Merge branch 'gt/utf8-width'
[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         "\033[m",       /* reset */
34         "",             /* PLAIN (normal) */
35         "\033[1m",      /* METAINFO (bold) */
36         "\033[36m",     /* FRAGINFO (cyan) */
37         "\033[31m",     /* OLD (red) */
38         "\033[32m",     /* NEW (green) */
39         "\033[33m",     /* COMMIT (yellow) */
40         "\033[41m",     /* WHITESPACE (red background) */
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 /*
66  * These are to give UI layer defaults.
67  * The core-level commands such as git-diff-files should
68  * never be affected by the setting of diff.renames
69  * the user happens to have in the configuration file.
70  */
71 int git_diff_ui_config(const char *var, const char *value, void *cb)
72 {
73         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
74                 diff_use_color_default = git_config_colorbool(var, value, -1);
75                 return 0;
76         }
77         if (!strcmp(var, "diff.renames")) {
78                 if (!value)
79                         diff_detect_rename_default = DIFF_DETECT_RENAME;
80                 else if (!strcasecmp(value, "copies") ||
81                          !strcasecmp(value, "copy"))
82                         diff_detect_rename_default = DIFF_DETECT_COPY;
83                 else if (git_config_bool(var,value))
84                         diff_detect_rename_default = DIFF_DETECT_RENAME;
85                 return 0;
86         }
87         if (!strcmp(var, "diff.autorefreshindex")) {
88                 diff_auto_refresh_index = git_config_bool(var, value);
89                 return 0;
90         }
91         if (!strcmp(var, "diff.mnemonicprefix")) {
92                 diff_mnemonic_prefix = git_config_bool(var, value);
93                 return 0;
94         }
95         if (!strcmp(var, "diff.external"))
96                 return git_config_string(&external_diff_cmd_cfg, var, value);
97         if (!strcmp(var, "diff.wordregex"))
98                 return git_config_string(&diff_word_regex_cfg, var, value);
99
100         return git_diff_basic_config(var, value, cb);
101 }
102
103 int git_diff_basic_config(const char *var, const char *value, void *cb)
104 {
105         if (!strcmp(var, "diff.renamelimit")) {
106                 diff_rename_limit_default = git_config_int(var, value);
107                 return 0;
108         }
109
110         switch (userdiff_config(var, value)) {
111                 case 0: break;
112                 case -1: return -1;
113                 default: return 0;
114         }
115
116         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
117                 int slot = parse_diff_color_slot(var, 11);
118                 if (!value)
119                         return config_error_nonbool(var);
120                 color_parse(value, var, diff_colors[slot]);
121                 return 0;
122         }
123
124         /* like GNU diff's --suppress-blank-empty option  */
125         if (!strcmp(var, "diff.suppressblankempty") ||
126                         /* for backwards compatibility */
127                         !strcmp(var, "diff.suppress-blank-empty")) {
128                 diff_suppress_blank_empty = git_config_bool(var, value);
129                 return 0;
130         }
131
132         return git_color_default_config(var, value, cb);
133 }
134
135 static char *quote_two(const char *one, const char *two)
136 {
137         int need_one = quote_c_style(one, NULL, NULL, 1);
138         int need_two = quote_c_style(two, NULL, NULL, 1);
139         struct strbuf res = STRBUF_INIT;
140
141         if (need_one + need_two) {
142                 strbuf_addch(&res, '"');
143                 quote_c_style(one, &res, NULL, 1);
144                 quote_c_style(two, &res, NULL, 1);
145                 strbuf_addch(&res, '"');
146         } else {
147                 strbuf_addstr(&res, one);
148                 strbuf_addstr(&res, two);
149         }
150         return strbuf_detach(&res, NULL);
151 }
152
153 static const char *external_diff(void)
154 {
155         static const char *external_diff_cmd = NULL;
156         static int done_preparing = 0;
157
158         if (done_preparing)
159                 return external_diff_cmd;
160         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
161         if (!external_diff_cmd)
162                 external_diff_cmd = external_diff_cmd_cfg;
163         done_preparing = 1;
164         return external_diff_cmd;
165 }
166
167 static struct diff_tempfile {
168         const char *name; /* filename external diff should read from */
169         char hex[41];
170         char mode[10];
171         char tmp_path[PATH_MAX];
172 } diff_temp[2];
173
174 static struct diff_tempfile *claim_diff_tempfile(void) {
175         int i;
176         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
177                 if (!diff_temp[i].name)
178                         return diff_temp + i;
179         die("BUG: diff is failing to clean up its tempfiles");
180 }
181
182 static int remove_tempfile_installed;
183
184 static void remove_tempfile(void)
185 {
186         int i;
187         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
188                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
189                         unlink(diff_temp[i].name);
190                         diff_temp[i].name = NULL;
191                 }
192 }
193
194 static void remove_tempfile_on_signal(int signo)
195 {
196         remove_tempfile();
197         sigchain_pop(signo);
198         raise(signo);
199 }
200
201 static int count_lines(const char *data, int size)
202 {
203         int count, ch, completely_empty = 1, nl_just_seen = 0;
204         count = 0;
205         while (0 < size--) {
206                 ch = *data++;
207                 if (ch == '\n') {
208                         count++;
209                         nl_just_seen = 1;
210                         completely_empty = 0;
211                 }
212                 else {
213                         nl_just_seen = 0;
214                         completely_empty = 0;
215                 }
216         }
217         if (completely_empty)
218                 return 0;
219         if (!nl_just_seen)
220                 count++; /* no trailing newline */
221         return count;
222 }
223
224 static void print_line_count(FILE *file, int count)
225 {
226         switch (count) {
227         case 0:
228                 fprintf(file, "0,0");
229                 break;
230         case 1:
231                 fprintf(file, "1");
232                 break;
233         default:
234                 fprintf(file, "1,%d", count);
235                 break;
236         }
237 }
238
239 static void copy_file_with_prefix(FILE *file,
240                                   int prefix, const char *data, int size,
241                                   const char *set, const char *reset)
242 {
243         int ch, nl_just_seen = 1;
244         while (0 < size--) {
245                 ch = *data++;
246                 if (nl_just_seen) {
247                         fputs(set, file);
248                         putc(prefix, file);
249                 }
250                 if (ch == '\n') {
251                         nl_just_seen = 1;
252                         fputs(reset, file);
253                 } else
254                         nl_just_seen = 0;
255                 putc(ch, file);
256         }
257         if (!nl_just_seen)
258                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
259 }
260
261 static void emit_rewrite_diff(const char *name_a,
262                               const char *name_b,
263                               struct diff_filespec *one,
264                               struct diff_filespec *two,
265                               const char *textconv_one,
266                               const char *textconv_two,
267                               struct diff_options *o)
268 {
269         int lc_a, lc_b;
270         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
271         const char *name_a_tab, *name_b_tab;
272         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
273         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
274         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
275         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
276         const char *reset = diff_get_color(color_diff, DIFF_RESET);
277         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
278         const char *a_prefix, *b_prefix;
279         const char *data_one, *data_two;
280         size_t size_one, size_two;
281
282         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
283                 a_prefix = o->b_prefix;
284                 b_prefix = o->a_prefix;
285         } else {
286                 a_prefix = o->a_prefix;
287                 b_prefix = o->b_prefix;
288         }
289
290         name_a += (*name_a == '/');
291         name_b += (*name_b == '/');
292         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
293         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
294
295         strbuf_reset(&a_name);
296         strbuf_reset(&b_name);
297         quote_two_c_style(&a_name, a_prefix, name_a, 0);
298         quote_two_c_style(&b_name, b_prefix, name_b, 0);
299
300         diff_populate_filespec(one, 0);
301         diff_populate_filespec(two, 0);
302         if (textconv_one) {
303                 data_one = run_textconv(textconv_one, one, &size_one);
304                 if (!data_one)
305                         die("unable to read files to diff");
306         }
307         else {
308                 data_one = one->data;
309                 size_one = one->size;
310         }
311         if (textconv_two) {
312                 data_two = run_textconv(textconv_two, two, &size_two);
313                 if (!data_two)
314                         die("unable to read files to diff");
315         }
316         else {
317                 data_two = two->data;
318                 size_two = two->size;
319         }
320
321         lc_a = count_lines(data_one, size_one);
322         lc_b = count_lines(data_two, size_two);
323         fprintf(o->file,
324                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
325                 metainfo, a_name.buf, name_a_tab, reset,
326                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
327         print_line_count(o->file, lc_a);
328         fprintf(o->file, " +");
329         print_line_count(o->file, lc_b);
330         fprintf(o->file, " @@%s\n", reset);
331         if (lc_a)
332                 copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
333         if (lc_b)
334                 copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
335 }
336
337 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
338 {
339         if (!DIFF_FILE_VALID(one)) {
340                 mf->ptr = (char *)""; /* does not matter */
341                 mf->size = 0;
342                 return 0;
343         }
344         else if (diff_populate_filespec(one, 0))
345                 return -1;
346
347         mf->ptr = one->data;
348         mf->size = one->size;
349         return 0;
350 }
351
352 struct diff_words_buffer {
353         mmfile_t text;
354         long alloc;
355         struct diff_words_orig {
356                 const char *begin, *end;
357         } *orig;
358         int orig_nr, orig_alloc;
359 };
360
361 static void diff_words_append(char *line, unsigned long len,
362                 struct diff_words_buffer *buffer)
363 {
364         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
365         line++;
366         len--;
367         memcpy(buffer->text.ptr + buffer->text.size, line, len);
368         buffer->text.size += len;
369         buffer->text.ptr[buffer->text.size] = '\0';
370 }
371
372 struct diff_words_data {
373         struct diff_words_buffer minus, plus;
374         const char *current_plus;
375         FILE *file;
376         regex_t *word_regex;
377 };
378
379 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
380 {
381         struct diff_words_data *diff_words = priv;
382         int minus_first, minus_len, plus_first, plus_len;
383         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
384
385         if (line[0] != '@' || parse_hunk_header(line, len,
386                         &minus_first, &minus_len, &plus_first, &plus_len))
387                 return;
388
389         /* POSIX requires that first be decremented by one if len == 0... */
390         if (minus_len) {
391                 minus_begin = diff_words->minus.orig[minus_first].begin;
392                 minus_end =
393                         diff_words->minus.orig[minus_first + minus_len - 1].end;
394         } else
395                 minus_begin = minus_end =
396                         diff_words->minus.orig[minus_first].end;
397
398         if (plus_len) {
399                 plus_begin = diff_words->plus.orig[plus_first].begin;
400                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
401         } else
402                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
403
404         if (diff_words->current_plus != plus_begin)
405                 fwrite(diff_words->current_plus,
406                                 plus_begin - diff_words->current_plus, 1,
407                                 diff_words->file);
408         if (minus_begin != minus_end)
409                 color_fwrite_lines(diff_words->file,
410                                 diff_get_color(1, DIFF_FILE_OLD),
411                                 minus_end - minus_begin, minus_begin);
412         if (plus_begin != plus_end)
413                 color_fwrite_lines(diff_words->file,
414                                 diff_get_color(1, DIFF_FILE_NEW),
415                                 plus_end - plus_begin, plus_begin);
416
417         diff_words->current_plus = plus_end;
418 }
419
420 /* This function starts looking at *begin, and returns 0 iff a word was found. */
421 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
422                 int *begin, int *end)
423 {
424         if (word_regex && *begin < buffer->size) {
425                 regmatch_t match[1];
426                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
427                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
428                                         '\n', match[0].rm_eo - match[0].rm_so);
429                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
430                         *begin += match[0].rm_so;
431                         return *begin >= *end;
432                 }
433                 return -1;
434         }
435
436         /* find the next word */
437         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
438                 (*begin)++;
439         if (*begin >= buffer->size)
440                 return -1;
441
442         /* find the end of the word */
443         *end = *begin + 1;
444         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
445                 (*end)++;
446
447         return 0;
448 }
449
450 /*
451  * This function splits the words in buffer->text, stores the list with
452  * newline separator into out, and saves the offsets of the original words
453  * in buffer->orig.
454  */
455 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
456                 regex_t *word_regex)
457 {
458         int i, j;
459         long alloc = 0;
460
461         out->size = 0;
462         out->ptr = NULL;
463
464         /* fake an empty "0th" word */
465         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
466         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
467         buffer->orig_nr = 1;
468
469         for (i = 0; i < buffer->text.size; i++) {
470                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
471                         return;
472
473                 /* store original boundaries */
474                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
475                                 buffer->orig_alloc);
476                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
477                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
478                 buffer->orig_nr++;
479
480                 /* store one word */
481                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
482                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
483                 out->ptr[out->size + j - i] = '\n';
484                 out->size += j - i + 1;
485
486                 i = j - 1;
487         }
488 }
489
490 /* this executes the word diff on the accumulated buffers */
491 static void diff_words_show(struct diff_words_data *diff_words)
492 {
493         xpparam_t xpp;
494         xdemitconf_t xecfg;
495         xdemitcb_t ecb;
496         mmfile_t minus, plus;
497
498         /* special case: only removal */
499         if (!diff_words->plus.text.size) {
500                 color_fwrite_lines(diff_words->file,
501                         diff_get_color(1, DIFF_FILE_OLD),
502                         diff_words->minus.text.size, diff_words->minus.text.ptr);
503                 diff_words->minus.text.size = 0;
504                 return;
505         }
506
507         diff_words->current_plus = diff_words->plus.text.ptr;
508
509         memset(&xpp, 0, sizeof(xpp));
510         memset(&xecfg, 0, sizeof(xecfg));
511         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
512         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
513         xpp.flags = XDF_NEED_MINIMAL;
514         /* as only the hunk header will be parsed, we need a 0-context */
515         xecfg.ctxlen = 0;
516         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
517                       &xpp, &xecfg, &ecb);
518         free(minus.ptr);
519         free(plus.ptr);
520         if (diff_words->current_plus != diff_words->plus.text.ptr +
521                         diff_words->plus.text.size)
522                 fwrite(diff_words->current_plus,
523                         diff_words->plus.text.ptr + diff_words->plus.text.size
524                         - diff_words->current_plus, 1,
525                         diff_words->file);
526         diff_words->minus.text.size = diff_words->plus.text.size = 0;
527 }
528
529 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
530
531 struct emit_callback {
532         int nparents, color_diff;
533         unsigned ws_rule;
534         sane_truncate_fn truncate;
535         const char **label_path;
536         struct diff_words_data *diff_words;
537         int *found_changesp;
538         FILE *file;
539 };
540
541 static void free_diff_words_data(struct emit_callback *ecbdata)
542 {
543         if (ecbdata->diff_words) {
544                 /* flush buffers */
545                 if (ecbdata->diff_words->minus.text.size ||
546                                 ecbdata->diff_words->plus.text.size)
547                         diff_words_show(ecbdata->diff_words);
548
549                 free (ecbdata->diff_words->minus.text.ptr);
550                 free (ecbdata->diff_words->minus.orig);
551                 free (ecbdata->diff_words->plus.text.ptr);
552                 free (ecbdata->diff_words->plus.orig);
553                 free(ecbdata->diff_words->word_regex);
554                 free(ecbdata->diff_words);
555                 ecbdata->diff_words = NULL;
556         }
557 }
558
559 const char *diff_get_color(int diff_use_color, enum color_diff ix)
560 {
561         if (diff_use_color)
562                 return diff_colors[ix];
563         return "";
564 }
565
566 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
567 {
568         int has_trailing_newline, has_trailing_carriage_return;
569
570         has_trailing_newline = (len > 0 && line[len-1] == '\n');
571         if (has_trailing_newline)
572                 len--;
573         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
574         if (has_trailing_carriage_return)
575                 len--;
576
577         fputs(set, file);
578         fwrite(line, len, 1, file);
579         fputs(reset, file);
580         if (has_trailing_carriage_return)
581                 fputc('\r', file);
582         if (has_trailing_newline)
583                 fputc('\n', file);
584 }
585
586 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
587 {
588         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
589         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
590
591         if (!*ws)
592                 emit_line(ecbdata->file, set, reset, line, len);
593         else {
594                 /* Emit just the prefix, then the rest. */
595                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
596                 ws_check_emit(line + ecbdata->nparents,
597                               len - ecbdata->nparents, ecbdata->ws_rule,
598                               ecbdata->file, set, reset, ws);
599         }
600 }
601
602 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
603 {
604         const char *cp;
605         unsigned long allot;
606         size_t l = len;
607
608         if (ecb->truncate)
609                 return ecb->truncate(line, len);
610         cp = line;
611         allot = l;
612         while (0 < l) {
613                 (void) utf8_width(&cp, &l);
614                 if (!cp)
615                         break; /* truncated in the middle? */
616         }
617         return allot - l;
618 }
619
620 static void fn_out_consume(void *priv, char *line, unsigned long len)
621 {
622         int i;
623         int color;
624         struct emit_callback *ecbdata = priv;
625         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
626         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
627         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
628
629         *(ecbdata->found_changesp) = 1;
630
631         if (ecbdata->label_path[0]) {
632                 const char *name_a_tab, *name_b_tab;
633
634                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
635                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
636
637                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
638                         meta, ecbdata->label_path[0], reset, name_a_tab);
639                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
640                         meta, ecbdata->label_path[1], reset, name_b_tab);
641                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
642         }
643
644         if (diff_suppress_blank_empty
645             && len == 2 && line[0] == ' ' && line[1] == '\n') {
646                 line[0] = '\n';
647                 len = 1;
648         }
649
650         /* This is not really necessary for now because
651          * this codepath only deals with two-way diffs.
652          */
653         for (i = 0; i < len && line[i] == '@'; i++)
654                 ;
655         if (2 <= i && i < len && line[i] == ' ') {
656                 ecbdata->nparents = i - 1;
657                 len = sane_truncate_line(ecbdata, line, len);
658                 emit_line(ecbdata->file,
659                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
660                           reset, line, len);
661                 if (line[len-1] != '\n')
662                         putc('\n', ecbdata->file);
663                 return;
664         }
665
666         if (len < ecbdata->nparents) {
667                 emit_line(ecbdata->file, reset, reset, line, len);
668                 return;
669         }
670
671         color = DIFF_PLAIN;
672         if (ecbdata->diff_words && ecbdata->nparents != 1)
673                 /* fall back to normal diff */
674                 free_diff_words_data(ecbdata);
675         if (ecbdata->diff_words) {
676                 if (line[0] == '-') {
677                         diff_words_append(line, len,
678                                           &ecbdata->diff_words->minus);
679                         return;
680                 } else if (line[0] == '+') {
681                         diff_words_append(line, len,
682                                           &ecbdata->diff_words->plus);
683                         return;
684                 }
685                 if (ecbdata->diff_words->minus.text.size ||
686                     ecbdata->diff_words->plus.text.size)
687                         diff_words_show(ecbdata->diff_words);
688                 line++;
689                 len--;
690                 emit_line(ecbdata->file, plain, reset, line, len);
691                 return;
692         }
693         for (i = 0; i < ecbdata->nparents && len; i++) {
694                 if (line[i] == '-')
695                         color = DIFF_FILE_OLD;
696                 else if (line[i] == '+')
697                         color = DIFF_FILE_NEW;
698         }
699
700         if (color != DIFF_FILE_NEW) {
701                 emit_line(ecbdata->file,
702                           diff_get_color(ecbdata->color_diff, color),
703                           reset, line, len);
704                 return;
705         }
706         emit_add_line(reset, ecbdata, line, len);
707 }
708
709 static char *pprint_rename(const char *a, const char *b)
710 {
711         const char *old = a;
712         const char *new = b;
713         struct strbuf name = STRBUF_INIT;
714         int pfx_length, sfx_length;
715         int len_a = strlen(a);
716         int len_b = strlen(b);
717         int a_midlen, b_midlen;
718         int qlen_a = quote_c_style(a, NULL, NULL, 0);
719         int qlen_b = quote_c_style(b, NULL, NULL, 0);
720
721         if (qlen_a || qlen_b) {
722                 quote_c_style(a, &name, NULL, 0);
723                 strbuf_addstr(&name, " => ");
724                 quote_c_style(b, &name, NULL, 0);
725                 return strbuf_detach(&name, NULL);
726         }
727
728         /* Find common prefix */
729         pfx_length = 0;
730         while (*old && *new && *old == *new) {
731                 if (*old == '/')
732                         pfx_length = old - a + 1;
733                 old++;
734                 new++;
735         }
736
737         /* Find common suffix */
738         old = a + len_a;
739         new = b + len_b;
740         sfx_length = 0;
741         while (a <= old && b <= new && *old == *new) {
742                 if (*old == '/')
743                         sfx_length = len_a - (old - a);
744                 old--;
745                 new--;
746         }
747
748         /*
749          * pfx{mid-a => mid-b}sfx
750          * {pfx-a => pfx-b}sfx
751          * pfx{sfx-a => sfx-b}
752          * name-a => name-b
753          */
754         a_midlen = len_a - pfx_length - sfx_length;
755         b_midlen = len_b - pfx_length - sfx_length;
756         if (a_midlen < 0)
757                 a_midlen = 0;
758         if (b_midlen < 0)
759                 b_midlen = 0;
760
761         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
762         if (pfx_length + sfx_length) {
763                 strbuf_add(&name, a, pfx_length);
764                 strbuf_addch(&name, '{');
765         }
766         strbuf_add(&name, a + pfx_length, a_midlen);
767         strbuf_addstr(&name, " => ");
768         strbuf_add(&name, b + pfx_length, b_midlen);
769         if (pfx_length + sfx_length) {
770                 strbuf_addch(&name, '}');
771                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
772         }
773         return strbuf_detach(&name, NULL);
774 }
775
776 struct diffstat_t {
777         int nr;
778         int alloc;
779         struct diffstat_file {
780                 char *from_name;
781                 char *name;
782                 char *print_name;
783                 unsigned is_unmerged:1;
784                 unsigned is_binary:1;
785                 unsigned is_renamed:1;
786                 unsigned int added, deleted;
787         } **files;
788 };
789
790 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
791                                           const char *name_a,
792                                           const char *name_b)
793 {
794         struct diffstat_file *x;
795         x = xcalloc(sizeof (*x), 1);
796         if (diffstat->nr == diffstat->alloc) {
797                 diffstat->alloc = alloc_nr(diffstat->alloc);
798                 diffstat->files = xrealloc(diffstat->files,
799                                 diffstat->alloc * sizeof(x));
800         }
801         diffstat->files[diffstat->nr++] = x;
802         if (name_b) {
803                 x->from_name = xstrdup(name_a);
804                 x->name = xstrdup(name_b);
805                 x->is_renamed = 1;
806         }
807         else {
808                 x->from_name = NULL;
809                 x->name = xstrdup(name_a);
810         }
811         return x;
812 }
813
814 static void diffstat_consume(void *priv, char *line, unsigned long len)
815 {
816         struct diffstat_t *diffstat = priv;
817         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
818
819         if (line[0] == '+')
820                 x->added++;
821         else if (line[0] == '-')
822                 x->deleted++;
823 }
824
825 const char mime_boundary_leader[] = "------------";
826
827 static int scale_linear(int it, int width, int max_change)
828 {
829         /*
830          * make sure that at least one '-' is printed if there were deletions,
831          * and likewise for '+'.
832          */
833         if (max_change < 2)
834                 return it;
835         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
836 }
837
838 static void show_name(FILE *file,
839                       const char *prefix, const char *name, int len,
840                       const char *reset, const char *set)
841 {
842         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
843 }
844
845 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
846 {
847         if (cnt <= 0)
848                 return;
849         fprintf(file, "%s", set);
850         while (cnt--)
851                 putc(ch, file);
852         fprintf(file, "%s", reset);
853 }
854
855 static void fill_print_name(struct diffstat_file *file)
856 {
857         char *pname;
858
859         if (file->print_name)
860                 return;
861
862         if (!file->is_renamed) {
863                 struct strbuf buf = STRBUF_INIT;
864                 if (quote_c_style(file->name, &buf, NULL, 0)) {
865                         pname = strbuf_detach(&buf, NULL);
866                 } else {
867                         pname = file->name;
868                         strbuf_release(&buf);
869                 }
870         } else {
871                 pname = pprint_rename(file->from_name, file->name);
872         }
873         file->print_name = pname;
874 }
875
876 static void show_stats(struct diffstat_t* data, struct diff_options *options)
877 {
878         int i, len, add, del, total, adds = 0, dels = 0;
879         int max_change = 0, max_len = 0;
880         int total_files = data->nr;
881         int width, name_width;
882         const char *reset, *set, *add_c, *del_c;
883
884         if (data->nr == 0)
885                 return;
886
887         width = options->stat_width ? options->stat_width : 80;
888         name_width = options->stat_name_width ? options->stat_name_width : 50;
889
890         /* Sanity: give at least 5 columns to the graph,
891          * but leave at least 10 columns for the name.
892          */
893         if (width < 25)
894                 width = 25;
895         if (name_width < 10)
896                 name_width = 10;
897         else if (width < name_width + 15)
898                 name_width = width - 15;
899
900         /* Find the longest filename and max number of changes */
901         reset = diff_get_color_opt(options, DIFF_RESET);
902         set   = diff_get_color_opt(options, DIFF_PLAIN);
903         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
904         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
905
906         for (i = 0; i < data->nr; i++) {
907                 struct diffstat_file *file = data->files[i];
908                 int change = file->added + file->deleted;
909                 fill_print_name(file);
910                 len = strlen(file->print_name);
911                 if (max_len < len)
912                         max_len = len;
913
914                 if (file->is_binary || file->is_unmerged)
915                         continue;
916                 if (max_change < change)
917                         max_change = change;
918         }
919
920         /* Compute the width of the graph part;
921          * 10 is for one blank at the beginning of the line plus
922          * " | count " between the name and the graph.
923          *
924          * From here on, name_width is the width of the name area,
925          * and width is the width of the graph area.
926          */
927         name_width = (name_width < max_len) ? name_width : max_len;
928         if (width < (name_width + 10) + max_change)
929                 width = width - (name_width + 10);
930         else
931                 width = max_change;
932
933         for (i = 0; i < data->nr; i++) {
934                 const char *prefix = "";
935                 char *name = data->files[i]->print_name;
936                 int added = data->files[i]->added;
937                 int deleted = data->files[i]->deleted;
938                 int name_len;
939
940                 /*
941                  * "scale" the filename
942                  */
943                 len = name_width;
944                 name_len = strlen(name);
945                 if (name_width < name_len) {
946                         char *slash;
947                         prefix = "...";
948                         len -= 3;
949                         name += name_len - len;
950                         slash = strchr(name, '/');
951                         if (slash)
952                                 name = slash;
953                 }
954
955                 if (data->files[i]->is_binary) {
956                         show_name(options->file, prefix, name, len, reset, set);
957                         fprintf(options->file, "  Bin ");
958                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
959                         fprintf(options->file, " -> ");
960                         fprintf(options->file, "%s%d%s", add_c, added, reset);
961                         fprintf(options->file, " bytes");
962                         fprintf(options->file, "\n");
963                         continue;
964                 }
965                 else if (data->files[i]->is_unmerged) {
966                         show_name(options->file, prefix, name, len, reset, set);
967                         fprintf(options->file, "  Unmerged\n");
968                         continue;
969                 }
970                 else if (!data->files[i]->is_renamed &&
971                          (added + deleted == 0)) {
972                         total_files--;
973                         continue;
974                 }
975
976                 /*
977                  * scale the add/delete
978                  */
979                 add = added;
980                 del = deleted;
981                 total = add + del;
982                 adds += add;
983                 dels += del;
984
985                 if (width <= max_change) {
986                         add = scale_linear(add, width, max_change);
987                         del = scale_linear(del, width, max_change);
988                         total = add + del;
989                 }
990                 show_name(options->file, prefix, name, len, reset, set);
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                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
999                set, total_files, adds, dels, reset);
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         /* We do not read the cache ourselves here, because the
1763          * benchmark with my previous version that always reads cache
1764          * shows that it makes things worse for diff-tree comparing
1765          * two linux-2.6 kernel trees in an already checked out work
1766          * tree.  This is because most diff-tree comparisons deal with
1767          * only a small number of files, while reading the cache is
1768          * expensive for a large project, and its cost outweighs the
1769          * savings we get by not inflating the object to a temporary
1770          * file.  Practically, this code only helps when we are used
1771          * by diff-cache --cached, which does read the cache before
1772          * calling us.
1773          */
1774         if (!active_cache)
1775                 return 0;
1776
1777         /* We want to avoid the working directory if our caller
1778          * doesn't need the data in a normal file, this system
1779          * is rather slow with its stat/open/mmap/close syscalls,
1780          * and the object is contained in a pack file.  The pack
1781          * is probably already open and will be faster to obtain
1782          * the data through than the working directory.  Loose
1783          * objects however would tend to be slower as they need
1784          * to be individually opened and inflated.
1785          */
1786         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1787                 return 0;
1788
1789         len = strlen(name);
1790         pos = cache_name_pos(name, len);
1791         if (pos < 0)
1792                 return 0;
1793         ce = active_cache[pos];
1794
1795         /*
1796          * This is not the sha1 we are looking for, or
1797          * unreusable because it is not a regular file.
1798          */
1799         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1800                 return 0;
1801
1802         /*
1803          * If ce matches the file in the work tree, we can reuse it.
1804          */
1805         if (ce_uptodate(ce) ||
1806             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1807                 return 1;
1808
1809         return 0;
1810 }
1811
1812 static int populate_from_stdin(struct diff_filespec *s)
1813 {
1814         struct strbuf buf = STRBUF_INIT;
1815         size_t size = 0;
1816
1817         if (strbuf_read(&buf, 0, 0) < 0)
1818                 return error("error while reading from stdin %s",
1819                                      strerror(errno));
1820
1821         s->should_munmap = 0;
1822         s->data = strbuf_detach(&buf, &size);
1823         s->size = size;
1824         s->should_free = 1;
1825         return 0;
1826 }
1827
1828 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1829 {
1830         int len;
1831         char *data = xmalloc(100);
1832         len = snprintf(data, 100,
1833                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1834         s->data = data;
1835         s->size = len;
1836         s->should_free = 1;
1837         if (size_only) {
1838                 s->data = NULL;
1839                 free(data);
1840         }
1841         return 0;
1842 }
1843
1844 /*
1845  * While doing rename detection and pickaxe operation, we may need to
1846  * grab the data for the blob (or file) for our own in-core comparison.
1847  * diff_filespec has data and size fields for this purpose.
1848  */
1849 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1850 {
1851         int err = 0;
1852         if (!DIFF_FILE_VALID(s))
1853                 die("internal error: asking to populate invalid file.");
1854         if (S_ISDIR(s->mode))
1855                 return -1;
1856
1857         if (s->data)
1858                 return 0;
1859
1860         if (size_only && 0 < s->size)
1861                 return 0;
1862
1863         if (S_ISGITLINK(s->mode))
1864                 return diff_populate_gitlink(s, size_only);
1865
1866         if (!s->sha1_valid ||
1867             reuse_worktree_file(s->path, s->sha1, 0)) {
1868                 struct strbuf buf = STRBUF_INIT;
1869                 struct stat st;
1870                 int fd;
1871
1872                 if (!strcmp(s->path, "-"))
1873                         return populate_from_stdin(s);
1874
1875                 if (lstat(s->path, &st) < 0) {
1876                         if (errno == ENOENT) {
1877                         err_empty:
1878                                 err = -1;
1879                         empty:
1880                                 s->data = (char *)"";
1881                                 s->size = 0;
1882                                 return err;
1883                         }
1884                 }
1885                 s->size = xsize_t(st.st_size);
1886                 if (!s->size)
1887                         goto empty;
1888                 if (S_ISLNK(st.st_mode)) {
1889                         struct strbuf sb = STRBUF_INIT;
1890
1891                         if (strbuf_readlink(&sb, s->path, s->size))
1892                                 goto err_empty;
1893                         s->size = sb.len;
1894                         s->data = strbuf_detach(&sb, NULL);
1895                         s->should_free = 1;
1896                         return 0;
1897                 }
1898                 if (size_only)
1899                         return 0;
1900                 fd = open(s->path, O_RDONLY);
1901                 if (fd < 0)
1902                         goto err_empty;
1903                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1904                 close(fd);
1905                 s->should_munmap = 1;
1906
1907                 /*
1908                  * Convert from working tree format to canonical git format
1909                  */
1910                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1911                         size_t size = 0;
1912                         munmap(s->data, s->size);
1913                         s->should_munmap = 0;
1914                         s->data = strbuf_detach(&buf, &size);
1915                         s->size = size;
1916                         s->should_free = 1;
1917                 }
1918         }
1919         else {
1920                 enum object_type type;
1921                 if (size_only)
1922                         type = sha1_object_info(s->sha1, &s->size);
1923                 else {
1924                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1925                         s->should_free = 1;
1926                 }
1927         }
1928         return 0;
1929 }
1930
1931 void diff_free_filespec_blob(struct diff_filespec *s)
1932 {
1933         if (s->should_free)
1934                 free(s->data);
1935         else if (s->should_munmap)
1936                 munmap(s->data, s->size);
1937
1938         if (s->should_free || s->should_munmap) {
1939                 s->should_free = s->should_munmap = 0;
1940                 s->data = NULL;
1941         }
1942 }
1943
1944 void diff_free_filespec_data(struct diff_filespec *s)
1945 {
1946         diff_free_filespec_blob(s);
1947         free(s->cnt_data);
1948         s->cnt_data = NULL;
1949 }
1950
1951 static void prep_temp_blob(struct diff_tempfile *temp,
1952                            void *blob,
1953                            unsigned long size,
1954                            const unsigned char *sha1,
1955                            int mode)
1956 {
1957         int fd;
1958
1959         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1960         if (fd < 0)
1961                 die("unable to create temp-file: %s", strerror(errno));
1962         if (write_in_full(fd, blob, size) != size)
1963                 die("unable to write temp-file");
1964         close(fd);
1965         temp->name = temp->tmp_path;
1966         strcpy(temp->hex, sha1_to_hex(sha1));
1967         temp->hex[40] = 0;
1968         sprintf(temp->mode, "%06o", mode);
1969 }
1970
1971 static struct diff_tempfile *prepare_temp_file(const char *name,
1972                 struct diff_filespec *one)
1973 {
1974         struct diff_tempfile *temp = claim_diff_tempfile();
1975
1976         if (!DIFF_FILE_VALID(one)) {
1977         not_a_valid_file:
1978                 /* A '-' entry produces this for file-2, and
1979                  * a '+' entry produces this for file-1.
1980                  */
1981                 temp->name = "/dev/null";
1982                 strcpy(temp->hex, ".");
1983                 strcpy(temp->mode, ".");
1984                 return temp;
1985         }
1986
1987         if (!remove_tempfile_installed) {
1988                 atexit(remove_tempfile);
1989                 sigchain_push_common(remove_tempfile_on_signal);
1990                 remove_tempfile_installed = 1;
1991         }
1992
1993         if (!one->sha1_valid ||
1994             reuse_worktree_file(name, one->sha1, 1)) {
1995                 struct stat st;
1996                 if (lstat(name, &st) < 0) {
1997                         if (errno == ENOENT)
1998                                 goto not_a_valid_file;
1999                         die("stat(%s): %s", name, strerror(errno));
2000                 }
2001                 if (S_ISLNK(st.st_mode)) {
2002                         int ret;
2003                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
2004                         ret = readlink(name, buf, sizeof(buf));
2005                         if (ret < 0)
2006                                 die("readlink(%s)", name);
2007                         if (ret == sizeof(buf))
2008                                 die("symlink too long: %s", name);
2009                         prep_temp_blob(temp, buf, ret,
2010                                        (one->sha1_valid ?
2011                                         one->sha1 : null_sha1),
2012                                        (one->sha1_valid ?
2013                                         one->mode : S_IFLNK));
2014                 }
2015                 else {
2016                         /* we can borrow from the file in the work tree */
2017                         temp->name = name;
2018                         if (!one->sha1_valid)
2019                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2020                         else
2021                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2022                         /* Even though we may sometimes borrow the
2023                          * contents from the work tree, we always want
2024                          * one->mode.  mode is trustworthy even when
2025                          * !(one->sha1_valid), as long as
2026                          * DIFF_FILE_VALID(one).
2027                          */
2028                         sprintf(temp->mode, "%06o", one->mode);
2029                 }
2030                 return temp;
2031         }
2032         else {
2033                 if (diff_populate_filespec(one, 0))
2034                         die("cannot read data blob for %s", one->path);
2035                 prep_temp_blob(temp, one->data, one->size,
2036                                one->sha1, one->mode);
2037         }
2038         return temp;
2039 }
2040
2041 /* An external diff command takes:
2042  *
2043  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2044  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2045  *
2046  */
2047 static void run_external_diff(const char *pgm,
2048                               const char *name,
2049                               const char *other,
2050                               struct diff_filespec *one,
2051                               struct diff_filespec *two,
2052                               const char *xfrm_msg,
2053                               int complete_rewrite)
2054 {
2055         const char *spawn_arg[10];
2056         int retval;
2057         const char **arg = &spawn_arg[0];
2058
2059         if (one && two) {
2060                 struct diff_tempfile *temp_one, *temp_two;
2061                 const char *othername = (other ? other : name);
2062                 temp_one = prepare_temp_file(name, one);
2063                 temp_two = prepare_temp_file(othername, two);
2064                 *arg++ = pgm;
2065                 *arg++ = name;
2066                 *arg++ = temp_one->name;
2067                 *arg++ = temp_one->hex;
2068                 *arg++ = temp_one->mode;
2069                 *arg++ = temp_two->name;
2070                 *arg++ = temp_two->hex;
2071                 *arg++ = temp_two->mode;
2072                 if (other) {
2073                         *arg++ = other;
2074                         *arg++ = xfrm_msg;
2075                 }
2076         } else {
2077                 *arg++ = pgm;
2078                 *arg++ = name;
2079         }
2080         *arg = NULL;
2081         fflush(NULL);
2082         retval = run_command_v_opt(spawn_arg, 0);
2083         remove_tempfile();
2084         if (retval) {
2085                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2086                 exit(1);
2087         }
2088 }
2089
2090 static int similarity_index(struct diff_filepair *p)
2091 {
2092         return p->score * 100 / MAX_SCORE;
2093 }
2094
2095 static void fill_metainfo(struct strbuf *msg,
2096                           const char *name,
2097                           const char *other,
2098                           struct diff_filespec *one,
2099                           struct diff_filespec *two,
2100                           struct diff_options *o,
2101                           struct diff_filepair *p)
2102 {
2103         strbuf_init(msg, PATH_MAX * 2 + 300);
2104         switch (p->status) {
2105         case DIFF_STATUS_COPIED:
2106                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2107                 strbuf_addstr(msg, "\ncopy from ");
2108                 quote_c_style(name, msg, NULL, 0);
2109                 strbuf_addstr(msg, "\ncopy to ");
2110                 quote_c_style(other, msg, NULL, 0);
2111                 strbuf_addch(msg, '\n');
2112                 break;
2113         case DIFF_STATUS_RENAMED:
2114                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2115                 strbuf_addstr(msg, "\nrename from ");
2116                 quote_c_style(name, msg, NULL, 0);
2117                 strbuf_addstr(msg, "\nrename to ");
2118                 quote_c_style(other, msg, NULL, 0);
2119                 strbuf_addch(msg, '\n');
2120                 break;
2121         case DIFF_STATUS_MODIFIED:
2122                 if (p->score) {
2123                         strbuf_addf(msg, "dissimilarity index %d%%\n",
2124                                     similarity_index(p));
2125                         break;
2126                 }
2127                 /* fallthru */
2128         default:
2129                 /* nothing */
2130                 ;
2131         }
2132         if (one && two && hashcmp(one->sha1, two->sha1)) {
2133                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2134
2135                 if (DIFF_OPT_TST(o, BINARY)) {
2136                         mmfile_t mf;
2137                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2138                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2139                                 abbrev = 40;
2140                 }
2141                 strbuf_addf(msg, "index %.*s..%.*s",
2142                             abbrev, sha1_to_hex(one->sha1),
2143                             abbrev, sha1_to_hex(two->sha1));
2144                 if (one->mode == two->mode)
2145                         strbuf_addf(msg, " %06o", one->mode);
2146                 strbuf_addch(msg, '\n');
2147         }
2148         if (msg->len)
2149                 strbuf_setlen(msg, msg->len - 1);
2150 }
2151
2152 static void run_diff_cmd(const char *pgm,
2153                          const char *name,
2154                          const char *other,
2155                          const char *attr_path,
2156                          struct diff_filespec *one,
2157                          struct diff_filespec *two,
2158                          struct strbuf *msg,
2159                          struct diff_options *o,
2160                          struct diff_filepair *p)
2161 {
2162         const char *xfrm_msg = NULL;
2163         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2164
2165         if (msg) {
2166                 fill_metainfo(msg, name, other, one, two, o, p);
2167                 xfrm_msg = msg->len ? msg->buf : NULL;
2168         }
2169
2170         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2171                 pgm = NULL;
2172         else {
2173                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2174                 if (drv && drv->external)
2175                         pgm = drv->external;
2176         }
2177
2178         if (pgm) {
2179                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2180                                   complete_rewrite);
2181                 return;
2182         }
2183         if (one && two)
2184                 builtin_diff(name, other ? other : name,
2185                              one, two, xfrm_msg, o, complete_rewrite);
2186         else
2187                 fprintf(o->file, "* Unmerged path %s\n", name);
2188 }
2189
2190 static void diff_fill_sha1_info(struct diff_filespec *one)
2191 {
2192         if (DIFF_FILE_VALID(one)) {
2193                 if (!one->sha1_valid) {
2194                         struct stat st;
2195                         if (!strcmp(one->path, "-")) {
2196                                 hashcpy(one->sha1, null_sha1);
2197                                 return;
2198                         }
2199                         if (lstat(one->path, &st) < 0)
2200                                 die("stat %s", one->path);
2201                         if (index_path(one->sha1, one->path, &st, 0))
2202                                 die("cannot hash %s", one->path);
2203                 }
2204         }
2205         else
2206                 hashclr(one->sha1);
2207 }
2208
2209 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2210 {
2211         /* Strip the prefix but do not molest /dev/null and absolute paths */
2212         if (*namep && **namep != '/')
2213                 *namep += prefix_length;
2214         if (*otherp && **otherp != '/')
2215                 *otherp += prefix_length;
2216 }
2217
2218 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2219 {
2220         const char *pgm = external_diff();
2221         struct strbuf msg;
2222         struct diff_filespec *one = p->one;
2223         struct diff_filespec *two = p->two;
2224         const char *name;
2225         const char *other;
2226         const char *attr_path;
2227
2228         name  = p->one->path;
2229         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2230         attr_path = name;
2231         if (o->prefix_length)
2232                 strip_prefix(o->prefix_length, &name, &other);
2233
2234         if (DIFF_PAIR_UNMERGED(p)) {
2235                 run_diff_cmd(pgm, name, NULL, attr_path,
2236                              NULL, NULL, NULL, o, p);
2237                 return;
2238         }
2239
2240         diff_fill_sha1_info(one);
2241         diff_fill_sha1_info(two);
2242
2243         if (!pgm &&
2244             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2245             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2246                 /*
2247                  * a filepair that changes between file and symlink
2248                  * needs to be split into deletion and creation.
2249                  */
2250                 struct diff_filespec *null = alloc_filespec(two->path);
2251                 run_diff_cmd(NULL, name, other, attr_path,
2252                              one, null, &msg, o, p);
2253                 free(null);
2254                 strbuf_release(&msg);
2255
2256                 null = alloc_filespec(one->path);
2257                 run_diff_cmd(NULL, name, other, attr_path,
2258                              null, two, &msg, o, p);
2259                 free(null);
2260         }
2261         else
2262                 run_diff_cmd(pgm, name, other, attr_path,
2263                              one, two, &msg, o, p);
2264
2265         strbuf_release(&msg);
2266 }
2267
2268 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2269                          struct diffstat_t *diffstat)
2270 {
2271         const char *name;
2272         const char *other;
2273         int complete_rewrite = 0;
2274
2275         if (DIFF_PAIR_UNMERGED(p)) {
2276                 /* unmerged */
2277                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2278                 return;
2279         }
2280
2281         name = p->one->path;
2282         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2283
2284         if (o->prefix_length)
2285                 strip_prefix(o->prefix_length, &name, &other);
2286
2287         diff_fill_sha1_info(p->one);
2288         diff_fill_sha1_info(p->two);
2289
2290         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2291                 complete_rewrite = 1;
2292         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2293 }
2294
2295 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2296 {
2297         const char *name;
2298         const char *other;
2299         const char *attr_path;
2300
2301         if (DIFF_PAIR_UNMERGED(p)) {
2302                 /* unmerged */
2303                 return;
2304         }
2305
2306         name = p->one->path;
2307         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2308         attr_path = other ? other : name;
2309
2310         if (o->prefix_length)
2311                 strip_prefix(o->prefix_length, &name, &other);
2312
2313         diff_fill_sha1_info(p->one);
2314         diff_fill_sha1_info(p->two);
2315
2316         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2317 }
2318
2319 void diff_setup(struct diff_options *options)
2320 {
2321         memset(options, 0, sizeof(*options));
2322
2323         options->file = stdout;
2324
2325         options->line_termination = '\n';
2326         options->break_opt = -1;
2327         options->rename_limit = -1;
2328         options->dirstat_percent = 3;
2329         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2330         options->context = 3;
2331
2332         options->change = diff_change;
2333         options->add_remove = diff_addremove;
2334         if (diff_use_color_default > 0)
2335                 DIFF_OPT_SET(options, COLOR_DIFF);
2336         else
2337                 DIFF_OPT_CLR(options, COLOR_DIFF);
2338         options->detect_rename = diff_detect_rename_default;
2339
2340         if (!diff_mnemonic_prefix) {
2341                 options->a_prefix = "a/";
2342                 options->b_prefix = "b/";
2343         }
2344 }
2345
2346 int diff_setup_done(struct diff_options *options)
2347 {
2348         int count = 0;
2349
2350         if (options->output_format & DIFF_FORMAT_NAME)
2351                 count++;
2352         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2353                 count++;
2354         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2355                 count++;
2356         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2357                 count++;
2358         if (count > 1)
2359                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2360
2361         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2362                 options->detect_rename = DIFF_DETECT_COPY;
2363
2364         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2365                 options->prefix = NULL;
2366         if (options->prefix)
2367                 options->prefix_length = strlen(options->prefix);
2368         else
2369                 options->prefix_length = 0;
2370
2371         if (options->output_format & (DIFF_FORMAT_NAME |
2372                                       DIFF_FORMAT_NAME_STATUS |
2373                                       DIFF_FORMAT_CHECKDIFF |
2374                                       DIFF_FORMAT_NO_OUTPUT))
2375                 options->output_format &= ~(DIFF_FORMAT_RAW |
2376                                             DIFF_FORMAT_NUMSTAT |
2377                                             DIFF_FORMAT_DIFFSTAT |
2378                                             DIFF_FORMAT_SHORTSTAT |
2379                                             DIFF_FORMAT_DIRSTAT |
2380                                             DIFF_FORMAT_SUMMARY |
2381                                             DIFF_FORMAT_PATCH);
2382
2383         /*
2384          * These cases always need recursive; we do not drop caller-supplied
2385          * recursive bits for other formats here.
2386          */
2387         if (options->output_format & (DIFF_FORMAT_PATCH |
2388                                       DIFF_FORMAT_NUMSTAT |
2389                                       DIFF_FORMAT_DIFFSTAT |
2390                                       DIFF_FORMAT_SHORTSTAT |
2391                                       DIFF_FORMAT_DIRSTAT |
2392                                       DIFF_FORMAT_SUMMARY |
2393                                       DIFF_FORMAT_CHECKDIFF))
2394                 DIFF_OPT_SET(options, RECURSIVE);
2395         /*
2396          * Also pickaxe would not work very well if you do not say recursive
2397          */
2398         if (options->pickaxe)
2399                 DIFF_OPT_SET(options, RECURSIVE);
2400
2401         if (options->detect_rename && options->rename_limit < 0)
2402                 options->rename_limit = diff_rename_limit_default;
2403         if (options->setup & DIFF_SETUP_USE_CACHE) {
2404                 if (!active_cache)
2405                         /* read-cache does not die even when it fails
2406                          * so it is safe for us to do this here.  Also
2407                          * it does not smudge active_cache or active_nr
2408                          * when it fails, so we do not have to worry about
2409                          * cleaning it up ourselves either.
2410                          */
2411                         read_cache();
2412         }
2413         if (options->abbrev <= 0 || 40 < options->abbrev)
2414                 options->abbrev = 40; /* full */
2415
2416         /*
2417          * It does not make sense to show the first hit we happened
2418          * to have found.  It does not make sense not to return with
2419          * exit code in such a case either.
2420          */
2421         if (DIFF_OPT_TST(options, QUIET)) {
2422                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2423                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2424         }
2425
2426         return 0;
2427 }
2428
2429 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2430 {
2431         char c, *eq;
2432         int len;
2433
2434         if (*arg != '-')
2435                 return 0;
2436         c = *++arg;
2437         if (!c)
2438                 return 0;
2439         if (c == arg_short) {
2440                 c = *++arg;
2441                 if (!c)
2442                         return 1;
2443                 if (val && isdigit(c)) {
2444                         char *end;
2445                         int n = strtoul(arg, &end, 10);
2446                         if (*end)
2447                                 return 0;
2448                         *val = n;
2449                         return 1;
2450                 }
2451                 return 0;
2452         }
2453         if (c != '-')
2454                 return 0;
2455         arg++;
2456         eq = strchr(arg, '=');
2457         if (eq)
2458                 len = eq - arg;
2459         else
2460                 len = strlen(arg);
2461         if (!len || strncmp(arg, arg_long, len))
2462                 return 0;
2463         if (eq) {
2464                 int n;
2465                 char *end;
2466                 if (!isdigit(*++eq))
2467                         return 0;
2468                 n = strtoul(eq, &end, 10);
2469                 if (*end)
2470                         return 0;
2471                 *val = n;
2472         }
2473         return 1;
2474 }
2475
2476 static int diff_scoreopt_parse(const char *opt);
2477
2478 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2479 {
2480         const char *arg = av[0];
2481
2482         /* Output format options */
2483         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2484                 options->output_format |= DIFF_FORMAT_PATCH;
2485         else if (opt_arg(arg, 'U', "unified", &options->context))
2486                 options->output_format |= DIFF_FORMAT_PATCH;
2487         else if (!strcmp(arg, "--raw"))
2488                 options->output_format |= DIFF_FORMAT_RAW;
2489         else if (!strcmp(arg, "--patch-with-raw"))
2490                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2491         else if (!strcmp(arg, "--numstat"))
2492                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2493         else if (!strcmp(arg, "--shortstat"))
2494                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2495         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2496                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2497         else if (!strcmp(arg, "--cumulative")) {
2498                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2499                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2500         } else if (opt_arg(arg, 0, "dirstat-by-file",
2501                            &options->dirstat_percent)) {
2502                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2503                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2504         }
2505         else if (!strcmp(arg, "--check"))
2506                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2507         else if (!strcmp(arg, "--summary"))
2508                 options->output_format |= DIFF_FORMAT_SUMMARY;
2509         else if (!strcmp(arg, "--patch-with-stat"))
2510                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2511         else if (!strcmp(arg, "--name-only"))
2512                 options->output_format |= DIFF_FORMAT_NAME;
2513         else if (!strcmp(arg, "--name-status"))
2514                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2515         else if (!strcmp(arg, "-s"))
2516                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2517         else if (!prefixcmp(arg, "--stat")) {
2518                 char *end;
2519                 int width = options->stat_width;
2520                 int name_width = options->stat_name_width;
2521                 arg += 6;
2522                 end = (char *)arg;
2523
2524                 switch (*arg) {
2525                 case '-':
2526                         if (!prefixcmp(arg, "-width="))
2527                                 width = strtoul(arg + 7, &end, 10);
2528                         else if (!prefixcmp(arg, "-name-width="))
2529                                 name_width = strtoul(arg + 12, &end, 10);
2530                         break;
2531                 case '=':
2532                         width = strtoul(arg+1, &end, 10);
2533                         if (*end == ',')
2534                                 name_width = strtoul(end+1, &end, 10);
2535                 }
2536
2537                 /* Important! This checks all the error cases! */
2538                 if (*end)
2539                         return 0;
2540                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2541                 options->stat_name_width = name_width;
2542                 options->stat_width = width;
2543         }
2544
2545         /* renames options */
2546         else if (!prefixcmp(arg, "-B")) {
2547                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2548                         return -1;
2549         }
2550         else if (!prefixcmp(arg, "-M")) {
2551                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2552                         return -1;
2553                 options->detect_rename = DIFF_DETECT_RENAME;
2554         }
2555         else if (!prefixcmp(arg, "-C")) {
2556                 if (options->detect_rename == DIFF_DETECT_COPY)
2557                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2558                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2559                         return -1;
2560                 options->detect_rename = DIFF_DETECT_COPY;
2561         }
2562         else if (!strcmp(arg, "--no-renames"))
2563                 options->detect_rename = 0;
2564         else if (!strcmp(arg, "--relative"))
2565                 DIFF_OPT_SET(options, RELATIVE_NAME);
2566         else if (!prefixcmp(arg, "--relative=")) {
2567                 DIFF_OPT_SET(options, RELATIVE_NAME);
2568                 options->prefix = arg + 11;
2569         }
2570
2571         /* xdiff options */
2572         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2573                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2574         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2575                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2576         else if (!strcmp(arg, "--ignore-space-at-eol"))
2577                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2578         else if (!strcmp(arg, "--patience"))
2579                 options->xdl_opts |= XDF_PATIENCE_DIFF;
2580
2581         /* flags options */
2582         else if (!strcmp(arg, "--binary")) {
2583                 options->output_format |= DIFF_FORMAT_PATCH;
2584                 DIFF_OPT_SET(options, BINARY);
2585         }
2586         else if (!strcmp(arg, "--full-index"))
2587                 DIFF_OPT_SET(options, FULL_INDEX);
2588         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2589                 DIFF_OPT_SET(options, TEXT);
2590         else if (!strcmp(arg, "-R"))
2591                 DIFF_OPT_SET(options, REVERSE_DIFF);
2592         else if (!strcmp(arg, "--find-copies-harder"))
2593                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2594         else if (!strcmp(arg, "--follow"))
2595                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2596         else if (!strcmp(arg, "--color"))
2597                 DIFF_OPT_SET(options, COLOR_DIFF);
2598         else if (!strcmp(arg, "--no-color"))
2599                 DIFF_OPT_CLR(options, COLOR_DIFF);
2600         else if (!strcmp(arg, "--color-words"))
2601                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2602         else if (!prefixcmp(arg, "--color-words=")) {
2603                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2604                 options->word_regex = arg + 14;
2605         }
2606         else if (!strcmp(arg, "--exit-code"))
2607                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2608         else if (!strcmp(arg, "--quiet"))
2609                 DIFF_OPT_SET(options, QUIET);
2610         else if (!strcmp(arg, "--ext-diff"))
2611                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2612         else if (!strcmp(arg, "--no-ext-diff"))
2613                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2614         else if (!strcmp(arg, "--textconv"))
2615                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2616         else if (!strcmp(arg, "--no-textconv"))
2617                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2618         else if (!strcmp(arg, "--ignore-submodules"))
2619                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2620
2621         /* misc options */
2622         else if (!strcmp(arg, "-z"))
2623                 options->line_termination = 0;
2624         else if (!prefixcmp(arg, "-l"))
2625                 options->rename_limit = strtoul(arg+2, NULL, 10);
2626         else if (!prefixcmp(arg, "-S"))
2627                 options->pickaxe = arg + 2;
2628         else if (!strcmp(arg, "--pickaxe-all"))
2629                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2630         else if (!strcmp(arg, "--pickaxe-regex"))
2631                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2632         else if (!prefixcmp(arg, "-O"))
2633                 options->orderfile = arg + 2;
2634         else if (!prefixcmp(arg, "--diff-filter="))
2635                 options->filter = arg + 14;
2636         else if (!strcmp(arg, "--abbrev"))
2637                 options->abbrev = DEFAULT_ABBREV;
2638         else if (!prefixcmp(arg, "--abbrev=")) {
2639                 options->abbrev = strtoul(arg + 9, NULL, 10);
2640                 if (options->abbrev < MINIMUM_ABBREV)
2641                         options->abbrev = MINIMUM_ABBREV;
2642                 else if (40 < options->abbrev)
2643                         options->abbrev = 40;
2644         }
2645         else if (!prefixcmp(arg, "--src-prefix="))
2646                 options->a_prefix = arg + 13;
2647         else if (!prefixcmp(arg, "--dst-prefix="))
2648                 options->b_prefix = arg + 13;
2649         else if (!strcmp(arg, "--no-prefix"))
2650                 options->a_prefix = options->b_prefix = "";
2651         else if (opt_arg(arg, '\0', "inter-hunk-context",
2652                          &options->interhunkcontext))
2653                 ;
2654         else if (!prefixcmp(arg, "--output=")) {
2655                 options->file = fopen(arg + strlen("--output="), "w");
2656                 options->close_file = 1;
2657         } else
2658                 return 0;
2659         return 1;
2660 }
2661
2662 static int parse_num(const char **cp_p)
2663 {
2664         unsigned long num, scale;
2665         int ch, dot;
2666         const char *cp = *cp_p;
2667
2668         num = 0;
2669         scale = 1;
2670         dot = 0;
2671         for(;;) {
2672                 ch = *cp;
2673                 if ( !dot && ch == '.' ) {
2674                         scale = 1;
2675                         dot = 1;
2676                 } else if ( ch == '%' ) {
2677                         scale = dot ? scale*100 : 100;
2678                         cp++;   /* % is always at the end */
2679                         break;
2680                 } else if ( ch >= '0' && ch <= '9' ) {
2681                         if ( scale < 100000 ) {
2682                                 scale *= 10;
2683                                 num = (num*10) + (ch-'0');
2684                         }
2685                 } else {
2686                         break;
2687                 }
2688                 cp++;
2689         }
2690         *cp_p = cp;
2691
2692         /* user says num divided by scale and we say internally that
2693          * is MAX_SCORE * num / scale.
2694          */
2695         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2696 }
2697
2698 static int diff_scoreopt_parse(const char *opt)
2699 {
2700         int opt1, opt2, cmd;
2701
2702         if (*opt++ != '-')
2703                 return -1;
2704         cmd = *opt++;
2705         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2706                 return -1; /* that is not a -M, -C nor -B option */
2707
2708         opt1 = parse_num(&opt);
2709         if (cmd != 'B')
2710                 opt2 = 0;
2711         else {
2712                 if (*opt == 0)
2713                         opt2 = 0;
2714                 else if (*opt != '/')
2715                         return -1; /* we expect -B80/99 or -B80 */
2716                 else {
2717                         opt++;
2718                         opt2 = parse_num(&opt);
2719                 }
2720         }
2721         if (*opt != 0)
2722                 return -1;
2723         return opt1 | (opt2 << 16);
2724 }
2725
2726 struct diff_queue_struct diff_queued_diff;
2727
2728 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2729 {
2730         if (queue->alloc <= queue->nr) {
2731                 queue->alloc = alloc_nr(queue->alloc);
2732                 queue->queue = xrealloc(queue->queue,
2733                                         sizeof(dp) * queue->alloc);
2734         }
2735         queue->queue[queue->nr++] = dp;
2736 }
2737
2738 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2739                                  struct diff_filespec *one,
2740                                  struct diff_filespec *two)
2741 {
2742         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2743         dp->one = one;
2744         dp->two = two;
2745         if (queue)
2746                 diff_q(queue, dp);
2747         return dp;
2748 }
2749
2750 void diff_free_filepair(struct diff_filepair *p)
2751 {
2752         free_filespec(p->one);
2753         free_filespec(p->two);
2754         free(p);
2755 }
2756
2757 /* This is different from find_unique_abbrev() in that
2758  * it stuffs the result with dots for alignment.
2759  */
2760 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2761 {
2762         int abblen;
2763         const char *abbrev;
2764         if (len == 40)
2765                 return sha1_to_hex(sha1);
2766
2767         abbrev = find_unique_abbrev(sha1, len);
2768         abblen = strlen(abbrev);
2769         if (abblen < 37) {
2770                 static char hex[41];
2771                 if (len < abblen && abblen <= len + 2)
2772                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2773                 else
2774                         sprintf(hex, "%s...", abbrev);
2775                 return hex;
2776         }
2777         return sha1_to_hex(sha1);
2778 }
2779
2780 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2781 {
2782         int line_termination = opt->line_termination;
2783         int inter_name_termination = line_termination ? '\t' : '\0';
2784
2785         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2786                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2787                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2788                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2789         }
2790         if (p->score) {
2791                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2792                         inter_name_termination);
2793         } else {
2794                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2795         }
2796
2797         if (p->status == DIFF_STATUS_COPIED ||
2798             p->status == DIFF_STATUS_RENAMED) {
2799                 const char *name_a, *name_b;
2800                 name_a = p->one->path;
2801                 name_b = p->two->path;
2802                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2803                 write_name_quoted(name_a, opt->file, inter_name_termination);
2804                 write_name_quoted(name_b, opt->file, line_termination);
2805         } else {
2806                 const char *name_a, *name_b;
2807                 name_a = p->one->mode ? p->one->path : p->two->path;
2808                 name_b = NULL;
2809                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2810                 write_name_quoted(name_a, opt->file, line_termination);
2811         }
2812 }
2813
2814 int diff_unmodified_pair(struct diff_filepair *p)
2815 {
2816         /* This function is written stricter than necessary to support
2817          * the currently implemented transformers, but the idea is to
2818          * let transformers to produce diff_filepairs any way they want,
2819          * and filter and clean them up here before producing the output.
2820          */
2821         struct diff_filespec *one = p->one, *two = p->two;
2822
2823         if (DIFF_PAIR_UNMERGED(p))
2824                 return 0; /* unmerged is interesting */
2825
2826         /* deletion, addition, mode or type change
2827          * and rename are all interesting.
2828          */
2829         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2830             DIFF_PAIR_MODE_CHANGED(p) ||
2831             strcmp(one->path, two->path))
2832                 return 0;
2833
2834         /* both are valid and point at the same path.  that is, we are
2835          * dealing with a change.
2836          */
2837         if (one->sha1_valid && two->sha1_valid &&
2838             !hashcmp(one->sha1, two->sha1))
2839                 return 1; /* no change */
2840         if (!one->sha1_valid && !two->sha1_valid)
2841                 return 1; /* both look at the same file on the filesystem. */
2842         return 0;
2843 }
2844
2845 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2846 {
2847         if (diff_unmodified_pair(p))
2848                 return;
2849
2850         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2851             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2852                 return; /* no tree diffs in patch format */
2853
2854         run_diff(p, o);
2855 }
2856
2857 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2858                             struct diffstat_t *diffstat)
2859 {
2860         if (diff_unmodified_pair(p))
2861                 return;
2862
2863         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2864             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2865                 return; /* no tree diffs in patch format */
2866
2867         run_diffstat(p, o, diffstat);
2868 }
2869
2870 static void diff_flush_checkdiff(struct diff_filepair *p,
2871                 struct diff_options *o)
2872 {
2873         if (diff_unmodified_pair(p))
2874                 return;
2875
2876         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2877             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2878                 return; /* no tree diffs in patch format */
2879
2880         run_checkdiff(p, o);
2881 }
2882
2883 int diff_queue_is_empty(void)
2884 {
2885         struct diff_queue_struct *q = &diff_queued_diff;
2886         int i;
2887         for (i = 0; i < q->nr; i++)
2888                 if (!diff_unmodified_pair(q->queue[i]))
2889                         return 0;
2890         return 1;
2891 }
2892
2893 #if DIFF_DEBUG
2894 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2895 {
2896         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2897                 x, one ? one : "",
2898                 s->path,
2899                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2900                 s->mode,
2901                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2902         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2903                 x, one ? one : "",
2904                 s->size, s->xfrm_flags);
2905 }
2906
2907 void diff_debug_filepair(const struct diff_filepair *p, int i)
2908 {
2909         diff_debug_filespec(p->one, i, "one");
2910         diff_debug_filespec(p->two, i, "two");
2911         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2912                 p->score, p->status ? p->status : '?',
2913                 p->one->rename_used, p->broken_pair);
2914 }
2915
2916 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2917 {
2918         int i;
2919         if (msg)
2920                 fprintf(stderr, "%s\n", msg);
2921         fprintf(stderr, "q->nr = %d\n", q->nr);
2922         for (i = 0; i < q->nr; i++) {
2923                 struct diff_filepair *p = q->queue[i];
2924                 diff_debug_filepair(p, i);
2925         }
2926 }
2927 #endif
2928
2929 static void diff_resolve_rename_copy(void)
2930 {
2931         int i;
2932         struct diff_filepair *p;
2933         struct diff_queue_struct *q = &diff_queued_diff;
2934
2935         diff_debug_queue("resolve-rename-copy", q);
2936
2937         for (i = 0; i < q->nr; i++) {
2938                 p = q->queue[i];
2939                 p->status = 0; /* undecided */
2940                 if (DIFF_PAIR_UNMERGED(p))
2941                         p->status = DIFF_STATUS_UNMERGED;
2942                 else if (!DIFF_FILE_VALID(p->one))
2943                         p->status = DIFF_STATUS_ADDED;
2944                 else if (!DIFF_FILE_VALID(p->two))
2945                         p->status = DIFF_STATUS_DELETED;
2946                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2947                         p->status = DIFF_STATUS_TYPE_CHANGED;
2948
2949                 /* from this point on, we are dealing with a pair
2950                  * whose both sides are valid and of the same type, i.e.
2951                  * either in-place edit or rename/copy edit.
2952                  */
2953                 else if (DIFF_PAIR_RENAME(p)) {
2954                         /*
2955                          * A rename might have re-connected a broken
2956                          * pair up, causing the pathnames to be the
2957                          * same again. If so, that's not a rename at
2958                          * all, just a modification..
2959                          *
2960                          * Otherwise, see if this source was used for
2961                          * multiple renames, in which case we decrement
2962                          * the count, and call it a copy.
2963                          */
2964                         if (!strcmp(p->one->path, p->two->path))
2965                                 p->status = DIFF_STATUS_MODIFIED;
2966                         else if (--p->one->rename_used > 0)
2967                                 p->status = DIFF_STATUS_COPIED;
2968                         else
2969                                 p->status = DIFF_STATUS_RENAMED;
2970                 }
2971                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2972                          p->one->mode != p->two->mode ||
2973                          is_null_sha1(p->one->sha1))
2974                         p->status = DIFF_STATUS_MODIFIED;
2975                 else {
2976                         /* This is a "no-change" entry and should not
2977                          * happen anymore, but prepare for broken callers.
2978                          */
2979                         error("feeding unmodified %s to diffcore",
2980                               p->one->path);
2981                         p->status = DIFF_STATUS_UNKNOWN;
2982                 }
2983         }
2984         diff_debug_queue("resolve-rename-copy done", q);
2985 }
2986
2987 static int check_pair_status(struct diff_filepair *p)
2988 {
2989         switch (p->status) {
2990         case DIFF_STATUS_UNKNOWN:
2991                 return 0;
2992         case 0:
2993                 die("internal error in diff-resolve-rename-copy");
2994         default:
2995                 return 1;
2996         }
2997 }
2998
2999 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3000 {
3001         int fmt = opt->output_format;
3002
3003         if (fmt & DIFF_FORMAT_CHECKDIFF)
3004                 diff_flush_checkdiff(p, opt);
3005         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3006                 diff_flush_raw(p, opt);
3007         else if (fmt & DIFF_FORMAT_NAME) {
3008                 const char *name_a, *name_b;
3009                 name_a = p->two->path;
3010                 name_b = NULL;
3011                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3012                 write_name_quoted(name_a, opt->file, opt->line_termination);
3013         }
3014 }
3015
3016 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3017 {
3018         if (fs->mode)
3019                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3020         else
3021                 fprintf(file, " %s ", newdelete);
3022         write_name_quoted(fs->path, file, '\n');
3023 }
3024
3025
3026 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3027 {
3028         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3029                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3030                         show_name ? ' ' : '\n');
3031                 if (show_name) {
3032                         write_name_quoted(p->two->path, file, '\n');
3033                 }
3034         }
3035 }
3036
3037 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3038 {
3039         char *names = pprint_rename(p->one->path, p->two->path);
3040
3041         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3042         free(names);
3043         show_mode_change(file, p, 0);
3044 }
3045
3046 static void diff_summary(FILE *file, struct diff_filepair *p)
3047 {
3048         switch(p->status) {
3049         case DIFF_STATUS_DELETED:
3050                 show_file_mode_name(file, "delete", p->one);
3051                 break;
3052         case DIFF_STATUS_ADDED:
3053                 show_file_mode_name(file, "create", p->two);
3054                 break;
3055         case DIFF_STATUS_COPIED:
3056                 show_rename_copy(file, "copy", p);
3057                 break;
3058         case DIFF_STATUS_RENAMED:
3059                 show_rename_copy(file, "rename", p);
3060                 break;
3061         default:
3062                 if (p->score) {
3063                         fputs(" rewrite ", file);
3064                         write_name_quoted(p->two->path, file, ' ');
3065                         fprintf(file, "(%d%%)\n", similarity_index(p));
3066                 }
3067                 show_mode_change(file, p, !p->score);
3068                 break;
3069         }
3070 }
3071
3072 struct patch_id_t {
3073         git_SHA_CTX *ctx;
3074         int patchlen;
3075 };
3076
3077 static int remove_space(char *line, int len)
3078 {
3079         int i;
3080         char *dst = line;
3081         unsigned char c;
3082
3083         for (i = 0; i < len; i++)
3084                 if (!isspace((c = line[i])))
3085                         *dst++ = c;
3086
3087         return dst - line;
3088 }
3089
3090 static void patch_id_consume(void *priv, char *line, unsigned long len)
3091 {
3092         struct patch_id_t *data = priv;
3093         int new_len;
3094
3095         /* Ignore line numbers when computing the SHA1 of the patch */
3096         if (!prefixcmp(line, "@@ -"))
3097                 return;
3098
3099         new_len = remove_space(line, len);
3100
3101         git_SHA1_Update(data->ctx, line, new_len);
3102         data->patchlen += new_len;
3103 }
3104
3105 /* returns 0 upon success, and writes result into sha1 */
3106 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3107 {
3108         struct diff_queue_struct *q = &diff_queued_diff;
3109         int i;
3110         git_SHA_CTX ctx;
3111         struct patch_id_t data;
3112         char buffer[PATH_MAX * 4 + 20];
3113
3114         git_SHA1_Init(&ctx);
3115         memset(&data, 0, sizeof(struct patch_id_t));
3116         data.ctx = &ctx;
3117
3118         for (i = 0; i < q->nr; i++) {
3119                 xpparam_t xpp;
3120                 xdemitconf_t xecfg;
3121                 xdemitcb_t ecb;
3122                 mmfile_t mf1, mf2;
3123                 struct diff_filepair *p = q->queue[i];
3124                 int len1, len2;
3125
3126                 memset(&xpp, 0, sizeof(xpp));
3127                 memset(&xecfg, 0, sizeof(xecfg));
3128                 if (p->status == 0)
3129                         return error("internal diff status error");
3130                 if (p->status == DIFF_STATUS_UNKNOWN)
3131                         continue;
3132                 if (diff_unmodified_pair(p))
3133                         continue;
3134                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3135                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3136                         continue;
3137                 if (DIFF_PAIR_UNMERGED(p))
3138                         continue;
3139
3140                 diff_fill_sha1_info(p->one);
3141                 diff_fill_sha1_info(p->two);
3142                 if (fill_mmfile(&mf1, p->one) < 0 ||
3143                                 fill_mmfile(&mf2, p->two) < 0)
3144                         return error("unable to read files to diff");
3145
3146                 len1 = remove_space(p->one->path, strlen(p->one->path));
3147                 len2 = remove_space(p->two->path, strlen(p->two->path));
3148                 if (p->one->mode == 0)
3149                         len1 = snprintf(buffer, sizeof(buffer),
3150                                         "diff--gita/%.*sb/%.*s"
3151                                         "newfilemode%06o"
3152                                         "---/dev/null"
3153                                         "+++b/%.*s",
3154                                         len1, p->one->path,
3155                                         len2, p->two->path,
3156                                         p->two->mode,
3157                                         len2, p->two->path);
3158                 else if (p->two->mode == 0)
3159                         len1 = snprintf(buffer, sizeof(buffer),
3160                                         "diff--gita/%.*sb/%.*s"
3161                                         "deletedfilemode%06o"
3162                                         "---a/%.*s"
3163                                         "+++/dev/null",
3164                                         len1, p->one->path,
3165                                         len2, p->two->path,
3166                                         p->one->mode,
3167                                         len1, p->one->path);
3168                 else
3169                         len1 = snprintf(buffer, sizeof(buffer),
3170                                         "diff--gita/%.*sb/%.*s"
3171                                         "---a/%.*s"
3172                                         "+++b/%.*s",
3173                                         len1, p->one->path,
3174                                         len2, p->two->path,
3175                                         len1, p->one->path,
3176                                         len2, p->two->path);
3177                 git_SHA1_Update(&ctx, buffer, len1);
3178
3179                 xpp.flags = XDF_NEED_MINIMAL;
3180                 xecfg.ctxlen = 3;
3181                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3182                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3183                               &xpp, &xecfg, &ecb);
3184         }
3185
3186         git_SHA1_Final(sha1, &ctx);
3187         return 0;
3188 }
3189
3190 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3191 {
3192         struct diff_queue_struct *q = &diff_queued_diff;
3193         int i;
3194         int result = diff_get_patch_id(options, sha1);
3195
3196         for (i = 0; i < q->nr; i++)
3197                 diff_free_filepair(q->queue[i]);
3198
3199         free(q->queue);
3200         q->queue = NULL;
3201         q->nr = q->alloc = 0;
3202
3203         return result;
3204 }
3205
3206 static int is_summary_empty(const struct diff_queue_struct *q)
3207 {
3208         int i;
3209
3210         for (i = 0; i < q->nr; i++) {
3211                 const struct diff_filepair *p = q->queue[i];
3212
3213                 switch (p->status) {
3214                 case DIFF_STATUS_DELETED:
3215                 case DIFF_STATUS_ADDED:
3216                 case DIFF_STATUS_COPIED:
3217                 case DIFF_STATUS_RENAMED:
3218                         return 0;
3219                 default:
3220                         if (p->score)
3221                                 return 0;
3222                         if (p->one->mode && p->two->mode &&
3223                             p->one->mode != p->two->mode)
3224                                 return 0;
3225                         break;
3226                 }
3227         }
3228         return 1;
3229 }
3230
3231 void diff_flush(struct diff_options *options)
3232 {
3233         struct diff_queue_struct *q = &diff_queued_diff;
3234         int i, output_format = options->output_format;
3235         int separator = 0;
3236
3237         /*
3238          * Order: raw, stat, summary, patch
3239          * or:    name/name-status/checkdiff (other bits clear)
3240          */
3241         if (!q->nr)
3242                 goto free_queue;
3243
3244         if (output_format & (DIFF_FORMAT_RAW |
3245                              DIFF_FORMAT_NAME |
3246                              DIFF_FORMAT_NAME_STATUS |
3247                              DIFF_FORMAT_CHECKDIFF)) {
3248                 for (i = 0; i < q->nr; i++) {
3249                         struct diff_filepair *p = q->queue[i];
3250                         if (check_pair_status(p))
3251                                 flush_one_pair(p, options);
3252                 }
3253                 separator++;
3254         }
3255
3256         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3257                 struct diffstat_t diffstat;
3258
3259                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3260                 for (i = 0; i < q->nr; i++) {
3261                         struct diff_filepair *p = q->queue[i];
3262                         if (check_pair_status(p))
3263                                 diff_flush_stat(p, options, &diffstat);
3264                 }
3265                 if (output_format & DIFF_FORMAT_NUMSTAT)
3266                         show_numstat(&diffstat, options);
3267                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3268                         show_stats(&diffstat, options);
3269                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3270                         show_shortstats(&diffstat, options);
3271                 free_diffstat_info(&diffstat);
3272                 separator++;
3273         }
3274         if (output_format & DIFF_FORMAT_DIRSTAT)
3275                 show_dirstat(options);
3276
3277         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3278                 for (i = 0; i < q->nr; i++)
3279                         diff_summary(options->file, q->queue[i]);
3280                 separator++;
3281         }
3282
3283         if (output_format & DIFF_FORMAT_PATCH) {
3284                 if (separator) {
3285                         putc(options->line_termination, options->file);
3286                         if (options->stat_sep) {
3287                                 /* attach patch instead of inline */
3288                                 fputs(options->stat_sep, options->file);
3289                         }
3290                 }
3291
3292                 for (i = 0; i < q->nr; i++) {
3293                         struct diff_filepair *p = q->queue[i];
3294                         if (check_pair_status(p))
3295                                 diff_flush_patch(p, options);
3296                 }
3297         }
3298
3299         if (output_format & DIFF_FORMAT_CALLBACK)
3300                 options->format_callback(q, options, options->format_callback_data);
3301
3302         for (i = 0; i < q->nr; i++)
3303                 diff_free_filepair(q->queue[i]);
3304 free_queue:
3305         free(q->queue);
3306         q->queue = NULL;
3307         q->nr = q->alloc = 0;
3308         if (options->close_file)
3309                 fclose(options->file);
3310 }
3311
3312 static void diffcore_apply_filter(const char *filter)
3313 {
3314         int i;
3315         struct diff_queue_struct *q = &diff_queued_diff;
3316         struct diff_queue_struct outq;
3317         outq.queue = NULL;
3318         outq.nr = outq.alloc = 0;
3319
3320         if (!filter)
3321                 return;
3322
3323         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3324                 int found;
3325                 for (i = found = 0; !found && i < q->nr; i++) {
3326                         struct diff_filepair *p = q->queue[i];
3327                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3328                              ((p->score &&
3329                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3330                               (!p->score &&
3331                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3332                             ((p->status != DIFF_STATUS_MODIFIED) &&
3333                              strchr(filter, p->status)))
3334                                 found++;
3335                 }
3336                 if (found)
3337                         return;
3338
3339                 /* otherwise we will clear the whole queue
3340                  * by copying the empty outq at the end of this
3341                  * function, but first clear the current entries
3342                  * in the queue.
3343                  */
3344                 for (i = 0; i < q->nr; i++)
3345                         diff_free_filepair(q->queue[i]);
3346         }
3347         else {
3348                 /* Only the matching ones */
3349                 for (i = 0; i < q->nr; i++) {
3350                         struct diff_filepair *p = q->queue[i];
3351
3352                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3353                              ((p->score &&
3354                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3355                               (!p->score &&
3356                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3357                             ((p->status != DIFF_STATUS_MODIFIED) &&
3358                              strchr(filter, p->status)))
3359                                 diff_q(&outq, p);
3360                         else
3361                                 diff_free_filepair(p);
3362                 }
3363         }
3364         free(q->queue);
3365         *q = outq;
3366 }
3367
3368 /* Check whether two filespecs with the same mode and size are identical */
3369 static int diff_filespec_is_identical(struct diff_filespec *one,
3370                                       struct diff_filespec *two)
3371 {
3372         if (S_ISGITLINK(one->mode))
3373                 return 0;
3374         if (diff_populate_filespec(one, 0))
3375                 return 0;
3376         if (diff_populate_filespec(two, 0))
3377                 return 0;
3378         return !memcmp(one->data, two->data, one->size);
3379 }
3380
3381 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3382 {
3383         int i;
3384         struct diff_queue_struct *q = &diff_queued_diff;
3385         struct diff_queue_struct outq;
3386         outq.queue = NULL;
3387         outq.nr = outq.alloc = 0;
3388
3389         for (i = 0; i < q->nr; i++) {
3390                 struct diff_filepair *p = q->queue[i];
3391
3392                 /*
3393                  * 1. Entries that come from stat info dirtyness
3394                  *    always have both sides (iow, not create/delete),
3395                  *    one side of the object name is unknown, with
3396                  *    the same mode and size.  Keep the ones that
3397                  *    do not match these criteria.  They have real
3398                  *    differences.
3399                  *
3400                  * 2. At this point, the file is known to be modified,
3401                  *    with the same mode and size, and the object
3402                  *    name of one side is unknown.  Need to inspect
3403                  *    the identical contents.
3404                  */
3405                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3406                     !DIFF_FILE_VALID(p->two) ||
3407                     (p->one->sha1_valid && p->two->sha1_valid) ||
3408                     (p->one->mode != p->two->mode) ||
3409                     diff_populate_filespec(p->one, 1) ||
3410                     diff_populate_filespec(p->two, 1) ||
3411                     (p->one->size != p->two->size) ||
3412                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3413                         diff_q(&outq, p);
3414                 else {
3415                         /*
3416                          * The caller can subtract 1 from skip_stat_unmatch
3417                          * to determine how many paths were dirty only
3418                          * due to stat info mismatch.
3419                          */
3420                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3421                                 diffopt->skip_stat_unmatch++;
3422                         diff_free_filepair(p);
3423                 }
3424         }
3425         free(q->queue);
3426         *q = outq;
3427 }
3428
3429 void diffcore_std(struct diff_options *options)
3430 {
3431         if (options->skip_stat_unmatch)
3432                 diffcore_skip_stat_unmatch(options);
3433         if (options->break_opt != -1)
3434                 diffcore_break(options->break_opt);
3435         if (options->detect_rename)
3436                 diffcore_rename(options);
3437         if (options->break_opt != -1)
3438                 diffcore_merge_broken();
3439         if (options->pickaxe)
3440                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3441         if (options->orderfile)
3442                 diffcore_order(options->orderfile);
3443         diff_resolve_rename_copy();
3444         diffcore_apply_filter(options->filter);
3445
3446         if (diff_queued_diff.nr)
3447                 DIFF_OPT_SET(options, HAS_CHANGES);
3448         else
3449                 DIFF_OPT_CLR(options, HAS_CHANGES);
3450 }
3451
3452 int diff_result_code(struct diff_options *opt, int status)
3453 {
3454         int result = 0;
3455         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3456             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3457                 return status;
3458         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3459             DIFF_OPT_TST(opt, HAS_CHANGES))
3460                 result |= 01;
3461         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3462             DIFF_OPT_TST(opt, CHECK_FAILED))
3463                 result |= 02;
3464         return result;
3465 }
3466
3467 void diff_addremove(struct diff_options *options,
3468                     int addremove, unsigned mode,
3469                     const unsigned char *sha1,
3470                     const char *concatpath)
3471 {
3472         struct diff_filespec *one, *two;
3473
3474         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3475                 return;
3476
3477         /* This may look odd, but it is a preparation for
3478          * feeding "there are unchanged files which should
3479          * not produce diffs, but when you are doing copy
3480          * detection you would need them, so here they are"
3481          * entries to the diff-core.  They will be prefixed
3482          * with something like '=' or '*' (I haven't decided
3483          * which but should not make any difference).
3484          * Feeding the same new and old to diff_change()
3485          * also has the same effect.
3486          * Before the final output happens, they are pruned after
3487          * merged into rename/copy pairs as appropriate.
3488          */
3489         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3490                 addremove = (addremove == '+' ? '-' :
3491                              addremove == '-' ? '+' : addremove);
3492
3493         if (options->prefix &&
3494             strncmp(concatpath, options->prefix, options->prefix_length))
3495                 return;
3496
3497         one = alloc_filespec(concatpath);
3498         two = alloc_filespec(concatpath);
3499
3500         if (addremove != '+')
3501                 fill_filespec(one, sha1, mode);
3502         if (addremove != '-')
3503                 fill_filespec(two, sha1, mode);
3504
3505         diff_queue(&diff_queued_diff, one, two);
3506         DIFF_OPT_SET(options, HAS_CHANGES);
3507 }
3508
3509 void diff_change(struct diff_options *options,
3510                  unsigned old_mode, unsigned new_mode,
3511                  const unsigned char *old_sha1,
3512                  const unsigned char *new_sha1,
3513                  const char *concatpath)
3514 {
3515         struct diff_filespec *one, *two;
3516
3517         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3518                         && S_ISGITLINK(new_mode))
3519                 return;
3520
3521         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3522                 unsigned tmp;
3523                 const unsigned char *tmp_c;
3524                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3525                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3526         }
3527
3528         if (options->prefix &&
3529             strncmp(concatpath, options->prefix, options->prefix_length))
3530                 return;
3531
3532         one = alloc_filespec(concatpath);
3533         two = alloc_filespec(concatpath);
3534         fill_filespec(one, old_sha1, old_mode);
3535         fill_filespec(two, new_sha1, new_mode);
3536
3537         diff_queue(&diff_queued_diff, one, two);
3538         DIFF_OPT_SET(options, HAS_CHANGES);
3539 }
3540
3541 void diff_unmerge(struct diff_options *options,
3542                   const char *path,
3543                   unsigned mode, const unsigned char *sha1)
3544 {
3545         struct diff_filespec *one, *two;
3546
3547         if (options->prefix &&
3548             strncmp(path, options->prefix, options->prefix_length))
3549                 return;
3550
3551         one = alloc_filespec(path);
3552         two = alloc_filespec(path);
3553         fill_filespec(one, sha1, mode);
3554         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3555 }
3556
3557 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3558                 size_t *outsize)
3559 {
3560         struct diff_tempfile *temp;
3561         const char *argv[3];
3562         const char **arg = argv;
3563         struct child_process child;
3564         struct strbuf buf = STRBUF_INIT;
3565
3566         temp = prepare_temp_file(spec->path, spec);
3567         *arg++ = pgm;
3568         *arg++ = temp->name;
3569         *arg = NULL;
3570
3571         memset(&child, 0, sizeof(child));
3572         child.argv = argv;
3573         child.out = -1;
3574         if (start_command(&child) != 0 ||
3575             strbuf_read(&buf, child.out, 0) < 0 ||
3576             finish_command(&child) != 0) {
3577                 remove_tempfile();
3578                 error("error running textconv command '%s'", pgm);
3579                 return NULL;
3580         }
3581         remove_tempfile();
3582
3583         return strbuf_detach(&buf, outsize);
3584 }