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