Merge branch 'maint'
[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 #include "submodule.h"
17
18 #ifdef NO_FAST_WORKING_DIRECTORY
19 #define FAST_WORKING_DIRECTORY 0
20 #else
21 #define FAST_WORKING_DIRECTORY 1
22 #endif
23
24 static int diff_detect_rename_default;
25 static int diff_rename_limit_default = 200;
26 static int diff_suppress_blank_empty;
27 int diff_use_color_default = -1;
28 static const char *diff_word_regex_cfg;
29 static const char *external_diff_cmd_cfg;
30 int diff_auto_refresh_index = 1;
31 static int diff_mnemonic_prefix;
32
33 static char diff_colors[][COLOR_MAXLEN] = {
34         GIT_COLOR_RESET,
35         GIT_COLOR_NORMAL,       /* PLAIN */
36         GIT_COLOR_BOLD,         /* METAINFO */
37         GIT_COLOR_CYAN,         /* FRAGINFO */
38         GIT_COLOR_RED,          /* OLD */
39         GIT_COLOR_GREEN,        /* NEW */
40         GIT_COLOR_YELLOW,       /* COMMIT */
41         GIT_COLOR_BG_RED,       /* WHITESPACE */
42         GIT_COLOR_NORMAL,       /* FUNCINFO */
43 };
44
45 static void diff_filespec_load_driver(struct diff_filespec *one);
46 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
47
48 static int parse_diff_color_slot(const char *var, int ofs)
49 {
50         if (!strcasecmp(var+ofs, "plain"))
51                 return DIFF_PLAIN;
52         if (!strcasecmp(var+ofs, "meta"))
53                 return DIFF_METAINFO;
54         if (!strcasecmp(var+ofs, "frag"))
55                 return DIFF_FRAGINFO;
56         if (!strcasecmp(var+ofs, "old"))
57                 return DIFF_FILE_OLD;
58         if (!strcasecmp(var+ofs, "new"))
59                 return DIFF_FILE_NEW;
60         if (!strcasecmp(var+ofs, "commit"))
61                 return DIFF_COMMIT;
62         if (!strcasecmp(var+ofs, "whitespace"))
63                 return DIFF_WHITESPACE;
64         if (!strcasecmp(var+ofs, "func"))
65                 return DIFF_FUNCINFO;
66         return -1;
67 }
68
69 static int git_config_rename(const char *var, const char *value)
70 {
71         if (!value)
72                 return DIFF_DETECT_RENAME;
73         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
74                 return  DIFF_DETECT_COPY;
75         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
76 }
77
78 /*
79  * These are to give UI layer defaults.
80  * The core-level commands such as git-diff-files should
81  * never be affected by the setting of diff.renames
82  * the user happens to have in the configuration file.
83  */
84 int git_diff_ui_config(const char *var, const char *value, void *cb)
85 {
86         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
87                 diff_use_color_default = git_config_colorbool(var, value, -1);
88                 return 0;
89         }
90         if (!strcmp(var, "diff.renames")) {
91                 diff_detect_rename_default = git_config_rename(var, value);
92                 return 0;
93         }
94         if (!strcmp(var, "diff.autorefreshindex")) {
95                 diff_auto_refresh_index = git_config_bool(var, value);
96                 return 0;
97         }
98         if (!strcmp(var, "diff.mnemonicprefix")) {
99                 diff_mnemonic_prefix = git_config_bool(var, value);
100                 return 0;
101         }
102         if (!strcmp(var, "diff.external"))
103                 return git_config_string(&external_diff_cmd_cfg, var, value);
104         if (!strcmp(var, "diff.wordregex"))
105                 return git_config_string(&diff_word_regex_cfg, var, value);
106
107         return git_diff_basic_config(var, value, cb);
108 }
109
110 int git_diff_basic_config(const char *var, const char *value, void *cb)
111 {
112         if (!strcmp(var, "diff.renamelimit")) {
113                 diff_rename_limit_default = git_config_int(var, value);
114                 return 0;
115         }
116
117         switch (userdiff_config(var, value)) {
118                 case 0: break;
119                 case -1: return -1;
120                 default: return 0;
121         }
122
123         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
124                 int slot = parse_diff_color_slot(var, 11);
125                 if (slot < 0)
126                         return 0;
127                 if (!value)
128                         return config_error_nonbool(var);
129                 color_parse(value, var, diff_colors[slot]);
130                 return 0;
131         }
132
133         /* like GNU diff's --suppress-blank-empty option  */
134         if (!strcmp(var, "diff.suppressblankempty") ||
135                         /* for backwards compatibility */
136                         !strcmp(var, "diff.suppress-blank-empty")) {
137                 diff_suppress_blank_empty = git_config_bool(var, value);
138                 return 0;
139         }
140
141         return git_color_default_config(var, value, cb);
142 }
143
144 static char *quote_two(const char *one, const char *two)
145 {
146         int need_one = quote_c_style(one, NULL, NULL, 1);
147         int need_two = quote_c_style(two, NULL, NULL, 1);
148         struct strbuf res = STRBUF_INIT;
149
150         if (need_one + need_two) {
151                 strbuf_addch(&res, '"');
152                 quote_c_style(one, &res, NULL, 1);
153                 quote_c_style(two, &res, NULL, 1);
154                 strbuf_addch(&res, '"');
155         } else {
156                 strbuf_addstr(&res, one);
157                 strbuf_addstr(&res, two);
158         }
159         return strbuf_detach(&res, NULL);
160 }
161
162 static const char *external_diff(void)
163 {
164         static const char *external_diff_cmd = NULL;
165         static int done_preparing = 0;
166
167         if (done_preparing)
168                 return external_diff_cmd;
169         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
170         if (!external_diff_cmd)
171                 external_diff_cmd = external_diff_cmd_cfg;
172         done_preparing = 1;
173         return external_diff_cmd;
174 }
175
176 static struct diff_tempfile {
177         const char *name; /* filename external diff should read from */
178         char hex[41];
179         char mode[10];
180         char tmp_path[PATH_MAX];
181 } diff_temp[2];
182
183 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
184
185 struct emit_callback {
186         int color_diff;
187         unsigned ws_rule;
188         int blank_at_eof_in_preimage;
189         int blank_at_eof_in_postimage;
190         int lno_in_preimage;
191         int lno_in_postimage;
192         sane_truncate_fn truncate;
193         const char **label_path;
194         struct diff_words_data *diff_words;
195         int *found_changesp;
196         FILE *file;
197         struct strbuf *header;
198 };
199
200 static int count_lines(const char *data, int size)
201 {
202         int count, ch, completely_empty = 1, nl_just_seen = 0;
203         count = 0;
204         while (0 < size--) {
205                 ch = *data++;
206                 if (ch == '\n') {
207                         count++;
208                         nl_just_seen = 1;
209                         completely_empty = 0;
210                 }
211                 else {
212                         nl_just_seen = 0;
213                         completely_empty = 0;
214                 }
215         }
216         if (completely_empty)
217                 return 0;
218         if (!nl_just_seen)
219                 count++; /* no trailing newline */
220         return count;
221 }
222
223 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
224 {
225         if (!DIFF_FILE_VALID(one)) {
226                 mf->ptr = (char *)""; /* does not matter */
227                 mf->size = 0;
228                 return 0;
229         }
230         else if (diff_populate_filespec(one, 0))
231                 return -1;
232
233         mf->ptr = one->data;
234         mf->size = one->size;
235         return 0;
236 }
237
238 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
239 {
240         char *ptr = mf->ptr;
241         long size = mf->size;
242         int cnt = 0;
243
244         if (!size)
245                 return cnt;
246         ptr += size - 1; /* pointing at the very end */
247         if (*ptr != '\n')
248                 ; /* incomplete line */
249         else
250                 ptr--; /* skip the last LF */
251         while (mf->ptr < ptr) {
252                 char *prev_eol;
253                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
254                         if (*prev_eol == '\n')
255                                 break;
256                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
257                         break;
258                 cnt++;
259                 ptr = prev_eol - 1;
260         }
261         return cnt;
262 }
263
264 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
265                                struct emit_callback *ecbdata)
266 {
267         int l1, l2, at;
268         unsigned ws_rule = ecbdata->ws_rule;
269         l1 = count_trailing_blank(mf1, ws_rule);
270         l2 = count_trailing_blank(mf2, ws_rule);
271         if (l2 <= l1) {
272                 ecbdata->blank_at_eof_in_preimage = 0;
273                 ecbdata->blank_at_eof_in_postimage = 0;
274                 return;
275         }
276         at = count_lines(mf1->ptr, mf1->size);
277         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
278
279         at = count_lines(mf2->ptr, mf2->size);
280         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
281 }
282
283 static void emit_line_0(FILE *file, const char *set, const char *reset,
284                         int first, const char *line, int len)
285 {
286         int has_trailing_newline, has_trailing_carriage_return;
287         int nofirst;
288
289         if (len == 0) {
290                 has_trailing_newline = (first == '\n');
291                 has_trailing_carriage_return = (!has_trailing_newline &&
292                                                 (first == '\r'));
293                 nofirst = has_trailing_newline || has_trailing_carriage_return;
294         } else {
295                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
296                 if (has_trailing_newline)
297                         len--;
298                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
299                 if (has_trailing_carriage_return)
300                         len--;
301                 nofirst = 0;
302         }
303
304         if (len || !nofirst) {
305                 fputs(set, file);
306                 if (!nofirst)
307                         fputc(first, file);
308                 fwrite(line, len, 1, file);
309                 fputs(reset, file);
310         }
311         if (has_trailing_carriage_return)
312                 fputc('\r', file);
313         if (has_trailing_newline)
314                 fputc('\n', file);
315 }
316
317 static void emit_line(FILE *file, const char *set, const char *reset,
318                       const char *line, int len)
319 {
320         emit_line_0(file, set, reset, line[0], line+1, len-1);
321 }
322
323 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
324 {
325         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
326               ecbdata->blank_at_eof_in_preimage &&
327               ecbdata->blank_at_eof_in_postimage &&
328               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
329               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
330                 return 0;
331         return ws_blank_line(line, len, ecbdata->ws_rule);
332 }
333
334 static void emit_add_line(const char *reset,
335                           struct emit_callback *ecbdata,
336                           const char *line, int len)
337 {
338         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
339         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
340
341         if (!*ws)
342                 emit_line_0(ecbdata->file, set, reset, '+', line, len);
343         else if (new_blank_line_at_eof(ecbdata, line, len))
344                 /* Blank line at EOF - paint '+' as well */
345                 emit_line_0(ecbdata->file, ws, reset, '+', line, len);
346         else {
347                 /* Emit just the prefix, then the rest. */
348                 emit_line_0(ecbdata->file, set, reset, '+', "", 0);
349                 ws_check_emit(line, len, ecbdata->ws_rule,
350                               ecbdata->file, set, reset, ws);
351         }
352 }
353
354 static void emit_hunk_header(struct emit_callback *ecbdata,
355                              const char *line, int len)
356 {
357         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
358         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
359         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
360         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
361         static const char atat[2] = { '@', '@' };
362         const char *cp, *ep;
363
364         /*
365          * As a hunk header must begin with "@@ -<old>, +<new> @@",
366          * it always is at least 10 bytes long.
367          */
368         if (len < 10 ||
369             memcmp(line, atat, 2) ||
370             !(ep = memmem(line + 2, len - 2, atat, 2))) {
371                 emit_line(ecbdata->file, plain, reset, line, len);
372                 return;
373         }
374         ep += 2; /* skip over @@ */
375
376         /* The hunk header in fraginfo color */
377         emit_line(ecbdata->file, frag, reset, line, ep - line);
378
379         /* blank before the func header */
380         for (cp = ep; ep - line < len; ep++)
381                 if (*ep != ' ' && *ep != '\t')
382                         break;
383         if (ep != cp)
384                 emit_line(ecbdata->file, plain, reset, cp, ep - cp);
385
386         if (ep < line + len)
387                 emit_line(ecbdata->file, func, reset, ep, line + len - ep);
388 }
389
390 static struct diff_tempfile *claim_diff_tempfile(void) {
391         int i;
392         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
393                 if (!diff_temp[i].name)
394                         return diff_temp + i;
395         die("BUG: diff is failing to clean up its tempfiles");
396 }
397
398 static int remove_tempfile_installed;
399
400 static void remove_tempfile(void)
401 {
402         int i;
403         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
404                 if (diff_temp[i].name == diff_temp[i].tmp_path)
405                         unlink_or_warn(diff_temp[i].name);
406                 diff_temp[i].name = NULL;
407         }
408 }
409
410 static void remove_tempfile_on_signal(int signo)
411 {
412         remove_tempfile();
413         sigchain_pop(signo);
414         raise(signo);
415 }
416
417 static void print_line_count(FILE *file, int count)
418 {
419         switch (count) {
420         case 0:
421                 fprintf(file, "0,0");
422                 break;
423         case 1:
424                 fprintf(file, "1");
425                 break;
426         default:
427                 fprintf(file, "1,%d", count);
428                 break;
429         }
430 }
431
432 static void emit_rewrite_lines(struct emit_callback *ecb,
433                                int prefix, const char *data, int size)
434 {
435         const char *endp = NULL;
436         static const char *nneof = " No newline at end of file\n";
437         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
438         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
439
440         while (0 < size) {
441                 int len;
442
443                 endp = memchr(data, '\n', size);
444                 len = endp ? (endp - data + 1) : size;
445                 if (prefix != '+') {
446                         ecb->lno_in_preimage++;
447                         emit_line_0(ecb->file, old, reset, '-',
448                                     data, len);
449                 } else {
450                         ecb->lno_in_postimage++;
451                         emit_add_line(reset, ecb, data, len);
452                 }
453                 size -= len;
454                 data += len;
455         }
456         if (!endp) {
457                 const char *plain = diff_get_color(ecb->color_diff,
458                                                    DIFF_PLAIN);
459                 emit_line_0(ecb->file, plain, reset, '\\',
460                             nneof, strlen(nneof));
461         }
462 }
463
464 static void emit_rewrite_diff(const char *name_a,
465                               const char *name_b,
466                               struct diff_filespec *one,
467                               struct diff_filespec *two,
468                               const char *textconv_one,
469                               const char *textconv_two,
470                               struct diff_options *o)
471 {
472         int lc_a, lc_b;
473         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
474         const char *name_a_tab, *name_b_tab;
475         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
476         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
477         const char *reset = diff_get_color(color_diff, DIFF_RESET);
478         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
479         const char *a_prefix, *b_prefix;
480         const char *data_one, *data_two;
481         size_t size_one, size_two;
482         struct emit_callback ecbdata;
483
484         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
485                 a_prefix = o->b_prefix;
486                 b_prefix = o->a_prefix;
487         } else {
488                 a_prefix = o->a_prefix;
489                 b_prefix = o->b_prefix;
490         }
491
492         name_a += (*name_a == '/');
493         name_b += (*name_b == '/');
494         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
495         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
496
497         strbuf_reset(&a_name);
498         strbuf_reset(&b_name);
499         quote_two_c_style(&a_name, a_prefix, name_a, 0);
500         quote_two_c_style(&b_name, b_prefix, name_b, 0);
501
502         diff_populate_filespec(one, 0);
503         diff_populate_filespec(two, 0);
504         if (textconv_one) {
505                 data_one = run_textconv(textconv_one, one, &size_one);
506                 if (!data_one)
507                         die("unable to read files to diff");
508         }
509         else {
510                 data_one = one->data;
511                 size_one = one->size;
512         }
513         if (textconv_two) {
514                 data_two = run_textconv(textconv_two, two, &size_two);
515                 if (!data_two)
516                         die("unable to read files to diff");
517         }
518         else {
519                 data_two = two->data;
520                 size_two = two->size;
521         }
522
523         memset(&ecbdata, 0, sizeof(ecbdata));
524         ecbdata.color_diff = color_diff;
525         ecbdata.found_changesp = &o->found_changes;
526         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
527         ecbdata.file = o->file;
528         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
529                 mmfile_t mf1, mf2;
530                 mf1.ptr = (char *)data_one;
531                 mf2.ptr = (char *)data_two;
532                 mf1.size = size_one;
533                 mf2.size = size_two;
534                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
535         }
536         ecbdata.lno_in_preimage = 1;
537         ecbdata.lno_in_postimage = 1;
538
539         lc_a = count_lines(data_one, size_one);
540         lc_b = count_lines(data_two, size_two);
541         fprintf(o->file,
542                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
543                 metainfo, a_name.buf, name_a_tab, reset,
544                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
545         print_line_count(o->file, lc_a);
546         fprintf(o->file, " +");
547         print_line_count(o->file, lc_b);
548         fprintf(o->file, " @@%s\n", reset);
549         if (lc_a)
550                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
551         if (lc_b)
552                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
553 }
554
555 struct diff_words_buffer {
556         mmfile_t text;
557         long alloc;
558         struct diff_words_orig {
559                 const char *begin, *end;
560         } *orig;
561         int orig_nr, orig_alloc;
562 };
563
564 static void diff_words_append(char *line, unsigned long len,
565                 struct diff_words_buffer *buffer)
566 {
567         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
568         line++;
569         len--;
570         memcpy(buffer->text.ptr + buffer->text.size, line, len);
571         buffer->text.size += len;
572         buffer->text.ptr[buffer->text.size] = '\0';
573 }
574
575 struct diff_words_data {
576         struct diff_words_buffer minus, plus;
577         const char *current_plus;
578         FILE *file;
579         regex_t *word_regex;
580 };
581
582 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
583 {
584         struct diff_words_data *diff_words = priv;
585         int minus_first, minus_len, plus_first, plus_len;
586         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
587
588         if (line[0] != '@' || parse_hunk_header(line, len,
589                         &minus_first, &minus_len, &plus_first, &plus_len))
590                 return;
591
592         /* POSIX requires that first be decremented by one if len == 0... */
593         if (minus_len) {
594                 minus_begin = diff_words->minus.orig[minus_first].begin;
595                 minus_end =
596                         diff_words->minus.orig[minus_first + minus_len - 1].end;
597         } else
598                 minus_begin = minus_end =
599                         diff_words->minus.orig[minus_first].end;
600
601         if (plus_len) {
602                 plus_begin = diff_words->plus.orig[plus_first].begin;
603                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
604         } else
605                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
606
607         if (diff_words->current_plus != plus_begin)
608                 fwrite(diff_words->current_plus,
609                                 plus_begin - diff_words->current_plus, 1,
610                                 diff_words->file);
611         if (minus_begin != minus_end)
612                 color_fwrite_lines(diff_words->file,
613                                 diff_get_color(1, DIFF_FILE_OLD),
614                                 minus_end - minus_begin, minus_begin);
615         if (plus_begin != plus_end)
616                 color_fwrite_lines(diff_words->file,
617                                 diff_get_color(1, DIFF_FILE_NEW),
618                                 plus_end - plus_begin, plus_begin);
619
620         diff_words->current_plus = plus_end;
621 }
622
623 /* This function starts looking at *begin, and returns 0 iff a word was found. */
624 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
625                 int *begin, int *end)
626 {
627         if (word_regex && *begin < buffer->size) {
628                 regmatch_t match[1];
629                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
630                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
631                                         '\n', match[0].rm_eo - match[0].rm_so);
632                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
633                         *begin += match[0].rm_so;
634                         return *begin >= *end;
635                 }
636                 return -1;
637         }
638
639         /* find the next word */
640         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
641                 (*begin)++;
642         if (*begin >= buffer->size)
643                 return -1;
644
645         /* find the end of the word */
646         *end = *begin + 1;
647         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
648                 (*end)++;
649
650         return 0;
651 }
652
653 /*
654  * This function splits the words in buffer->text, stores the list with
655  * newline separator into out, and saves the offsets of the original words
656  * in buffer->orig.
657  */
658 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
659                 regex_t *word_regex)
660 {
661         int i, j;
662         long alloc = 0;
663
664         out->size = 0;
665         out->ptr = NULL;
666
667         /* fake an empty "0th" word */
668         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
669         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
670         buffer->orig_nr = 1;
671
672         for (i = 0; i < buffer->text.size; i++) {
673                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
674                         return;
675
676                 /* store original boundaries */
677                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
678                                 buffer->orig_alloc);
679                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
680                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
681                 buffer->orig_nr++;
682
683                 /* store one word */
684                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
685                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
686                 out->ptr[out->size + j - i] = '\n';
687                 out->size += j - i + 1;
688
689                 i = j - 1;
690         }
691 }
692
693 /* this executes the word diff on the accumulated buffers */
694 static void diff_words_show(struct diff_words_data *diff_words)
695 {
696         xpparam_t xpp;
697         xdemitconf_t xecfg;
698         xdemitcb_t ecb;
699         mmfile_t minus, plus;
700
701         /* special case: only removal */
702         if (!diff_words->plus.text.size) {
703                 color_fwrite_lines(diff_words->file,
704                         diff_get_color(1, DIFF_FILE_OLD),
705                         diff_words->minus.text.size, diff_words->minus.text.ptr);
706                 diff_words->minus.text.size = 0;
707                 return;
708         }
709
710         diff_words->current_plus = diff_words->plus.text.ptr;
711
712         memset(&xpp, 0, sizeof(xpp));
713         memset(&xecfg, 0, sizeof(xecfg));
714         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
715         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
716         xpp.flags = XDF_NEED_MINIMAL;
717         /* as only the hunk header will be parsed, we need a 0-context */
718         xecfg.ctxlen = 0;
719         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
720                       &xpp, &xecfg, &ecb);
721         free(minus.ptr);
722         free(plus.ptr);
723         if (diff_words->current_plus != diff_words->plus.text.ptr +
724                         diff_words->plus.text.size)
725                 fwrite(diff_words->current_plus,
726                         diff_words->plus.text.ptr + diff_words->plus.text.size
727                         - diff_words->current_plus, 1,
728                         diff_words->file);
729         diff_words->minus.text.size = diff_words->plus.text.size = 0;
730 }
731
732 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
733 static void diff_words_flush(struct emit_callback *ecbdata)
734 {
735         if (ecbdata->diff_words->minus.text.size ||
736             ecbdata->diff_words->plus.text.size)
737                 diff_words_show(ecbdata->diff_words);
738 }
739
740 static void free_diff_words_data(struct emit_callback *ecbdata)
741 {
742         if (ecbdata->diff_words) {
743                 diff_words_flush(ecbdata);
744                 free (ecbdata->diff_words->minus.text.ptr);
745                 free (ecbdata->diff_words->minus.orig);
746                 free (ecbdata->diff_words->plus.text.ptr);
747                 free (ecbdata->diff_words->plus.orig);
748                 free(ecbdata->diff_words->word_regex);
749                 free(ecbdata->diff_words);
750                 ecbdata->diff_words = NULL;
751         }
752 }
753
754 const char *diff_get_color(int diff_use_color, enum color_diff ix)
755 {
756         if (diff_use_color)
757                 return diff_colors[ix];
758         return "";
759 }
760
761 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
762 {
763         const char *cp;
764         unsigned long allot;
765         size_t l = len;
766
767         if (ecb->truncate)
768                 return ecb->truncate(line, len);
769         cp = line;
770         allot = l;
771         while (0 < l) {
772                 (void) utf8_width(&cp, &l);
773                 if (!cp)
774                         break; /* truncated in the middle? */
775         }
776         return allot - l;
777 }
778
779 static void find_lno(const char *line, struct emit_callback *ecbdata)
780 {
781         const char *p;
782         ecbdata->lno_in_preimage = 0;
783         ecbdata->lno_in_postimage = 0;
784         p = strchr(line, '-');
785         if (!p)
786                 return; /* cannot happen */
787         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
788         p = strchr(p, '+');
789         if (!p)
790                 return; /* cannot happen */
791         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
792 }
793
794 static void fn_out_consume(void *priv, char *line, unsigned long len)
795 {
796         struct emit_callback *ecbdata = priv;
797         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
798         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
799         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
800
801         if (ecbdata->header) {
802                 fprintf(ecbdata->file, "%s", ecbdata->header->buf);
803                 strbuf_reset(ecbdata->header);
804                 ecbdata->header = NULL;
805         }
806         *(ecbdata->found_changesp) = 1;
807
808         if (ecbdata->label_path[0]) {
809                 const char *name_a_tab, *name_b_tab;
810
811                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
812                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
813
814                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
815                         meta, ecbdata->label_path[0], reset, name_a_tab);
816                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
817                         meta, ecbdata->label_path[1], reset, name_b_tab);
818                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
819         }
820
821         if (diff_suppress_blank_empty
822             && len == 2 && line[0] == ' ' && line[1] == '\n') {
823                 line[0] = '\n';
824                 len = 1;
825         }
826
827         if (line[0] == '@') {
828                 if (ecbdata->diff_words)
829                         diff_words_flush(ecbdata);
830                 len = sane_truncate_line(ecbdata, line, len);
831                 find_lno(line, ecbdata);
832                 emit_hunk_header(ecbdata, line, len);
833                 if (line[len-1] != '\n')
834                         putc('\n', ecbdata->file);
835                 return;
836         }
837
838         if (len < 1) {
839                 emit_line(ecbdata->file, reset, reset, line, len);
840                 return;
841         }
842
843         if (ecbdata->diff_words) {
844                 if (line[0] == '-') {
845                         diff_words_append(line, len,
846                                           &ecbdata->diff_words->minus);
847                         return;
848                 } else if (line[0] == '+') {
849                         diff_words_append(line, len,
850                                           &ecbdata->diff_words->plus);
851                         return;
852                 }
853                 diff_words_flush(ecbdata);
854                 line++;
855                 len--;
856                 emit_line(ecbdata->file, plain, reset, line, len);
857                 return;
858         }
859
860         if (line[0] != '+') {
861                 const char *color =
862                         diff_get_color(ecbdata->color_diff,
863                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
864                 ecbdata->lno_in_preimage++;
865                 if (line[0] == ' ')
866                         ecbdata->lno_in_postimage++;
867                 emit_line(ecbdata->file, color, reset, line, len);
868         } else {
869                 ecbdata->lno_in_postimage++;
870                 emit_add_line(reset, ecbdata, line + 1, len - 1);
871         }
872 }
873
874 static char *pprint_rename(const char *a, const char *b)
875 {
876         const char *old = a;
877         const char *new = b;
878         struct strbuf name = STRBUF_INIT;
879         int pfx_length, sfx_length;
880         int len_a = strlen(a);
881         int len_b = strlen(b);
882         int a_midlen, b_midlen;
883         int qlen_a = quote_c_style(a, NULL, NULL, 0);
884         int qlen_b = quote_c_style(b, NULL, NULL, 0);
885
886         if (qlen_a || qlen_b) {
887                 quote_c_style(a, &name, NULL, 0);
888                 strbuf_addstr(&name, " => ");
889                 quote_c_style(b, &name, NULL, 0);
890                 return strbuf_detach(&name, NULL);
891         }
892
893         /* Find common prefix */
894         pfx_length = 0;
895         while (*old && *new && *old == *new) {
896                 if (*old == '/')
897                         pfx_length = old - a + 1;
898                 old++;
899                 new++;
900         }
901
902         /* Find common suffix */
903         old = a + len_a;
904         new = b + len_b;
905         sfx_length = 0;
906         while (a <= old && b <= new && *old == *new) {
907                 if (*old == '/')
908                         sfx_length = len_a - (old - a);
909                 old--;
910                 new--;
911         }
912
913         /*
914          * pfx{mid-a => mid-b}sfx
915          * {pfx-a => pfx-b}sfx
916          * pfx{sfx-a => sfx-b}
917          * name-a => name-b
918          */
919         a_midlen = len_a - pfx_length - sfx_length;
920         b_midlen = len_b - pfx_length - sfx_length;
921         if (a_midlen < 0)
922                 a_midlen = 0;
923         if (b_midlen < 0)
924                 b_midlen = 0;
925
926         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
927         if (pfx_length + sfx_length) {
928                 strbuf_add(&name, a, pfx_length);
929                 strbuf_addch(&name, '{');
930         }
931         strbuf_add(&name, a + pfx_length, a_midlen);
932         strbuf_addstr(&name, " => ");
933         strbuf_add(&name, b + pfx_length, b_midlen);
934         if (pfx_length + sfx_length) {
935                 strbuf_addch(&name, '}');
936                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
937         }
938         return strbuf_detach(&name, NULL);
939 }
940
941 struct diffstat_t {
942         int nr;
943         int alloc;
944         struct diffstat_file {
945                 char *from_name;
946                 char *name;
947                 char *print_name;
948                 unsigned is_unmerged:1;
949                 unsigned is_binary:1;
950                 unsigned is_renamed:1;
951                 unsigned int added, deleted;
952         } **files;
953 };
954
955 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
956                                           const char *name_a,
957                                           const char *name_b)
958 {
959         struct diffstat_file *x;
960         x = xcalloc(sizeof (*x), 1);
961         if (diffstat->nr == diffstat->alloc) {
962                 diffstat->alloc = alloc_nr(diffstat->alloc);
963                 diffstat->files = xrealloc(diffstat->files,
964                                 diffstat->alloc * sizeof(x));
965         }
966         diffstat->files[diffstat->nr++] = x;
967         if (name_b) {
968                 x->from_name = xstrdup(name_a);
969                 x->name = xstrdup(name_b);
970                 x->is_renamed = 1;
971         }
972         else {
973                 x->from_name = NULL;
974                 x->name = xstrdup(name_a);
975         }
976         return x;
977 }
978
979 static void diffstat_consume(void *priv, char *line, unsigned long len)
980 {
981         struct diffstat_t *diffstat = priv;
982         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
983
984         if (line[0] == '+')
985                 x->added++;
986         else if (line[0] == '-')
987                 x->deleted++;
988 }
989
990 const char mime_boundary_leader[] = "------------";
991
992 static int scale_linear(int it, int width, int max_change)
993 {
994         /*
995          * make sure that at least one '-' is printed if there were deletions,
996          * and likewise for '+'.
997          */
998         if (max_change < 2)
999                 return it;
1000         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1001 }
1002
1003 static void show_name(FILE *file,
1004                       const char *prefix, const char *name, int len)
1005 {
1006         fprintf(file, " %s%-*s |", prefix, len, name);
1007 }
1008
1009 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1010 {
1011         if (cnt <= 0)
1012                 return;
1013         fprintf(file, "%s", set);
1014         while (cnt--)
1015                 putc(ch, file);
1016         fprintf(file, "%s", reset);
1017 }
1018
1019 static void fill_print_name(struct diffstat_file *file)
1020 {
1021         char *pname;
1022
1023         if (file->print_name)
1024                 return;
1025
1026         if (!file->is_renamed) {
1027                 struct strbuf buf = STRBUF_INIT;
1028                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1029                         pname = strbuf_detach(&buf, NULL);
1030                 } else {
1031                         pname = file->name;
1032                         strbuf_release(&buf);
1033                 }
1034         } else {
1035                 pname = pprint_rename(file->from_name, file->name);
1036         }
1037         file->print_name = pname;
1038 }
1039
1040 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1041 {
1042         int i, len, add, del, adds = 0, dels = 0;
1043         int max_change = 0, max_len = 0;
1044         int total_files = data->nr;
1045         int width, name_width;
1046         const char *reset, *set, *add_c, *del_c;
1047
1048         if (data->nr == 0)
1049                 return;
1050
1051         width = options->stat_width ? options->stat_width : 80;
1052         name_width = options->stat_name_width ? options->stat_name_width : 50;
1053
1054         /* Sanity: give at least 5 columns to the graph,
1055          * but leave at least 10 columns for the name.
1056          */
1057         if (width < 25)
1058                 width = 25;
1059         if (name_width < 10)
1060                 name_width = 10;
1061         else if (width < name_width + 15)
1062                 name_width = width - 15;
1063
1064         /* Find the longest filename and max number of changes */
1065         reset = diff_get_color_opt(options, DIFF_RESET);
1066         set   = diff_get_color_opt(options, DIFF_PLAIN);
1067         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1068         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1069
1070         for (i = 0; i < data->nr; i++) {
1071                 struct diffstat_file *file = data->files[i];
1072                 int change = file->added + file->deleted;
1073                 fill_print_name(file);
1074                 len = strlen(file->print_name);
1075                 if (max_len < len)
1076                         max_len = len;
1077
1078                 if (file->is_binary || file->is_unmerged)
1079                         continue;
1080                 if (max_change < change)
1081                         max_change = change;
1082         }
1083
1084         /* Compute the width of the graph part;
1085          * 10 is for one blank at the beginning of the line plus
1086          * " | count " between the name and the graph.
1087          *
1088          * From here on, name_width is the width of the name area,
1089          * and width is the width of the graph area.
1090          */
1091         name_width = (name_width < max_len) ? name_width : max_len;
1092         if (width < (name_width + 10) + max_change)
1093                 width = width - (name_width + 10);
1094         else
1095                 width = max_change;
1096
1097         for (i = 0; i < data->nr; i++) {
1098                 const char *prefix = "";
1099                 char *name = data->files[i]->print_name;
1100                 int added = data->files[i]->added;
1101                 int deleted = data->files[i]->deleted;
1102                 int name_len;
1103
1104                 /*
1105                  * "scale" the filename
1106                  */
1107                 len = name_width;
1108                 name_len = strlen(name);
1109                 if (name_width < name_len) {
1110                         char *slash;
1111                         prefix = "...";
1112                         len -= 3;
1113                         name += name_len - len;
1114                         slash = strchr(name, '/');
1115                         if (slash)
1116                                 name = slash;
1117                 }
1118
1119                 if (data->files[i]->is_binary) {
1120                         show_name(options->file, prefix, name, len);
1121                         fprintf(options->file, "  Bin ");
1122                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
1123                         fprintf(options->file, " -> ");
1124                         fprintf(options->file, "%s%d%s", add_c, added, reset);
1125                         fprintf(options->file, " bytes");
1126                         fprintf(options->file, "\n");
1127                         continue;
1128                 }
1129                 else if (data->files[i]->is_unmerged) {
1130                         show_name(options->file, prefix, name, len);
1131                         fprintf(options->file, "  Unmerged\n");
1132                         continue;
1133                 }
1134                 else if (!data->files[i]->is_renamed &&
1135                          (added + deleted == 0)) {
1136                         total_files--;
1137                         continue;
1138                 }
1139
1140                 /*
1141                  * scale the add/delete
1142                  */
1143                 add = added;
1144                 del = deleted;
1145                 adds += add;
1146                 dels += del;
1147
1148                 if (width <= max_change) {
1149                         add = scale_linear(add, width, max_change);
1150                         del = scale_linear(del, width, max_change);
1151                 }
1152                 show_name(options->file, prefix, name, len);
1153                 fprintf(options->file, "%5d%s", added + deleted,
1154                                 added + deleted ? " " : "");
1155                 show_graph(options->file, '+', add, add_c, reset);
1156                 show_graph(options->file, '-', del, del_c, reset);
1157                 fprintf(options->file, "\n");
1158         }
1159         fprintf(options->file,
1160                " %d files changed, %d insertions(+), %d deletions(-)\n",
1161                total_files, adds, dels);
1162 }
1163
1164 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1165 {
1166         int i, adds = 0, dels = 0, total_files = data->nr;
1167
1168         if (data->nr == 0)
1169                 return;
1170
1171         for (i = 0; i < data->nr; i++) {
1172                 if (!data->files[i]->is_binary &&
1173                     !data->files[i]->is_unmerged) {
1174                         int added = data->files[i]->added;
1175                         int deleted= data->files[i]->deleted;
1176                         if (!data->files[i]->is_renamed &&
1177                             (added + deleted == 0)) {
1178                                 total_files--;
1179                         } else {
1180                                 adds += added;
1181                                 dels += deleted;
1182                         }
1183                 }
1184         }
1185         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1186                total_files, adds, dels);
1187 }
1188
1189 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1190 {
1191         int i;
1192
1193         if (data->nr == 0)
1194                 return;
1195
1196         for (i = 0; i < data->nr; i++) {
1197                 struct diffstat_file *file = data->files[i];
1198
1199                 if (file->is_binary)
1200                         fprintf(options->file, "-\t-\t");
1201                 else
1202                         fprintf(options->file,
1203                                 "%d\t%d\t", file->added, file->deleted);
1204                 if (options->line_termination) {
1205                         fill_print_name(file);
1206                         if (!file->is_renamed)
1207                                 write_name_quoted(file->name, options->file,
1208                                                   options->line_termination);
1209                         else {
1210                                 fputs(file->print_name, options->file);
1211                                 putc(options->line_termination, options->file);
1212                         }
1213                 } else {
1214                         if (file->is_renamed) {
1215                                 putc('\0', options->file);
1216                                 write_name_quoted(file->from_name, options->file, '\0');
1217                         }
1218                         write_name_quoted(file->name, options->file, '\0');
1219                 }
1220         }
1221 }
1222
1223 struct dirstat_file {
1224         const char *name;
1225         unsigned long changed;
1226 };
1227
1228 struct dirstat_dir {
1229         struct dirstat_file *files;
1230         int alloc, nr, percent, cumulative;
1231 };
1232
1233 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1234 {
1235         unsigned long this_dir = 0;
1236         unsigned int sources = 0;
1237
1238         while (dir->nr) {
1239                 struct dirstat_file *f = dir->files;
1240                 int namelen = strlen(f->name);
1241                 unsigned long this;
1242                 char *slash;
1243
1244                 if (namelen < baselen)
1245                         break;
1246                 if (memcmp(f->name, base, baselen))
1247                         break;
1248                 slash = strchr(f->name + baselen, '/');
1249                 if (slash) {
1250                         int newbaselen = slash + 1 - f->name;
1251                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1252                         sources++;
1253                 } else {
1254                         this = f->changed;
1255                         dir->files++;
1256                         dir->nr--;
1257                         sources += 2;
1258                 }
1259                 this_dir += this;
1260         }
1261
1262         /*
1263          * We don't report dirstat's for
1264          *  - the top level
1265          *  - or cases where everything came from a single directory
1266          *    under this directory (sources == 1).
1267          */
1268         if (baselen && sources != 1) {
1269                 int permille = this_dir * 1000 / changed;
1270                 if (permille) {
1271                         int percent = permille / 10;
1272                         if (percent >= dir->percent) {
1273                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1274                                 if (!dir->cumulative)
1275                                         return 0;
1276                         }
1277                 }
1278         }
1279         return this_dir;
1280 }
1281
1282 static int dirstat_compare(const void *_a, const void *_b)
1283 {
1284         const struct dirstat_file *a = _a;
1285         const struct dirstat_file *b = _b;
1286         return strcmp(a->name, b->name);
1287 }
1288
1289 static void show_dirstat(struct diff_options *options)
1290 {
1291         int i;
1292         unsigned long changed;
1293         struct dirstat_dir dir;
1294         struct diff_queue_struct *q = &diff_queued_diff;
1295
1296         dir.files = NULL;
1297         dir.alloc = 0;
1298         dir.nr = 0;
1299         dir.percent = options->dirstat_percent;
1300         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1301
1302         changed = 0;
1303         for (i = 0; i < q->nr; i++) {
1304                 struct diff_filepair *p = q->queue[i];
1305                 const char *name;
1306                 unsigned long copied, added, damage;
1307
1308                 name = p->one->path ? p->one->path : p->two->path;
1309
1310                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1311                         diff_populate_filespec(p->one, 0);
1312                         diff_populate_filespec(p->two, 0);
1313                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1314                                                &copied, &added);
1315                         diff_free_filespec_data(p->one);
1316                         diff_free_filespec_data(p->two);
1317                 } else if (DIFF_FILE_VALID(p->one)) {
1318                         diff_populate_filespec(p->one, 1);
1319                         copied = added = 0;
1320                         diff_free_filespec_data(p->one);
1321                 } else if (DIFF_FILE_VALID(p->two)) {
1322                         diff_populate_filespec(p->two, 1);
1323                         copied = 0;
1324                         added = p->two->size;
1325                         diff_free_filespec_data(p->two);
1326                 } else
1327                         continue;
1328
1329                 /*
1330                  * Original minus copied is the removed material,
1331                  * added is the new material.  They are both damages
1332                  * made to the preimage. In --dirstat-by-file mode, count
1333                  * damaged files, not damaged lines. This is done by
1334                  * counting only a single damaged line per file.
1335                  */
1336                 damage = (p->one->size - copied) + added;
1337                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1338                         damage = 1;
1339
1340                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1341                 dir.files[dir.nr].name = name;
1342                 dir.files[dir.nr].changed = damage;
1343                 changed += damage;
1344                 dir.nr++;
1345         }
1346
1347         /* This can happen even with many files, if everything was renames */
1348         if (!changed)
1349                 return;
1350
1351         /* Show all directories with more than x% of the changes */
1352         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1353         gather_dirstat(options->file, &dir, changed, "", 0);
1354 }
1355
1356 static void free_diffstat_info(struct diffstat_t *diffstat)
1357 {
1358         int i;
1359         for (i = 0; i < diffstat->nr; i++) {
1360                 struct diffstat_file *f = diffstat->files[i];
1361                 if (f->name != f->print_name)
1362                         free(f->print_name);
1363                 free(f->name);
1364                 free(f->from_name);
1365                 free(f);
1366         }
1367         free(diffstat->files);
1368 }
1369
1370 struct checkdiff_t {
1371         const char *filename;
1372         int lineno;
1373         struct diff_options *o;
1374         unsigned ws_rule;
1375         unsigned status;
1376 };
1377
1378 static int is_conflict_marker(const char *line, unsigned long len)
1379 {
1380         char firstchar;
1381         int cnt;
1382
1383         if (len < 8)
1384                 return 0;
1385         firstchar = line[0];
1386         switch (firstchar) {
1387         case '=': case '>': case '<':
1388                 break;
1389         default:
1390                 return 0;
1391         }
1392         for (cnt = 1; cnt < 7; cnt++)
1393                 if (line[cnt] != firstchar)
1394                         return 0;
1395         /* line[0] thru line[6] are same as firstchar */
1396         if (firstchar == '=') {
1397                 /* divider between ours and theirs? */
1398                 if (len != 8 || line[7] != '\n')
1399                         return 0;
1400         } else if (len < 8 || !isspace(line[7])) {
1401                 /* not divider before ours nor after theirs */
1402                 return 0;
1403         }
1404         return 1;
1405 }
1406
1407 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1408 {
1409         struct checkdiff_t *data = priv;
1410         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1411         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1412         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1413         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1414         char *err;
1415
1416         if (line[0] == '+') {
1417                 unsigned bad;
1418                 data->lineno++;
1419                 if (is_conflict_marker(line + 1, len - 1)) {
1420                         data->status |= 1;
1421                         fprintf(data->o->file,
1422                                 "%s:%d: leftover conflict marker\n",
1423                                 data->filename, data->lineno);
1424                 }
1425                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1426                 if (!bad)
1427                         return;
1428                 data->status |= bad;
1429                 err = whitespace_error_string(bad);
1430                 fprintf(data->o->file, "%s:%d: %s.\n",
1431                         data->filename, data->lineno, err);
1432                 free(err);
1433                 emit_line(data->o->file, set, reset, line, 1);
1434                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1435                               data->o->file, set, reset, ws);
1436         } else if (line[0] == ' ') {
1437                 data->lineno++;
1438         } else if (line[0] == '@') {
1439                 char *plus = strchr(line, '+');
1440                 if (plus)
1441                         data->lineno = strtol(plus, NULL, 10) - 1;
1442                 else
1443                         die("invalid diff");
1444         }
1445 }
1446
1447 static unsigned char *deflate_it(char *data,
1448                                  unsigned long size,
1449                                  unsigned long *result_size)
1450 {
1451         int bound;
1452         unsigned char *deflated;
1453         z_stream stream;
1454
1455         memset(&stream, 0, sizeof(stream));
1456         deflateInit(&stream, zlib_compression_level);
1457         bound = deflateBound(&stream, size);
1458         deflated = xmalloc(bound);
1459         stream.next_out = deflated;
1460         stream.avail_out = bound;
1461
1462         stream.next_in = (unsigned char *)data;
1463         stream.avail_in = size;
1464         while (deflate(&stream, Z_FINISH) == Z_OK)
1465                 ; /* nothing */
1466         deflateEnd(&stream);
1467         *result_size = stream.total_out;
1468         return deflated;
1469 }
1470
1471 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1472 {
1473         void *cp;
1474         void *delta;
1475         void *deflated;
1476         void *data;
1477         unsigned long orig_size;
1478         unsigned long delta_size;
1479         unsigned long deflate_size;
1480         unsigned long data_size;
1481
1482         /* We could do deflated delta, or we could do just deflated two,
1483          * whichever is smaller.
1484          */
1485         delta = NULL;
1486         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1487         if (one->size && two->size) {
1488                 delta = diff_delta(one->ptr, one->size,
1489                                    two->ptr, two->size,
1490                                    &delta_size, deflate_size);
1491                 if (delta) {
1492                         void *to_free = delta;
1493                         orig_size = delta_size;
1494                         delta = deflate_it(delta, delta_size, &delta_size);
1495                         free(to_free);
1496                 }
1497         }
1498
1499         if (delta && delta_size < deflate_size) {
1500                 fprintf(file, "delta %lu\n", orig_size);
1501                 free(deflated);
1502                 data = delta;
1503                 data_size = delta_size;
1504         }
1505         else {
1506                 fprintf(file, "literal %lu\n", two->size);
1507                 free(delta);
1508                 data = deflated;
1509                 data_size = deflate_size;
1510         }
1511
1512         /* emit data encoded in base85 */
1513         cp = data;
1514         while (data_size) {
1515                 int bytes = (52 < data_size) ? 52 : data_size;
1516                 char line[70];
1517                 data_size -= bytes;
1518                 if (bytes <= 26)
1519                         line[0] = bytes + 'A' - 1;
1520                 else
1521                         line[0] = bytes - 26 + 'a' - 1;
1522                 encode_85(line + 1, cp, bytes);
1523                 cp = (char *) cp + bytes;
1524                 fputs(line, file);
1525                 fputc('\n', file);
1526         }
1527         fprintf(file, "\n");
1528         free(data);
1529 }
1530
1531 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1532 {
1533         fprintf(file, "GIT binary patch\n");
1534         emit_binary_diff_body(file, one, two);
1535         emit_binary_diff_body(file, two, one);
1536 }
1537
1538 static void diff_filespec_load_driver(struct diff_filespec *one)
1539 {
1540         if (!one->driver)
1541                 one->driver = userdiff_find_by_path(one->path);
1542         if (!one->driver)
1543                 one->driver = userdiff_find_by_name("default");
1544 }
1545
1546 int diff_filespec_is_binary(struct diff_filespec *one)
1547 {
1548         if (one->is_binary == -1) {
1549                 diff_filespec_load_driver(one);
1550                 if (one->driver->binary != -1)
1551                         one->is_binary = one->driver->binary;
1552                 else {
1553                         if (!one->data && DIFF_FILE_VALID(one))
1554                                 diff_populate_filespec(one, 0);
1555                         if (one->data)
1556                                 one->is_binary = buffer_is_binary(one->data,
1557                                                 one->size);
1558                         if (one->is_binary == -1)
1559                                 one->is_binary = 0;
1560                 }
1561         }
1562         return one->is_binary;
1563 }
1564
1565 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1566 {
1567         diff_filespec_load_driver(one);
1568         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1569 }
1570
1571 static const char *userdiff_word_regex(struct diff_filespec *one)
1572 {
1573         diff_filespec_load_driver(one);
1574         return one->driver->word_regex;
1575 }
1576
1577 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1578 {
1579         if (!options->a_prefix)
1580                 options->a_prefix = a;
1581         if (!options->b_prefix)
1582                 options->b_prefix = b;
1583 }
1584
1585 static const char *get_textconv(struct diff_filespec *one)
1586 {
1587         if (!DIFF_FILE_VALID(one))
1588                 return NULL;
1589         if (!S_ISREG(one->mode))
1590                 return NULL;
1591         diff_filespec_load_driver(one);
1592         return one->driver->textconv;
1593 }
1594
1595 static void builtin_diff(const char *name_a,
1596                          const char *name_b,
1597                          struct diff_filespec *one,
1598                          struct diff_filespec *two,
1599                          const char *xfrm_msg,
1600                          struct diff_options *o,
1601                          int complete_rewrite)
1602 {
1603         mmfile_t mf1, mf2;
1604         const char *lbl[2];
1605         char *a_one, *b_two;
1606         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1607         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1608         const char *a_prefix, *b_prefix;
1609         const char *textconv_one = NULL, *textconv_two = NULL;
1610         struct strbuf header = STRBUF_INIT;
1611
1612         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1613                         (!one->mode || S_ISGITLINK(one->mode)) &&
1614                         (!two->mode || S_ISGITLINK(two->mode))) {
1615                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1616                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1617                 show_submodule_summary(o->file, one ? one->path : two->path,
1618                                 one->sha1, two->sha1, two->dirty_submodule,
1619                                 del, add, reset);
1620                 return;
1621         }
1622
1623         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1624                 textconv_one = get_textconv(one);
1625                 textconv_two = get_textconv(two);
1626         }
1627
1628         diff_set_mnemonic_prefix(o, "a/", "b/");
1629         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1630                 a_prefix = o->b_prefix;
1631                 b_prefix = o->a_prefix;
1632         } else {
1633                 a_prefix = o->a_prefix;
1634                 b_prefix = o->b_prefix;
1635         }
1636
1637         /* Never use a non-valid filename anywhere if at all possible */
1638         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1639         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1640
1641         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1642         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1643         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1644         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1645         strbuf_addf(&header, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1646         if (lbl[0][0] == '/') {
1647                 /* /dev/null */
1648                 strbuf_addf(&header, "%snew file mode %06o%s\n", set, two->mode, reset);
1649                 if (xfrm_msg && xfrm_msg[0])
1650                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1651         }
1652         else if (lbl[1][0] == '/') {
1653                 strbuf_addf(&header, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1654                 if (xfrm_msg && xfrm_msg[0])
1655                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1656         }
1657         else {
1658                 if (one->mode != two->mode) {
1659                         strbuf_addf(&header, "%sold mode %06o%s\n", set, one->mode, reset);
1660                         strbuf_addf(&header, "%snew mode %06o%s\n", set, two->mode, reset);
1661                 }
1662                 if (xfrm_msg && xfrm_msg[0])
1663                         strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1664
1665                 /*
1666                  * we do not run diff between different kind
1667                  * of objects.
1668                  */
1669                 if ((one->mode ^ two->mode) & S_IFMT)
1670                         goto free_ab_and_return;
1671                 if (complete_rewrite &&
1672                     (textconv_one || !diff_filespec_is_binary(one)) &&
1673                     (textconv_two || !diff_filespec_is_binary(two))) {
1674                         fprintf(o->file, "%s", header.buf);
1675                         strbuf_reset(&header);
1676                         emit_rewrite_diff(name_a, name_b, one, two,
1677                                                 textconv_one, textconv_two, o);
1678                         o->found_changes = 1;
1679                         goto free_ab_and_return;
1680                 }
1681         }
1682
1683         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1684                 die("unable to read files to diff");
1685
1686         if (!DIFF_OPT_TST(o, TEXT) &&
1687             ( (diff_filespec_is_binary(one) && !textconv_one) ||
1688               (diff_filespec_is_binary(two) && !textconv_two) )) {
1689                 /* Quite common confusing case */
1690                 if (mf1.size == mf2.size &&
1691                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1692                         goto free_ab_and_return;
1693                 fprintf(o->file, "%s", header.buf);
1694                 strbuf_reset(&header);
1695                 if (DIFF_OPT_TST(o, BINARY))
1696                         emit_binary_diff(o->file, &mf1, &mf2);
1697                 else
1698                         fprintf(o->file, "Binary files %s and %s differ\n",
1699                                 lbl[0], lbl[1]);
1700                 o->found_changes = 1;
1701         }
1702         else {
1703                 /* Crazy xdl interfaces.. */
1704                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1705                 xpparam_t xpp;
1706                 xdemitconf_t xecfg;
1707                 xdemitcb_t ecb;
1708                 struct emit_callback ecbdata;
1709                 const struct userdiff_funcname *pe;
1710
1711                 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS)) {
1712                         fprintf(o->file, "%s", header.buf);
1713                         strbuf_reset(&header);
1714                 }
1715
1716                 if (textconv_one) {
1717                         size_t size;
1718                         mf1.ptr = run_textconv(textconv_one, one, &size);
1719                         if (!mf1.ptr)
1720                                 die("unable to read files to diff");
1721                         mf1.size = size;
1722                 }
1723                 if (textconv_two) {
1724                         size_t size;
1725                         mf2.ptr = run_textconv(textconv_two, two, &size);
1726                         if (!mf2.ptr)
1727                                 die("unable to read files to diff");
1728                         mf2.size = size;
1729                 }
1730
1731                 pe = diff_funcname_pattern(one);
1732                 if (!pe)
1733                         pe = diff_funcname_pattern(two);
1734
1735                 memset(&xpp, 0, sizeof(xpp));
1736                 memset(&xecfg, 0, sizeof(xecfg));
1737                 memset(&ecbdata, 0, sizeof(ecbdata));
1738                 ecbdata.label_path = lbl;
1739                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1740                 ecbdata.found_changesp = &o->found_changes;
1741                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1742                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1743                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1744                 ecbdata.file = o->file;
1745                 ecbdata.header = header.len ? &header : NULL;
1746                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1747                 xecfg.ctxlen = o->context;
1748                 xecfg.interhunkctxlen = o->interhunkcontext;
1749                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1750                 if (pe)
1751                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1752                 if (!diffopts)
1753                         ;
1754                 else if (!prefixcmp(diffopts, "--unified="))
1755                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1756                 else if (!prefixcmp(diffopts, "-u"))
1757                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1758                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1759                         ecbdata.diff_words =
1760                                 xcalloc(1, sizeof(struct diff_words_data));
1761                         ecbdata.diff_words->file = o->file;
1762                         if (!o->word_regex)
1763                                 o->word_regex = userdiff_word_regex(one);
1764                         if (!o->word_regex)
1765                                 o->word_regex = userdiff_word_regex(two);
1766                         if (!o->word_regex)
1767                                 o->word_regex = diff_word_regex_cfg;
1768                         if (o->word_regex) {
1769                                 ecbdata.diff_words->word_regex = (regex_t *)
1770                                         xmalloc(sizeof(regex_t));
1771                                 if (regcomp(ecbdata.diff_words->word_regex,
1772                                                 o->word_regex,
1773                                                 REG_EXTENDED | REG_NEWLINE))
1774                                         die ("Invalid regular expression: %s",
1775                                                         o->word_regex);
1776                         }
1777                 }
1778                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1779                               &xpp, &xecfg, &ecb);
1780                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1781                         free_diff_words_data(&ecbdata);
1782                 if (textconv_one)
1783                         free(mf1.ptr);
1784                 if (textconv_two)
1785                         free(mf2.ptr);
1786                 xdiff_clear_find_func(&xecfg);
1787         }
1788
1789  free_ab_and_return:
1790         strbuf_release(&header);
1791         diff_free_filespec_data(one);
1792         diff_free_filespec_data(two);
1793         free(a_one);
1794         free(b_two);
1795         return;
1796 }
1797
1798 static void builtin_diffstat(const char *name_a, const char *name_b,
1799                              struct diff_filespec *one,
1800                              struct diff_filespec *two,
1801                              struct diffstat_t *diffstat,
1802                              struct diff_options *o,
1803                              int complete_rewrite)
1804 {
1805         mmfile_t mf1, mf2;
1806         struct diffstat_file *data;
1807
1808         data = diffstat_add(diffstat, name_a, name_b);
1809
1810         if (!one || !two) {
1811                 data->is_unmerged = 1;
1812                 return;
1813         }
1814         if (complete_rewrite) {
1815                 diff_populate_filespec(one, 0);
1816                 diff_populate_filespec(two, 0);
1817                 data->deleted = count_lines(one->data, one->size);
1818                 data->added = count_lines(two->data, two->size);
1819                 goto free_and_return;
1820         }
1821         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1822                 die("unable to read files to diff");
1823
1824         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1825                 data->is_binary = 1;
1826                 data->added = mf2.size;
1827                 data->deleted = mf1.size;
1828         } else {
1829                 /* Crazy xdl interfaces.. */
1830                 xpparam_t xpp;
1831                 xdemitconf_t xecfg;
1832                 xdemitcb_t ecb;
1833
1834                 memset(&xpp, 0, sizeof(xpp));
1835                 memset(&xecfg, 0, sizeof(xecfg));
1836                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1837                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1838                               &xpp, &xecfg, &ecb);
1839         }
1840
1841  free_and_return:
1842         diff_free_filespec_data(one);
1843         diff_free_filespec_data(two);
1844 }
1845
1846 static void builtin_checkdiff(const char *name_a, const char *name_b,
1847                               const char *attr_path,
1848                               struct diff_filespec *one,
1849                               struct diff_filespec *two,
1850                               struct diff_options *o)
1851 {
1852         mmfile_t mf1, mf2;
1853         struct checkdiff_t data;
1854
1855         if (!two)
1856                 return;
1857
1858         memset(&data, 0, sizeof(data));
1859         data.filename = name_b ? name_b : name_a;
1860         data.lineno = 0;
1861         data.o = o;
1862         data.ws_rule = whitespace_rule(attr_path);
1863
1864         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1865                 die("unable to read files to diff");
1866
1867         /*
1868          * All the other codepaths check both sides, but not checking
1869          * the "old" side here is deliberate.  We are checking the newly
1870          * introduced changes, and as long as the "new" side is text, we
1871          * can and should check what it introduces.
1872          */
1873         if (diff_filespec_is_binary(two))
1874                 goto free_and_return;
1875         else {
1876                 /* Crazy xdl interfaces.. */
1877                 xpparam_t xpp;
1878                 xdemitconf_t xecfg;
1879                 xdemitcb_t ecb;
1880
1881                 memset(&xpp, 0, sizeof(xpp));
1882                 memset(&xecfg, 0, sizeof(xecfg));
1883                 xecfg.ctxlen = 1; /* at least one context line */
1884                 xpp.flags = XDF_NEED_MINIMAL;
1885                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1886                               &xpp, &xecfg, &ecb);
1887
1888                 if (data.ws_rule & WS_BLANK_AT_EOF) {
1889                         struct emit_callback ecbdata;
1890                         int blank_at_eof;
1891
1892                         ecbdata.ws_rule = data.ws_rule;
1893                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
1894                         blank_at_eof = ecbdata.blank_at_eof_in_preimage;
1895
1896                         if (blank_at_eof) {
1897                                 static char *err;
1898                                 if (!err)
1899                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
1900                                 fprintf(o->file, "%s:%d: %s.\n",
1901                                         data.filename, blank_at_eof, err);
1902                                 data.status = 1; /* report errors */
1903                         }
1904                 }
1905         }
1906  free_and_return:
1907         diff_free_filespec_data(one);
1908         diff_free_filespec_data(two);
1909         if (data.status)
1910                 DIFF_OPT_SET(o, CHECK_FAILED);
1911 }
1912
1913 struct diff_filespec *alloc_filespec(const char *path)
1914 {
1915         int namelen = strlen(path);
1916         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1917
1918         memset(spec, 0, sizeof(*spec));
1919         spec->path = (char *)(spec + 1);
1920         memcpy(spec->path, path, namelen+1);
1921         spec->count = 1;
1922         spec->is_binary = -1;
1923         return spec;
1924 }
1925
1926 void free_filespec(struct diff_filespec *spec)
1927 {
1928         if (!--spec->count) {
1929                 diff_free_filespec_data(spec);
1930                 free(spec);
1931         }
1932 }
1933
1934 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1935                    unsigned short mode)
1936 {
1937         if (mode) {
1938                 spec->mode = canon_mode(mode);
1939                 hashcpy(spec->sha1, sha1);
1940                 spec->sha1_valid = !is_null_sha1(sha1);
1941         }
1942 }
1943
1944 /*
1945  * Given a name and sha1 pair, if the index tells us the file in
1946  * the work tree has that object contents, return true, so that
1947  * prepare_temp_file() does not have to inflate and extract.
1948  */
1949 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1950 {
1951         struct cache_entry *ce;
1952         struct stat st;
1953         int pos, len;
1954
1955         /*
1956          * We do not read the cache ourselves here, because the
1957          * benchmark with my previous version that always reads cache
1958          * shows that it makes things worse for diff-tree comparing
1959          * two linux-2.6 kernel trees in an already checked out work
1960          * tree.  This is because most diff-tree comparisons deal with
1961          * only a small number of files, while reading the cache is
1962          * expensive for a large project, and its cost outweighs the
1963          * savings we get by not inflating the object to a temporary
1964          * file.  Practically, this code only helps when we are used
1965          * by diff-cache --cached, which does read the cache before
1966          * calling us.
1967          */
1968         if (!active_cache)
1969                 return 0;
1970
1971         /* We want to avoid the working directory if our caller
1972          * doesn't need the data in a normal file, this system
1973          * is rather slow with its stat/open/mmap/close syscalls,
1974          * and the object is contained in a pack file.  The pack
1975          * is probably already open and will be faster to obtain
1976          * the data through than the working directory.  Loose
1977          * objects however would tend to be slower as they need
1978          * to be individually opened and inflated.
1979          */
1980         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1981                 return 0;
1982
1983         len = strlen(name);
1984         pos = cache_name_pos(name, len);
1985         if (pos < 0)
1986                 return 0;
1987         ce = active_cache[pos];
1988
1989         /*
1990          * This is not the sha1 we are looking for, or
1991          * unreusable because it is not a regular file.
1992          */
1993         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1994                 return 0;
1995
1996         /*
1997          * If ce is marked as "assume unchanged", there is no
1998          * guarantee that work tree matches what we are looking for.
1999          */
2000         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2001                 return 0;
2002
2003         /*
2004          * If ce matches the file in the work tree, we can reuse it.
2005          */
2006         if (ce_uptodate(ce) ||
2007             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2008                 return 1;
2009
2010         return 0;
2011 }
2012
2013 static int populate_from_stdin(struct diff_filespec *s)
2014 {
2015         struct strbuf buf = STRBUF_INIT;
2016         size_t size = 0;
2017
2018         if (strbuf_read(&buf, 0, 0) < 0)
2019                 return error("error while reading from stdin %s",
2020                                      strerror(errno));
2021
2022         s->should_munmap = 0;
2023         s->data = strbuf_detach(&buf, &size);
2024         s->size = size;
2025         s->should_free = 1;
2026         return 0;
2027 }
2028
2029 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2030 {
2031         int len;
2032         char *data = xmalloc(100), *dirty = "";
2033
2034         /* Are we looking at the work tree? */
2035         if (!s->sha1_valid && s->dirty_submodule)
2036                 dirty = "-dirty";
2037
2038         len = snprintf(data, 100,
2039                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2040         s->data = data;
2041         s->size = len;
2042         s->should_free = 1;
2043         if (size_only) {
2044                 s->data = NULL;
2045                 free(data);
2046         }
2047         return 0;
2048 }
2049
2050 /*
2051  * While doing rename detection and pickaxe operation, we may need to
2052  * grab the data for the blob (or file) for our own in-core comparison.
2053  * diff_filespec has data and size fields for this purpose.
2054  */
2055 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2056 {
2057         int err = 0;
2058         if (!DIFF_FILE_VALID(s))
2059                 die("internal error: asking to populate invalid file.");
2060         if (S_ISDIR(s->mode))
2061                 return -1;
2062
2063         if (s->data)
2064                 return 0;
2065
2066         if (size_only && 0 < s->size)
2067                 return 0;
2068
2069         if (S_ISGITLINK(s->mode))
2070                 return diff_populate_gitlink(s, size_only);
2071
2072         if (!s->sha1_valid ||
2073             reuse_worktree_file(s->path, s->sha1, 0)) {
2074                 struct strbuf buf = STRBUF_INIT;
2075                 struct stat st;
2076                 int fd;
2077
2078                 if (!strcmp(s->path, "-"))
2079                         return populate_from_stdin(s);
2080
2081                 if (lstat(s->path, &st) < 0) {
2082                         if (errno == ENOENT) {
2083                         err_empty:
2084                                 err = -1;
2085                         empty:
2086                                 s->data = (char *)"";
2087                                 s->size = 0;
2088                                 return err;
2089                         }
2090                 }
2091                 s->size = xsize_t(st.st_size);
2092                 if (!s->size)
2093                         goto empty;
2094                 if (S_ISLNK(st.st_mode)) {
2095                         struct strbuf sb = STRBUF_INIT;
2096
2097                         if (strbuf_readlink(&sb, s->path, s->size))
2098                                 goto err_empty;
2099                         s->size = sb.len;
2100                         s->data = strbuf_detach(&sb, NULL);
2101                         s->should_free = 1;
2102                         return 0;
2103                 }
2104                 if (size_only)
2105                         return 0;
2106                 fd = open(s->path, O_RDONLY);
2107                 if (fd < 0)
2108                         goto err_empty;
2109                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2110                 close(fd);
2111                 s->should_munmap = 1;
2112
2113                 /*
2114                  * Convert from working tree format to canonical git format
2115                  */
2116                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2117                         size_t size = 0;
2118                         munmap(s->data, s->size);
2119                         s->should_munmap = 0;
2120                         s->data = strbuf_detach(&buf, &size);
2121                         s->size = size;
2122                         s->should_free = 1;
2123                 }
2124         }
2125         else {
2126                 enum object_type type;
2127                 if (size_only)
2128                         type = sha1_object_info(s->sha1, &s->size);
2129                 else {
2130                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2131                         s->should_free = 1;
2132                 }
2133         }
2134         return 0;
2135 }
2136
2137 void diff_free_filespec_blob(struct diff_filespec *s)
2138 {
2139         if (s->should_free)
2140                 free(s->data);
2141         else if (s->should_munmap)
2142                 munmap(s->data, s->size);
2143
2144         if (s->should_free || s->should_munmap) {
2145                 s->should_free = s->should_munmap = 0;
2146                 s->data = NULL;
2147         }
2148 }
2149
2150 void diff_free_filespec_data(struct diff_filespec *s)
2151 {
2152         diff_free_filespec_blob(s);
2153         free(s->cnt_data);
2154         s->cnt_data = NULL;
2155 }
2156
2157 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2158                            void *blob,
2159                            unsigned long size,
2160                            const unsigned char *sha1,
2161                            int mode)
2162 {
2163         int fd;
2164         struct strbuf buf = STRBUF_INIT;
2165         struct strbuf template = STRBUF_INIT;
2166         char *path_dup = xstrdup(path);
2167         const char *base = basename(path_dup);
2168
2169         /* Generate "XXXXXX_basename.ext" */
2170         strbuf_addstr(&template, "XXXXXX_");
2171         strbuf_addstr(&template, base);
2172
2173         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2174                         strlen(base) + 1);
2175         if (fd < 0)
2176                 die_errno("unable to create temp-file");
2177         if (convert_to_working_tree(path,
2178                         (const char *)blob, (size_t)size, &buf)) {
2179                 blob = buf.buf;
2180                 size = buf.len;
2181         }
2182         if (write_in_full(fd, blob, size) != size)
2183                 die_errno("unable to write temp-file");
2184         close(fd);
2185         temp->name = temp->tmp_path;
2186         strcpy(temp->hex, sha1_to_hex(sha1));
2187         temp->hex[40] = 0;
2188         sprintf(temp->mode, "%06o", mode);
2189         strbuf_release(&buf);
2190         strbuf_release(&template);
2191         free(path_dup);
2192 }
2193
2194 static struct diff_tempfile *prepare_temp_file(const char *name,
2195                 struct diff_filespec *one)
2196 {
2197         struct diff_tempfile *temp = claim_diff_tempfile();
2198
2199         if (!DIFF_FILE_VALID(one)) {
2200         not_a_valid_file:
2201                 /* A '-' entry produces this for file-2, and
2202                  * a '+' entry produces this for file-1.
2203                  */
2204                 temp->name = "/dev/null";
2205                 strcpy(temp->hex, ".");
2206                 strcpy(temp->mode, ".");
2207                 return temp;
2208         }
2209
2210         if (!remove_tempfile_installed) {
2211                 atexit(remove_tempfile);
2212                 sigchain_push_common(remove_tempfile_on_signal);
2213                 remove_tempfile_installed = 1;
2214         }
2215
2216         if (!one->sha1_valid ||
2217             reuse_worktree_file(name, one->sha1, 1)) {
2218                 struct stat st;
2219                 if (lstat(name, &st) < 0) {
2220                         if (errno == ENOENT)
2221                                 goto not_a_valid_file;
2222                         die_errno("stat(%s)", name);
2223                 }
2224                 if (S_ISLNK(st.st_mode)) {
2225                         struct strbuf sb = STRBUF_INIT;
2226                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2227                                 die_errno("readlink(%s)", name);
2228                         prep_temp_blob(name, temp, sb.buf, sb.len,
2229                                        (one->sha1_valid ?
2230                                         one->sha1 : null_sha1),
2231                                        (one->sha1_valid ?
2232                                         one->mode : S_IFLNK));
2233                         strbuf_release(&sb);
2234                 }
2235                 else {
2236                         /* we can borrow from the file in the work tree */
2237                         temp->name = name;
2238                         if (!one->sha1_valid)
2239                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2240                         else
2241                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2242                         /* Even though we may sometimes borrow the
2243                          * contents from the work tree, we always want
2244                          * one->mode.  mode is trustworthy even when
2245                          * !(one->sha1_valid), as long as
2246                          * DIFF_FILE_VALID(one).
2247                          */
2248                         sprintf(temp->mode, "%06o", one->mode);
2249                 }
2250                 return temp;
2251         }
2252         else {
2253                 if (diff_populate_filespec(one, 0))
2254                         die("cannot read data blob for %s", one->path);
2255                 prep_temp_blob(name, temp, one->data, one->size,
2256                                one->sha1, one->mode);
2257         }
2258         return temp;
2259 }
2260
2261 /* An external diff command takes:
2262  *
2263  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2264  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2265  *
2266  */
2267 static void run_external_diff(const char *pgm,
2268                               const char *name,
2269                               const char *other,
2270                               struct diff_filespec *one,
2271                               struct diff_filespec *two,
2272                               const char *xfrm_msg,
2273                               int complete_rewrite)
2274 {
2275         const char *spawn_arg[10];
2276         int retval;
2277         const char **arg = &spawn_arg[0];
2278
2279         if (one && two) {
2280                 struct diff_tempfile *temp_one, *temp_two;
2281                 const char *othername = (other ? other : name);
2282                 temp_one = prepare_temp_file(name, one);
2283                 temp_two = prepare_temp_file(othername, two);
2284                 *arg++ = pgm;
2285                 *arg++ = name;
2286                 *arg++ = temp_one->name;
2287                 *arg++ = temp_one->hex;
2288                 *arg++ = temp_one->mode;
2289                 *arg++ = temp_two->name;
2290                 *arg++ = temp_two->hex;
2291                 *arg++ = temp_two->mode;
2292                 if (other) {
2293                         *arg++ = other;
2294                         *arg++ = xfrm_msg;
2295                 }
2296         } else {
2297                 *arg++ = pgm;
2298                 *arg++ = name;
2299         }
2300         *arg = NULL;
2301         fflush(NULL);
2302         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2303         remove_tempfile();
2304         if (retval) {
2305                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2306                 exit(1);
2307         }
2308 }
2309
2310 static int similarity_index(struct diff_filepair *p)
2311 {
2312         return p->score * 100 / MAX_SCORE;
2313 }
2314
2315 static void fill_metainfo(struct strbuf *msg,
2316                           const char *name,
2317                           const char *other,
2318                           struct diff_filespec *one,
2319                           struct diff_filespec *two,
2320                           struct diff_options *o,
2321                           struct diff_filepair *p)
2322 {
2323         strbuf_init(msg, PATH_MAX * 2 + 300);
2324         switch (p->status) {
2325         case DIFF_STATUS_COPIED:
2326                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2327                 strbuf_addstr(msg, "\ncopy from ");
2328                 quote_c_style(name, msg, NULL, 0);
2329                 strbuf_addstr(msg, "\ncopy to ");
2330                 quote_c_style(other, msg, NULL, 0);
2331                 strbuf_addch(msg, '\n');
2332                 break;
2333         case DIFF_STATUS_RENAMED:
2334                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2335                 strbuf_addstr(msg, "\nrename from ");
2336                 quote_c_style(name, msg, NULL, 0);
2337                 strbuf_addstr(msg, "\nrename to ");
2338                 quote_c_style(other, msg, NULL, 0);
2339                 strbuf_addch(msg, '\n');
2340                 break;
2341         case DIFF_STATUS_MODIFIED:
2342                 if (p->score) {
2343                         strbuf_addf(msg, "dissimilarity index %d%%\n",
2344                                     similarity_index(p));
2345                         break;
2346                 }
2347                 /* fallthru */
2348         default:
2349                 /* nothing */
2350                 ;
2351         }
2352         if (one && two && hashcmp(one->sha1, two->sha1)) {
2353                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2354
2355                 if (DIFF_OPT_TST(o, BINARY)) {
2356                         mmfile_t mf;
2357                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2358                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2359                                 abbrev = 40;
2360                 }
2361                 strbuf_addf(msg, "index %.*s..%.*s",
2362                             abbrev, sha1_to_hex(one->sha1),
2363                             abbrev, sha1_to_hex(two->sha1));
2364                 if (one->mode == two->mode)
2365                         strbuf_addf(msg, " %06o", one->mode);
2366                 strbuf_addch(msg, '\n');
2367         }
2368         if (msg->len)
2369                 strbuf_setlen(msg, msg->len - 1);
2370 }
2371
2372 static void run_diff_cmd(const char *pgm,
2373                          const char *name,
2374                          const char *other,
2375                          const char *attr_path,
2376                          struct diff_filespec *one,
2377                          struct diff_filespec *two,
2378                          struct strbuf *msg,
2379                          struct diff_options *o,
2380                          struct diff_filepair *p)
2381 {
2382         const char *xfrm_msg = NULL;
2383         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2384
2385         if (msg) {
2386                 fill_metainfo(msg, name, other, one, two, o, p);
2387                 xfrm_msg = msg->len ? msg->buf : NULL;
2388         }
2389
2390         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2391                 pgm = NULL;
2392         else {
2393                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2394                 if (drv && drv->external)
2395                         pgm = drv->external;
2396         }
2397
2398         if (pgm) {
2399                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2400                                   complete_rewrite);
2401                 return;
2402         }
2403         if (one && two)
2404                 builtin_diff(name, other ? other : name,
2405                              one, two, xfrm_msg, o, complete_rewrite);
2406         else
2407                 fprintf(o->file, "* Unmerged path %s\n", name);
2408 }
2409
2410 static void diff_fill_sha1_info(struct diff_filespec *one)
2411 {
2412         if (DIFF_FILE_VALID(one)) {
2413                 if (!one->sha1_valid) {
2414                         struct stat st;
2415                         if (!strcmp(one->path, "-")) {
2416                                 hashcpy(one->sha1, null_sha1);
2417                                 return;
2418                         }
2419                         if (lstat(one->path, &st) < 0)
2420                                 die_errno("stat '%s'", one->path);
2421                         if (index_path(one->sha1, one->path, &st, 0))
2422                                 die("cannot hash %s", one->path);
2423                 }
2424         }
2425         else
2426                 hashclr(one->sha1);
2427 }
2428
2429 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2430 {
2431         /* Strip the prefix but do not molest /dev/null and absolute paths */
2432         if (*namep && **namep != '/')
2433                 *namep += prefix_length;
2434         if (*otherp && **otherp != '/')
2435                 *otherp += prefix_length;
2436 }
2437
2438 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2439 {
2440         const char *pgm = external_diff();
2441         struct strbuf msg;
2442         struct diff_filespec *one = p->one;
2443         struct diff_filespec *two = p->two;
2444         const char *name;
2445         const char *other;
2446         const char *attr_path;
2447
2448         name  = p->one->path;
2449         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2450         attr_path = name;
2451         if (o->prefix_length)
2452                 strip_prefix(o->prefix_length, &name, &other);
2453
2454         if (DIFF_PAIR_UNMERGED(p)) {
2455                 run_diff_cmd(pgm, name, NULL, attr_path,
2456                              NULL, NULL, NULL, o, p);
2457                 return;
2458         }
2459
2460         diff_fill_sha1_info(one);
2461         diff_fill_sha1_info(two);
2462
2463         if (!pgm &&
2464             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2465             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2466                 /*
2467                  * a filepair that changes between file and symlink
2468                  * needs to be split into deletion and creation.
2469                  */
2470                 struct diff_filespec *null = alloc_filespec(two->path);
2471                 run_diff_cmd(NULL, name, other, attr_path,
2472                              one, null, &msg, o, p);
2473                 free(null);
2474                 strbuf_release(&msg);
2475
2476                 null = alloc_filespec(one->path);
2477                 run_diff_cmd(NULL, name, other, attr_path,
2478                              null, two, &msg, o, p);
2479                 free(null);
2480         }
2481         else
2482                 run_diff_cmd(pgm, name, other, attr_path,
2483                              one, two, &msg, o, p);
2484
2485         strbuf_release(&msg);
2486 }
2487
2488 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2489                          struct diffstat_t *diffstat)
2490 {
2491         const char *name;
2492         const char *other;
2493         int complete_rewrite = 0;
2494
2495         if (DIFF_PAIR_UNMERGED(p)) {
2496                 /* unmerged */
2497                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2498                 return;
2499         }
2500
2501         name = p->one->path;
2502         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2503
2504         if (o->prefix_length)
2505                 strip_prefix(o->prefix_length, &name, &other);
2506
2507         diff_fill_sha1_info(p->one);
2508         diff_fill_sha1_info(p->two);
2509
2510         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2511                 complete_rewrite = 1;
2512         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2513 }
2514
2515 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2516 {
2517         const char *name;
2518         const char *other;
2519         const char *attr_path;
2520
2521         if (DIFF_PAIR_UNMERGED(p)) {
2522                 /* unmerged */
2523                 return;
2524         }
2525
2526         name = p->one->path;
2527         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2528         attr_path = other ? other : name;
2529
2530         if (o->prefix_length)
2531                 strip_prefix(o->prefix_length, &name, &other);
2532
2533         diff_fill_sha1_info(p->one);
2534         diff_fill_sha1_info(p->two);
2535
2536         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2537 }
2538
2539 void diff_setup(struct diff_options *options)
2540 {
2541         memset(options, 0, sizeof(*options));
2542
2543         options->file = stdout;
2544
2545         options->line_termination = '\n';
2546         options->break_opt = -1;
2547         options->rename_limit = -1;
2548         options->dirstat_percent = 3;
2549         options->context = 3;
2550
2551         options->change = diff_change;
2552         options->add_remove = diff_addremove;
2553         if (diff_use_color_default > 0)
2554                 DIFF_OPT_SET(options, COLOR_DIFF);
2555         options->detect_rename = diff_detect_rename_default;
2556
2557         if (!diff_mnemonic_prefix) {
2558                 options->a_prefix = "a/";
2559                 options->b_prefix = "b/";
2560         }
2561 }
2562
2563 int diff_setup_done(struct diff_options *options)
2564 {
2565         int count = 0;
2566
2567         if (options->output_format & DIFF_FORMAT_NAME)
2568                 count++;
2569         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2570                 count++;
2571         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2572                 count++;
2573         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2574                 count++;
2575         if (count > 1)
2576                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2577
2578         /*
2579          * Most of the time we can say "there are changes"
2580          * only by checking if there are changed paths, but
2581          * --ignore-whitespace* options force us to look
2582          * inside contents.
2583          */
2584
2585         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2586             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2587             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2588                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2589         else
2590                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2591
2592         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2593                 options->detect_rename = DIFF_DETECT_COPY;
2594
2595         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2596                 options->prefix = NULL;
2597         if (options->prefix)
2598                 options->prefix_length = strlen(options->prefix);
2599         else
2600                 options->prefix_length = 0;
2601
2602         if (options->output_format & (DIFF_FORMAT_NAME |
2603                                       DIFF_FORMAT_NAME_STATUS |
2604                                       DIFF_FORMAT_CHECKDIFF |
2605                                       DIFF_FORMAT_NO_OUTPUT))
2606                 options->output_format &= ~(DIFF_FORMAT_RAW |
2607                                             DIFF_FORMAT_NUMSTAT |
2608                                             DIFF_FORMAT_DIFFSTAT |
2609                                             DIFF_FORMAT_SHORTSTAT |
2610                                             DIFF_FORMAT_DIRSTAT |
2611                                             DIFF_FORMAT_SUMMARY |
2612                                             DIFF_FORMAT_PATCH);
2613
2614         /*
2615          * These cases always need recursive; we do not drop caller-supplied
2616          * recursive bits for other formats here.
2617          */
2618         if (options->output_format & (DIFF_FORMAT_PATCH |
2619                                       DIFF_FORMAT_NUMSTAT |
2620                                       DIFF_FORMAT_DIFFSTAT |
2621                                       DIFF_FORMAT_SHORTSTAT |
2622                                       DIFF_FORMAT_DIRSTAT |
2623                                       DIFF_FORMAT_SUMMARY |
2624                                       DIFF_FORMAT_CHECKDIFF))
2625                 DIFF_OPT_SET(options, RECURSIVE);
2626         /*
2627          * Also pickaxe would not work very well if you do not say recursive
2628          */
2629         if (options->pickaxe)
2630                 DIFF_OPT_SET(options, RECURSIVE);
2631
2632         if (options->detect_rename && options->rename_limit < 0)
2633                 options->rename_limit = diff_rename_limit_default;
2634         if (options->setup & DIFF_SETUP_USE_CACHE) {
2635                 if (!active_cache)
2636                         /* read-cache does not die even when it fails
2637                          * so it is safe for us to do this here.  Also
2638                          * it does not smudge active_cache or active_nr
2639                          * when it fails, so we do not have to worry about
2640                          * cleaning it up ourselves either.
2641                          */
2642                         read_cache();
2643         }
2644         if (options->abbrev <= 0 || 40 < options->abbrev)
2645                 options->abbrev = 40; /* full */
2646
2647         /*
2648          * It does not make sense to show the first hit we happened
2649          * to have found.  It does not make sense not to return with
2650          * exit code in such a case either.
2651          */
2652         if (DIFF_OPT_TST(options, QUICK)) {
2653                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2654                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2655         }
2656
2657         return 0;
2658 }
2659
2660 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2661 {
2662         char c, *eq;
2663         int len;
2664
2665         if (*arg != '-')
2666                 return 0;
2667         c = *++arg;
2668         if (!c)
2669                 return 0;
2670         if (c == arg_short) {
2671                 c = *++arg;
2672                 if (!c)
2673                         return 1;
2674                 if (val && isdigit(c)) {
2675                         char *end;
2676                         int n = strtoul(arg, &end, 10);
2677                         if (*end)
2678                                 return 0;
2679                         *val = n;
2680                         return 1;
2681                 }
2682                 return 0;
2683         }
2684         if (c != '-')
2685                 return 0;
2686         arg++;
2687         eq = strchr(arg, '=');
2688         if (eq)
2689                 len = eq - arg;
2690         else
2691                 len = strlen(arg);
2692         if (!len || strncmp(arg, arg_long, len))
2693                 return 0;
2694         if (eq) {
2695                 int n;
2696                 char *end;
2697                 if (!isdigit(*++eq))
2698                         return 0;
2699                 n = strtoul(eq, &end, 10);
2700                 if (*end)
2701                         return 0;
2702                 *val = n;
2703         }
2704         return 1;
2705 }
2706
2707 static int diff_scoreopt_parse(const char *opt);
2708
2709 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2710 {
2711         const char *arg = av[0];
2712
2713         /* Output format options */
2714         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2715                 options->output_format |= DIFF_FORMAT_PATCH;
2716         else if (opt_arg(arg, 'U', "unified", &options->context))
2717                 options->output_format |= DIFF_FORMAT_PATCH;
2718         else if (!strcmp(arg, "--raw"))
2719                 options->output_format |= DIFF_FORMAT_RAW;
2720         else if (!strcmp(arg, "--patch-with-raw"))
2721                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2722         else if (!strcmp(arg, "--numstat"))
2723                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2724         else if (!strcmp(arg, "--shortstat"))
2725                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2726         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2727                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2728         else if (!strcmp(arg, "--cumulative")) {
2729                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2730                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2731         } else if (opt_arg(arg, 0, "dirstat-by-file",
2732                            &options->dirstat_percent)) {
2733                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2734                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2735         }
2736         else if (!strcmp(arg, "--check"))
2737                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2738         else if (!strcmp(arg, "--summary"))
2739                 options->output_format |= DIFF_FORMAT_SUMMARY;
2740         else if (!strcmp(arg, "--patch-with-stat"))
2741                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2742         else if (!strcmp(arg, "--name-only"))
2743                 options->output_format |= DIFF_FORMAT_NAME;
2744         else if (!strcmp(arg, "--name-status"))
2745                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2746         else if (!strcmp(arg, "-s"))
2747                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2748         else if (!prefixcmp(arg, "--stat")) {
2749                 char *end;
2750                 int width = options->stat_width;
2751                 int name_width = options->stat_name_width;
2752                 arg += 6;
2753                 end = (char *)arg;
2754
2755                 switch (*arg) {
2756                 case '-':
2757                         if (!prefixcmp(arg, "-width="))
2758                                 width = strtoul(arg + 7, &end, 10);
2759                         else if (!prefixcmp(arg, "-name-width="))
2760                                 name_width = strtoul(arg + 12, &end, 10);
2761                         break;
2762                 case '=':
2763                         width = strtoul(arg+1, &end, 10);
2764                         if (*end == ',')
2765                                 name_width = strtoul(end+1, &end, 10);
2766                 }
2767
2768                 /* Important! This checks all the error cases! */
2769                 if (*end)
2770                         return 0;
2771                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2772                 options->stat_name_width = name_width;
2773                 options->stat_width = width;
2774         }
2775
2776         /* renames options */
2777         else if (!prefixcmp(arg, "-B")) {
2778                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2779                         return -1;
2780         }
2781         else if (!prefixcmp(arg, "-M")) {
2782                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2783                         return -1;
2784                 options->detect_rename = DIFF_DETECT_RENAME;
2785         }
2786         else if (!prefixcmp(arg, "-C")) {
2787                 if (options->detect_rename == DIFF_DETECT_COPY)
2788                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2789                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2790                         return -1;
2791                 options->detect_rename = DIFF_DETECT_COPY;
2792         }
2793         else if (!strcmp(arg, "--no-renames"))
2794                 options->detect_rename = 0;
2795         else if (!strcmp(arg, "--relative"))
2796                 DIFF_OPT_SET(options, RELATIVE_NAME);
2797         else if (!prefixcmp(arg, "--relative=")) {
2798                 DIFF_OPT_SET(options, RELATIVE_NAME);
2799                 options->prefix = arg + 11;
2800         }
2801
2802         /* xdiff options */
2803         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2804                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2805         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2806                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2807         else if (!strcmp(arg, "--ignore-space-at-eol"))
2808                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2809         else if (!strcmp(arg, "--patience"))
2810                 DIFF_XDL_SET(options, PATIENCE_DIFF);
2811
2812         /* flags options */
2813         else if (!strcmp(arg, "--binary")) {
2814                 options->output_format |= DIFF_FORMAT_PATCH;
2815                 DIFF_OPT_SET(options, BINARY);
2816         }
2817         else if (!strcmp(arg, "--full-index"))
2818                 DIFF_OPT_SET(options, FULL_INDEX);
2819         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2820                 DIFF_OPT_SET(options, TEXT);
2821         else if (!strcmp(arg, "-R"))
2822                 DIFF_OPT_SET(options, REVERSE_DIFF);
2823         else if (!strcmp(arg, "--find-copies-harder"))
2824                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2825         else if (!strcmp(arg, "--follow"))
2826                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2827         else if (!strcmp(arg, "--color"))
2828                 DIFF_OPT_SET(options, COLOR_DIFF);
2829         else if (!strcmp(arg, "--no-color"))
2830                 DIFF_OPT_CLR(options, COLOR_DIFF);
2831         else if (!strcmp(arg, "--color-words")) {
2832                 DIFF_OPT_SET(options, COLOR_DIFF);
2833                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2834         }
2835         else if (!prefixcmp(arg, "--color-words=")) {
2836                 DIFF_OPT_SET(options, COLOR_DIFF);
2837                 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2838                 options->word_regex = arg + 14;
2839         }
2840         else if (!strcmp(arg, "--exit-code"))
2841                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2842         else if (!strcmp(arg, "--quiet"))
2843                 DIFF_OPT_SET(options, QUICK);
2844         else if (!strcmp(arg, "--ext-diff"))
2845                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2846         else if (!strcmp(arg, "--no-ext-diff"))
2847                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2848         else if (!strcmp(arg, "--textconv"))
2849                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2850         else if (!strcmp(arg, "--no-textconv"))
2851                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2852         else if (!strcmp(arg, "--ignore-submodules"))
2853                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2854         else if (!strcmp(arg, "--submodule"))
2855                 DIFF_OPT_SET(options, SUBMODULE_LOG);
2856         else if (!prefixcmp(arg, "--submodule=")) {
2857                 if (!strcmp(arg + 12, "log"))
2858                         DIFF_OPT_SET(options, SUBMODULE_LOG);
2859         }
2860
2861         /* misc options */
2862         else if (!strcmp(arg, "-z"))
2863                 options->line_termination = 0;
2864         else if (!prefixcmp(arg, "-l"))
2865                 options->rename_limit = strtoul(arg+2, NULL, 10);
2866         else if (!prefixcmp(arg, "-S"))
2867                 options->pickaxe = arg + 2;
2868         else if (!strcmp(arg, "--pickaxe-all"))
2869                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2870         else if (!strcmp(arg, "--pickaxe-regex"))
2871                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2872         else if (!prefixcmp(arg, "-O"))
2873                 options->orderfile = arg + 2;
2874         else if (!prefixcmp(arg, "--diff-filter="))
2875                 options->filter = arg + 14;
2876         else if (!strcmp(arg, "--abbrev"))
2877                 options->abbrev = DEFAULT_ABBREV;
2878         else if (!prefixcmp(arg, "--abbrev=")) {
2879                 options->abbrev = strtoul(arg + 9, NULL, 10);
2880                 if (options->abbrev < MINIMUM_ABBREV)
2881                         options->abbrev = MINIMUM_ABBREV;
2882                 else if (40 < options->abbrev)
2883                         options->abbrev = 40;
2884         }
2885         else if (!prefixcmp(arg, "--src-prefix="))
2886                 options->a_prefix = arg + 13;
2887         else if (!prefixcmp(arg, "--dst-prefix="))
2888                 options->b_prefix = arg + 13;
2889         else if (!strcmp(arg, "--no-prefix"))
2890                 options->a_prefix = options->b_prefix = "";
2891         else if (opt_arg(arg, '\0', "inter-hunk-context",
2892                          &options->interhunkcontext))
2893                 ;
2894         else if (!prefixcmp(arg, "--output=")) {
2895                 options->file = fopen(arg + strlen("--output="), "w");
2896                 if (!options->file)
2897                         die_errno("Could not open '%s'", arg + strlen("--output="));
2898                 options->close_file = 1;
2899         } else
2900                 return 0;
2901         return 1;
2902 }
2903
2904 static int parse_num(const char **cp_p)
2905 {
2906         unsigned long num, scale;
2907         int ch, dot;
2908         const char *cp = *cp_p;
2909
2910         num = 0;
2911         scale = 1;
2912         dot = 0;
2913         for (;;) {
2914                 ch = *cp;
2915                 if ( !dot && ch == '.' ) {
2916                         scale = 1;
2917                         dot = 1;
2918                 } else if ( ch == '%' ) {
2919                         scale = dot ? scale*100 : 100;
2920                         cp++;   /* % is always at the end */
2921                         break;
2922                 } else if ( ch >= '0' && ch <= '9' ) {
2923                         if ( scale < 100000 ) {
2924                                 scale *= 10;
2925                                 num = (num*10) + (ch-'0');
2926                         }
2927                 } else {
2928                         break;
2929                 }
2930                 cp++;
2931         }
2932         *cp_p = cp;
2933
2934         /* user says num divided by scale and we say internally that
2935          * is MAX_SCORE * num / scale.
2936          */
2937         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2938 }
2939
2940 static int diff_scoreopt_parse(const char *opt)
2941 {
2942         int opt1, opt2, cmd;
2943
2944         if (*opt++ != '-')
2945                 return -1;
2946         cmd = *opt++;
2947         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2948                 return -1; /* that is not a -M, -C nor -B option */
2949
2950         opt1 = parse_num(&opt);
2951         if (cmd != 'B')
2952                 opt2 = 0;
2953         else {
2954                 if (*opt == 0)
2955                         opt2 = 0;
2956                 else if (*opt != '/')
2957                         return -1; /* we expect -B80/99 or -B80 */
2958                 else {
2959                         opt++;
2960                         opt2 = parse_num(&opt);
2961                 }
2962         }
2963         if (*opt != 0)
2964                 return -1;
2965         return opt1 | (opt2 << 16);
2966 }
2967
2968 struct diff_queue_struct diff_queued_diff;
2969
2970 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2971 {
2972         if (queue->alloc <= queue->nr) {
2973                 queue->alloc = alloc_nr(queue->alloc);
2974                 queue->queue = xrealloc(queue->queue,
2975                                         sizeof(dp) * queue->alloc);
2976         }
2977         queue->queue[queue->nr++] = dp;
2978 }
2979
2980 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2981                                  struct diff_filespec *one,
2982                                  struct diff_filespec *two)
2983 {
2984         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2985         dp->one = one;
2986         dp->two = two;
2987         if (queue)
2988                 diff_q(queue, dp);
2989         return dp;
2990 }
2991
2992 void diff_free_filepair(struct diff_filepair *p)
2993 {
2994         free_filespec(p->one);
2995         free_filespec(p->two);
2996         free(p);
2997 }
2998
2999 /* This is different from find_unique_abbrev() in that
3000  * it stuffs the result with dots for alignment.
3001  */
3002 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3003 {
3004         int abblen;
3005         const char *abbrev;
3006         if (len == 40)
3007                 return sha1_to_hex(sha1);
3008
3009         abbrev = find_unique_abbrev(sha1, len);
3010         abblen = strlen(abbrev);
3011         if (abblen < 37) {
3012                 static char hex[41];
3013                 if (len < abblen && abblen <= len + 2)
3014                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3015                 else
3016                         sprintf(hex, "%s...", abbrev);
3017                 return hex;
3018         }
3019         return sha1_to_hex(sha1);
3020 }
3021
3022 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3023 {
3024         int line_termination = opt->line_termination;
3025         int inter_name_termination = line_termination ? '\t' : '\0';
3026
3027         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3028                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3029                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3030                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3031         }
3032         if (p->score) {
3033                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3034                         inter_name_termination);
3035         } else {
3036                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3037         }
3038
3039         if (p->status == DIFF_STATUS_COPIED ||
3040             p->status == DIFF_STATUS_RENAMED) {
3041                 const char *name_a, *name_b;
3042                 name_a = p->one->path;
3043                 name_b = p->two->path;
3044                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3045                 write_name_quoted(name_a, opt->file, inter_name_termination);
3046                 write_name_quoted(name_b, opt->file, line_termination);
3047         } else {
3048                 const char *name_a, *name_b;
3049                 name_a = p->one->mode ? p->one->path : p->two->path;
3050                 name_b = NULL;
3051                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3052                 write_name_quoted(name_a, opt->file, line_termination);
3053         }
3054 }
3055
3056 int diff_unmodified_pair(struct diff_filepair *p)
3057 {
3058         /* This function is written stricter than necessary to support
3059          * the currently implemented transformers, but the idea is to
3060          * let transformers to produce diff_filepairs any way they want,
3061          * and filter and clean them up here before producing the output.
3062          */
3063         struct diff_filespec *one = p->one, *two = p->two;
3064
3065         if (DIFF_PAIR_UNMERGED(p))
3066                 return 0; /* unmerged is interesting */
3067
3068         /* deletion, addition, mode or type change
3069          * and rename are all interesting.
3070          */
3071         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3072             DIFF_PAIR_MODE_CHANGED(p) ||
3073             strcmp(one->path, two->path))
3074                 return 0;
3075
3076         /* both are valid and point at the same path.  that is, we are
3077          * dealing with a change.
3078          */
3079         if (one->sha1_valid && two->sha1_valid &&
3080             !hashcmp(one->sha1, two->sha1))
3081                 return 1; /* no change */
3082         if (!one->sha1_valid && !two->sha1_valid)
3083                 return 1; /* both look at the same file on the filesystem. */
3084         return 0;
3085 }
3086
3087 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3088 {
3089         if (diff_unmodified_pair(p))
3090                 return;
3091
3092         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3093             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3094                 return; /* no tree diffs in patch format */
3095
3096         run_diff(p, o);
3097 }
3098
3099 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3100                             struct diffstat_t *diffstat)
3101 {
3102         if (diff_unmodified_pair(p))
3103                 return;
3104
3105         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3106             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3107                 return; /* no tree diffs in patch format */
3108
3109         run_diffstat(p, o, diffstat);
3110 }
3111
3112 static void diff_flush_checkdiff(struct diff_filepair *p,
3113                 struct diff_options *o)
3114 {
3115         if (diff_unmodified_pair(p))
3116                 return;
3117
3118         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3119             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3120                 return; /* no tree diffs in patch format */
3121
3122         run_checkdiff(p, o);
3123 }
3124
3125 int diff_queue_is_empty(void)
3126 {
3127         struct diff_queue_struct *q = &diff_queued_diff;
3128         int i;
3129         for (i = 0; i < q->nr; i++)
3130                 if (!diff_unmodified_pair(q->queue[i]))
3131                         return 0;
3132         return 1;
3133 }
3134
3135 #if DIFF_DEBUG
3136 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3137 {
3138         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3139                 x, one ? one : "",
3140                 s->path,
3141                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3142                 s->mode,
3143                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3144         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3145                 x, one ? one : "",
3146                 s->size, s->xfrm_flags);
3147 }
3148
3149 void diff_debug_filepair(const struct diff_filepair *p, int i)
3150 {
3151         diff_debug_filespec(p->one, i, "one");
3152         diff_debug_filespec(p->two, i, "two");
3153         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3154                 p->score, p->status ? p->status : '?',
3155                 p->one->rename_used, p->broken_pair);
3156 }
3157
3158 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3159 {
3160         int i;
3161         if (msg)
3162                 fprintf(stderr, "%s\n", msg);
3163         fprintf(stderr, "q->nr = %d\n", q->nr);
3164         for (i = 0; i < q->nr; i++) {
3165                 struct diff_filepair *p = q->queue[i];
3166                 diff_debug_filepair(p, i);
3167         }
3168 }
3169 #endif
3170
3171 static void diff_resolve_rename_copy(void)
3172 {
3173         int i;
3174         struct diff_filepair *p;
3175         struct diff_queue_struct *q = &diff_queued_diff;
3176
3177         diff_debug_queue("resolve-rename-copy", q);
3178
3179         for (i = 0; i < q->nr; i++) {
3180                 p = q->queue[i];
3181                 p->status = 0; /* undecided */
3182                 if (DIFF_PAIR_UNMERGED(p))
3183                         p->status = DIFF_STATUS_UNMERGED;
3184                 else if (!DIFF_FILE_VALID(p->one))
3185                         p->status = DIFF_STATUS_ADDED;
3186                 else if (!DIFF_FILE_VALID(p->two))
3187                         p->status = DIFF_STATUS_DELETED;
3188                 else if (DIFF_PAIR_TYPE_CHANGED(p))
3189                         p->status = DIFF_STATUS_TYPE_CHANGED;
3190
3191                 /* from this point on, we are dealing with a pair
3192                  * whose both sides are valid and of the same type, i.e.
3193                  * either in-place edit or rename/copy edit.
3194                  */
3195                 else if (DIFF_PAIR_RENAME(p)) {
3196                         /*
3197                          * A rename might have re-connected a broken
3198                          * pair up, causing the pathnames to be the
3199                          * same again. If so, that's not a rename at
3200                          * all, just a modification..
3201                          *
3202                          * Otherwise, see if this source was used for
3203                          * multiple renames, in which case we decrement
3204                          * the count, and call it a copy.
3205                          */
3206                         if (!strcmp(p->one->path, p->two->path))
3207                                 p->status = DIFF_STATUS_MODIFIED;
3208                         else if (--p->one->rename_used > 0)
3209                                 p->status = DIFF_STATUS_COPIED;
3210                         else
3211                                 p->status = DIFF_STATUS_RENAMED;
3212                 }
3213                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3214                          p->one->mode != p->two->mode ||
3215                          is_null_sha1(p->one->sha1))
3216                         p->status = DIFF_STATUS_MODIFIED;
3217                 else {
3218                         /* This is a "no-change" entry and should not
3219                          * happen anymore, but prepare for broken callers.
3220                          */
3221                         error("feeding unmodified %s to diffcore",
3222                               p->one->path);
3223                         p->status = DIFF_STATUS_UNKNOWN;
3224                 }
3225         }
3226         diff_debug_queue("resolve-rename-copy done", q);
3227 }
3228
3229 static int check_pair_status(struct diff_filepair *p)
3230 {
3231         switch (p->status) {
3232         case DIFF_STATUS_UNKNOWN:
3233                 return 0;
3234         case 0:
3235                 die("internal error in diff-resolve-rename-copy");
3236         default:
3237                 return 1;
3238         }
3239 }
3240
3241 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3242 {
3243         int fmt = opt->output_format;
3244
3245         if (fmt & DIFF_FORMAT_CHECKDIFF)
3246                 diff_flush_checkdiff(p, opt);
3247         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3248                 diff_flush_raw(p, opt);
3249         else if (fmt & DIFF_FORMAT_NAME) {
3250                 const char *name_a, *name_b;
3251                 name_a = p->two->path;
3252                 name_b = NULL;
3253                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3254                 write_name_quoted(name_a, opt->file, opt->line_termination);
3255         }
3256 }
3257
3258 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3259 {
3260         if (fs->mode)
3261                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3262         else
3263                 fprintf(file, " %s ", newdelete);
3264         write_name_quoted(fs->path, file, '\n');
3265 }
3266
3267
3268 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3269 {
3270         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3271                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3272                         show_name ? ' ' : '\n');
3273                 if (show_name) {
3274                         write_name_quoted(p->two->path, file, '\n');
3275                 }
3276         }
3277 }
3278
3279 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3280 {
3281         char *names = pprint_rename(p->one->path, p->two->path);
3282
3283         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3284         free(names);
3285         show_mode_change(file, p, 0);
3286 }
3287
3288 static void diff_summary(FILE *file, struct diff_filepair *p)
3289 {
3290         switch(p->status) {
3291         case DIFF_STATUS_DELETED:
3292                 show_file_mode_name(file, "delete", p->one);
3293                 break;
3294         case DIFF_STATUS_ADDED:
3295                 show_file_mode_name(file, "create", p->two);
3296                 break;
3297         case DIFF_STATUS_COPIED:
3298                 show_rename_copy(file, "copy", p);
3299                 break;
3300         case DIFF_STATUS_RENAMED:
3301                 show_rename_copy(file, "rename", p);
3302                 break;
3303         default:
3304                 if (p->score) {
3305                         fputs(" rewrite ", file);
3306                         write_name_quoted(p->two->path, file, ' ');
3307                         fprintf(file, "(%d%%)\n", similarity_index(p));
3308                 }
3309                 show_mode_change(file, p, !p->score);
3310                 break;
3311         }
3312 }
3313
3314 struct patch_id_t {
3315         git_SHA_CTX *ctx;
3316         int patchlen;
3317 };
3318
3319 static int remove_space(char *line, int len)
3320 {
3321         int i;
3322         char *dst = line;
3323         unsigned char c;
3324
3325         for (i = 0; i < len; i++)
3326                 if (!isspace((c = line[i])))
3327                         *dst++ = c;
3328
3329         return dst - line;
3330 }
3331
3332 static void patch_id_consume(void *priv, char *line, unsigned long len)
3333 {
3334         struct patch_id_t *data = priv;
3335         int new_len;
3336
3337         /* Ignore line numbers when computing the SHA1 of the patch */
3338         if (!prefixcmp(line, "@@ -"))
3339                 return;
3340
3341         new_len = remove_space(line, len);
3342
3343         git_SHA1_Update(data->ctx, line, new_len);
3344         data->patchlen += new_len;
3345 }
3346
3347 /* returns 0 upon success, and writes result into sha1 */
3348 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3349 {
3350         struct diff_queue_struct *q = &diff_queued_diff;
3351         int i;
3352         git_SHA_CTX ctx;
3353         struct patch_id_t data;
3354         char buffer[PATH_MAX * 4 + 20];
3355
3356         git_SHA1_Init(&ctx);
3357         memset(&data, 0, sizeof(struct patch_id_t));
3358         data.ctx = &ctx;
3359
3360         for (i = 0; i < q->nr; i++) {
3361                 xpparam_t xpp;
3362                 xdemitconf_t xecfg;
3363                 xdemitcb_t ecb;
3364                 mmfile_t mf1, mf2;
3365                 struct diff_filepair *p = q->queue[i];
3366                 int len1, len2;
3367
3368                 memset(&xpp, 0, sizeof(xpp));
3369                 memset(&xecfg, 0, sizeof(xecfg));
3370                 if (p->status == 0)
3371                         return error("internal diff status error");
3372                 if (p->status == DIFF_STATUS_UNKNOWN)
3373                         continue;
3374                 if (diff_unmodified_pair(p))
3375                         continue;
3376                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3377                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3378                         continue;
3379                 if (DIFF_PAIR_UNMERGED(p))
3380                         continue;
3381
3382                 diff_fill_sha1_info(p->one);
3383                 diff_fill_sha1_info(p->two);
3384                 if (fill_mmfile(&mf1, p->one) < 0 ||
3385                                 fill_mmfile(&mf2, p->two) < 0)
3386                         return error("unable to read files to diff");
3387
3388                 len1 = remove_space(p->one->path, strlen(p->one->path));
3389                 len2 = remove_space(p->two->path, strlen(p->two->path));
3390                 if (p->one->mode == 0)
3391                         len1 = snprintf(buffer, sizeof(buffer),
3392                                         "diff--gita/%.*sb/%.*s"
3393                                         "newfilemode%06o"
3394                                         "---/dev/null"
3395                                         "+++b/%.*s",
3396                                         len1, p->one->path,
3397                                         len2, p->two->path,
3398                                         p->two->mode,
3399                                         len2, p->two->path);
3400                 else if (p->two->mode == 0)
3401                         len1 = snprintf(buffer, sizeof(buffer),
3402                                         "diff--gita/%.*sb/%.*s"
3403                                         "deletedfilemode%06o"
3404                                         "---a/%.*s"
3405                                         "+++/dev/null",
3406                                         len1, p->one->path,
3407                                         len2, p->two->path,
3408                                         p->one->mode,
3409                                         len1, p->one->path);
3410                 else
3411                         len1 = snprintf(buffer, sizeof(buffer),
3412                                         "diff--gita/%.*sb/%.*s"
3413                                         "---a/%.*s"
3414                                         "+++b/%.*s",
3415                                         len1, p->one->path,
3416                                         len2, p->two->path,
3417                                         len1, p->one->path,
3418                                         len2, p->two->path);
3419                 git_SHA1_Update(&ctx, buffer, len1);
3420
3421                 xpp.flags = XDF_NEED_MINIMAL;
3422                 xecfg.ctxlen = 3;
3423                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3424                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3425                               &xpp, &xecfg, &ecb);
3426         }
3427
3428         git_SHA1_Final(sha1, &ctx);
3429         return 0;
3430 }
3431
3432 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3433 {
3434         struct diff_queue_struct *q = &diff_queued_diff;
3435         int i;
3436         int result = diff_get_patch_id(options, sha1);
3437
3438         for (i = 0; i < q->nr; i++)
3439                 diff_free_filepair(q->queue[i]);
3440
3441         free(q->queue);
3442         q->queue = NULL;
3443         q->nr = q->alloc = 0;
3444
3445         return result;
3446 }
3447
3448 static int is_summary_empty(const struct diff_queue_struct *q)
3449 {
3450         int i;
3451
3452         for (i = 0; i < q->nr; i++) {
3453                 const struct diff_filepair *p = q->queue[i];
3454
3455                 switch (p->status) {
3456                 case DIFF_STATUS_DELETED:
3457                 case DIFF_STATUS_ADDED:
3458                 case DIFF_STATUS_COPIED:
3459                 case DIFF_STATUS_RENAMED:
3460                         return 0;
3461                 default:
3462                         if (p->score)
3463                                 return 0;
3464                         if (p->one->mode && p->two->mode &&
3465                             p->one->mode != p->two->mode)
3466                                 return 0;
3467                         break;
3468                 }
3469         }
3470         return 1;
3471 }
3472
3473 void diff_flush(struct diff_options *options)
3474 {
3475         struct diff_queue_struct *q = &diff_queued_diff;
3476         int i, output_format = options->output_format;
3477         int separator = 0;
3478
3479         /*
3480          * Order: raw, stat, summary, patch
3481          * or:    name/name-status/checkdiff (other bits clear)
3482          */
3483         if (!q->nr)
3484                 goto free_queue;
3485
3486         if (output_format & (DIFF_FORMAT_RAW |
3487                              DIFF_FORMAT_NAME |
3488                              DIFF_FORMAT_NAME_STATUS |
3489                              DIFF_FORMAT_CHECKDIFF)) {
3490                 for (i = 0; i < q->nr; i++) {
3491                         struct diff_filepair *p = q->queue[i];
3492                         if (check_pair_status(p))
3493                                 flush_one_pair(p, options);
3494                 }
3495                 separator++;
3496         }
3497
3498         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3499                 struct diffstat_t diffstat;
3500
3501                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3502                 for (i = 0; i < q->nr; i++) {
3503                         struct diff_filepair *p = q->queue[i];
3504                         if (check_pair_status(p))
3505                                 diff_flush_stat(p, options, &diffstat);
3506                 }
3507                 if (output_format & DIFF_FORMAT_NUMSTAT)
3508                         show_numstat(&diffstat, options);
3509                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3510                         show_stats(&diffstat, options);
3511                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3512                         show_shortstats(&diffstat, options);
3513                 free_diffstat_info(&diffstat);
3514                 separator++;
3515         }
3516         if (output_format & DIFF_FORMAT_DIRSTAT)
3517                 show_dirstat(options);
3518
3519         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3520                 for (i = 0; i < q->nr; i++)
3521                         diff_summary(options->file, q->queue[i]);
3522                 separator++;
3523         }
3524
3525         if (output_format & DIFF_FORMAT_PATCH) {
3526                 if (separator) {
3527                         putc(options->line_termination, options->file);
3528                         if (options->stat_sep) {
3529                                 /* attach patch instead of inline */
3530                                 fputs(options->stat_sep, options->file);
3531                         }
3532                 }
3533
3534                 for (i = 0; i < q->nr; i++) {
3535                         struct diff_filepair *p = q->queue[i];
3536                         if (check_pair_status(p))
3537                                 diff_flush_patch(p, options);
3538                 }
3539         }
3540
3541         if (output_format & DIFF_FORMAT_CALLBACK)
3542                 options->format_callback(q, options, options->format_callback_data);
3543
3544         for (i = 0; i < q->nr; i++)
3545                 diff_free_filepair(q->queue[i]);
3546 free_queue:
3547         free(q->queue);
3548         q->queue = NULL;
3549         q->nr = q->alloc = 0;
3550         if (options->close_file)
3551                 fclose(options->file);
3552
3553         /*
3554          * Report the content-level differences with HAS_CHANGES;
3555          * diff_addremove/diff_change does not set the bit when
3556          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3557          */
3558         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3559                 if (options->found_changes)
3560                         DIFF_OPT_SET(options, HAS_CHANGES);
3561                 else
3562                         DIFF_OPT_CLR(options, HAS_CHANGES);
3563         }
3564 }
3565
3566 static void diffcore_apply_filter(const char *filter)
3567 {
3568         int i;
3569         struct diff_queue_struct *q = &diff_queued_diff;
3570         struct diff_queue_struct outq;
3571         outq.queue = NULL;
3572         outq.nr = outq.alloc = 0;
3573
3574         if (!filter)
3575                 return;
3576
3577         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3578                 int found;
3579                 for (i = found = 0; !found && i < q->nr; i++) {
3580                         struct diff_filepair *p = q->queue[i];
3581                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3582                              ((p->score &&
3583                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3584                               (!p->score &&
3585                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3586                             ((p->status != DIFF_STATUS_MODIFIED) &&
3587                              strchr(filter, p->status)))
3588                                 found++;
3589                 }
3590                 if (found)
3591                         return;
3592
3593                 /* otherwise we will clear the whole queue
3594                  * by copying the empty outq at the end of this
3595                  * function, but first clear the current entries
3596                  * in the queue.
3597                  */
3598                 for (i = 0; i < q->nr; i++)
3599                         diff_free_filepair(q->queue[i]);
3600         }
3601         else {
3602                 /* Only the matching ones */
3603                 for (i = 0; i < q->nr; i++) {
3604                         struct diff_filepair *p = q->queue[i];
3605
3606                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3607                              ((p->score &&
3608                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3609                               (!p->score &&
3610                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3611                             ((p->status != DIFF_STATUS_MODIFIED) &&
3612                              strchr(filter, p->status)))
3613                                 diff_q(&outq, p);
3614                         else
3615                                 diff_free_filepair(p);
3616                 }
3617         }
3618         free(q->queue);
3619         *q = outq;
3620 }
3621
3622 /* Check whether two filespecs with the same mode and size are identical */
3623 static int diff_filespec_is_identical(struct diff_filespec *one,
3624                                       struct diff_filespec *two)
3625 {
3626         if (S_ISGITLINK(one->mode))
3627                 return 0;
3628         if (diff_populate_filespec(one, 0))
3629                 return 0;
3630         if (diff_populate_filespec(two, 0))
3631                 return 0;
3632         return !memcmp(one->data, two->data, one->size);
3633 }
3634
3635 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3636 {
3637         int i;
3638         struct diff_queue_struct *q = &diff_queued_diff;
3639         struct diff_queue_struct outq;
3640         outq.queue = NULL;
3641         outq.nr = outq.alloc = 0;
3642
3643         for (i = 0; i < q->nr; i++) {
3644                 struct diff_filepair *p = q->queue[i];
3645
3646                 /*
3647                  * 1. Entries that come from stat info dirtiness
3648                  *    always have both sides (iow, not create/delete),
3649                  *    one side of the object name is unknown, with
3650                  *    the same mode and size.  Keep the ones that
3651                  *    do not match these criteria.  They have real
3652                  *    differences.
3653                  *
3654                  * 2. At this point, the file is known to be modified,
3655                  *    with the same mode and size, and the object
3656                  *    name of one side is unknown.  Need to inspect
3657                  *    the identical contents.
3658                  */
3659                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3660                     !DIFF_FILE_VALID(p->two) ||
3661                     (p->one->sha1_valid && p->two->sha1_valid) ||
3662                     (p->one->mode != p->two->mode) ||
3663                     diff_populate_filespec(p->one, 1) ||
3664                     diff_populate_filespec(p->two, 1) ||
3665                     (p->one->size != p->two->size) ||
3666                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3667                         diff_q(&outq, p);
3668                 else {
3669                         /*
3670                          * The caller can subtract 1 from skip_stat_unmatch
3671                          * to determine how many paths were dirty only
3672                          * due to stat info mismatch.
3673                          */
3674                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3675                                 diffopt->skip_stat_unmatch++;
3676                         diff_free_filepair(p);
3677                 }
3678         }
3679         free(q->queue);
3680         *q = outq;
3681 }
3682
3683 static int diffnamecmp(const void *a_, const void *b_)
3684 {
3685         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
3686         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
3687         const char *name_a, *name_b;
3688
3689         name_a = a->one ? a->one->path : a->two->path;
3690         name_b = b->one ? b->one->path : b->two->path;
3691         return strcmp(name_a, name_b);
3692 }
3693
3694 void diffcore_fix_diff_index(struct diff_options *options)
3695 {
3696         struct diff_queue_struct *q = &diff_queued_diff;
3697         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
3698 }
3699
3700 void diffcore_std(struct diff_options *options)
3701 {
3702         if (options->skip_stat_unmatch)
3703                 diffcore_skip_stat_unmatch(options);
3704         if (options->break_opt != -1)
3705                 diffcore_break(options->break_opt);
3706         if (options->detect_rename)
3707                 diffcore_rename(options);
3708         if (options->break_opt != -1)
3709                 diffcore_merge_broken();
3710         if (options->pickaxe)
3711                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3712         if (options->orderfile)
3713                 diffcore_order(options->orderfile);
3714         diff_resolve_rename_copy();
3715         diffcore_apply_filter(options->filter);
3716
3717         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3718                 DIFF_OPT_SET(options, HAS_CHANGES);
3719         else
3720                 DIFF_OPT_CLR(options, HAS_CHANGES);
3721 }
3722
3723 int diff_result_code(struct diff_options *opt, int status)
3724 {
3725         int result = 0;
3726         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3727             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3728                 return status;
3729         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3730             DIFF_OPT_TST(opt, HAS_CHANGES))
3731                 result |= 01;
3732         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3733             DIFF_OPT_TST(opt, CHECK_FAILED))
3734                 result |= 02;
3735         return result;
3736 }
3737
3738 void diff_addremove(struct diff_options *options,
3739                     int addremove, unsigned mode,
3740                     const unsigned char *sha1,
3741                     const char *concatpath, unsigned dirty_submodule)
3742 {
3743         struct diff_filespec *one, *two;
3744
3745         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3746                 return;
3747
3748         /* This may look odd, but it is a preparation for
3749          * feeding "there are unchanged files which should
3750          * not produce diffs, but when you are doing copy
3751          * detection you would need them, so here they are"
3752          * entries to the diff-core.  They will be prefixed
3753          * with something like '=' or '*' (I haven't decided
3754          * which but should not make any difference).
3755          * Feeding the same new and old to diff_change()
3756          * also has the same effect.
3757          * Before the final output happens, they are pruned after
3758          * merged into rename/copy pairs as appropriate.
3759          */
3760         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3761                 addremove = (addremove == '+' ? '-' :
3762                              addremove == '-' ? '+' : addremove);
3763
3764         if (options->prefix &&
3765             strncmp(concatpath, options->prefix, options->prefix_length))
3766                 return;
3767
3768         one = alloc_filespec(concatpath);
3769         two = alloc_filespec(concatpath);
3770
3771         if (addremove != '+')
3772                 fill_filespec(one, sha1, mode);
3773         if (addremove != '-') {
3774                 fill_filespec(two, sha1, mode);
3775                 two->dirty_submodule = dirty_submodule;
3776         }
3777
3778         diff_queue(&diff_queued_diff, one, two);
3779         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3780                 DIFF_OPT_SET(options, HAS_CHANGES);
3781 }
3782
3783 void diff_change(struct diff_options *options,
3784                  unsigned old_mode, unsigned new_mode,
3785                  const unsigned char *old_sha1,
3786                  const unsigned char *new_sha1,
3787                  const char *concatpath,
3788                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
3789 {
3790         struct diff_filespec *one, *two;
3791
3792         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3793                         && S_ISGITLINK(new_mode))
3794                 return;
3795
3796         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3797                 unsigned tmp;
3798                 const unsigned char *tmp_c;
3799                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3800                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3801                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
3802                         new_dirty_submodule = tmp;
3803         }
3804
3805         if (options->prefix &&
3806             strncmp(concatpath, options->prefix, options->prefix_length))
3807                 return;
3808
3809         one = alloc_filespec(concatpath);
3810         two = alloc_filespec(concatpath);
3811         fill_filespec(one, old_sha1, old_mode);
3812         fill_filespec(two, new_sha1, new_mode);
3813         one->dirty_submodule = old_dirty_submodule;
3814         two->dirty_submodule = new_dirty_submodule;
3815
3816         diff_queue(&diff_queued_diff, one, two);
3817         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3818                 DIFF_OPT_SET(options, HAS_CHANGES);
3819 }
3820
3821 void diff_unmerge(struct diff_options *options,
3822                   const char *path,
3823                   unsigned mode, const unsigned char *sha1)
3824 {
3825         struct diff_filespec *one, *two;
3826
3827         if (options->prefix &&
3828             strncmp(path, options->prefix, options->prefix_length))
3829                 return;
3830
3831         one = alloc_filespec(path);
3832         two = alloc_filespec(path);
3833         fill_filespec(one, sha1, mode);
3834         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3835 }
3836
3837 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3838                 size_t *outsize)
3839 {
3840         struct diff_tempfile *temp;
3841         const char *argv[3];
3842         const char **arg = argv;
3843         struct child_process child;
3844         struct strbuf buf = STRBUF_INIT;
3845
3846         temp = prepare_temp_file(spec->path, spec);
3847         *arg++ = pgm;
3848         *arg++ = temp->name;
3849         *arg = NULL;
3850
3851         memset(&child, 0, sizeof(child));
3852         child.use_shell = 1;
3853         child.argv = argv;
3854         child.out = -1;
3855         if (start_command(&child) != 0 ||
3856             strbuf_read(&buf, child.out, 0) < 0 ||
3857             finish_command(&child) != 0) {
3858                 close(child.out);
3859                 strbuf_release(&buf);
3860                 remove_tempfile();
3861                 error("error running textconv command '%s'", pgm);
3862                 return NULL;
3863         }
3864         close(child.out);
3865         remove_tempfile();
3866
3867         return strbuf_detach(&buf, outsize);
3868 }