                     (p->one->sha1_valid && p->two->sha1_valid) ||
3430                     (p->one->mode != p->two->mode) ||
3431                     diff_populate_filespec(p->one, 1) ||
3432                     diff_populate_filespec(p->two, 1) ||
3433                     (p->one->size != p->two->size) ||
3434                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3435                         diff_q(&outq, p);
3436                 else {
3437                         /*
3438                          * The caller can subtract 1 from skip_stat_unmatch
3439                          * to determine how many paths were dirty only
3440                          * due to stat info mismatch.
3441                          */
3442                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3443                                 diffopt->skip_stat_unmatch++;
3444                         diff_free_filepair(p);
3445                 }
3446         }
3447         free(q->queue);
3448         *q = outq;
3449 }
3450
3451 void diffcore_std(struct diff_options *options)
3452 {
3453         if (options->skip_stat_unmatch)
3454                 diffcore_skip_stat_unmatch(options);
3455         if (options->break_opt != -1)
3456                 diffcore_break(options->break_opt);
3457         if (options->detect_rename)
3458                 diffcore_rename(options);
3459         if (options->break_opt != -1)
3460                 diffcore_merge_broken();
3461         if (options->pickaxe)
3462                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3463         if (options->orderfile)
3464                 diffcore_order(options->orderfile);
3465         diff_resolve_rename_copy();
3466         diffcore_apply_filter(options->filter);
3467
3468         if (diff_queued_diff.nr)
3469                 DIFF_OPT_SET(options, HAS_CHANGES);
3470         else
3471                 DIFF_OPT_CLR(options, HAS_CHANGES);
3472 }
3473
3474 int diff_result_code(struct diff_options *opt, int status)
3475 {
3476         int result = 0;
3477         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3478             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3479                 return status;
3480         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3481             DIFF_OPT_TST(opt, HAS_CHANGES))
3482                 result |= 01;
3483         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3484             DIFF_OPT_TST(opt, CHECK_FAILED))
3485                 result |= 02;
3486         return result;
3487 }
3488
3489 void diff_addremove(struct diff_options *options,
3490                     int addremove, unsigned mode,
3491                     const unsigned char *sha1,
3492                     const char *concatpath)
3493 {
3494         struct diff_filespec *one, *two;
3495
3496         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3497                 return;
3498
3499         /* This may look odd, but it is a preparation for
3500          * feeding "there are unchanged files which should
3501          * not produce diffs, but when you are doing copy
3502          * detection you would need them, so here they are"
3503          * entries to the diff-core.  They will be prefixed
3504          * with something like '=' or '*' (I haven't decided
3505          * which but should not make any difference).
3506          * Feeding the same new and old to diff_change()
3507          * also has the same effect.
3508          * Before the final output happens, they are pruned after
3509          * merged into rename/copy pairs as appropriate.
3510          */
3511         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3512                 addremove = (addremove == '+' ? '-' :
3513                              addremove == '-' ? '+' : addremove);
3514
3515         if (options->prefix &&
3516             strncmp(concatpath, options->prefix, options->prefix_length))
3517                 return;
3518
3519         one = alloc_filespec(concatpath);
3520         two = alloc_filespec(concatpath);
3521
3522         if (addremove != '+')
3523                 fill_filespec(one, sha1, mode);
3524         if (addremove != '-')
3525                 fill_filespec(two, sha1, mode);
3526
3527         diff_queue(&diff_queued_diff, one, two);
3528         DIFF_OPT_SET(options, HAS_CHANGES);
3529 }
3530
3531 void diff_change(struct diff_options *options,
3532                  unsigned old_mode, unsigned new_mode,
3533                  const unsigned char *old_sha1,
3534                  const unsigned char *new_sha1,
3535                  const char *concatpath)
3536 {
3537         struct diff_filespec *one, *two;
3538
3539         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3540                         && S_ISGITLINK(new_mode))
3541                 return;
3542
3543         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3544                 unsigned tmp;
3545                 const unsigned char *tmp_c;
3546                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3547                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3548         }
3549
3550         if (options->prefix &&
3551             strncmp(concatpath, options->prefix, options->prefix_length))
3552                 return;
3553
3554         one = alloc_filespec(concatpath);
3555         two = alloc_filespec(concatpath);
3556         fill_filespec(one, old_sha1, old_mode);
3557         fill_filespec(two, new_sha1, new_mode);
3558
3559         diff_queue(&diff_queued_diff, one, two);
3560         DIFF_OPT_SET(options, HAS_CHANGES);
3561 }
3562
3563 void diff_unmerge(struct diff_options *options,
3564                   const char *path,
3565                   unsigned mode, const unsigned char *sha1)
3566 {
3567         struct diff_filespec *one, *two;
3568
3569         if (options->prefix &&
3570             strncmp(path, options->prefix, options->prefix_length))
3571                 return;
3572
3573         one = alloc_filespec(path);
3574         two = alloc_filespec(path);
3575         fill_filespec(one, sha1, mode);
3576         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3577 }
3578
3579 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3580                 size_t *outsize)
3581 {
3582         struct diff_tempfile *temp;
3583         const char *argv[3];
3584         const char **arg = argv;
3585         struct child_process child;
3586         struct strbuf buf = STRBUF_INIT;
3587
3588         temp = prepare_temp_file(spec->path, spec);
3589         *arg++ = pgm;
3590         *arg++ = temp->name;
3591         *arg = NULL;
3592
3593         memset(&child, 0, sizeof(child));
3594         child.argv = argv;
3595         child.out = -1;
3596         if (start_command(&child) != 0 ||
3597             strbuf_read(&buf, child.out, 0) < 0 ||
3598             finish_command(&child) != 0) {
3599                 strbuf_release(&buf);
3600                 remove_tempfile();
3601                 error("error running textconv command '%s'", pgm);
3602                 return NULL;
3603         }
3604         remove_tempfile();
3605
3606         return strbuf_detach(&buf, outsize);
3607 }