Merge branch 'sh/write-pack-file-warning-message-fix' into sh/finish-tmp-packfile
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
14 #include "userdiff.h"
15 #include "sigchain.h"
16 #include "submodule.h"
17 #include "ll-merge.h"
18 #include "string-list.h"
19
20 #ifdef NO_FAST_WORKING_DIRECTORY
21 #define FAST_WORKING_DIRECTORY 0
22 #else
23 #define FAST_WORKING_DIRECTORY 1
24 #endif
25
26 static int diff_detect_rename_default;
27 static int diff_rename_limit_default = 400;
28 static int diff_suppress_blank_empty;
29 static int diff_use_color_default = -1;
30 static int diff_context_default = 3;
31 static const char *diff_word_regex_cfg;
32 static const char *external_diff_cmd_cfg;
33 static const char *diff_order_file_cfg;
34 int diff_auto_refresh_index = 1;
35 static int diff_mnemonic_prefix;
36 static int diff_no_prefix;
37 static int diff_stat_graph_width;
38 static int diff_dirstat_permille_default = 30;
39 static struct diff_options default_diff_options;
40 static long diff_algorithm;
41
42 static char diff_colors[][COLOR_MAXLEN] = {
43         GIT_COLOR_RESET,
44         GIT_COLOR_NORMAL,       /* PLAIN */
45         GIT_COLOR_BOLD,         /* METAINFO */
46         GIT_COLOR_CYAN,         /* FRAGINFO */
47         GIT_COLOR_RED,          /* OLD */
48         GIT_COLOR_GREEN,        /* NEW */
49         GIT_COLOR_YELLOW,       /* COMMIT */
50         GIT_COLOR_BG_RED,       /* WHITESPACE */
51         GIT_COLOR_NORMAL,       /* FUNCINFO */
52 };
53
54 static int parse_diff_color_slot(const char *var, int ofs)
55 {
56         if (!strcasecmp(var+ofs, "plain"))
57                 return DIFF_PLAIN;
58         if (!strcasecmp(var+ofs, "meta"))
59                 return DIFF_METAINFO;
60         if (!strcasecmp(var+ofs, "frag"))
61                 return DIFF_FRAGINFO;
62         if (!strcasecmp(var+ofs, "old"))
63                 return DIFF_FILE_OLD;
64         if (!strcasecmp(var+ofs, "new"))
65                 return DIFF_FILE_NEW;
66         if (!strcasecmp(var+ofs, "commit"))
67                 return DIFF_COMMIT;
68         if (!strcasecmp(var+ofs, "whitespace"))
69                 return DIFF_WHITESPACE;
70         if (!strcasecmp(var+ofs, "func"))
71                 return DIFF_FUNCINFO;
72         return -1;
73 }
74
75 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
76                                 struct strbuf *errmsg)
77 {
78         char *params_copy = xstrdup(params_string);
79         struct string_list params = STRING_LIST_INIT_NODUP;
80         int ret = 0;
81         int i;
82
83         if (*params_copy)
84                 string_list_split_in_place(&params, params_copy, ',', -1);
85         for (i = 0; i < params.nr; i++) {
86                 const char *p = params.items[i].string;
87                 if (!strcmp(p, "changes")) {
88                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
89                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
90                 } else if (!strcmp(p, "lines")) {
91                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
92                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
93                 } else if (!strcmp(p, "files")) {
94                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
95                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
96                 } else if (!strcmp(p, "noncumulative")) {
97                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
98                 } else if (!strcmp(p, "cumulative")) {
99                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
100                 } else if (isdigit(*p)) {
101                         char *end;
102                         int permille = strtoul(p, &end, 10) * 10;
103                         if (*end == '.' && isdigit(*++end)) {
104                                 /* only use first digit */
105                                 permille += *end - '0';
106                                 /* .. and ignore any further digits */
107                                 while (isdigit(*++end))
108                                         ; /* nothing */
109                         }
110                         if (!*end)
111                                 options->dirstat_permille = permille;
112                         else {
113                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
114                                             p);
115                                 ret++;
116                         }
117                 } else {
118                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
119                         ret++;
120                 }
121
122         }
123         string_list_clear(&params, 0);
124         free(params_copy);
125         return ret;
126 }
127
128 static int parse_submodule_params(struct diff_options *options, const char *value)
129 {
130         if (!strcmp(value, "log"))
131                 DIFF_OPT_SET(options, SUBMODULE_LOG);
132         else if (!strcmp(value, "short"))
133                 DIFF_OPT_CLR(options, SUBMODULE_LOG);
134         else
135                 return -1;
136         return 0;
137 }
138
139 static int git_config_rename(const char *var, const char *value)
140 {
141         if (!value)
142                 return DIFF_DETECT_RENAME;
143         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
144                 return  DIFF_DETECT_COPY;
145         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
146 }
147
148 long parse_algorithm_value(const char *value)
149 {
150         if (!value)
151                 return -1;
152         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
153                 return 0;
154         else if (!strcasecmp(value, "minimal"))
155                 return XDF_NEED_MINIMAL;
156         else if (!strcasecmp(value, "patience"))
157                 return XDF_PATIENCE_DIFF;
158         else if (!strcasecmp(value, "histogram"))
159                 return XDF_HISTOGRAM_DIFF;
160         return -1;
161 }
162
163 /*
164  * These are to give UI layer defaults.
165  * The core-level commands such as git-diff-files should
166  * never be affected by the setting of diff.renames
167  * the user happens to have in the configuration file.
168  */
169 int git_diff_ui_config(const char *var, const char *value, void *cb)
170 {
171         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
172                 diff_use_color_default = git_config_colorbool(var, value);
173                 return 0;
174         }
175         if (!strcmp(var, "diff.context")) {
176                 diff_context_default = git_config_int(var, value);
177                 if (diff_context_default < 0)
178                         return -1;
179                 return 0;
180         }
181         if (!strcmp(var, "diff.renames")) {
182                 diff_detect_rename_default = git_config_rename(var, value);
183                 return 0;
184         }
185         if (!strcmp(var, "diff.autorefreshindex")) {
186                 diff_auto_refresh_index = git_config_bool(var, value);
187                 return 0;
188         }
189         if (!strcmp(var, "diff.mnemonicprefix")) {
190                 diff_mnemonic_prefix = git_config_bool(var, value);
191                 return 0;
192         }
193         if (!strcmp(var, "diff.noprefix")) {
194                 diff_no_prefix = git_config_bool(var, value);
195                 return 0;
196         }
197         if (!strcmp(var, "diff.statgraphwidth")) {
198                 diff_stat_graph_width = git_config_int(var, value);
199                 return 0;
200         }
201         if (!strcmp(var, "diff.external"))
202                 return git_config_string(&external_diff_cmd_cfg, var, value);
203         if (!strcmp(var, "diff.wordregex"))
204                 return git_config_string(&diff_word_regex_cfg, var, value);
205         if (!strcmp(var, "diff.orderfile"))
206                 return git_config_pathname(&diff_order_file_cfg, var, value);
207
208         if (!strcmp(var, "diff.ignoresubmodules"))
209                 handle_ignore_submodules_arg(&default_diff_options, value);
210
211         if (!strcmp(var, "diff.submodule")) {
212                 if (parse_submodule_params(&default_diff_options, value))
213                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
214                                 value);
215                 return 0;
216         }
217
218         if (!strcmp(var, "diff.algorithm")) {
219                 diff_algorithm = parse_algorithm_value(value);
220                 if (diff_algorithm < 0)
221                         return -1;
222                 return 0;
223         }
224
225         if (git_color_config(var, value, cb) < 0)
226                 return -1;
227
228         return git_diff_basic_config(var, value, cb);
229 }
230
231 int git_diff_basic_config(const char *var, const char *value, void *cb)
232 {
233         if (!strcmp(var, "diff.renamelimit")) {
234                 diff_rename_limit_default = git_config_int(var, value);
235                 return 0;
236         }
237
238         if (userdiff_config(var, value) < 0)
239                 return -1;
240
241         if (starts_with(var, "diff.color.") || starts_with(var, "color.diff.")) {
242                 int slot = parse_diff_color_slot(var, 11);
243                 if (slot < 0)
244                         return 0;
245                 if (!value)
246                         return config_error_nonbool(var);
247                 color_parse(value, var, diff_colors[slot]);
248                 return 0;
249         }
250
251         /* like GNU diff's --suppress-blank-empty option  */
252         if (!strcmp(var, "diff.suppressblankempty") ||
253                         /* for backwards compatibility */
254                         !strcmp(var, "diff.suppress-blank-empty")) {
255                 diff_suppress_blank_empty = git_config_bool(var, value);
256                 return 0;
257         }
258
259         if (!strcmp(var, "diff.dirstat")) {
260                 struct strbuf errmsg = STRBUF_INIT;
261                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
262                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
263                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
264                                 errmsg.buf);
265                 strbuf_release(&errmsg);
266                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
267                 return 0;
268         }
269
270         if (starts_with(var, "submodule."))
271                 return parse_submodule_config_option(var, value);
272
273         return git_default_config(var, value, cb);
274 }
275
276 static char *quote_two(const char *one, const char *two)
277 {
278         int need_one = quote_c_style(one, NULL, NULL, 1);
279         int need_two = quote_c_style(two, NULL, NULL, 1);
280         struct strbuf res = STRBUF_INIT;
281
282         if (need_one + need_two) {
283                 strbuf_addch(&res, '"');
284                 quote_c_style(one, &res, NULL, 1);
285                 quote_c_style(two, &res, NULL, 1);
286                 strbuf_addch(&res, '"');
287         } else {
288                 strbuf_addstr(&res, one);
289                 strbuf_addstr(&res, two);
290         }
291         return strbuf_detach(&res, NULL);
292 }
293
294 static const char *external_diff(void)
295 {
296         static const char *external_diff_cmd = NULL;
297         static int done_preparing = 0;
298
299         if (done_preparing)
300                 return external_diff_cmd;
301         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
302         if (!external_diff_cmd)
303                 external_diff_cmd = external_diff_cmd_cfg;
304         done_preparing = 1;
305         return external_diff_cmd;
306 }
307
308 static struct diff_tempfile {
309         const char *name; /* filename external diff should read from */
310         char hex[41];
311         char mode[10];
312         char tmp_path[PATH_MAX];
313 } diff_temp[2];
314
315 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
316
317 struct emit_callback {
318         int color_diff;
319         unsigned ws_rule;
320         int blank_at_eof_in_preimage;
321         int blank_at_eof_in_postimage;
322         int lno_in_preimage;
323         int lno_in_postimage;
324         sane_truncate_fn truncate;
325         const char **label_path;
326         struct diff_words_data *diff_words;
327         struct diff_options *opt;
328         int *found_changesp;
329         struct strbuf *header;
330 };
331
332 static int count_lines(const char *data, int size)
333 {
334         int count, ch, completely_empty = 1, nl_just_seen = 0;
335         count = 0;
336         while (0 < size--) {
337                 ch = *data++;
338                 if (ch == '\n') {
339                         count++;
340                         nl_just_seen = 1;
341                         completely_empty = 0;
342                 }
343                 else {
344                         nl_just_seen = 0;
345                         completely_empty = 0;
346                 }
347         }
348         if (completely_empty)
349                 return 0;
350         if (!nl_just_seen)
351                 count++; /* no trailing newline */
352         return count;
353 }
354
355 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
356 {
357         if (!DIFF_FILE_VALID(one)) {
358                 mf->ptr = (char *)""; /* does not matter */
359                 mf->size = 0;
360                 return 0;
361         }
362         else if (diff_populate_filespec(one, 0))
363                 return -1;
364
365         mf->ptr = one->data;
366         mf->size = one->size;
367         return 0;
368 }
369
370 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
371 static unsigned long diff_filespec_size(struct diff_filespec *one)
372 {
373         if (!DIFF_FILE_VALID(one))
374                 return 0;
375         diff_populate_filespec(one, 1);
376         return one->size;
377 }
378
379 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
380 {
381         char *ptr = mf->ptr;
382         long size = mf->size;
383         int cnt = 0;
384
385         if (!size)
386                 return cnt;
387         ptr += size - 1; /* pointing at the very end */
388         if (*ptr != '\n')
389                 ; /* incomplete line */
390         else
391                 ptr--; /* skip the last LF */
392         while (mf->ptr < ptr) {
393                 char *prev_eol;
394                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
395                         if (*prev_eol == '\n')
396                                 break;
397                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
398                         break;
399                 cnt++;
400                 ptr = prev_eol - 1;
401         }
402         return cnt;
403 }
404
405 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
406                                struct emit_callback *ecbdata)
407 {
408         int l1, l2, at;
409         unsigned ws_rule = ecbdata->ws_rule;
410         l1 = count_trailing_blank(mf1, ws_rule);
411         l2 = count_trailing_blank(mf2, ws_rule);
412         if (l2 <= l1) {
413                 ecbdata->blank_at_eof_in_preimage = 0;
414                 ecbdata->blank_at_eof_in_postimage = 0;
415                 return;
416         }
417         at = count_lines(mf1->ptr, mf1->size);
418         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
419
420         at = count_lines(mf2->ptr, mf2->size);
421         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
422 }
423
424 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
425                         int first, const char *line, int len)
426 {
427         int has_trailing_newline, has_trailing_carriage_return;
428         int nofirst;
429         FILE *file = o->file;
430
431         fputs(diff_line_prefix(o), file);
432
433         if (len == 0) {
434                 has_trailing_newline = (first == '\n');
435                 has_trailing_carriage_return = (!has_trailing_newline &&
436                                                 (first == '\r'));
437                 nofirst = has_trailing_newline || has_trailing_carriage_return;
438         } else {
439                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
440                 if (has_trailing_newline)
441                         len--;
442                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
443                 if (has_trailing_carriage_return)
444                         len--;
445                 nofirst = 0;
446         }
447
448         if (len || !nofirst) {
449                 fputs(set, file);
450                 if (!nofirst)
451                         fputc(first, file);
452                 fwrite(line, len, 1, file);
453                 fputs(reset, file);
454         }
455         if (has_trailing_carriage_return)
456                 fputc('\r', file);
457         if (has_trailing_newline)
458                 fputc('\n', file);
459 }
460
461 static void emit_line(struct diff_options *o, const char *set, const char *reset,
462                       const char *line, int len)
463 {
464         emit_line_0(o, set, reset, line[0], line+1, len-1);
465 }
466
467 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
468 {
469         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
470               ecbdata->blank_at_eof_in_preimage &&
471               ecbdata->blank_at_eof_in_postimage &&
472               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
473               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
474                 return 0;
475         return ws_blank_line(line, len, ecbdata->ws_rule);
476 }
477
478 static void emit_add_line(const char *reset,
479                           struct emit_callback *ecbdata,
480                           const char *line, int len)
481 {
482         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
483         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
484
485         if (!*ws)
486                 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
487         else if (new_blank_line_at_eof(ecbdata, line, len))
488                 /* Blank line at EOF - paint '+' as well */
489                 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
490         else {
491                 /* Emit just the prefix, then the rest. */
492                 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
493                 ws_check_emit(line, len, ecbdata->ws_rule,
494                               ecbdata->opt->file, set, reset, ws);
495         }
496 }
497
498 static void emit_hunk_header(struct emit_callback *ecbdata,
499                              const char *line, int len)
500 {
501         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
502         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
503         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
504         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
505         static const char atat[2] = { '@', '@' };
506         const char *cp, *ep;
507         struct strbuf msgbuf = STRBUF_INIT;
508         int org_len = len;
509         int i = 1;
510
511         /*
512          * As a hunk header must begin with "@@ -<old>, +<new> @@",
513          * it always is at least 10 bytes long.
514          */
515         if (len < 10 ||
516             memcmp(line, atat, 2) ||
517             !(ep = memmem(line + 2, len - 2, atat, 2))) {
518                 emit_line(ecbdata->opt, plain, reset, line, len);
519                 return;
520         }
521         ep += 2; /* skip over @@ */
522
523         /* The hunk header in fraginfo color */
524         strbuf_add(&msgbuf, frag, strlen(frag));
525         strbuf_add(&msgbuf, line, ep - line);
526         strbuf_add(&msgbuf, reset, strlen(reset));
527
528         /*
529          * trailing "\r\n"
530          */
531         for ( ; i < 3; i++)
532                 if (line[len - i] == '\r' || line[len - i] == '\n')
533                         len--;
534
535         /* blank before the func header */
536         for (cp = ep; ep - line < len; ep++)
537                 if (*ep != ' ' && *ep != '\t')
538                         break;
539         if (ep != cp) {
540                 strbuf_add(&msgbuf, plain, strlen(plain));
541                 strbuf_add(&msgbuf, cp, ep - cp);
542                 strbuf_add(&msgbuf, reset, strlen(reset));
543         }
544
545         if (ep < line + len) {
546                 strbuf_add(&msgbuf, func, strlen(func));
547                 strbuf_add(&msgbuf, ep, line + len - ep);
548                 strbuf_add(&msgbuf, reset, strlen(reset));
549         }
550
551         strbuf_add(&msgbuf, line + len, org_len - len);
552         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
553         strbuf_release(&msgbuf);
554 }
555
556 static struct diff_tempfile *claim_diff_tempfile(void) {
557         int i;
558         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
559                 if (!diff_temp[i].name)
560                         return diff_temp + i;
561         die("BUG: diff is failing to clean up its tempfiles");
562 }
563
564 static int remove_tempfile_installed;
565
566 static void remove_tempfile(void)
567 {
568         int i;
569         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
570                 if (diff_temp[i].name == diff_temp[i].tmp_path)
571                         unlink_or_warn(diff_temp[i].name);
572                 diff_temp[i].name = NULL;
573         }
574 }
575
576 static void remove_tempfile_on_signal(int signo)
577 {
578         remove_tempfile();
579         sigchain_pop(signo);
580         raise(signo);
581 }
582
583 static void print_line_count(FILE *file, int count)
584 {
585         switch (count) {
586         case 0:
587                 fprintf(file, "0,0");
588                 break;
589         case 1:
590                 fprintf(file, "1");
591                 break;
592         default:
593                 fprintf(file, "1,%d", count);
594                 break;
595         }
596 }
597
598 static void emit_rewrite_lines(struct emit_callback *ecb,
599                                int prefix, const char *data, int size)
600 {
601         const char *endp = NULL;
602         static const char *nneof = " No newline at end of file\n";
603         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
604         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
605
606         while (0 < size) {
607                 int len;
608
609                 endp = memchr(data, '\n', size);
610                 len = endp ? (endp - data + 1) : size;
611                 if (prefix != '+') {
612                         ecb->lno_in_preimage++;
613                         emit_line_0(ecb->opt, old, reset, '-',
614                                     data, len);
615                 } else {
616                         ecb->lno_in_postimage++;
617                         emit_add_line(reset, ecb, data, len);
618                 }
619                 size -= len;
620                 data += len;
621         }
622         if (!endp) {
623                 const char *plain = diff_get_color(ecb->color_diff,
624                                                    DIFF_PLAIN);
625                 putc('\n', ecb->opt->file);
626                 emit_line_0(ecb->opt, plain, reset, '\\',
627                             nneof, strlen(nneof));
628         }
629 }
630
631 static void emit_rewrite_diff(const char *name_a,
632                               const char *name_b,
633                               struct diff_filespec *one,
634                               struct diff_filespec *two,
635                               struct userdiff_driver *textconv_one,
636                               struct userdiff_driver *textconv_two,
637                               struct diff_options *o)
638 {
639         int lc_a, lc_b;
640         const char *name_a_tab, *name_b_tab;
641         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
642         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
643         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
644         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
645         const char *a_prefix, *b_prefix;
646         char *data_one, *data_two;
647         size_t size_one, size_two;
648         struct emit_callback ecbdata;
649         const char *line_prefix = diff_line_prefix(o);
650
651         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
652                 a_prefix = o->b_prefix;
653                 b_prefix = o->a_prefix;
654         } else {
655                 a_prefix = o->a_prefix;
656                 b_prefix = o->b_prefix;
657         }
658
659         name_a += (*name_a == '/');
660         name_b += (*name_b == '/');
661         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
662         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
663
664         strbuf_reset(&a_name);
665         strbuf_reset(&b_name);
666         quote_two_c_style(&a_name, a_prefix, name_a, 0);
667         quote_two_c_style(&b_name, b_prefix, name_b, 0);
668
669         size_one = fill_textconv(textconv_one, one, &data_one);
670         size_two = fill_textconv(textconv_two, two, &data_two);
671
672         memset(&ecbdata, 0, sizeof(ecbdata));
673         ecbdata.color_diff = want_color(o->use_color);
674         ecbdata.found_changesp = &o->found_changes;
675         ecbdata.ws_rule = whitespace_rule(name_b);
676         ecbdata.opt = o;
677         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
678                 mmfile_t mf1, mf2;
679                 mf1.ptr = (char *)data_one;
680                 mf2.ptr = (char *)data_two;
681                 mf1.size = size_one;
682                 mf2.size = size_two;
683                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
684         }
685         ecbdata.lno_in_preimage = 1;
686         ecbdata.lno_in_postimage = 1;
687
688         lc_a = count_lines(data_one, size_one);
689         lc_b = count_lines(data_two, size_two);
690         fprintf(o->file,
691                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
692                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
693                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
694                 line_prefix, fraginfo);
695         if (!o->irreversible_delete)
696                 print_line_count(o->file, lc_a);
697         else
698                 fprintf(o->file, "?,?");
699         fprintf(o->file, " +");
700         print_line_count(o->file, lc_b);
701         fprintf(o->file, " @@%s\n", reset);
702         if (lc_a && !o->irreversible_delete)
703                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
704         if (lc_b)
705                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
706         if (textconv_one)
707                 free((char *)data_one);
708         if (textconv_two)
709                 free((char *)data_two);
710 }
711
712 struct diff_words_buffer {
713         mmfile_t text;
714         long alloc;
715         struct diff_words_orig {
716                 const char *begin, *end;
717         } *orig;
718         int orig_nr, orig_alloc;
719 };
720
721 static void diff_words_append(char *line, unsigned long len,
722                 struct diff_words_buffer *buffer)
723 {
724         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
725         line++;
726         len--;
727         memcpy(buffer->text.ptr + buffer->text.size, line, len);
728         buffer->text.size += len;
729         buffer->text.ptr[buffer->text.size] = '\0';
730 }
731
732 struct diff_words_style_elem {
733         const char *prefix;
734         const char *suffix;
735         const char *color; /* NULL; filled in by the setup code if
736                             * color is enabled */
737 };
738
739 struct diff_words_style {
740         enum diff_words_type type;
741         struct diff_words_style_elem new, old, ctx;
742         const char *newline;
743 };
744
745 static struct diff_words_style diff_words_styles[] = {
746         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
747         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
748         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
749 };
750
751 struct diff_words_data {
752         struct diff_words_buffer minus, plus;
753         const char *current_plus;
754         int last_minus;
755         struct diff_options *opt;
756         regex_t *word_regex;
757         enum diff_words_type type;
758         struct diff_words_style *style;
759 };
760
761 static int fn_out_diff_words_write_helper(FILE *fp,
762                                           struct diff_words_style_elem *st_el,
763                                           const char *newline,
764                                           size_t count, const char *buf,
765                                           const char *line_prefix)
766 {
767         int print = 0;
768
769         while (count) {
770                 char *p = memchr(buf, '\n', count);
771                 if (print)
772                         fputs(line_prefix, fp);
773                 if (p != buf) {
774                         if (st_el->color && fputs(st_el->color, fp) < 0)
775                                 return -1;
776                         if (fputs(st_el->prefix, fp) < 0 ||
777                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
778                             fputs(st_el->suffix, fp) < 0)
779                                 return -1;
780                         if (st_el->color && *st_el->color
781                             && fputs(GIT_COLOR_RESET, fp) < 0)
782                                 return -1;
783                 }
784                 if (!p)
785                         return 0;
786                 if (fputs(newline, fp) < 0)
787                         return -1;
788                 count -= p + 1 - buf;
789                 buf = p + 1;
790                 print = 1;
791         }
792         return 0;
793 }
794
795 /*
796  * '--color-words' algorithm can be described as:
797  *
798  *   1. collect a the minus/plus lines of a diff hunk, divided into
799  *      minus-lines and plus-lines;
800  *
801  *   2. break both minus-lines and plus-lines into words and
802  *      place them into two mmfile_t with one word for each line;
803  *
804  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
805  *
806  * And for the common parts of the both file, we output the plus side text.
807  * diff_words->current_plus is used to trace the current position of the plus file
808  * which printed. diff_words->last_minus is used to trace the last minus word
809  * printed.
810  *
811  * For '--graph' to work with '--color-words', we need to output the graph prefix
812  * on each line of color words output. Generally, there are two conditions on
813  * which we should output the prefix.
814  *
815  *   1. diff_words->last_minus == 0 &&
816  *      diff_words->current_plus == diff_words->plus.text.ptr
817  *
818  *      that is: the plus text must start as a new line, and if there is no minus
819  *      word printed, a graph prefix must be printed.
820  *
821  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
822  *      *(diff_words->current_plus - 1) == '\n'
823  *
824  *      that is: a graph prefix must be printed following a '\n'
825  */
826 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
827 {
828         if ((diff_words->last_minus == 0 &&
829                 diff_words->current_plus == diff_words->plus.text.ptr) ||
830                 (diff_words->current_plus > diff_words->plus.text.ptr &&
831                 *(diff_words->current_plus - 1) == '\n')) {
832                 return 1;
833         } else {
834                 return 0;
835         }
836 }
837
838 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
839 {
840         struct diff_words_data *diff_words = priv;
841         struct diff_words_style *style = diff_words->style;
842         int minus_first, minus_len, plus_first, plus_len;
843         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
844         struct diff_options *opt = diff_words->opt;
845         const char *line_prefix;
846
847         if (line[0] != '@' || parse_hunk_header(line, len,
848                         &minus_first, &minus_len, &plus_first, &plus_len))
849                 return;
850
851         assert(opt);
852         line_prefix = diff_line_prefix(opt);
853
854         /* POSIX requires that first be decremented by one if len == 0... */
855         if (minus_len) {
856                 minus_begin = diff_words->minus.orig[minus_first].begin;
857                 minus_end =
858                         diff_words->minus.orig[minus_first + minus_len - 1].end;
859         } else
860                 minus_begin = minus_end =
861                         diff_words->minus.orig[minus_first].end;
862
863         if (plus_len) {
864                 plus_begin = diff_words->plus.orig[plus_first].begin;
865                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
866         } else
867                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
868
869         if (color_words_output_graph_prefix(diff_words)) {
870                 fputs(line_prefix, diff_words->opt->file);
871         }
872         if (diff_words->current_plus != plus_begin) {
873                 fn_out_diff_words_write_helper(diff_words->opt->file,
874                                 &style->ctx, style->newline,
875                                 plus_begin - diff_words->current_plus,
876                                 diff_words->current_plus, line_prefix);
877                 if (*(plus_begin - 1) == '\n')
878                         fputs(line_prefix, diff_words->opt->file);
879         }
880         if (minus_begin != minus_end) {
881                 fn_out_diff_words_write_helper(diff_words->opt->file,
882                                 &style->old, style->newline,
883                                 minus_end - minus_begin, minus_begin,
884                                 line_prefix);
885         }
886         if (plus_begin != plus_end) {
887                 fn_out_diff_words_write_helper(diff_words->opt->file,
888                                 &style->new, style->newline,
889                                 plus_end - plus_begin, plus_begin,
890                                 line_prefix);
891         }
892
893         diff_words->current_plus = plus_end;
894         diff_words->last_minus = minus_first;
895 }
896
897 /* This function starts looking at *begin, and returns 0 iff a word was found. */
898 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
899                 int *begin, int *end)
900 {
901         if (word_regex && *begin < buffer->size) {
902                 regmatch_t match[1];
903                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
904                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
905                                         '\n', match[0].rm_eo - match[0].rm_so);
906                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
907                         *begin += match[0].rm_so;
908                         return *begin >= *end;
909                 }
910                 return -1;
911         }
912
913         /* find the next word */
914         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
915                 (*begin)++;
916         if (*begin >= buffer->size)
917                 return -1;
918
919         /* find the end of the word */
920         *end = *begin + 1;
921         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
922                 (*end)++;
923
924         return 0;
925 }
926
927 /*
928  * This function splits the words in buffer->text, stores the list with
929  * newline separator into out, and saves the offsets of the original words
930  * in buffer->orig.
931  */
932 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
933                 regex_t *word_regex)
934 {
935         int i, j;
936         long alloc = 0;
937
938         out->size = 0;
939         out->ptr = NULL;
940
941         /* fake an empty "0th" word */
942         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
943         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
944         buffer->orig_nr = 1;
945
946         for (i = 0; i < buffer->text.size; i++) {
947                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
948                         return;
949
950                 /* store original boundaries */
951                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
952                                 buffer->orig_alloc);
953                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
954                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
955                 buffer->orig_nr++;
956
957                 /* store one word */
958                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
959                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
960                 out->ptr[out->size + j - i] = '\n';
961                 out->size += j - i + 1;
962
963                 i = j - 1;
964         }
965 }
966
967 /* this executes the word diff on the accumulated buffers */
968 static void diff_words_show(struct diff_words_data *diff_words)
969 {
970         xpparam_t xpp;
971         xdemitconf_t xecfg;
972         mmfile_t minus, plus;
973         struct diff_words_style *style = diff_words->style;
974
975         struct diff_options *opt = diff_words->opt;
976         const char *line_prefix;
977
978         assert(opt);
979         line_prefix = diff_line_prefix(opt);
980
981         /* special case: only removal */
982         if (!diff_words->plus.text.size) {
983                 fputs(line_prefix, diff_words->opt->file);
984                 fn_out_diff_words_write_helper(diff_words->opt->file,
985                         &style->old, style->newline,
986                         diff_words->minus.text.size,
987                         diff_words->minus.text.ptr, line_prefix);
988                 diff_words->minus.text.size = 0;
989                 return;
990         }
991
992         diff_words->current_plus = diff_words->plus.text.ptr;
993         diff_words->last_minus = 0;
994
995         memset(&xpp, 0, sizeof(xpp));
996         memset(&xecfg, 0, sizeof(xecfg));
997         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
998         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
999         xpp.flags = 0;
1000         /* as only the hunk header will be parsed, we need a 0-context */
1001         xecfg.ctxlen = 0;
1002         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1003                       &xpp, &xecfg);
1004         free(minus.ptr);
1005         free(plus.ptr);
1006         if (diff_words->current_plus != diff_words->plus.text.ptr +
1007                         diff_words->plus.text.size) {
1008                 if (color_words_output_graph_prefix(diff_words))
1009                         fputs(line_prefix, diff_words->opt->file);
1010                 fn_out_diff_words_write_helper(diff_words->opt->file,
1011                         &style->ctx, style->newline,
1012                         diff_words->plus.text.ptr + diff_words->plus.text.size
1013                         - diff_words->current_plus, diff_words->current_plus,
1014                         line_prefix);
1015         }
1016         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1017 }
1018
1019 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1020 static void diff_words_flush(struct emit_callback *ecbdata)
1021 {
1022         if (ecbdata->diff_words->minus.text.size ||
1023             ecbdata->diff_words->plus.text.size)
1024                 diff_words_show(ecbdata->diff_words);
1025 }
1026
1027 static void diff_filespec_load_driver(struct diff_filespec *one)
1028 {
1029         /* Use already-loaded driver */
1030         if (one->driver)
1031                 return;
1032
1033         if (S_ISREG(one->mode))
1034                 one->driver = userdiff_find_by_path(one->path);
1035
1036         /* Fallback to default settings */
1037         if (!one->driver)
1038                 one->driver = userdiff_find_by_name("default");
1039 }
1040
1041 static const char *userdiff_word_regex(struct diff_filespec *one)
1042 {
1043         diff_filespec_load_driver(one);
1044         return one->driver->word_regex;
1045 }
1046
1047 static void init_diff_words_data(struct emit_callback *ecbdata,
1048                                  struct diff_options *orig_opts,
1049                                  struct diff_filespec *one,
1050                                  struct diff_filespec *two)
1051 {
1052         int i;
1053         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1054         memcpy(o, orig_opts, sizeof(struct diff_options));
1055
1056         ecbdata->diff_words =
1057                 xcalloc(1, sizeof(struct diff_words_data));
1058         ecbdata->diff_words->type = o->word_diff;
1059         ecbdata->diff_words->opt = o;
1060         if (!o->word_regex)
1061                 o->word_regex = userdiff_word_regex(one);
1062         if (!o->word_regex)
1063                 o->word_regex = userdiff_word_regex(two);
1064         if (!o->word_regex)
1065                 o->word_regex = diff_word_regex_cfg;
1066         if (o->word_regex) {
1067                 ecbdata->diff_words->word_regex = (regex_t *)
1068                         xmalloc(sizeof(regex_t));
1069                 if (regcomp(ecbdata->diff_words->word_regex,
1070                             o->word_regex,
1071                             REG_EXTENDED | REG_NEWLINE))
1072                         die ("Invalid regular expression: %s",
1073                              o->word_regex);
1074         }
1075         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1076                 if (o->word_diff == diff_words_styles[i].type) {
1077                         ecbdata->diff_words->style =
1078                                 &diff_words_styles[i];
1079                         break;
1080                 }
1081         }
1082         if (want_color(o->use_color)) {
1083                 struct diff_words_style *st = ecbdata->diff_words->style;
1084                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1085                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1086                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1087         }
1088 }
1089
1090 static void free_diff_words_data(struct emit_callback *ecbdata)
1091 {
1092         if (ecbdata->diff_words) {
1093                 diff_words_flush(ecbdata);
1094                 free (ecbdata->diff_words->opt);
1095                 free (ecbdata->diff_words->minus.text.ptr);
1096                 free (ecbdata->diff_words->minus.orig);
1097                 free (ecbdata->diff_words->plus.text.ptr);
1098                 free (ecbdata->diff_words->plus.orig);
1099                 if (ecbdata->diff_words->word_regex) {
1100                         regfree(ecbdata->diff_words->word_regex);
1101                         free(ecbdata->diff_words->word_regex);
1102                 }
1103                 free(ecbdata->diff_words);
1104                 ecbdata->diff_words = NULL;
1105         }
1106 }
1107
1108 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1109 {
1110         if (want_color(diff_use_color))
1111                 return diff_colors[ix];
1112         return "";
1113 }
1114
1115 const char *diff_line_prefix(struct diff_options *opt)
1116 {
1117         struct strbuf *msgbuf;
1118         if (!opt->output_prefix)
1119                 return "";
1120
1121         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1122         return msgbuf->buf;
1123 }
1124
1125 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1126 {
1127         const char *cp;
1128         unsigned long allot;
1129         size_t l = len;
1130
1131         if (ecb->truncate)
1132                 return ecb->truncate(line, len);
1133         cp = line;
1134         allot = l;
1135         while (0 < l) {
1136                 (void) utf8_width(&cp, &l);
1137                 if (!cp)
1138                         break; /* truncated in the middle? */
1139         }
1140         return allot - l;
1141 }
1142
1143 static void find_lno(const char *line, struct emit_callback *ecbdata)
1144 {
1145         const char *p;
1146         ecbdata->lno_in_preimage = 0;
1147         ecbdata->lno_in_postimage = 0;
1148         p = strchr(line, '-');
1149         if (!p)
1150                 return; /* cannot happen */
1151         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1152         p = strchr(p, '+');
1153         if (!p)
1154                 return; /* cannot happen */
1155         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1156 }
1157
1158 static void fn_out_consume(void *priv, char *line, unsigned long len)
1159 {
1160         struct emit_callback *ecbdata = priv;
1161         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1162         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1163         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1164         struct diff_options *o = ecbdata->opt;
1165         const char *line_prefix = diff_line_prefix(o);
1166
1167         if (ecbdata->header) {
1168                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1169                 strbuf_reset(ecbdata->header);
1170                 ecbdata->header = NULL;
1171         }
1172         *(ecbdata->found_changesp) = 1;
1173
1174         if (ecbdata->label_path[0]) {
1175                 const char *name_a_tab, *name_b_tab;
1176
1177                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1178                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1179
1180                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1181                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1182                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1183                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1184                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1185         }
1186
1187         if (diff_suppress_blank_empty
1188             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1189                 line[0] = '\n';
1190                 len = 1;
1191         }
1192
1193         if (line[0] == '@') {
1194                 if (ecbdata->diff_words)
1195                         diff_words_flush(ecbdata);
1196                 len = sane_truncate_line(ecbdata, line, len);
1197                 find_lno(line, ecbdata);
1198                 emit_hunk_header(ecbdata, line, len);
1199                 if (line[len-1] != '\n')
1200                         putc('\n', ecbdata->opt->file);
1201                 return;
1202         }
1203
1204         if (len < 1) {
1205                 emit_line(ecbdata->opt, reset, reset, line, len);
1206                 if (ecbdata->diff_words
1207                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1208                         fputs("~\n", ecbdata->opt->file);
1209                 return;
1210         }
1211
1212         if (ecbdata->diff_words) {
1213                 if (line[0] == '-') {
1214                         diff_words_append(line, len,
1215                                           &ecbdata->diff_words->minus);
1216                         return;
1217                 } else if (line[0] == '+') {
1218                         diff_words_append(line, len,
1219                                           &ecbdata->diff_words->plus);
1220                         return;
1221                 } else if (starts_with(line, "\\ ")) {
1222                         /*
1223                          * Eat the "no newline at eof" marker as if we
1224                          * saw a "+" or "-" line with nothing on it,
1225                          * and return without diff_words_flush() to
1226                          * defer processing. If this is the end of
1227                          * preimage, more "+" lines may come after it.
1228                          */
1229                         return;
1230                 }
1231                 diff_words_flush(ecbdata);
1232                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1233                         emit_line(ecbdata->opt, plain, reset, line, len);
1234                         fputs("~\n", ecbdata->opt->file);
1235                 } else {
1236                         /*
1237                          * Skip the prefix character, if any.  With
1238                          * diff_suppress_blank_empty, there may be
1239                          * none.
1240                          */
1241                         if (line[0] != '\n') {
1242                               line++;
1243                               len--;
1244                         }
1245                         emit_line(ecbdata->opt, plain, reset, line, len);
1246                 }
1247                 return;
1248         }
1249
1250         if (line[0] != '+') {
1251                 const char *color =
1252                         diff_get_color(ecbdata->color_diff,
1253                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1254                 ecbdata->lno_in_preimage++;
1255                 if (line[0] == ' ')
1256                         ecbdata->lno_in_postimage++;
1257                 emit_line(ecbdata->opt, color, reset, line, len);
1258         } else {
1259                 ecbdata->lno_in_postimage++;
1260                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1261         }
1262 }
1263
1264 static char *pprint_rename(const char *a, const char *b)
1265 {
1266         const char *old = a;
1267         const char *new = b;
1268         struct strbuf name = STRBUF_INIT;
1269         int pfx_length, sfx_length;
1270         int pfx_adjust_for_slash;
1271         int len_a = strlen(a);
1272         int len_b = strlen(b);
1273         int a_midlen, b_midlen;
1274         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1275         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1276
1277         if (qlen_a || qlen_b) {
1278                 quote_c_style(a, &name, NULL, 0);
1279                 strbuf_addstr(&name, " => ");
1280                 quote_c_style(b, &name, NULL, 0);
1281                 return strbuf_detach(&name, NULL);
1282         }
1283
1284         /* Find common prefix */
1285         pfx_length = 0;
1286         while (*old && *new && *old == *new) {
1287                 if (*old == '/')
1288                         pfx_length = old - a + 1;
1289                 old++;
1290                 new++;
1291         }
1292
1293         /* Find common suffix */
1294         old = a + len_a;
1295         new = b + len_b;
1296         sfx_length = 0;
1297         /*
1298          * If there is a common prefix, it must end in a slash.  In
1299          * that case we let this loop run 1 into the prefix to see the
1300          * same slash.
1301          *
1302          * If there is no common prefix, we cannot do this as it would
1303          * underrun the input strings.
1304          */
1305         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1306         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1307                b + pfx_length - pfx_adjust_for_slash <= new &&
1308                *old == *new) {
1309                 if (*old == '/')
1310                         sfx_length = len_a - (old - a);
1311                 old--;
1312                 new--;
1313         }
1314
1315         /*
1316          * pfx{mid-a => mid-b}sfx
1317          * {pfx-a => pfx-b}sfx
1318          * pfx{sfx-a => sfx-b}
1319          * name-a => name-b
1320          */
1321         a_midlen = len_a - pfx_length - sfx_length;
1322         b_midlen = len_b - pfx_length - sfx_length;
1323         if (a_midlen < 0)
1324                 a_midlen = 0;
1325         if (b_midlen < 0)
1326                 b_midlen = 0;
1327
1328         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1329         if (pfx_length + sfx_length) {
1330                 strbuf_add(&name, a, pfx_length);
1331                 strbuf_addch(&name, '{');
1332         }
1333         strbuf_add(&name, a + pfx_length, a_midlen);
1334         strbuf_addstr(&name, " => ");
1335         strbuf_add(&name, b + pfx_length, b_midlen);
1336         if (pfx_length + sfx_length) {
1337                 strbuf_addch(&name, '}');
1338                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1339         }
1340         return strbuf_detach(&name, NULL);
1341 }
1342
1343 struct diffstat_t {
1344         int nr;
1345         int alloc;
1346         struct diffstat_file {
1347                 char *from_name;
1348                 char *name;
1349                 char *print_name;
1350                 unsigned is_unmerged:1;
1351                 unsigned is_binary:1;
1352                 unsigned is_renamed:1;
1353                 unsigned is_interesting:1;
1354                 uintmax_t added, deleted;
1355         } **files;
1356 };
1357
1358 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1359                                           const char *name_a,
1360                                           const char *name_b)
1361 {
1362         struct diffstat_file *x;
1363         x = xcalloc(sizeof (*x), 1);
1364         if (diffstat->nr == diffstat->alloc) {
1365                 diffstat->alloc = alloc_nr(diffstat->alloc);
1366                 diffstat->files = xrealloc(diffstat->files,
1367                                 diffstat->alloc * sizeof(x));
1368         }
1369         diffstat->files[diffstat->nr++] = x;
1370         if (name_b) {
1371                 x->from_name = xstrdup(name_a);
1372                 x->name = xstrdup(name_b);
1373                 x->is_renamed = 1;
1374         }
1375         else {
1376                 x->from_name = NULL;
1377                 x->name = xstrdup(name_a);
1378         }
1379         return x;
1380 }
1381
1382 static void diffstat_consume(void *priv, char *line, unsigned long len)
1383 {
1384         struct diffstat_t *diffstat = priv;
1385         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1386
1387         if (line[0] == '+')
1388                 x->added++;
1389         else if (line[0] == '-')
1390                 x->deleted++;
1391 }
1392
1393 const char mime_boundary_leader[] = "------------";
1394
1395 static int scale_linear(int it, int width, int max_change)
1396 {
1397         if (!it)
1398                 return 0;
1399         /*
1400          * make sure that at least one '-' or '+' is printed if
1401          * there is any change to this path. The easiest way is to
1402          * scale linearly as if the alloted width is one column shorter
1403          * than it is, and then add 1 to the result.
1404          */
1405         return 1 + (it * (width - 1) / max_change);
1406 }
1407
1408 static void show_name(FILE *file,
1409                       const char *prefix, const char *name, int len)
1410 {
1411         fprintf(file, " %s%-*s |", prefix, len, name);
1412 }
1413
1414 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1415 {
1416         if (cnt <= 0)
1417                 return;
1418         fprintf(file, "%s", set);
1419         while (cnt--)
1420                 putc(ch, file);
1421         fprintf(file, "%s", reset);
1422 }
1423
1424 static void fill_print_name(struct diffstat_file *file)
1425 {
1426         char *pname;
1427
1428         if (file->print_name)
1429                 return;
1430
1431         if (!file->is_renamed) {
1432                 struct strbuf buf = STRBUF_INIT;
1433                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1434                         pname = strbuf_detach(&buf, NULL);
1435                 } else {
1436                         pname = file->name;
1437                         strbuf_release(&buf);
1438                 }
1439         } else {
1440                 pname = pprint_rename(file->from_name, file->name);
1441         }
1442         file->print_name = pname;
1443 }
1444
1445 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1446 {
1447         struct strbuf sb = STRBUF_INIT;
1448         int ret;
1449
1450         if (!files) {
1451                 assert(insertions == 0 && deletions == 0);
1452                 return fprintf(fp, "%s\n", " 0 files changed");
1453         }
1454
1455         strbuf_addf(&sb,
1456                     (files == 1) ? " %d file changed" : " %d files changed",
1457                     files);
1458
1459         /*
1460          * For binary diff, the caller may want to print "x files
1461          * changed" with insertions == 0 && deletions == 0.
1462          *
1463          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1464          * is probably less confusing (i.e skip over "2 files changed
1465          * but nothing about added/removed lines? Is this a bug in Git?").
1466          */
1467         if (insertions || deletions == 0) {
1468                 /*
1469                  * TRANSLATORS: "+" in (+) is a line addition marker;
1470                  * do not translate it.
1471                  */
1472                 strbuf_addf(&sb,
1473                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1474                             insertions);
1475         }
1476
1477         if (deletions || insertions == 0) {
1478                 /*
1479                  * TRANSLATORS: "-" in (-) is a line removal marker;
1480                  * do not translate it.
1481                  */
1482                 strbuf_addf(&sb,
1483                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1484                             deletions);
1485         }
1486         strbuf_addch(&sb, '\n');
1487         ret = fputs(sb.buf, fp);
1488         strbuf_release(&sb);
1489         return ret;
1490 }
1491
1492 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1493 {
1494         int i, len, add, del, adds = 0, dels = 0;
1495         uintmax_t max_change = 0, max_len = 0;
1496         int total_files = data->nr, count;
1497         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1498         const char *reset, *add_c, *del_c;
1499         const char *line_prefix = "";
1500         int extra_shown = 0;
1501
1502         if (data->nr == 0)
1503                 return;
1504
1505         line_prefix = diff_line_prefix(options);
1506         count = options->stat_count ? options->stat_count : data->nr;
1507
1508         reset = diff_get_color_opt(options, DIFF_RESET);
1509         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1510         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1511
1512         /*
1513          * Find the longest filename and max number of changes
1514          */
1515         for (i = 0; (i < count) && (i < data->nr); i++) {
1516                 struct diffstat_file *file = data->files[i];
1517                 uintmax_t change = file->added + file->deleted;
1518
1519                 if (!file->is_interesting && (change == 0)) {
1520                         count++; /* not shown == room for one more */
1521                         continue;
1522                 }
1523                 fill_print_name(file);
1524                 len = strlen(file->print_name);
1525                 if (max_len < len)
1526                         max_len = len;
1527
1528                 if (file->is_unmerged) {
1529                         /* "Unmerged" is 8 characters */
1530                         bin_width = bin_width < 8 ? 8 : bin_width;
1531                         continue;
1532                 }
1533                 if (file->is_binary) {
1534                         /* "Bin XXX -> YYY bytes" */
1535                         int w = 14 + decimal_width(file->added)
1536                                 + decimal_width(file->deleted);
1537                         bin_width = bin_width < w ? w : bin_width;
1538                         /* Display change counts aligned with "Bin" */
1539                         number_width = 3;
1540                         continue;
1541                 }
1542
1543                 if (max_change < change)
1544                         max_change = change;
1545         }
1546         count = i; /* where we can stop scanning in data->files[] */
1547
1548         /*
1549          * We have width = stat_width or term_columns() columns total.
1550          * We want a maximum of min(max_len, stat_name_width) for the name part.
1551          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1552          * We also need 1 for " " and 4 + decimal_width(max_change)
1553          * for " | NNNN " and one the empty column at the end, altogether
1554          * 6 + decimal_width(max_change).
1555          *
1556          * If there's not enough space, we will use the smaller of
1557          * stat_name_width (if set) and 5/8*width for the filename,
1558          * and the rest for constant elements + graph part, but no more
1559          * than stat_graph_width for the graph part.
1560          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1561          * for the standard terminal size).
1562          *
1563          * In other words: stat_width limits the maximum width, and
1564          * stat_name_width fixes the maximum width of the filename,
1565          * and is also used to divide available columns if there
1566          * aren't enough.
1567          *
1568          * Binary files are displayed with "Bin XXX -> YYY bytes"
1569          * instead of the change count and graph. This part is treated
1570          * similarly to the graph part, except that it is not
1571          * "scaled". If total width is too small to accommodate the
1572          * guaranteed minimum width of the filename part and the
1573          * separators and this message, this message will "overflow"
1574          * making the line longer than the maximum width.
1575          */
1576
1577         if (options->stat_width == -1)
1578                 width = term_columns() - options->output_prefix_length;
1579         else
1580                 width = options->stat_width ? options->stat_width : 80;
1581         number_width = decimal_width(max_change) > number_width ?
1582                 decimal_width(max_change) : number_width;
1583
1584         if (options->stat_graph_width == -1)
1585                 options->stat_graph_width = diff_stat_graph_width;
1586
1587         /*
1588          * Guarantee 3/8*16==6 for the graph part
1589          * and 5/8*16==10 for the filename part
1590          */
1591         if (width < 16 + 6 + number_width)
1592                 width = 16 + 6 + number_width;
1593
1594         /*
1595          * First assign sizes that are wanted, ignoring available width.
1596          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1597          * starting from "XXX" should fit in graph_width.
1598          */
1599         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1600         if (options->stat_graph_width &&
1601             options->stat_graph_width < graph_width)
1602                 graph_width = options->stat_graph_width;
1603
1604         name_width = (options->stat_name_width > 0 &&
1605                       options->stat_name_width < max_len) ?
1606                 options->stat_name_width : max_len;
1607
1608         /*
1609          * Adjust adjustable widths not to exceed maximum width
1610          */
1611         if (name_width + number_width + 6 + graph_width > width) {
1612                 if (graph_width > width * 3/8 - number_width - 6) {
1613                         graph_width = width * 3/8 - number_width - 6;
1614                         if (graph_width < 6)
1615                                 graph_width = 6;
1616                 }
1617
1618                 if (options->stat_graph_width &&
1619                     graph_width > options->stat_graph_width)
1620                         graph_width = options->stat_graph_width;
1621                 if (name_width > width - number_width - 6 - graph_width)
1622                         name_width = width - number_width - 6 - graph_width;
1623                 else
1624                         graph_width = width - number_width - 6 - name_width;
1625         }
1626
1627         /*
1628          * From here name_width is the width of the name area,
1629          * and graph_width is the width of the graph area.
1630          * max_change is used to scale graph properly.
1631          */
1632         for (i = 0; i < count; i++) {
1633                 const char *prefix = "";
1634                 struct diffstat_file *file = data->files[i];
1635                 char *name = file->print_name;
1636                 uintmax_t added = file->added;
1637                 uintmax_t deleted = file->deleted;
1638                 int name_len;
1639
1640                 if (!file->is_interesting && (added + deleted == 0))
1641                         continue;
1642
1643                 /*
1644                  * "scale" the filename
1645                  */
1646                 len = name_width;
1647                 name_len = strlen(name);
1648                 if (name_width < name_len) {
1649                         char *slash;
1650                         prefix = "...";
1651                         len -= 3;
1652                         name += name_len - len;
1653                         slash = strchr(name, '/');
1654                         if (slash)
1655                                 name = slash;
1656                 }
1657
1658                 if (file->is_binary) {
1659                         fprintf(options->file, "%s", line_prefix);
1660                         show_name(options->file, prefix, name, len);
1661                         fprintf(options->file, " %*s", number_width, "Bin");
1662                         if (!added && !deleted) {
1663                                 putc('\n', options->file);
1664                                 continue;
1665                         }
1666                         fprintf(options->file, " %s%"PRIuMAX"%s",
1667                                 del_c, deleted, reset);
1668                         fprintf(options->file, " -> ");
1669                         fprintf(options->file, "%s%"PRIuMAX"%s",
1670                                 add_c, added, reset);
1671                         fprintf(options->file, " bytes");
1672                         fprintf(options->file, "\n");
1673                         continue;
1674                 }
1675                 else if (file->is_unmerged) {
1676                         fprintf(options->file, "%s", line_prefix);
1677                         show_name(options->file, prefix, name, len);
1678                         fprintf(options->file, " Unmerged\n");
1679                         continue;
1680                 }
1681
1682                 /*
1683                  * scale the add/delete
1684                  */
1685                 add = added;
1686                 del = deleted;
1687
1688                 if (graph_width <= max_change) {
1689                         int total = scale_linear(add + del, graph_width, max_change);
1690                         if (total < 2 && add && del)
1691                                 /* width >= 2 due to the sanity check */
1692                                 total = 2;
1693                         if (add < del) {
1694                                 add = scale_linear(add, graph_width, max_change);
1695                                 del = total - add;
1696                         } else {
1697                                 del = scale_linear(del, graph_width, max_change);
1698                                 add = total - del;
1699                         }
1700                 }
1701                 fprintf(options->file, "%s", line_prefix);
1702                 show_name(options->file, prefix, name, len);
1703                 fprintf(options->file, " %*"PRIuMAX"%s",
1704                         number_width, added + deleted,
1705                         added + deleted ? " " : "");
1706                 show_graph(options->file, '+', add, add_c, reset);
1707                 show_graph(options->file, '-', del, del_c, reset);
1708                 fprintf(options->file, "\n");
1709         }
1710
1711         for (i = 0; i < data->nr; i++) {
1712                 struct diffstat_file *file = data->files[i];
1713                 uintmax_t added = file->added;
1714                 uintmax_t deleted = file->deleted;
1715
1716                 if (file->is_unmerged ||
1717                     (!file->is_interesting && (added + deleted == 0))) {
1718                         total_files--;
1719                         continue;
1720                 }
1721
1722                 if (!file->is_binary) {
1723                         adds += added;
1724                         dels += deleted;
1725                 }
1726                 if (i < count)
1727                         continue;
1728                 if (!extra_shown)
1729                         fprintf(options->file, "%s ...\n", line_prefix);
1730                 extra_shown = 1;
1731         }
1732         fprintf(options->file, "%s", line_prefix);
1733         print_stat_summary(options->file, total_files, adds, dels);
1734 }
1735
1736 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1737 {
1738         int i, adds = 0, dels = 0, total_files = data->nr;
1739
1740         if (data->nr == 0)
1741                 return;
1742
1743         for (i = 0; i < data->nr; i++) {
1744                 int added = data->files[i]->added;
1745                 int deleted= data->files[i]->deleted;
1746
1747                 if (data->files[i]->is_unmerged ||
1748                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1749                         total_files--;
1750                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1751                         adds += added;
1752                         dels += deleted;
1753                 }
1754         }
1755         fprintf(options->file, "%s", diff_line_prefix(options));
1756         print_stat_summary(options->file, total_files, adds, dels);
1757 }
1758
1759 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1760 {
1761         int i;
1762
1763         if (data->nr == 0)
1764                 return;
1765
1766         for (i = 0; i < data->nr; i++) {
1767                 struct diffstat_file *file = data->files[i];
1768
1769                 fprintf(options->file, "%s", diff_line_prefix(options));
1770
1771                 if (file->is_binary)
1772                         fprintf(options->file, "-\t-\t");
1773                 else
1774                         fprintf(options->file,
1775                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1776                                 file->added, file->deleted);
1777                 if (options->line_termination) {
1778                         fill_print_name(file);
1779                         if (!file->is_renamed)
1780                                 write_name_quoted(file->name, options->file,
1781                                                   options->line_termination);
1782                         else {
1783                                 fputs(file->print_name, options->file);
1784                                 putc(options->line_termination, options->file);
1785                         }
1786                 } else {
1787                         if (file->is_renamed) {
1788                                 putc('\0', options->file);
1789                                 write_name_quoted(file->from_name, options->file, '\0');
1790                         }
1791                         write_name_quoted(file->name, options->file, '\0');
1792                 }
1793         }
1794 }
1795
1796 struct dirstat_file {
1797         const char *name;
1798         unsigned long changed;
1799 };
1800
1801 struct dirstat_dir {
1802         struct dirstat_file *files;
1803         int alloc, nr, permille, cumulative;
1804 };
1805
1806 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1807                 unsigned long changed, const char *base, int baselen)
1808 {
1809         unsigned long this_dir = 0;
1810         unsigned int sources = 0;
1811         const char *line_prefix = diff_line_prefix(opt);
1812
1813         while (dir->nr) {
1814                 struct dirstat_file *f = dir->files;
1815                 int namelen = strlen(f->name);
1816                 unsigned long this;
1817                 char *slash;
1818
1819                 if (namelen < baselen)
1820                         break;
1821                 if (memcmp(f->name, base, baselen))
1822                         break;
1823                 slash = strchr(f->name + baselen, '/');
1824                 if (slash) {
1825                         int newbaselen = slash + 1 - f->name;
1826                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1827                         sources++;
1828                 } else {
1829                         this = f->changed;
1830                         dir->files++;
1831                         dir->nr--;
1832                         sources += 2;
1833                 }
1834                 this_dir += this;
1835         }
1836
1837         /*
1838          * We don't report dirstat's for
1839          *  - the top level
1840          *  - or cases where everything came from a single directory
1841          *    under this directory (sources == 1).
1842          */
1843         if (baselen && sources != 1) {
1844                 if (this_dir) {
1845                         int permille = this_dir * 1000 / changed;
1846                         if (permille >= dir->permille) {
1847                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1848                                         permille / 10, permille % 10, baselen, base);
1849                                 if (!dir->cumulative)
1850                                         return 0;
1851                         }
1852                 }
1853         }
1854         return this_dir;
1855 }
1856
1857 static int dirstat_compare(const void *_a, const void *_b)
1858 {
1859         const struct dirstat_file *a = _a;
1860         const struct dirstat_file *b = _b;
1861         return strcmp(a->name, b->name);
1862 }
1863
1864 static void show_dirstat(struct diff_options *options)
1865 {
1866         int i;
1867         unsigned long changed;
1868         struct dirstat_dir dir;
1869         struct diff_queue_struct *q = &diff_queued_diff;
1870
1871         dir.files = NULL;
1872         dir.alloc = 0;
1873         dir.nr = 0;
1874         dir.permille = options->dirstat_permille;
1875         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1876
1877         changed = 0;
1878         for (i = 0; i < q->nr; i++) {
1879                 struct diff_filepair *p = q->queue[i];
1880                 const char *name;
1881                 unsigned long copied, added, damage;
1882                 int content_changed;
1883
1884                 name = p->two->path ? p->two->path : p->one->path;
1885
1886                 if (p->one->sha1_valid && p->two->sha1_valid)
1887                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1888                 else
1889                         content_changed = 1;
1890
1891                 if (!content_changed) {
1892                         /*
1893                          * The SHA1 has not changed, so pre-/post-content is
1894                          * identical. We can therefore skip looking at the
1895                          * file contents altogether.
1896                          */
1897                         damage = 0;
1898                         goto found_damage;
1899                 }
1900
1901                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1902                         /*
1903                          * In --dirstat-by-file mode, we don't really need to
1904                          * look at the actual file contents at all.
1905                          * The fact that the SHA1 changed is enough for us to
1906                          * add this file to the list of results
1907                          * (with each file contributing equal damage).
1908                          */
1909                         damage = 1;
1910                         goto found_damage;
1911                 }
1912
1913                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1914                         diff_populate_filespec(p->one, 0);
1915                         diff_populate_filespec(p->two, 0);
1916                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1917                                                &copied, &added);
1918                         diff_free_filespec_data(p->one);
1919                         diff_free_filespec_data(p->two);
1920                 } else if (DIFF_FILE_VALID(p->one)) {
1921                         diff_populate_filespec(p->one, 1);
1922                         copied = added = 0;
1923                         diff_free_filespec_data(p->one);
1924                 } else if (DIFF_FILE_VALID(p->two)) {
1925                         diff_populate_filespec(p->two, 1);
1926                         copied = 0;
1927                         added = p->two->size;
1928                         diff_free_filespec_data(p->two);
1929                 } else
1930                         continue;
1931
1932                 /*
1933                  * Original minus copied is the removed material,
1934                  * added is the new material.  They are both damages
1935                  * made to the preimage.
1936                  * If the resulting damage is zero, we know that
1937                  * diffcore_count_changes() considers the two entries to
1938                  * be identical, but since content_changed is true, we
1939                  * know that there must have been _some_ kind of change,
1940                  * so we force all entries to have damage > 0.
1941                  */
1942                 damage = (p->one->size - copied) + added;
1943                 if (!damage)
1944                         damage = 1;
1945
1946 found_damage:
1947                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1948                 dir.files[dir.nr].name = name;
1949                 dir.files[dir.nr].changed = damage;
1950                 changed += damage;
1951                 dir.nr++;
1952         }
1953
1954         /* This can happen even with many files, if everything was renames */
1955         if (!changed)
1956                 return;
1957
1958         /* Show all directories with more than x% of the changes */
1959         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1960         gather_dirstat(options, &dir, changed, "", 0);
1961 }
1962
1963 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1964 {
1965         int i;
1966         unsigned long changed;
1967         struct dirstat_dir dir;
1968
1969         if (data->nr == 0)
1970                 return;
1971
1972         dir.files = NULL;
1973         dir.alloc = 0;
1974         dir.nr = 0;
1975         dir.permille = options->dirstat_permille;
1976         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1977
1978         changed = 0;
1979         for (i = 0; i < data->nr; i++) {
1980                 struct diffstat_file *file = data->files[i];
1981                 unsigned long damage = file->added + file->deleted;
1982                 if (file->is_binary)
1983                         /*
1984                          * binary files counts bytes, not lines. Must find some
1985                          * way to normalize binary bytes vs. textual lines.
1986                          * The following heuristic assumes that there are 64
1987                          * bytes per "line".
1988                          * This is stupid and ugly, but very cheap...
1989                          */
1990                         damage = (damage + 63) / 64;
1991                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1992                 dir.files[dir.nr].name = file->name;
1993                 dir.files[dir.nr].changed = damage;
1994                 changed += damage;
1995                 dir.nr++;
1996         }
1997
1998         /* This can happen even with many files, if everything was renames */
1999         if (!changed)
2000                 return;
2001
2002         /* Show all directories with more than x% of the changes */
2003         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2004         gather_dirstat(options, &dir, changed, "", 0);
2005 }
2006
2007 static void free_diffstat_info(struct diffstat_t *diffstat)
2008 {
2009         int i;
2010         for (i = 0; i < diffstat->nr; i++) {
2011                 struct diffstat_file *f = diffstat->files[i];
2012                 if (f->name != f->print_name)
2013                         free(f->print_name);
2014                 free(f->name);
2015                 free(f->from_name);
2016                 free(f);
2017         }
2018         free(diffstat->files);
2019 }
2020
2021 struct checkdiff_t {
2022         const char *filename;
2023         int lineno;
2024         int conflict_marker_size;
2025         struct diff_options *o;
2026         unsigned ws_rule;
2027         unsigned status;
2028 };
2029
2030 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2031 {
2032         char firstchar;
2033         int cnt;
2034
2035         if (len < marker_size + 1)
2036                 return 0;
2037         firstchar = line[0];
2038         switch (firstchar) {
2039         case '=': case '>': case '<': case '|':
2040                 break;
2041         default:
2042                 return 0;
2043         }
2044         for (cnt = 1; cnt < marker_size; cnt++)
2045                 if (line[cnt] != firstchar)
2046                         return 0;
2047         /* line[1] thru line[marker_size-1] are same as firstchar */
2048         if (len < marker_size + 1 || !isspace(line[marker_size]))
2049                 return 0;
2050         return 1;
2051 }
2052
2053 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2054 {
2055         struct checkdiff_t *data = priv;
2056         int marker_size = data->conflict_marker_size;
2057         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2058         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2059         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2060         char *err;
2061         const char *line_prefix;
2062
2063         assert(data->o);
2064         line_prefix = diff_line_prefix(data->o);
2065
2066         if (line[0] == '+') {
2067                 unsigned bad;
2068                 data->lineno++;
2069                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2070                         data->status |= 1;
2071                         fprintf(data->o->file,
2072                                 "%s%s:%d: leftover conflict marker\n",
2073                                 line_prefix, data->filename, data->lineno);
2074                 }
2075                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2076                 if (!bad)
2077                         return;
2078                 data->status |= bad;
2079                 err = whitespace_error_string(bad);
2080                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2081                         line_prefix, data->filename, data->lineno, err);
2082                 free(err);
2083                 emit_line(data->o, set, reset, line, 1);
2084                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2085                               data->o->file, set, reset, ws);
2086         } else if (line[0] == ' ') {
2087                 data->lineno++;
2088         } else if (line[0] == '@') {
2089                 char *plus = strchr(line, '+');
2090                 if (plus)
2091                         data->lineno = strtol(plus, NULL, 10) - 1;
2092                 else
2093                         die("invalid diff");
2094         }
2095 }
2096
2097 static unsigned char *deflate_it(char *data,
2098                                  unsigned long size,
2099                                  unsigned long *result_size)
2100 {
2101         int bound;
2102         unsigned char *deflated;
2103         git_zstream stream;
2104
2105         memset(&stream, 0, sizeof(stream));
2106         git_deflate_init(&stream, zlib_compression_level);
2107         bound = git_deflate_bound(&stream, size);
2108         deflated = xmalloc(bound);
2109         stream.next_out = deflated;
2110         stream.avail_out = bound;
2111
2112         stream.next_in = (unsigned char *)data;
2113         stream.avail_in = size;
2114         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2115                 ; /* nothing */
2116         git_deflate_end(&stream);
2117         *result_size = stream.total_out;
2118         return deflated;
2119 }
2120
2121 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2122                                   const char *prefix)
2123 {
2124         void *cp;
2125         void *delta;
2126         void *deflated;
2127         void *data;
2128         unsigned long orig_size;
2129         unsigned long delta_size;
2130         unsigned long deflate_size;
2131         unsigned long data_size;
2132
2133         /* We could do deflated delta, or we could do just deflated two,
2134          * whichever is smaller.
2135          */
2136         delta = NULL;
2137         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2138         if (one->size && two->size) {
2139                 delta = diff_delta(one->ptr, one->size,
2140                                    two->ptr, two->size,
2141                                    &delta_size, deflate_size);
2142                 if (delta) {
2143                         void *to_free = delta;
2144                         orig_size = delta_size;
2145                         delta = deflate_it(delta, delta_size, &delta_size);
2146                         free(to_free);
2147                 }
2148         }
2149
2150         if (delta && delta_size < deflate_size) {
2151                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2152                 free(deflated);
2153                 data = delta;
2154                 data_size = delta_size;
2155         }
2156         else {
2157                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2158                 free(delta);
2159                 data = deflated;
2160                 data_size = deflate_size;
2161         }
2162
2163         /* emit data encoded in base85 */
2164         cp = data;
2165         while (data_size) {
2166                 int bytes = (52 < data_size) ? 52 : data_size;
2167                 char line[70];
2168                 data_size -= bytes;
2169                 if (bytes <= 26)
2170                         line[0] = bytes + 'A' - 1;
2171                 else
2172                         line[0] = bytes - 26 + 'a' - 1;
2173                 encode_85(line + 1, cp, bytes);
2174                 cp = (char *) cp + bytes;
2175                 fprintf(file, "%s", prefix);
2176                 fputs(line, file);
2177                 fputc('\n', file);
2178         }
2179         fprintf(file, "%s\n", prefix);
2180         free(data);
2181 }
2182
2183 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2184                              const char *prefix)
2185 {
2186         fprintf(file, "%sGIT binary patch\n", prefix);
2187         emit_binary_diff_body(file, one, two, prefix);
2188         emit_binary_diff_body(file, two, one, prefix);
2189 }
2190
2191 int diff_filespec_is_binary(struct diff_filespec *one)
2192 {
2193         if (one->is_binary == -1) {
2194                 diff_filespec_load_driver(one);
2195                 if (one->driver->binary != -1)
2196                         one->is_binary = one->driver->binary;
2197                 else {
2198                         if (!one->data && DIFF_FILE_VALID(one))
2199                                 diff_populate_filespec(one, 0);
2200                         if (one->data)
2201                                 one->is_binary = buffer_is_binary(one->data,
2202                                                 one->size);
2203                         if (one->is_binary == -1)
2204                                 one->is_binary = 0;
2205                 }
2206         }
2207         return one->is_binary;
2208 }
2209
2210 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2211 {
2212         diff_filespec_load_driver(one);
2213         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2214 }
2215
2216 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2217 {
2218         if (!options->a_prefix)
2219                 options->a_prefix = a;
2220         if (!options->b_prefix)
2221                 options->b_prefix = b;
2222 }
2223
2224 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2225 {
2226         if (!DIFF_FILE_VALID(one))
2227                 return NULL;
2228
2229         diff_filespec_load_driver(one);
2230         return userdiff_get_textconv(one->driver);
2231 }
2232
2233 static void builtin_diff(const char *name_a,
2234                          const char *name_b,
2235                          struct diff_filespec *one,
2236                          struct diff_filespec *two,
2237                          const char *xfrm_msg,
2238                          int must_show_header,
2239                          struct diff_options *o,
2240                          int complete_rewrite)
2241 {
2242         mmfile_t mf1, mf2;
2243         const char *lbl[2];
2244         char *a_one, *b_two;
2245         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2246         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2247         const char *a_prefix, *b_prefix;
2248         struct userdiff_driver *textconv_one = NULL;
2249         struct userdiff_driver *textconv_two = NULL;
2250         struct strbuf header = STRBUF_INIT;
2251         const char *line_prefix = diff_line_prefix(o);
2252
2253         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2254                         (!one->mode || S_ISGITLINK(one->mode)) &&
2255                         (!two->mode || S_ISGITLINK(two->mode))) {
2256                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2257                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2258                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2259                                 line_prefix,
2260                                 one->sha1, two->sha1, two->dirty_submodule,
2261                                 meta, del, add, reset);
2262                 return;
2263         }
2264
2265         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2266                 textconv_one = get_textconv(one);
2267                 textconv_two = get_textconv(two);
2268         }
2269
2270         diff_set_mnemonic_prefix(o, "a/", "b/");
2271         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2272                 a_prefix = o->b_prefix;
2273                 b_prefix = o->a_prefix;
2274         } else {
2275                 a_prefix = o->a_prefix;
2276                 b_prefix = o->b_prefix;
2277         }
2278
2279         /* Never use a non-valid filename anywhere if at all possible */
2280         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2281         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2282
2283         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2284         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2285         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2286         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2287         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2288         if (lbl[0][0] == '/') {
2289                 /* /dev/null */
2290                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2291                 if (xfrm_msg)
2292                         strbuf_addstr(&header, xfrm_msg);
2293                 must_show_header = 1;
2294         }
2295         else if (lbl[1][0] == '/') {
2296                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2297                 if (xfrm_msg)
2298                         strbuf_addstr(&header, xfrm_msg);
2299                 must_show_header = 1;
2300         }
2301         else {
2302                 if (one->mode != two->mode) {
2303                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2304                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2305                         must_show_header = 1;
2306                 }
2307                 if (xfrm_msg)
2308                         strbuf_addstr(&header, xfrm_msg);
2309
2310                 /*
2311                  * we do not run diff between different kind
2312                  * of objects.
2313                  */
2314                 if ((one->mode ^ two->mode) & S_IFMT)
2315                         goto free_ab_and_return;
2316                 if (complete_rewrite &&
2317                     (textconv_one || !diff_filespec_is_binary(one)) &&
2318                     (textconv_two || !diff_filespec_is_binary(two))) {
2319                         fprintf(o->file, "%s", header.buf);
2320                         strbuf_reset(&header);
2321                         emit_rewrite_diff(name_a, name_b, one, two,
2322                                                 textconv_one, textconv_two, o);
2323                         o->found_changes = 1;
2324                         goto free_ab_and_return;
2325                 }
2326         }
2327
2328         if (o->irreversible_delete && lbl[1][0] == '/') {
2329                 fprintf(o->file, "%s", header.buf);
2330                 strbuf_reset(&header);
2331                 goto free_ab_and_return;
2332         } else if (!DIFF_OPT_TST(o, TEXT) &&
2333             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2334               (!textconv_two && diff_filespec_is_binary(two)) )) {
2335                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2336                         die("unable to read files to diff");
2337                 /* Quite common confusing case */
2338                 if (mf1.size == mf2.size &&
2339                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2340                         if (must_show_header)
2341                                 fprintf(o->file, "%s", header.buf);
2342                         goto free_ab_and_return;
2343                 }
2344                 fprintf(o->file, "%s", header.buf);
2345                 strbuf_reset(&header);
2346                 if (DIFF_OPT_TST(o, BINARY))
2347                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2348                 else
2349                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2350                                 line_prefix, lbl[0], lbl[1]);
2351                 o->found_changes = 1;
2352         } else {
2353                 /* Crazy xdl interfaces.. */
2354                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2355                 xpparam_t xpp;
2356                 xdemitconf_t xecfg;
2357                 struct emit_callback ecbdata;
2358                 const struct userdiff_funcname *pe;
2359
2360                 if (must_show_header) {
2361                         fprintf(o->file, "%s", header.buf);
2362                         strbuf_reset(&header);
2363                 }
2364
2365                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2366                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2367
2368                 pe = diff_funcname_pattern(one);
2369                 if (!pe)
2370                         pe = diff_funcname_pattern(two);
2371
2372                 memset(&xpp, 0, sizeof(xpp));
2373                 memset(&xecfg, 0, sizeof(xecfg));
2374                 memset(&ecbdata, 0, sizeof(ecbdata));
2375                 ecbdata.label_path = lbl;
2376                 ecbdata.color_diff = want_color(o->use_color);
2377                 ecbdata.found_changesp = &o->found_changes;
2378                 ecbdata.ws_rule = whitespace_rule(name_b);
2379                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2380                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2381                 ecbdata.opt = o;
2382                 ecbdata.header = header.len ? &header : NULL;
2383                 xpp.flags = o->xdl_opts;
2384                 xecfg.ctxlen = o->context;
2385                 xecfg.interhunkctxlen = o->interhunkcontext;
2386                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2387                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2388                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2389                 if (pe)
2390                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2391                 if (!diffopts)
2392                         ;
2393                 else if (starts_with(diffopts, "--unified="))
2394                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2395                 else if (starts_with(diffopts, "-u"))
2396                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2397                 if (o->word_diff)
2398                         init_diff_words_data(&ecbdata, o, one, two);
2399                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2400                               &xpp, &xecfg);
2401                 if (o->word_diff)
2402                         free_diff_words_data(&ecbdata);
2403                 if (textconv_one)
2404                         free(mf1.ptr);
2405                 if (textconv_two)
2406                         free(mf2.ptr);
2407                 xdiff_clear_find_func(&xecfg);
2408         }
2409
2410  free_ab_and_return:
2411         strbuf_release(&header);
2412         diff_free_filespec_data(one);
2413         diff_free_filespec_data(two);
2414         free(a_one);
2415         free(b_two);
2416         return;
2417 }
2418
2419 static void builtin_diffstat(const char *name_a, const char *name_b,
2420                              struct diff_filespec *one,
2421                              struct diff_filespec *two,
2422                              struct diffstat_t *diffstat,
2423                              struct diff_options *o,
2424                              struct diff_filepair *p)
2425 {
2426         mmfile_t mf1, mf2;
2427         struct diffstat_file *data;
2428         int same_contents;
2429         int complete_rewrite = 0;
2430
2431         if (!DIFF_PAIR_UNMERGED(p)) {
2432                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2433                         complete_rewrite = 1;
2434         }
2435
2436         data = diffstat_add(diffstat, name_a, name_b);
2437         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2438
2439         if (!one || !two) {
2440                 data->is_unmerged = 1;
2441                 return;
2442         }
2443
2444         same_contents = !hashcmp(one->sha1, two->sha1);
2445
2446         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2447                 data->is_binary = 1;
2448                 if (same_contents) {
2449                         data->added = 0;
2450                         data->deleted = 0;
2451                 } else {
2452                         data->added = diff_filespec_size(two);
2453                         data->deleted = diff_filespec_size(one);
2454                 }
2455         }
2456
2457         else if (complete_rewrite) {
2458                 diff_populate_filespec(one, 0);
2459                 diff_populate_filespec(two, 0);
2460                 data->deleted = count_lines(one->data, one->size);
2461                 data->added = count_lines(two->data, two->size);
2462         }
2463
2464         else if (!same_contents) {
2465                 /* Crazy xdl interfaces.. */
2466                 xpparam_t xpp;
2467                 xdemitconf_t xecfg;
2468
2469                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2470                         die("unable to read files to diff");
2471
2472                 memset(&xpp, 0, sizeof(xpp));
2473                 memset(&xecfg, 0, sizeof(xecfg));
2474                 xpp.flags = o->xdl_opts;
2475                 xecfg.ctxlen = o->context;
2476                 xecfg.interhunkctxlen = o->interhunkcontext;
2477                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2478                               &xpp, &xecfg);
2479         }
2480
2481         diff_free_filespec_data(one);
2482         diff_free_filespec_data(two);
2483 }
2484
2485 static void builtin_checkdiff(const char *name_a, const char *name_b,
2486                               const char *attr_path,
2487                               struct diff_filespec *one,
2488                               struct diff_filespec *two,
2489                               struct diff_options *o)
2490 {
2491         mmfile_t mf1, mf2;
2492         struct checkdiff_t data;
2493
2494         if (!two)
2495                 return;
2496
2497         memset(&data, 0, sizeof(data));
2498         data.filename = name_b ? name_b : name_a;
2499         data.lineno = 0;
2500         data.o = o;
2501         data.ws_rule = whitespace_rule(attr_path);
2502         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2503
2504         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2505                 die("unable to read files to diff");
2506
2507         /*
2508          * All the other codepaths check both sides, but not checking
2509          * the "old" side here is deliberate.  We are checking the newly
2510          * introduced changes, and as long as the "new" side is text, we
2511          * can and should check what it introduces.
2512          */
2513         if (diff_filespec_is_binary(two))
2514                 goto free_and_return;
2515         else {
2516                 /* Crazy xdl interfaces.. */
2517                 xpparam_t xpp;
2518                 xdemitconf_t xecfg;
2519
2520                 memset(&xpp, 0, sizeof(xpp));
2521                 memset(&xecfg, 0, sizeof(xecfg));
2522                 xecfg.ctxlen = 1; /* at least one context line */
2523                 xpp.flags = 0;
2524                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2525                               &xpp, &xecfg);
2526
2527                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2528                         struct emit_callback ecbdata;
2529                         int blank_at_eof;
2530
2531                         ecbdata.ws_rule = data.ws_rule;
2532                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2533                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2534
2535                         if (blank_at_eof) {
2536                                 static char *err;
2537                                 if (!err)
2538                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2539                                 fprintf(o->file, "%s:%d: %s.\n",
2540                                         data.filename, blank_at_eof, err);
2541                                 data.status = 1; /* report errors */
2542                         }
2543                 }
2544         }
2545  free_and_return:
2546         diff_free_filespec_data(one);
2547         diff_free_filespec_data(two);
2548         if (data.status)
2549                 DIFF_OPT_SET(o, CHECK_FAILED);
2550 }
2551
2552 struct diff_filespec *alloc_filespec(const char *path)
2553 {
2554         int namelen = strlen(path);
2555         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2556
2557         memset(spec, 0, sizeof(*spec));
2558         spec->path = (char *)(spec + 1);
2559         memcpy(spec->path, path, namelen+1);
2560         spec->count = 1;
2561         spec->is_binary = -1;
2562         return spec;
2563 }
2564
2565 void free_filespec(struct diff_filespec *spec)
2566 {
2567         if (!--spec->count) {
2568                 diff_free_filespec_data(spec);
2569                 free(spec);
2570         }
2571 }
2572
2573 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2574                    int sha1_valid, unsigned short mode)
2575 {
2576         if (mode) {
2577                 spec->mode = canon_mode(mode);
2578                 hashcpy(spec->sha1, sha1);
2579                 spec->sha1_valid = sha1_valid;
2580         }
2581 }
2582
2583 /*
2584  * Given a name and sha1 pair, if the index tells us the file in
2585  * the work tree has that object contents, return true, so that
2586  * prepare_temp_file() does not have to inflate and extract.
2587  */
2588 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2589 {
2590         const struct cache_entry *ce;
2591         struct stat st;
2592         int pos, len;
2593
2594         /*
2595          * We do not read the cache ourselves here, because the
2596          * benchmark with my previous version that always reads cache
2597          * shows that it makes things worse for diff-tree comparing
2598          * two linux-2.6 kernel trees in an already checked out work
2599          * tree.  This is because most diff-tree comparisons deal with
2600          * only a small number of files, while reading the cache is
2601          * expensive for a large project, and its cost outweighs the
2602          * savings we get by not inflating the object to a temporary
2603          * file.  Practically, this code only helps when we are used
2604          * by diff-cache --cached, which does read the cache before
2605          * calling us.
2606          */
2607         if (!active_cache)
2608                 return 0;
2609
2610         /* We want to avoid the working directory if our caller
2611          * doesn't need the data in a normal file, this system
2612          * is rather slow with its stat/open/mmap/close syscalls,
2613          * and the object is contained in a pack file.  The pack
2614          * is probably already open and will be faster to obtain
2615          * the data through than the working directory.  Loose
2616          * objects however would tend to be slower as they need
2617          * to be individually opened and inflated.
2618          */
2619         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2620                 return 0;
2621
2622         len = strlen(name);
2623         pos = cache_name_pos(name, len);
2624         if (pos < 0)
2625                 return 0;
2626         ce = active_cache[pos];
2627
2628         /*
2629          * This is not the sha1 we are looking for, or
2630          * unreusable because it is not a regular file.
2631          */
2632         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2633                 return 0;
2634
2635         /*
2636          * If ce is marked as "assume unchanged", there is no
2637          * guarantee that work tree matches what we are looking for.
2638          */
2639         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2640                 return 0;
2641
2642         /*
2643          * If ce matches the file in the work tree, we can reuse it.
2644          */
2645         if (ce_uptodate(ce) ||
2646             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2647                 return 1;
2648
2649         return 0;
2650 }
2651
2652 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2653 {
2654         int len;
2655         char *data = xmalloc(100), *dirty = "";
2656
2657         /* Are we looking at the work tree? */
2658         if (s->dirty_submodule)
2659                 dirty = "-dirty";
2660
2661         len = snprintf(data, 100,
2662                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2663         s->data = data;
2664         s->size = len;
2665         s->should_free = 1;
2666         if (size_only) {
2667                 s->data = NULL;
2668                 free(data);
2669         }
2670         return 0;
2671 }
2672
2673 /*
2674  * While doing rename detection and pickaxe operation, we may need to
2675  * grab the data for the blob (or file) for our own in-core comparison.
2676  * diff_filespec has data and size fields for this purpose.
2677  */
2678 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2679 {
2680         int err = 0;
2681         /*
2682          * demote FAIL to WARN to allow inspecting the situation
2683          * instead of refusing.
2684          */
2685         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2686                                     ? SAFE_CRLF_WARN
2687                                     : safe_crlf);
2688
2689         if (!DIFF_FILE_VALID(s))
2690                 die("internal error: asking to populate invalid file.");
2691         if (S_ISDIR(s->mode))
2692                 return -1;
2693
2694         if (s->data)
2695                 return 0;
2696
2697         if (size_only && 0 < s->size)
2698                 return 0;
2699
2700         if (S_ISGITLINK(s->mode))
2701                 return diff_populate_gitlink(s, size_only);
2702
2703         if (!s->sha1_valid ||
2704             reuse_worktree_file(s->path, s->sha1, 0)) {
2705                 struct strbuf buf = STRBUF_INIT;
2706                 struct stat st;
2707                 int fd;
2708
2709                 if (lstat(s->path, &st) < 0) {
2710                         if (errno == ENOENT) {
2711                         err_empty:
2712                                 err = -1;
2713                         empty:
2714                                 s->data = (char *)"";
2715                                 s->size = 0;
2716                                 return err;
2717                         }
2718                 }
2719                 s->size = xsize_t(st.st_size);
2720                 if (!s->size)
2721                         goto empty;
2722                 if (S_ISLNK(st.st_mode)) {
2723                         struct strbuf sb = STRBUF_INIT;
2724
2725                         if (strbuf_readlink(&sb, s->path, s->size))
2726                                 goto err_empty;
2727                         s->size = sb.len;
2728                         s->data = strbuf_detach(&sb, NULL);
2729                         s->should_free = 1;
2730                         return 0;
2731                 }
2732                 if (size_only)
2733                         return 0;
2734                 fd = open(s->path, O_RDONLY);
2735                 if (fd < 0)
2736                         goto err_empty;
2737                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2738                 close(fd);
2739                 s->should_munmap = 1;
2740
2741                 /*
2742                  * Convert from working tree format to canonical git format
2743                  */
2744                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2745                         size_t size = 0;
2746                         munmap(s->data, s->size);
2747                         s->should_munmap = 0;
2748                         s->data = strbuf_detach(&buf, &size);
2749                         s->size = size;
2750                         s->should_free = 1;
2751                 }
2752         }
2753         else {
2754                 enum object_type type;
2755                 if (size_only) {
2756                         type = sha1_object_info(s->sha1, &s->size);
2757                         if (type < 0)
2758                                 die("unable to read %s", sha1_to_hex(s->sha1));
2759                 } else {
2760                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2761                         if (!s->data)
2762                                 die("unable to read %s", sha1_to_hex(s->sha1));
2763                         s->should_free = 1;
2764                 }
2765         }
2766         return 0;
2767 }
2768
2769 void diff_free_filespec_blob(struct diff_filespec *s)
2770 {
2771         if (s->should_free)
2772                 free(s->data);
2773         else if (s->should_munmap)
2774                 munmap(s->data, s->size);
2775
2776         if (s->should_free || s->should_munmap) {
2777                 s->should_free = s->should_munmap = 0;
2778                 s->data = NULL;
2779         }
2780 }
2781
2782 void diff_free_filespec_data(struct diff_filespec *s)
2783 {
2784         diff_free_filespec_blob(s);
2785         free(s->cnt_data);
2786         s->cnt_data = NULL;
2787 }
2788
2789 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2790                            void *blob,
2791                            unsigned long size,
2792                            const unsigned char *sha1,
2793                            int mode)
2794 {
2795         int fd;
2796         struct strbuf buf = STRBUF_INIT;
2797         struct strbuf template = STRBUF_INIT;
2798         char *path_dup = xstrdup(path);
2799         const char *base = basename(path_dup);
2800
2801         /* Generate "XXXXXX_basename.ext" */
2802         strbuf_addstr(&template, "XXXXXX_");
2803         strbuf_addstr(&template, base);
2804
2805         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2806                         strlen(base) + 1);
2807         if (fd < 0)
2808                 die_errno("unable to create temp-file");
2809         if (convert_to_working_tree(path,
2810                         (const char *)blob, (size_t)size, &buf)) {
2811                 blob = buf.buf;
2812                 size = buf.len;
2813         }
2814         if (write_in_full(fd, blob, size) != size)
2815                 die_errno("unable to write temp-file");
2816         close(fd);
2817         temp->name = temp->tmp_path;
2818         strcpy(temp->hex, sha1_to_hex(sha1));
2819         temp->hex[40] = 0;
2820         sprintf(temp->mode, "%06o", mode);
2821         strbuf_release(&buf);
2822         strbuf_release(&template);
2823         free(path_dup);
2824 }
2825
2826 static struct diff_tempfile *prepare_temp_file(const char *name,
2827                 struct diff_filespec *one)
2828 {
2829         struct diff_tempfile *temp = claim_diff_tempfile();
2830
2831         if (!DIFF_FILE_VALID(one)) {
2832         not_a_valid_file:
2833                 /* A '-' entry produces this for file-2, and
2834                  * a '+' entry produces this for file-1.
2835                  */
2836                 temp->name = "/dev/null";
2837                 strcpy(temp->hex, ".");
2838                 strcpy(temp->mode, ".");
2839                 return temp;
2840         }
2841
2842         if (!remove_tempfile_installed) {
2843                 atexit(remove_tempfile);
2844                 sigchain_push_common(remove_tempfile_on_signal);
2845                 remove_tempfile_installed = 1;
2846         }
2847
2848         if (!one->sha1_valid ||
2849             reuse_worktree_file(name, one->sha1, 1)) {
2850                 struct stat st;
2851                 if (lstat(name, &st) < 0) {
2852                         if (errno == ENOENT)
2853                                 goto not_a_valid_file;
2854                         die_errno("stat(%s)", name);
2855                 }
2856                 if (S_ISLNK(st.st_mode)) {
2857                         struct strbuf sb = STRBUF_INIT;
2858                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2859                                 die_errno("readlink(%s)", name);
2860                         prep_temp_blob(name, temp, sb.buf, sb.len,
2861                                        (one->sha1_valid ?
2862                                         one->sha1 : null_sha1),
2863                                        (one->sha1_valid ?
2864                                         one->mode : S_IFLNK));
2865                         strbuf_release(&sb);
2866                 }
2867                 else {
2868                         /* we can borrow from the file in the work tree */
2869                         temp->name = name;
2870                         if (!one->sha1_valid)
2871                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2872                         else
2873                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2874                         /* Even though we may sometimes borrow the
2875                          * contents from the work tree, we always want
2876                          * one->mode.  mode is trustworthy even when
2877                          * !(one->sha1_valid), as long as
2878                          * DIFF_FILE_VALID(one).
2879                          */
2880                         sprintf(temp->mode, "%06o", one->mode);
2881                 }
2882                 return temp;
2883         }
2884         else {
2885                 if (diff_populate_filespec(one, 0))
2886                         die("cannot read data blob for %s", one->path);
2887                 prep_temp_blob(name, temp, one->data, one->size,
2888                                one->sha1, one->mode);
2889         }
2890         return temp;
2891 }
2892
2893 /* An external diff command takes:
2894  *
2895  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2896  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2897  *
2898  */
2899 static void run_external_diff(const char *pgm,
2900                               const char *name,
2901                               const char *other,
2902                               struct diff_filespec *one,
2903                               struct diff_filespec *two,
2904                               const char *xfrm_msg,
2905                               int complete_rewrite,
2906                               struct diff_options *o)
2907 {
2908         const char *spawn_arg[10];
2909         int retval;
2910         const char **arg = &spawn_arg[0];
2911         struct diff_queue_struct *q = &diff_queued_diff;
2912         const char *env[3] = { NULL };
2913         char env_counter[50];
2914         char env_total[50];
2915
2916         if (one && two) {
2917                 struct diff_tempfile *temp_one, *temp_two;
2918                 const char *othername = (other ? other : name);
2919                 temp_one = prepare_temp_file(name, one);
2920                 temp_two = prepare_temp_file(othername, two);
2921                 *arg++ = pgm;
2922                 *arg++ = name;
2923                 *arg++ = temp_one->name;
2924                 *arg++ = temp_one->hex;
2925                 *arg++ = temp_one->mode;
2926                 *arg++ = temp_two->name;
2927                 *arg++ = temp_two->hex;
2928                 *arg++ = temp_two->mode;
2929                 if (other) {
2930                         *arg++ = other;
2931                         *arg++ = xfrm_msg;
2932                 }
2933         } else {
2934                 *arg++ = pgm;
2935                 *arg++ = name;
2936         }
2937         *arg = NULL;
2938         fflush(NULL);
2939
2940         env[0] = env_counter;
2941         snprintf(env_counter, sizeof(env_counter), "GIT_DIFF_PATH_COUNTER=%d",
2942                  ++o->diff_path_counter);
2943         env[1] = env_total;
2944         snprintf(env_total, sizeof(env_total), "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2945
2946         retval = run_command_v_opt_cd_env(spawn_arg, RUN_USING_SHELL, NULL, env);
2947         remove_tempfile();
2948         if (retval) {
2949                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2950                 exit(1);
2951         }
2952 }
2953
2954 static int similarity_index(struct diff_filepair *p)
2955 {
2956         return p->score * 100 / MAX_SCORE;
2957 }
2958
2959 static void fill_metainfo(struct strbuf *msg,
2960                           const char *name,
2961                           const char *other,
2962                           struct diff_filespec *one,
2963                           struct diff_filespec *two,
2964                           struct diff_options *o,
2965                           struct diff_filepair *p,
2966                           int *must_show_header,
2967                           int use_color)
2968 {
2969         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2970         const char *reset = diff_get_color(use_color, DIFF_RESET);
2971         const char *line_prefix = diff_line_prefix(o);
2972
2973         *must_show_header = 1;
2974         strbuf_init(msg, PATH_MAX * 2 + 300);
2975         switch (p->status) {
2976         case DIFF_STATUS_COPIED:
2977                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2978                             line_prefix, set, similarity_index(p));
2979                 strbuf_addf(msg, "%s\n%s%scopy from ",
2980                             reset,  line_prefix, set);
2981                 quote_c_style(name, msg, NULL, 0);
2982                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2983                 quote_c_style(other, msg, NULL, 0);
2984                 strbuf_addf(msg, "%s\n", reset);
2985                 break;
2986         case DIFF_STATUS_RENAMED:
2987                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2988                             line_prefix, set, similarity_index(p));
2989                 strbuf_addf(msg, "%s\n%s%srename from ",
2990                             reset, line_prefix, set);
2991                 quote_c_style(name, msg, NULL, 0);
2992                 strbuf_addf(msg, "%s\n%s%srename to ",
2993                             reset, line_prefix, set);
2994                 quote_c_style(other, msg, NULL, 0);
2995                 strbuf_addf(msg, "%s\n", reset);
2996                 break;
2997         case DIFF_STATUS_MODIFIED:
2998                 if (p->score) {
2999                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3000                                     line_prefix,
3001                                     set, similarity_index(p), reset);
3002                         break;
3003                 }
3004                 /* fallthru */
3005         default:
3006                 *must_show_header = 0;
3007         }
3008         if (one && two && hashcmp(one->sha1, two->sha1)) {
3009                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3010
3011                 if (DIFF_OPT_TST(o, BINARY)) {
3012                         mmfile_t mf;
3013                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3014                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3015                                 abbrev = 40;
3016                 }
3017                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3018                             find_unique_abbrev(one->sha1, abbrev));
3019                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3020                 if (one->mode == two->mode)
3021                         strbuf_addf(msg, " %06o", one->mode);
3022                 strbuf_addf(msg, "%s\n", reset);
3023         }
3024 }
3025
3026 static void run_diff_cmd(const char *pgm,
3027                          const char *name,
3028                          const char *other,
3029                          const char *attr_path,
3030                          struct diff_filespec *one,
3031                          struct diff_filespec *two,
3032                          struct strbuf *msg,
3033                          struct diff_options *o,
3034                          struct diff_filepair *p)
3035 {
3036         const char *xfrm_msg = NULL;
3037         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3038         int must_show_header = 0;
3039
3040
3041         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3042                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3043                 if (drv && drv->external)
3044                         pgm = drv->external;
3045         }
3046
3047         if (msg) {
3048                 /*
3049                  * don't use colors when the header is intended for an
3050                  * external diff driver
3051                  */
3052                 fill_metainfo(msg, name, other, one, two, o, p,
3053                               &must_show_header,
3054                               want_color(o->use_color) && !pgm);
3055                 xfrm_msg = msg->len ? msg->buf : NULL;
3056         }
3057
3058         if (pgm) {
3059                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3060                                   complete_rewrite, o);
3061                 return;
3062         }
3063         if (one && two)
3064                 builtin_diff(name, other ? other : name,
3065                              one, two, xfrm_msg, must_show_header,
3066                              o, complete_rewrite);
3067         else
3068                 fprintf(o->file, "* Unmerged path %s\n", name);
3069 }
3070
3071 static void diff_fill_sha1_info(struct diff_filespec *one)
3072 {
3073         if (DIFF_FILE_VALID(one)) {
3074                 if (!one->sha1_valid) {
3075                         struct stat st;
3076                         if (one->is_stdin) {
3077                                 hashcpy(one->sha1, null_sha1);
3078                                 return;
3079                         }
3080                         if (lstat(one->path, &st) < 0)
3081                                 die_errno("stat '%s'", one->path);
3082                         if (index_path(one->sha1, one->path, &st, 0))
3083                                 die("cannot hash %s", one->path);
3084                 }
3085         }
3086         else
3087                 hashclr(one->sha1);
3088 }
3089
3090 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3091 {
3092         /* Strip the prefix but do not molest /dev/null and absolute paths */
3093         if (*namep && **namep != '/') {
3094                 *namep += prefix_length;
3095                 if (**namep == '/')
3096                         ++*namep;
3097         }
3098         if (*otherp && **otherp != '/') {
3099                 *otherp += prefix_length;
3100                 if (**otherp == '/')
3101                         ++*otherp;
3102         }
3103 }
3104
3105 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3106 {
3107         const char *pgm = external_diff();
3108         struct strbuf msg;
3109         struct diff_filespec *one = p->one;
3110         struct diff_filespec *two = p->two;
3111         const char *name;
3112         const char *other;
3113         const char *attr_path;
3114
3115         name  = p->one->path;
3116         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3117         attr_path = name;
3118         if (o->prefix_length)
3119                 strip_prefix(o->prefix_length, &name, &other);
3120
3121         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3122                 pgm = NULL;
3123
3124         if (DIFF_PAIR_UNMERGED(p)) {
3125                 run_diff_cmd(pgm, name, NULL, attr_path,
3126                              NULL, NULL, NULL, o, p);
3127                 return;
3128         }
3129
3130         diff_fill_sha1_info(one);
3131         diff_fill_sha1_info(two);
3132
3133         if (!pgm &&
3134             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3135             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3136                 /*
3137                  * a filepair that changes between file and symlink
3138                  * needs to be split into deletion and creation.
3139                  */
3140                 struct diff_filespec *null = alloc_filespec(two->path);
3141                 run_diff_cmd(NULL, name, other, attr_path,
3142                              one, null, &msg, o, p);
3143                 free(null);
3144                 strbuf_release(&msg);
3145
3146                 null = alloc_filespec(one->path);
3147                 run_diff_cmd(NULL, name, other, attr_path,
3148                              null, two, &msg, o, p);
3149                 free(null);
3150         }
3151         else
3152                 run_diff_cmd(pgm, name, other, attr_path,
3153                              one, two, &msg, o, p);
3154
3155         strbuf_release(&msg);
3156 }
3157
3158 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3159                          struct diffstat_t *diffstat)
3160 {
3161         const char *name;
3162         const char *other;
3163
3164         if (DIFF_PAIR_UNMERGED(p)) {
3165                 /* unmerged */
3166                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3167                 return;
3168         }
3169
3170         name = p->one->path;
3171         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3172
3173         if (o->prefix_length)
3174                 strip_prefix(o->prefix_length, &name, &other);
3175
3176         diff_fill_sha1_info(p->one);
3177         diff_fill_sha1_info(p->two);
3178
3179         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3180 }
3181
3182 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3183 {
3184         const char *name;
3185         const char *other;
3186         const char *attr_path;
3187
3188         if (DIFF_PAIR_UNMERGED(p)) {
3189                 /* unmerged */
3190                 return;
3191         }
3192
3193         name = p->one->path;
3194         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3195         attr_path = other ? other : name;
3196
3197         if (o->prefix_length)
3198                 strip_prefix(o->prefix_length, &name, &other);
3199
3200         diff_fill_sha1_info(p->one);
3201         diff_fill_sha1_info(p->two);
3202
3203         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3204 }
3205
3206 void diff_setup(struct diff_options *options)
3207 {
3208         memcpy(options, &default_diff_options, sizeof(*options));
3209
3210         options->file = stdout;
3211
3212         options->line_termination = '\n';
3213         options->break_opt = -1;
3214         options->rename_limit = -1;
3215         options->dirstat_permille = diff_dirstat_permille_default;
3216         options->context = diff_context_default;
3217         DIFF_OPT_SET(options, RENAME_EMPTY);
3218
3219         options->change = diff_change;
3220         options->add_remove = diff_addremove;
3221         options->use_color = diff_use_color_default;
3222         options->detect_rename = diff_detect_rename_default;
3223         options->xdl_opts |= diff_algorithm;
3224
3225         options->orderfile = diff_order_file_cfg;
3226
3227         if (diff_no_prefix) {
3228                 options->a_prefix = options->b_prefix = "";
3229         } else if (!diff_mnemonic_prefix) {
3230                 options->a_prefix = "a/";
3231                 options->b_prefix = "b/";
3232         }
3233 }
3234
3235 void diff_setup_done(struct diff_options *options)
3236 {
3237         int count = 0;
3238
3239         if (options->set_default)
3240                 options->set_default(options);
3241
3242         if (options->output_format & DIFF_FORMAT_NAME)
3243                 count++;
3244         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3245                 count++;
3246         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3247                 count++;
3248         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3249                 count++;
3250         if (count > 1)
3251                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3252
3253         /*
3254          * Most of the time we can say "there are changes"
3255          * only by checking if there are changed paths, but
3256          * --ignore-whitespace* options force us to look
3257          * inside contents.
3258          */
3259
3260         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3261             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3262             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3263                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3264         else
3265                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3266
3267         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3268                 options->detect_rename = DIFF_DETECT_COPY;
3269
3270         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3271                 options->prefix = NULL;
3272         if (options->prefix)
3273                 options->prefix_length = strlen(options->prefix);
3274         else
3275                 options->prefix_length = 0;
3276
3277         if (options->output_format & (DIFF_FORMAT_NAME |
3278                                       DIFF_FORMAT_NAME_STATUS |
3279                                       DIFF_FORMAT_CHECKDIFF |
3280                                       DIFF_FORMAT_NO_OUTPUT))
3281                 options->output_format &= ~(DIFF_FORMAT_RAW |
3282                                             DIFF_FORMAT_NUMSTAT |
3283                                             DIFF_FORMAT_DIFFSTAT |
3284                                             DIFF_FORMAT_SHORTSTAT |
3285                                             DIFF_FORMAT_DIRSTAT |
3286                                             DIFF_FORMAT_SUMMARY |
3287                                             DIFF_FORMAT_PATCH);
3288
3289         /*
3290          * These cases always need recursive; we do not drop caller-supplied
3291          * recursive bits for other formats here.
3292          */
3293         if (options->output_format & (DIFF_FORMAT_PATCH |
3294                                       DIFF_FORMAT_NUMSTAT |
3295                                       DIFF_FORMAT_DIFFSTAT |
3296                                       DIFF_FORMAT_SHORTSTAT |
3297                                       DIFF_FORMAT_DIRSTAT |
3298                                       DIFF_FORMAT_SUMMARY |
3299                                       DIFF_FORMAT_CHECKDIFF))
3300                 DIFF_OPT_SET(options, RECURSIVE);
3301         /*
3302          * Also pickaxe would not work very well if you do not say recursive
3303          */
3304         if (options->pickaxe)
3305                 DIFF_OPT_SET(options, RECURSIVE);
3306         /*
3307          * When patches are generated, submodules diffed against the work tree
3308          * must be checked for dirtiness too so it can be shown in the output
3309          */
3310         if (options->output_format & DIFF_FORMAT_PATCH)
3311                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3312
3313         if (options->detect_rename && options->rename_limit < 0)
3314                 options->rename_limit = diff_rename_limit_default;
3315         if (options->setup & DIFF_SETUP_USE_CACHE) {
3316                 if (!active_cache)
3317                         /* read-cache does not die even when it fails
3318                          * so it is safe for us to do this here.  Also
3319                          * it does not smudge active_cache or active_nr
3320                          * when it fails, so we do not have to worry about
3321                          * cleaning it up ourselves either.
3322                          */
3323                         read_cache();
3324         }
3325         if (options->abbrev <= 0 || 40 < options->abbrev)
3326                 options->abbrev = 40; /* full */
3327
3328         /*
3329          * It does not make sense to show the first hit we happened
3330          * to have found.  It does not make sense not to return with
3331          * exit code in such a case either.
3332          */
3333         if (DIFF_OPT_TST(options, QUICK)) {
3334                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3335                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3336         }
3337
3338         options->diff_path_counter = 0;
3339 }
3340
3341 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3342 {
3343         char c, *eq;
3344         int len;
3345
3346         if (*arg != '-')
3347                 return 0;
3348         c = *++arg;
3349         if (!c)
3350                 return 0;
3351         if (c == arg_short) {
3352                 c = *++arg;
3353                 if (!c)
3354                         return 1;
3355                 if (val && isdigit(c)) {
3356                         char *end;
3357                         int n = strtoul(arg, &end, 10);
3358                         if (*end)
3359                                 return 0;
3360                         *val = n;
3361                         return 1;
3362                 }
3363                 return 0;
3364         }
3365         if (c != '-')
3366                 return 0;
3367         arg++;
3368         eq = strchr(arg, '=');
3369         if (eq)
3370                 len = eq - arg;
3371         else
3372                 len = strlen(arg);
3373         if (!len || strncmp(arg, arg_long, len))
3374                 return 0;
3375         if (eq) {
3376                 int n;
3377                 char *end;
3378                 if (!isdigit(*++eq))
3379                         return 0;
3380                 n = strtoul(eq, &end, 10);
3381                 if (*end)
3382                         return 0;
3383                 *val = n;
3384         }
3385         return 1;
3386 }
3387
3388 static int diff_scoreopt_parse(const char *opt);
3389
3390 static inline int short_opt(char opt, const char **argv,
3391                             const char **optarg)
3392 {
3393         const char *arg = argv[0];
3394         if (arg[0] != '-' || arg[1] != opt)
3395                 return 0;
3396         if (arg[2] != '\0') {
3397                 *optarg = arg + 2;
3398                 return 1;
3399         }
3400         if (!argv[1])
3401                 die("Option '%c' requires a value", opt);
3402         *optarg = argv[1];
3403         return 2;
3404 }
3405
3406 int parse_long_opt(const char *opt, const char **argv,
3407                    const char **optarg)
3408 {
3409         const char *arg = argv[0];
3410         if (arg[0] != '-' || arg[1] != '-')
3411                 return 0;
3412         arg += strlen("--");
3413         if (!starts_with(arg, opt))
3414                 return 0;
3415         arg += strlen(opt);
3416         if (*arg == '=') { /* stuck form: --option=value */
3417                 *optarg = arg + 1;
3418                 return 1;
3419         }
3420         if (*arg != '\0')
3421                 return 0;
3422         /* separate form: --option value */
3423         if (!argv[1])
3424                 die("Option '--%s' requires a value", opt);
3425         *optarg = argv[1];
3426         return 2;
3427 }
3428
3429 static int stat_opt(struct diff_options *options, const char **av)
3430 {
3431         const char *arg = av[0];
3432         char *end;
3433         int width = options->stat_width;
3434         int name_width = options->stat_name_width;
3435         int graph_width = options->stat_graph_width;
3436         int count = options->stat_count;
3437         int argcount = 1;
3438
3439         arg += strlen("--stat");
3440         end = (char *)arg;
3441
3442         switch (*arg) {
3443         case '-':
3444                 if (starts_with(arg, "-width")) {
3445                         arg += strlen("-width");
3446                         if (*arg == '=')
3447                                 width = strtoul(arg + 1, &end, 10);
3448                         else if (!*arg && !av[1])
3449                                 die("Option '--stat-width' requires a value");
3450                         else if (!*arg) {
3451                                 width = strtoul(av[1], &end, 10);
3452                                 argcount = 2;
3453                         }
3454                 } else if (starts_with(arg, "-name-width")) {
3455                         arg += strlen("-name-width");
3456                         if (*arg == '=')
3457                                 name_width = strtoul(arg + 1, &end, 10);
3458                         else if (!*arg && !av[1])
3459                                 die("Option '--stat-name-width' requires a value");
3460                         else if (!*arg) {
3461                                 name_width = strtoul(av[1], &end, 10);
3462                                 argcount = 2;
3463                         }
3464                 } else if (starts_with(arg, "-graph-width")) {
3465                         arg += strlen("-graph-width");
3466                         if (*arg == '=')
3467                                 graph_width = strtoul(arg + 1, &end, 10);
3468                         else if (!*arg && !av[1])
3469                                 die("Option '--stat-graph-width' requires a value");
3470                         else if (!*arg) {
3471                                 graph_width = strtoul(av[1], &end, 10);
3472                                 argcount = 2;
3473                         }
3474                 } else if (starts_with(arg, "-count")) {
3475                         arg += strlen("-count");
3476                         if (*arg == '=')
3477                                 count = strtoul(arg + 1, &end, 10);
3478                         else if (!*arg && !av[1])
3479                                 die("Option '--stat-count' requires a value");
3480                         else if (!*arg) {
3481                                 count = strtoul(av[1], &end, 10);
3482                                 argcount = 2;
3483                         }
3484                 }
3485                 break;
3486         case '=':
3487                 width = strtoul(arg+1, &end, 10);
3488                 if (*end == ',')
3489                         name_width = strtoul(end+1, &end, 10);
3490                 if (*end == ',')
3491                         count = strtoul(end+1, &end, 10);
3492         }
3493
3494         /* Important! This checks all the error cases! */
3495         if (*end)
3496                 return 0;
3497         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3498         options->stat_name_width = name_width;
3499         options->stat_graph_width = graph_width;
3500         options->stat_width = width;
3501         options->stat_count = count;
3502         return argcount;
3503 }
3504
3505 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3506 {
3507         struct strbuf errmsg = STRBUF_INIT;
3508         if (parse_dirstat_params(options, params, &errmsg))
3509                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3510                     errmsg.buf);
3511         strbuf_release(&errmsg);
3512         /*
3513          * The caller knows a dirstat-related option is given from the command
3514          * line; allow it to say "return this_function();"
3515          */
3516         options->output_format |= DIFF_FORMAT_DIRSTAT;
3517         return 1;
3518 }
3519
3520 static int parse_submodule_opt(struct diff_options *options, const char *value)
3521 {
3522         if (parse_submodule_params(options, value))
3523                 die(_("Failed to parse --submodule option parameter: '%s'"),
3524                         value);
3525         return 1;
3526 }
3527
3528 static const char diff_status_letters[] = {
3529         DIFF_STATUS_ADDED,
3530         DIFF_STATUS_COPIED,
3531         DIFF_STATUS_DELETED,
3532         DIFF_STATUS_MODIFIED,
3533         DIFF_STATUS_RENAMED,
3534         DIFF_STATUS_TYPE_CHANGED,
3535         DIFF_STATUS_UNKNOWN,
3536         DIFF_STATUS_UNMERGED,
3537         DIFF_STATUS_FILTER_AON,
3538         DIFF_STATUS_FILTER_BROKEN,
3539         '\0',
3540 };
3541
3542 static unsigned int filter_bit['Z' + 1];
3543
3544 static void prepare_filter_bits(void)
3545 {
3546         int i;
3547
3548         if (!filter_bit[DIFF_STATUS_ADDED]) {
3549                 for (i = 0; diff_status_letters[i]; i++)
3550                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3551         }
3552 }
3553
3554 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3555 {
3556         return opt->filter & filter_bit[(int) status];
3557 }
3558
3559 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3560 {
3561         int i, optch;
3562
3563         prepare_filter_bits();
3564
3565         /*
3566          * If there is a negation e.g. 'd' in the input, and we haven't
3567          * initialized the filter field with another --diff-filter, start
3568          * from full set of bits, except for AON.
3569          */
3570         if (!opt->filter) {
3571                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3572                         if (optch < 'a' || 'z' < optch)
3573                                 continue;
3574                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3575                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3576                         break;
3577                 }
3578         }
3579
3580         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3581                 unsigned int bit;
3582                 int negate;
3583
3584                 if ('a' <= optch && optch <= 'z') {
3585                         negate = 1;
3586                         optch = toupper(optch);
3587                 } else {
3588                         negate = 0;
3589                 }
3590
3591                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3592                 if (!bit)
3593                         return optarg[i];
3594                 if (negate)
3595                         opt->filter &= ~bit;
3596                 else
3597                         opt->filter |= bit;
3598         }
3599         return 0;
3600 }
3601
3602 /* Used only by "diff-files" and "diff --no-index" */
3603 void handle_deprecated_show_diff_q(struct diff_options *opt)
3604 {
3605         warning("'diff -q' and 'diff-files -q' are deprecated.");
3606         warning("Use 'diff --diff-filter=d' instead to ignore deleted filepairs.");
3607         parse_diff_filter_opt("d", opt);
3608 }
3609
3610 static void enable_patch_output(int *fmt) {
3611         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3612         *fmt |= DIFF_FORMAT_PATCH;
3613 }
3614
3615 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3616 {
3617         const char *arg = av[0];
3618         const char *optarg;
3619         int argcount;
3620
3621         /* Output format options */
3622         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3623             || opt_arg(arg, 'U', "unified", &options->context))
3624                 enable_patch_output(&options->output_format);
3625         else if (!strcmp(arg, "--raw"))
3626                 options->output_format |= DIFF_FORMAT_RAW;
3627         else if (!strcmp(arg, "--patch-with-raw")) {
3628                 enable_patch_output(&options->output_format);
3629                 options->output_format |= DIFF_FORMAT_RAW;
3630         } else if (!strcmp(arg, "--numstat"))
3631                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3632         else if (!strcmp(arg, "--shortstat"))
3633                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3634         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3635                 return parse_dirstat_opt(options, "");
3636         else if (starts_with(arg, "-X"))
3637                 return parse_dirstat_opt(options, arg + 2);
3638         else if (starts_with(arg, "--dirstat="))
3639                 return parse_dirstat_opt(options, arg + 10);
3640         else if (!strcmp(arg, "--cumulative"))
3641                 return parse_dirstat_opt(options, "cumulative");
3642         else if (!strcmp(arg, "--dirstat-by-file"))
3643                 return parse_dirstat_opt(options, "files");
3644         else if (starts_with(arg, "--dirstat-by-file=")) {
3645                 parse_dirstat_opt(options, "files");
3646                 return parse_dirstat_opt(options, arg + 18);
3647         }
3648         else if (!strcmp(arg, "--check"))
3649                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3650         else if (!strcmp(arg, "--summary"))
3651                 options->output_format |= DIFF_FORMAT_SUMMARY;
3652         else if (!strcmp(arg, "--patch-with-stat")) {
3653                 enable_patch_output(&options->output_format);
3654                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3655         } else if (!strcmp(arg, "--name-only"))
3656                 options->output_format |= DIFF_FORMAT_NAME;
3657         else if (!strcmp(arg, "--name-status"))
3658                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3659         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3660                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3661         else if (starts_with(arg, "--stat"))
3662                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3663                 return stat_opt(options, av);
3664
3665         /* renames options */
3666         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3667                  !strcmp(arg, "--break-rewrites")) {
3668                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3669                         return error("invalid argument to -B: %s", arg+2);
3670         }
3671         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3672                  !strcmp(arg, "--find-renames")) {
3673                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3674                         return error("invalid argument to -M: %s", arg+2);
3675                 options->detect_rename = DIFF_DETECT_RENAME;
3676         }
3677         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3678                 options->irreversible_delete = 1;
3679         }
3680         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3681                  !strcmp(arg, "--find-copies")) {
3682                 if (options->detect_rename == DIFF_DETECT_COPY)
3683                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3684                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3685                         return error("invalid argument to -C: %s", arg+2);
3686                 options->detect_rename = DIFF_DETECT_COPY;
3687         }
3688         else if (!strcmp(arg, "--no-renames"))
3689                 options->detect_rename = 0;
3690         else if (!strcmp(arg, "--rename-empty"))
3691                 DIFF_OPT_SET(options, RENAME_EMPTY);
3692         else if (!strcmp(arg, "--no-rename-empty"))
3693                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3694         else if (!strcmp(arg, "--relative"))
3695                 DIFF_OPT_SET(options, RELATIVE_NAME);
3696         else if (starts_with(arg, "--relative=")) {
3697                 DIFF_OPT_SET(options, RELATIVE_NAME);
3698                 options->prefix = arg + 11;
3699         }
3700
3701         /* xdiff options */
3702         else if (!strcmp(arg, "--minimal"))
3703                 DIFF_XDL_SET(options, NEED_MINIMAL);
3704         else if (!strcmp(arg, "--no-minimal"))
3705                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3706         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3707                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3708         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3709                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3710         else if (!strcmp(arg, "--ignore-space-at-eol"))
3711                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3712         else if (!strcmp(arg, "--ignore-blank-lines"))
3713                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3714         else if (!strcmp(arg, "--patience"))
3715                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3716         else if (!strcmp(arg, "--histogram"))
3717                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3718         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3719                 long value = parse_algorithm_value(optarg);
3720                 if (value < 0)
3721                         return error("option diff-algorithm accepts \"myers\", "
3722                                      "\"minimal\", \"patience\" and \"histogram\"");
3723                 /* clear out previous settings */
3724                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3725                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3726                 options->xdl_opts |= value;
3727                 return argcount;
3728         }
3729
3730         /* flags options */
3731         else if (!strcmp(arg, "--binary")) {
3732                 enable_patch_output(&options->output_format);
3733                 DIFF_OPT_SET(options, BINARY);
3734         }
3735         else if (!strcmp(arg, "--full-index"))
3736                 DIFF_OPT_SET(options, FULL_INDEX);
3737         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3738                 DIFF_OPT_SET(options, TEXT);
3739         else if (!strcmp(arg, "-R"))
3740                 DIFF_OPT_SET(options, REVERSE_DIFF);
3741         else if (!strcmp(arg, "--find-copies-harder"))
3742                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3743         else if (!strcmp(arg, "--follow"))
3744                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3745         else if (!strcmp(arg, "--no-follow"))
3746                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3747         else if (!strcmp(arg, "--color"))
3748                 options->use_color = 1;
3749         else if (starts_with(arg, "--color=")) {
3750                 int value = git_config_colorbool(NULL, arg+8);
3751                 if (value < 0)
3752                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3753                 options->use_color = value;
3754         }
3755         else if (!strcmp(arg, "--no-color"))
3756                 options->use_color = 0;
3757         else if (!strcmp(arg, "--color-words")) {
3758                 options->use_color = 1;
3759                 options->word_diff = DIFF_WORDS_COLOR;
3760         }
3761         else if (starts_with(arg, "--color-words=")) {
3762                 options->use_color = 1;
3763                 options->word_diff = DIFF_WORDS_COLOR;
3764                 options->word_regex = arg + 14;
3765         }
3766         else if (!strcmp(arg, "--word-diff")) {
3767                 if (options->word_diff == DIFF_WORDS_NONE)
3768                         options->word_diff = DIFF_WORDS_PLAIN;
3769         }
3770         else if (starts_with(arg, "--word-diff=")) {
3771                 const char *type = arg + 12;
3772                 if (!strcmp(type, "plain"))
3773                         options->word_diff = DIFF_WORDS_PLAIN;
3774                 else if (!strcmp(type, "color")) {
3775                         options->use_color = 1;
3776                         options->word_diff = DIFF_WORDS_COLOR;
3777                 }
3778                 else if (!strcmp(type, "porcelain"))
3779                         options->word_diff = DIFF_WORDS_PORCELAIN;
3780                 else if (!strcmp(type, "none"))
3781                         options->word_diff = DIFF_WORDS_NONE;
3782                 else
3783                         die("bad --word-diff argument: %s", type);
3784         }
3785         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3786                 if (options->word_diff == DIFF_WORDS_NONE)
3787                         options->word_diff = DIFF_WORDS_PLAIN;
3788                 options->word_regex = optarg;
3789                 return argcount;
3790         }
3791         else if (!strcmp(arg, "--exit-code"))
3792                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3793         else if (!strcmp(arg, "--quiet"))
3794                 DIFF_OPT_SET(options, QUICK);
3795         else if (!strcmp(arg, "--ext-diff"))
3796                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3797         else if (!strcmp(arg, "--no-ext-diff"))
3798                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3799         else if (!strcmp(arg, "--textconv"))
3800                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3801         else if (!strcmp(arg, "--no-textconv"))
3802                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3803         else if (!strcmp(arg, "--ignore-submodules")) {
3804                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3805                 handle_ignore_submodules_arg(options, "all");
3806         } else if (starts_with(arg, "--ignore-submodules=")) {
3807                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3808                 handle_ignore_submodules_arg(options, arg + 20);
3809         } else if (!strcmp(arg, "--submodule"))
3810                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3811         else if (starts_with(arg, "--submodule="))
3812                 return parse_submodule_opt(options, arg + 12);
3813
3814         /* misc options */
3815         else if (!strcmp(arg, "-z"))
3816                 options->line_termination = 0;
3817         else if ((argcount = short_opt('l', av, &optarg))) {
3818                 options->rename_limit = strtoul(optarg, NULL, 10);
3819                 return argcount;
3820         }
3821         else if ((argcount = short_opt('S', av, &optarg))) {
3822                 options->pickaxe = optarg;
3823                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3824                 return argcount;
3825         } else if ((argcount = short_opt('G', av, &optarg))) {
3826                 options->pickaxe = optarg;
3827                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3828                 return argcount;
3829         }
3830         else if (!strcmp(arg, "--pickaxe-all"))
3831                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3832         else if (!strcmp(arg, "--pickaxe-regex"))
3833                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3834         else if ((argcount = short_opt('O', av, &optarg))) {
3835                 options->orderfile = optarg;
3836                 return argcount;
3837         }
3838         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3839                 int offending = parse_diff_filter_opt(optarg, options);
3840                 if (offending)
3841                         die("unknown change class '%c' in --diff-filter=%s",
3842                             offending, optarg);
3843                 return argcount;
3844         }
3845         else if (!strcmp(arg, "--abbrev"))
3846                 options->abbrev = DEFAULT_ABBREV;
3847         else if (starts_with(arg, "--abbrev=")) {
3848                 options->abbrev = strtoul(arg + 9, NULL, 10);
3849                 if (options->abbrev < MINIMUM_ABBREV)
3850                         options->abbrev = MINIMUM_ABBREV;
3851                 else if (40 < options->abbrev)
3852                         options->abbrev = 40;
3853         }
3854         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3855                 options->a_prefix = optarg;
3856                 return argcount;
3857         }
3858         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3859                 options->b_prefix = optarg;
3860                 return argcount;
3861         }
3862         else if (!strcmp(arg, "--no-prefix"))
3863                 options->a_prefix = options->b_prefix = "";
3864         else if (opt_arg(arg, '\0', "inter-hunk-context",
3865                          &options->interhunkcontext))
3866                 ;
3867         else if (!strcmp(arg, "-W"))
3868                 DIFF_OPT_SET(options, FUNCCONTEXT);
3869         else if (!strcmp(arg, "--function-context"))
3870                 DIFF_OPT_SET(options, FUNCCONTEXT);
3871         else if (!strcmp(arg, "--no-function-context"))
3872                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3873         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3874                 options->file = fopen(optarg, "w");
3875                 if (!options->file)
3876                         die_errno("Could not open '%s'", optarg);
3877                 options->close_file = 1;
3878                 return argcount;
3879         } else
3880                 return 0;
3881         return 1;
3882 }
3883
3884 int parse_rename_score(const char **cp_p)
3885 {
3886         unsigned long num, scale;
3887         int ch, dot;
3888         const char *cp = *cp_p;
3889
3890         num = 0;
3891         scale = 1;
3892         dot = 0;
3893         for (;;) {
3894                 ch = *cp;
3895                 if ( !dot && ch == '.' ) {
3896                         scale = 1;
3897                         dot = 1;
3898                 } else if ( ch == '%' ) {
3899                         scale = dot ? scale*100 : 100;
3900                         cp++;   /* % is always at the end */
3901                         break;
3902                 } else if ( ch >= '0' && ch <= '9' ) {
3903                         if ( scale < 100000 ) {
3904                                 scale *= 10;
3905                                 num = (num*10) + (ch-'0');
3906                         }
3907                 } else {
3908                         break;
3909                 }
3910                 cp++;
3911         }
3912         *cp_p = cp;
3913
3914         /* user says num divided by scale and we say internally that
3915          * is MAX_SCORE * num / scale.
3916          */
3917         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3918 }
3919
3920 static int diff_scoreopt_parse(const char *opt)
3921 {
3922         int opt1, opt2, cmd;
3923
3924         if (*opt++ != '-')
3925                 return -1;
3926         cmd = *opt++;
3927         if (cmd == '-') {
3928                 /* convert the long-form arguments into short-form versions */
3929                 if (starts_with(opt, "break-rewrites")) {
3930                         opt += strlen("break-rewrites");
3931                         if (*opt == 0 || *opt++ == '=')
3932                                 cmd = 'B';
3933                 } else if (starts_with(opt, "find-copies")) {
3934                         opt += strlen("find-copies");
3935                         if (*opt == 0 || *opt++ == '=')
3936                                 cmd = 'C';
3937                 } else if (starts_with(opt, "find-renames")) {
3938                         opt += strlen("find-renames");
3939                         if (*opt == 0 || *opt++ == '=')
3940                                 cmd = 'M';
3941                 }
3942         }
3943         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3944                 return -1; /* that is not a -M, -C nor -B option */
3945
3946         opt1 = parse_rename_score(&opt);
3947         if (cmd != 'B')
3948                 opt2 = 0;
3949         else {
3950                 if (*opt == 0)
3951                         opt2 = 0;
3952                 else if (*opt != '/')
3953                         return -1; /* we expect -B80/99 or -B80 */
3954                 else {
3955                         opt++;
3956                         opt2 = parse_rename_score(&opt);
3957                 }
3958         }
3959         if (*opt != 0)
3960                 return -1;
3961         return opt1 | (opt2 << 16);
3962 }
3963
3964 struct diff_queue_struct diff_queued_diff;
3965
3966 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3967 {
3968         if (queue->alloc <= queue->nr) {
3969                 queue->alloc = alloc_nr(queue->alloc);
3970                 queue->queue = xrealloc(queue->queue,
3971                                         sizeof(dp) * queue->alloc);
3972         }
3973         queue->queue[queue->nr++] = dp;
3974 }
3975
3976 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3977                                  struct diff_filespec *one,
3978                                  struct diff_filespec *two)
3979 {
3980         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3981         dp->one = one;
3982         dp->two = two;
3983         if (queue)
3984                 diff_q(queue, dp);
3985         return dp;
3986 }
3987
3988 void diff_free_filepair(struct diff_filepair *p)
3989 {
3990         free_filespec(p->one);
3991         free_filespec(p->two);
3992         free(p);
3993 }
3994
3995 /* This is different from find_unique_abbrev() in that
3996  * it stuffs the result with dots for alignment.
3997  */
3998 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3999 {
4000         int abblen;
4001         const char *abbrev;
4002         if (len == 40)
4003                 return sha1_to_hex(sha1);
4004
4005         abbrev = find_unique_abbrev(sha1, len);
4006         abblen = strlen(abbrev);
4007         if (abblen < 37) {
4008                 static char hex[41];
4009                 if (len < abblen && abblen <= len + 2)
4010                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
4011                 else
4012                         sprintf(hex, "%s...", abbrev);
4013                 return hex;
4014         }
4015         return sha1_to_hex(sha1);
4016 }
4017
4018 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4019 {
4020         int line_termination = opt->line_termination;
4021         int inter_name_termination = line_termination ? '\t' : '\0';
4022
4023         fprintf(opt->file, "%s", diff_line_prefix(opt));
4024         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4025                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4026                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
4027                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4028         }
4029         if (p->score) {
4030                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4031                         inter_name_termination);
4032         } else {
4033                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4034         }
4035
4036         if (p->status == DIFF_STATUS_COPIED ||
4037             p->status == DIFF_STATUS_RENAMED) {
4038                 const char *name_a, *name_b;
4039                 name_a = p->one->path;
4040                 name_b = p->two->path;
4041                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4042                 write_name_quoted(name_a, opt->file, inter_name_termination);
4043                 write_name_quoted(name_b, opt->file, line_termination);
4044         } else {
4045                 const char *name_a, *name_b;
4046                 name_a = p->one->mode ? p->one->path : p->two->path;
4047                 name_b = NULL;
4048                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4049                 write_name_quoted(name_a, opt->file, line_termination);
4050         }
4051 }
4052
4053 int diff_unmodified_pair(struct diff_filepair *p)
4054 {
4055         /* This function is written stricter than necessary to support
4056          * the currently implemented transformers, but the idea is to
4057          * let transformers to produce diff_filepairs any way they want,
4058          * and filter and clean them up here before producing the output.
4059          */
4060         struct diff_filespec *one = p->one, *two = p->two;
4061
4062         if (DIFF_PAIR_UNMERGED(p))
4063                 return 0; /* unmerged is interesting */
4064
4065         /* deletion, addition, mode or type change
4066          * and rename are all interesting.
4067          */
4068         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4069             DIFF_PAIR_MODE_CHANGED(p) ||
4070             strcmp(one->path, two->path))
4071                 return 0;
4072
4073         /* both are valid and point at the same path.  that is, we are
4074          * dealing with a change.
4075          */
4076         if (one->sha1_valid && two->sha1_valid &&
4077             !hashcmp(one->sha1, two->sha1) &&
4078             !one->dirty_submodule && !two->dirty_submodule)
4079                 return 1; /* no change */
4080         if (!one->sha1_valid && !two->sha1_valid)
4081                 return 1; /* both look at the same file on the filesystem. */
4082         return 0;
4083 }
4084
4085 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4086 {
4087         if (diff_unmodified_pair(p))
4088                 return;
4089
4090         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4091             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4092                 return; /* no tree diffs in patch format */
4093
4094         run_diff(p, o);
4095 }
4096
4097 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4098                             struct diffstat_t *diffstat)
4099 {
4100         if (diff_unmodified_pair(p))
4101                 return;
4102
4103         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4104             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4105                 return; /* no useful stat for tree diffs */
4106
4107         run_diffstat(p, o, diffstat);
4108 }
4109
4110 static void diff_flush_checkdiff(struct diff_filepair *p,
4111                 struct diff_options *o)
4112 {
4113         if (diff_unmodified_pair(p))
4114                 return;
4115
4116         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4117             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4118                 return; /* nothing to check in tree diffs */
4119
4120         run_checkdiff(p, o);
4121 }
4122
4123 int diff_queue_is_empty(void)
4124 {
4125         struct diff_queue_struct *q = &diff_queued_diff;
4126         int i;
4127         for (i = 0; i < q->nr; i++)
4128                 if (!diff_unmodified_pair(q->queue[i]))
4129                         return 0;
4130         return 1;
4131 }
4132
4133 #if DIFF_DEBUG
4134 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4135 {
4136         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4137                 x, one ? one : "",
4138                 s->path,
4139                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4140                 s->mode,
4141                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4142         fprintf(stderr, "queue[%d] %s size %lu\n",
4143                 x, one ? one : "",
4144                 s->size);
4145 }
4146
4147 void diff_debug_filepair(const struct diff_filepair *p, int i)
4148 {
4149         diff_debug_filespec(p->one, i, "one");
4150         diff_debug_filespec(p->two, i, "two");
4151         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4152                 p->score, p->status ? p->status : '?',
4153                 p->one->rename_used, p->broken_pair);
4154 }
4155
4156 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4157 {
4158         int i;
4159         if (msg)
4160                 fprintf(stderr, "%s\n", msg);
4161         fprintf(stderr, "q->nr = %d\n", q->nr);
4162         for (i = 0; i < q->nr; i++) {
4163                 struct diff_filepair *p = q->queue[i];
4164                 diff_debug_filepair(p, i);
4165         }
4166 }
4167 #endif
4168
4169 static void diff_resolve_rename_copy(void)
4170 {
4171         int i;
4172         struct diff_filepair *p;
4173         struct diff_queue_struct *q = &diff_queued_diff;
4174
4175         diff_debug_queue("resolve-rename-copy", q);
4176
4177         for (i = 0; i < q->nr; i++) {
4178                 p = q->queue[i];
4179                 p->status = 0; /* undecided */
4180                 if (DIFF_PAIR_UNMERGED(p))
4181                         p->status = DIFF_STATUS_UNMERGED;
4182                 else if (!DIFF_FILE_VALID(p->one))
4183                         p->status = DIFF_STATUS_ADDED;
4184                 else if (!DIFF_FILE_VALID(p->two))
4185                         p->status = DIFF_STATUS_DELETED;
4186                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4187                         p->status = DIFF_STATUS_TYPE_CHANGED;
4188
4189                 /* from this point on, we are dealing with a pair
4190                  * whose both sides are valid and of the same type, i.e.
4191                  * either in-place edit or rename/copy edit.
4192                  */
4193                 else if (DIFF_PAIR_RENAME(p)) {
4194                         /*
4195                          * A rename might have re-connected a broken
4196                          * pair up, causing the pathnames to be the
4197                          * same again. If so, that's not a rename at
4198                          * all, just a modification..
4199                          *
4200                          * Otherwise, see if this source was used for
4201                          * multiple renames, in which case we decrement
4202                          * the count, and call it a copy.
4203                          */
4204                         if (!strcmp(p->one->path, p->two->path))
4205                                 p->status = DIFF_STATUS_MODIFIED;
4206                         else if (--p->one->rename_used > 0)
4207                                 p->status = DIFF_STATUS_COPIED;
4208                         else
4209                                 p->status = DIFF_STATUS_RENAMED;
4210                 }
4211                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4212                          p->one->mode != p->two->mode ||
4213                          p->one->dirty_submodule ||
4214                          p->two->dirty_submodule ||
4215                          is_null_sha1(p->one->sha1))
4216                         p->status = DIFF_STATUS_MODIFIED;
4217                 else {
4218                         /* This is a "no-change" entry and should not
4219                          * happen anymore, but prepare for broken callers.
4220                          */
4221                         error("feeding unmodified %s to diffcore",
4222                               p->one->path);
4223                         p->status = DIFF_STATUS_UNKNOWN;
4224                 }
4225         }
4226         diff_debug_queue("resolve-rename-copy done", q);
4227 }
4228
4229 static int check_pair_status(struct diff_filepair *p)
4230 {
4231         switch (p->status) {
4232         case DIFF_STATUS_UNKNOWN:
4233                 return 0;
4234         case 0:
4235                 die("internal error in diff-resolve-rename-copy");
4236         default:
4237                 return 1;
4238         }
4239 }
4240
4241 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4242 {
4243         int fmt = opt->output_format;
4244
4245         if (fmt & DIFF_FORMAT_CHECKDIFF)
4246                 diff_flush_checkdiff(p, opt);
4247         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4248                 diff_flush_raw(p, opt);
4249         else if (fmt & DIFF_FORMAT_NAME) {
4250                 const char *name_a, *name_b;
4251                 name_a = p->two->path;
4252                 name_b = NULL;
4253                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4254                 write_name_quoted(name_a, opt->file, opt->line_termination);
4255         }
4256 }
4257
4258 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4259 {
4260         if (fs->mode)
4261                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4262         else
4263                 fprintf(file, " %s ", newdelete);
4264         write_name_quoted(fs->path, file, '\n');
4265 }
4266
4267
4268 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4269                 const char *line_prefix)
4270 {
4271         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4272                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4273                         p->two->mode, show_name ? ' ' : '\n');
4274                 if (show_name) {
4275                         write_name_quoted(p->two->path, file, '\n');
4276                 }
4277         }
4278 }
4279
4280 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4281                         const char *line_prefix)
4282 {
4283         char *names = pprint_rename(p->one->path, p->two->path);
4284
4285         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4286         free(names);
4287         show_mode_change(file, p, 0, line_prefix);
4288 }
4289
4290 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4291 {
4292         FILE *file = opt->file;
4293         const char *line_prefix = diff_line_prefix(opt);
4294
4295         switch(p->status) {
4296         case DIFF_STATUS_DELETED:
4297                 fputs(line_prefix, file);
4298                 show_file_mode_name(file, "delete", p->one);
4299                 break;
4300         case DIFF_STATUS_ADDED:
4301                 fputs(line_prefix, file);
4302                 show_file_mode_name(file, "create", p->two);
4303                 break;
4304         case DIFF_STATUS_COPIED:
4305                 fputs(line_prefix, file);
4306                 show_rename_copy(file, "copy", p, line_prefix);
4307                 break;
4308         case DIFF_STATUS_RENAMED:
4309                 fputs(line_prefix, file);
4310                 show_rename_copy(file, "rename", p, line_prefix);
4311                 break;
4312         default:
4313                 if (p->score) {
4314                         fprintf(file, "%s rewrite ", line_prefix);
4315                         write_name_quoted(p->two->path, file, ' ');
4316                         fprintf(file, "(%d%%)\n", similarity_index(p));
4317                 }
4318                 show_mode_change(file, p, !p->score, line_prefix);
4319                 break;
4320         }
4321 }
4322
4323 struct patch_id_t {
4324         git_SHA_CTX *ctx;
4325         int patchlen;
4326 };
4327
4328 static int remove_space(char *line, int len)
4329 {
4330         int i;
4331         char *dst = line;
4332         unsigned char c;
4333
4334         for (i = 0; i < len; i++)
4335                 if (!isspace((c = line[i])))
4336                         *dst++ = c;
4337
4338         return dst - line;
4339 }
4340
4341 static void patch_id_consume(void *priv, char *line, unsigned long len)
4342 {
4343         struct patch_id_t *data = priv;
4344         int new_len;
4345
4346         /* Ignore line numbers when computing the SHA1 of the patch */
4347         if (starts_with(line, "@@ -"))
4348                 return;
4349
4350         new_len = remove_space(line, len);
4351
4352         git_SHA1_Update(data->ctx, line, new_len);
4353         data->patchlen += new_len;
4354 }
4355
4356 /* returns 0 upon success, and writes result into sha1 */
4357 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4358 {
4359         struct diff_queue_struct *q = &diff_queued_diff;
4360         int i;
4361         git_SHA_CTX ctx;
4362         struct patch_id_t data;
4363         char buffer[PATH_MAX * 4 + 20];
4364
4365         git_SHA1_Init(&ctx);
4366         memset(&data, 0, sizeof(struct patch_id_t));
4367         data.ctx = &ctx;
4368
4369         for (i = 0; i < q->nr; i++) {
4370                 xpparam_t xpp;
4371                 xdemitconf_t xecfg;
4372                 mmfile_t mf1, mf2;
4373                 struct diff_filepair *p = q->queue[i];
4374                 int len1, len2;
4375
4376                 memset(&xpp, 0, sizeof(xpp));
4377                 memset(&xecfg, 0, sizeof(xecfg));
4378                 if (p->status == 0)
4379                         return error("internal diff status error");
4380                 if (p->status == DIFF_STATUS_UNKNOWN)
4381                         continue;
4382                 if (diff_unmodified_pair(p))
4383                         continue;
4384                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4385                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4386                         continue;
4387                 if (DIFF_PAIR_UNMERGED(p))
4388                         continue;
4389
4390                 diff_fill_sha1_info(p->one);
4391                 diff_fill_sha1_info(p->two);
4392                 if (fill_mmfile(&mf1, p->one) < 0 ||
4393                                 fill_mmfile(&mf2, p->two) < 0)
4394                         return error("unable to read files to diff");
4395
4396                 len1 = remove_space(p->one->path, strlen(p->one->path));
4397                 len2 = remove_space(p->two->path, strlen(p->two->path));
4398                 if (p->one->mode == 0)
4399                         len1 = snprintf(buffer, sizeof(buffer),
4400                                         "diff--gita/%.*sb/%.*s"
4401                                         "newfilemode%06o"
4402                                         "---/dev/null"
4403                                         "+++b/%.*s",
4404                                         len1, p->one->path,
4405                                         len2, p->two->path,
4406                                         p->two->mode,
4407                                         len2, p->two->path);
4408                 else if (p->two->mode == 0)
4409                         len1 = snprintf(buffer, sizeof(buffer),
4410                                         "diff--gita/%.*sb/%.*s"
4411                                         "deletedfilemode%06o"
4412                                         "---a/%.*s"
4413                                         "+++/dev/null",
4414                                         len1, p->one->path,
4415                                         len2, p->two->path,
4416                                         p->one->mode,
4417                                         len1, p->one->path);
4418                 else
4419                         len1 = snprintf(buffer, sizeof(buffer),
4420                                         "diff--gita/%.*sb/%.*s"
4421                                         "---a/%.*s"
4422                                         "+++b/%.*s",
4423                                         len1, p->one->path,
4424                                         len2, p->two->path,
4425                                         len1, p->one->path,
4426                                         len2, p->two->path);
4427                 git_SHA1_Update(&ctx, buffer, len1);
4428
4429                 if (diff_filespec_is_binary(p->one) ||
4430                     diff_filespec_is_binary(p->two)) {
4431                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4432                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4433                         continue;
4434                 }
4435
4436                 xpp.flags = 0;
4437                 xecfg.ctxlen = 3;
4438                 xecfg.flags = 0;
4439                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4440                               &xpp, &xecfg);
4441         }
4442
4443         git_SHA1_Final(sha1, &ctx);
4444         return 0;
4445 }
4446
4447 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4448 {
4449         struct diff_queue_struct *q = &diff_queued_diff;
4450         int i;
4451         int result = diff_get_patch_id(options, sha1);
4452
4453         for (i = 0; i < q->nr; i++)
4454                 diff_free_filepair(q->queue[i]);
4455
4456         free(q->queue);
4457         DIFF_QUEUE_CLEAR(q);
4458
4459         return result;
4460 }
4461
4462 static int is_summary_empty(const struct diff_queue_struct *q)
4463 {
4464         int i;
4465
4466         for (i = 0; i < q->nr; i++) {
4467                 const struct diff_filepair *p = q->queue[i];
4468
4469                 switch (p->status) {
4470                 case DIFF_STATUS_DELETED:
4471                 case DIFF_STATUS_ADDED:
4472                 case DIFF_STATUS_COPIED:
4473                 case DIFF_STATUS_RENAMED:
4474                         return 0;
4475                 default:
4476                         if (p->score)
4477                                 return 0;
4478                         if (p->one->mode && p->two->mode &&
4479                             p->one->mode != p->two->mode)
4480                                 return 0;
4481                         break;
4482                 }
4483         }
4484         return 1;
4485 }
4486
4487 static const char rename_limit_warning[] =
4488 "inexact rename detection was skipped due to too many files.";
4489
4490 static const char degrade_cc_to_c_warning[] =
4491 "only found copies from modified paths due to too many files.";
4492
4493 static const char rename_limit_advice[] =
4494 "you may want to set your %s variable to at least "
4495 "%d and retry the command.";
4496
4497 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4498 {
4499         if (degraded_cc)
4500                 warning(degrade_cc_to_c_warning);
4501         else if (needed)
4502                 warning(rename_limit_warning);
4503         else
4504                 return;
4505         if (0 < needed && needed < 32767)
4506                 warning(rename_limit_advice, varname, needed);
4507 }
4508
4509 void diff_flush(struct diff_options *options)
4510 {
4511         struct diff_queue_struct *q = &diff_queued_diff;
4512         int i, output_format = options->output_format;
4513         int separator = 0;
4514         int dirstat_by_line = 0;
4515
4516         /*
4517          * Order: raw, stat, summary, patch
4518          * or:    name/name-status/checkdiff (other bits clear)
4519          */
4520         if (!q->nr)
4521                 goto free_queue;
4522
4523         if (output_format & (DIFF_FORMAT_RAW |
4524                              DIFF_FORMAT_NAME |
4525                              DIFF_FORMAT_NAME_STATUS |
4526                              DIFF_FORMAT_CHECKDIFF)) {
4527                 for (i = 0; i < q->nr; i++) {
4528                         struct diff_filepair *p = q->queue[i];
4529                         if (check_pair_status(p))
4530                                 flush_one_pair(p, options);
4531                 }
4532                 separator++;
4533         }
4534
4535         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4536                 dirstat_by_line = 1;
4537
4538         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4539             dirstat_by_line) {
4540                 struct diffstat_t diffstat;
4541
4542                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4543                 for (i = 0; i < q->nr; i++) {
4544                         struct diff_filepair *p = q->queue[i];
4545                         if (check_pair_status(p))
4546                                 diff_flush_stat(p, options, &diffstat);
4547                 }
4548                 if (output_format & DIFF_FORMAT_NUMSTAT)
4549                         show_numstat(&diffstat, options);
4550                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4551                         show_stats(&diffstat, options);
4552                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4553                         show_shortstats(&diffstat, options);
4554                 if (output_format & DIFF_FORMAT_DIRSTAT)
4555                         show_dirstat_by_line(&diffstat, options);
4556                 free_diffstat_info(&diffstat);
4557                 separator++;
4558         }
4559         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4560                 show_dirstat(options);
4561
4562         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4563                 for (i = 0; i < q->nr; i++) {
4564                         diff_summary(options, q->queue[i]);
4565                 }
4566                 separator++;
4567         }
4568
4569         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4570             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4571             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4572                 /*
4573                  * run diff_flush_patch for the exit status. setting
4574                  * options->file to /dev/null should be safe, because we
4575                  * aren't supposed to produce any output anyway.
4576                  */
4577                 if (options->close_file)
4578                         fclose(options->file);
4579                 options->file = fopen("/dev/null", "w");
4580                 if (!options->file)
4581                         die_errno("Could not open /dev/null");
4582                 options->close_file = 1;
4583                 for (i = 0; i < q->nr; i++) {
4584                         struct diff_filepair *p = q->queue[i];
4585                         if (check_pair_status(p))
4586                                 diff_flush_patch(p, options);
4587                         if (options->found_changes)
4588                                 break;
4589                 }
4590         }
4591
4592         if (output_format & DIFF_FORMAT_PATCH) {
4593                 if (separator) {
4594                         fprintf(options->file, "%s%c",
4595                                 diff_line_prefix(options),
4596                                 options->line_termination);
4597                         if (options->stat_sep) {
4598                                 /* attach patch instead of inline */
4599                                 fputs(options->stat_sep, options->file);
4600                         }
4601                 }
4602
4603                 for (i = 0; i < q->nr; i++) {
4604                         struct diff_filepair *p = q->queue[i];
4605                         if (check_pair_status(p))
4606                                 diff_flush_patch(p, options);
4607                 }
4608         }
4609
4610         if (output_format & DIFF_FORMAT_CALLBACK)
4611                 options->format_callback(q, options, options->format_callback_data);
4612
4613         for (i = 0; i < q->nr; i++)
4614                 diff_free_filepair(q->queue[i]);
4615 free_queue:
4616         free(q->queue);
4617         DIFF_QUEUE_CLEAR(q);
4618         if (options->close_file)
4619                 fclose(options->file);
4620
4621         /*
4622          * Report the content-level differences with HAS_CHANGES;
4623          * diff_addremove/diff_change does not set the bit when
4624          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4625          */
4626         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4627                 if (options->found_changes)
4628                         DIFF_OPT_SET(options, HAS_CHANGES);
4629                 else
4630                         DIFF_OPT_CLR(options, HAS_CHANGES);
4631         }
4632 }
4633
4634 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4635 {
4636         return (((p->status == DIFF_STATUS_MODIFIED) &&
4637                  ((p->score &&
4638                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4639                   (!p->score &&
4640                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4641                 ((p->status != DIFF_STATUS_MODIFIED) &&
4642                  filter_bit_tst(p->status, options)));
4643 }
4644
4645 static void diffcore_apply_filter(struct diff_options *options)
4646 {
4647         int i;
4648         struct diff_queue_struct *q = &diff_queued_diff;
4649         struct diff_queue_struct outq;
4650
4651         DIFF_QUEUE_CLEAR(&outq);
4652
4653         if (!options->filter)
4654                 return;
4655
4656         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4657                 int found;
4658                 for (i = found = 0; !found && i < q->nr; i++) {
4659                         if (match_filter(options, q->queue[i]))
4660                                 found++;
4661                 }
4662                 if (found)
4663                         return;
4664
4665                 /* otherwise we will clear the whole queue
4666                  * by copying the empty outq at the end of this
4667                  * function, but first clear the current entries
4668                  * in the queue.
4669                  */
4670                 for (i = 0; i < q->nr; i++)
4671                         diff_free_filepair(q->queue[i]);
4672         }
4673         else {
4674                 /* Only the matching ones */
4675                 for (i = 0; i < q->nr; i++) {
4676                         struct diff_filepair *p = q->queue[i];
4677                         if (match_filter(options, p))
4678                                 diff_q(&outq, p);
4679                         else
4680                                 diff_free_filepair(p);
4681                 }
4682         }
4683         free(q->queue);
4684         *q = outq;
4685 }
4686
4687 /* Check whether two filespecs with the same mode and size are identical */
4688 static int diff_filespec_is_identical(struct diff_filespec *one,
4689                                       struct diff_filespec *two)
4690 {
4691         if (S_ISGITLINK(one->mode))
4692                 return 0;
4693         if (diff_populate_filespec(one, 0))
4694                 return 0;
4695         if (diff_populate_filespec(two, 0))
4696                 return 0;
4697         return !memcmp(one->data, two->data, one->size);
4698 }
4699
4700 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4701 {
4702         if (p->done_skip_stat_unmatch)
4703                 return p->skip_stat_unmatch_result;
4704
4705         p->done_skip_stat_unmatch = 1;
4706         p->skip_stat_unmatch_result = 0;
4707         /*
4708          * 1. Entries that come from stat info dirtiness
4709          *    always have both sides (iow, not create/delete),
4710          *    one side of the object name is unknown, with
4711          *    the same mode and size.  Keep the ones that
4712          *    do not match these criteria.  They have real
4713          *    differences.
4714          *
4715          * 2. At this point, the file is known to be modified,
4716          *    with the same mode and size, and the object
4717          *    name of one side is unknown.  Need to inspect
4718          *    the identical contents.
4719          */
4720         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4721             !DIFF_FILE_VALID(p->two) ||
4722             (p->one->sha1_valid && p->two->sha1_valid) ||
4723             (p->one->mode != p->two->mode) ||
4724             diff_populate_filespec(p->one, 1) ||
4725             diff_populate_filespec(p->two, 1) ||
4726             (p->one->size != p->two->size) ||
4727             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4728                 p->skip_stat_unmatch_result = 1;
4729         return p->skip_stat_unmatch_result;
4730 }
4731
4732 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4733 {
4734         int i;
4735         struct diff_queue_struct *q = &diff_queued_diff;
4736         struct diff_queue_struct outq;
4737         DIFF_QUEUE_CLEAR(&outq);
4738
4739         for (i = 0; i < q->nr; i++) {
4740                 struct diff_filepair *p = q->queue[i];
4741
4742                 if (diff_filespec_check_stat_unmatch(p))
4743                         diff_q(&outq, p);
4744                 else {
4745                         /*
4746                          * The caller can subtract 1 from skip_stat_unmatch
4747                          * to determine how many paths were dirty only
4748                          * due to stat info mismatch.
4749                          */
4750                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4751                                 diffopt->skip_stat_unmatch++;
4752                         diff_free_filepair(p);
4753                 }
4754         }
4755         free(q->queue);
4756         *q = outq;
4757 }
4758
4759 static int diffnamecmp(const void *a_, const void *b_)
4760 {
4761         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4762         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4763         const char *name_a, *name_b;
4764
4765         name_a = a->one ? a->one->path : a->two->path;
4766         name_b = b->one ? b->one->path : b->two->path;
4767         return strcmp(name_a, name_b);
4768 }
4769
4770 void diffcore_fix_diff_index(struct diff_options *options)
4771 {
4772         struct diff_queue_struct *q = &diff_queued_diff;
4773         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4774 }
4775
4776 void diffcore_std(struct diff_options *options)
4777 {
4778         if (options->skip_stat_unmatch)
4779                 diffcore_skip_stat_unmatch(options);
4780         if (!options->found_follow) {
4781                 /* See try_to_follow_renames() in tree-diff.c */
4782                 if (options->break_opt != -1)
4783                         diffcore_break(options->break_opt);
4784                 if (options->detect_rename)
4785                         diffcore_rename(options);
4786                 if (options->break_opt != -1)
4787                         diffcore_merge_broken();
4788         }
4789         if (options->pickaxe)
4790                 diffcore_pickaxe(options);
4791         if (options->orderfile)
4792                 diffcore_order(options->orderfile);
4793         if (!options->found_follow)
4794                 /* See try_to_follow_renames() in tree-diff.c */
4795                 diff_resolve_rename_copy();
4796         diffcore_apply_filter(options);
4797
4798         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4799                 DIFF_OPT_SET(options, HAS_CHANGES);
4800         else
4801                 DIFF_OPT_CLR(options, HAS_CHANGES);
4802
4803         options->found_follow = 0;
4804 }
4805
4806 int diff_result_code(struct diff_options *opt, int status)
4807 {
4808         int result = 0;
4809
4810         diff_warn_rename_limit("diff.renameLimit",
4811                                opt->needed_rename_limit,
4812                                opt->degraded_cc_to_c);
4813         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4814             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4815                 return status;
4816         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4817             DIFF_OPT_TST(opt, HAS_CHANGES))
4818                 result |= 01;
4819         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4820             DIFF_OPT_TST(opt, CHECK_FAILED))
4821                 result |= 02;
4822         return result;
4823 }
4824
4825 int diff_can_quit_early(struct diff_options *opt)
4826 {
4827         return (DIFF_OPT_TST(opt, QUICK) &&
4828                 !opt->filter &&
4829                 DIFF_OPT_TST(opt, HAS_CHANGES));
4830 }
4831
4832 /*
4833  * Shall changes to this submodule be ignored?
4834  *
4835  * Submodule changes can be configured to be ignored separately for each path,
4836  * but that configuration can be overridden from the command line.
4837  */
4838 static int is_submodule_ignored(const char *path, struct diff_options *options)
4839 {
4840         int ignored = 0;
4841         unsigned orig_flags = options->flags;
4842         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4843                 set_diffopt_flags_from_submodule_config(options, path);
4844         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4845                 ignored = 1;
4846         options->flags = orig_flags;
4847         return ignored;
4848 }
4849
4850 void diff_addremove(struct diff_options *options,
4851                     int addremove, unsigned mode,
4852                     const unsigned char *sha1,
4853                     int sha1_valid,
4854                     const char *concatpath, unsigned dirty_submodule)
4855 {
4856         struct diff_filespec *one, *two;
4857
4858         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4859                 return;
4860
4861         /* This may look odd, but it is a preparation for
4862          * feeding "there are unchanged files which should
4863          * not produce diffs, but when you are doing copy
4864          * detection you would need them, so here they are"
4865          * entries to the diff-core.  They will be prefixed
4866          * with something like '=' or '*' (I haven't decided
4867          * which but should not make any difference).
4868          * Feeding the same new and old to diff_change()
4869          * also has the same effect.
4870          * Before the final output happens, they are pruned after
4871          * merged into rename/copy pairs as appropriate.
4872          */
4873         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4874                 addremove = (addremove == '+' ? '-' :
4875                              addremove == '-' ? '+' : addremove);
4876
4877         if (options->prefix &&
4878             strncmp(concatpath, options->prefix, options->prefix_length))
4879                 return;
4880
4881         one = alloc_filespec(concatpath);
4882         two = alloc_filespec(concatpath);
4883
4884         if (addremove != '+')
4885                 fill_filespec(one, sha1, sha1_valid, mode);
4886         if (addremove != '-') {
4887                 fill_filespec(two, sha1, sha1_valid, mode);
4888                 two->dirty_submodule = dirty_submodule;
4889         }
4890
4891         diff_queue(&diff_queued_diff, one, two);
4892         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4893                 DIFF_OPT_SET(options, HAS_CHANGES);
4894 }
4895
4896 void diff_change(struct diff_options *options,
4897                  unsigned old_mode, unsigned new_mode,
4898                  const unsigned char *old_sha1,
4899                  const unsigned char *new_sha1,
4900                  int old_sha1_valid, int new_sha1_valid,
4901                  const char *concatpath,
4902                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4903 {
4904         struct diff_filespec *one, *two;
4905         struct diff_filepair *p;
4906
4907         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4908             is_submodule_ignored(concatpath, options))
4909                 return;
4910
4911         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4912                 unsigned tmp;
4913                 const unsigned char *tmp_c;
4914                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4915                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4916                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4917                         new_sha1_valid = tmp;
4918                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4919                         new_dirty_submodule = tmp;
4920         }
4921
4922         if (options->prefix &&
4923             strncmp(concatpath, options->prefix, options->prefix_length))
4924                 return;
4925
4926         one = alloc_filespec(concatpath);
4927         two = alloc_filespec(concatpath);
4928         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4929         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4930         one->dirty_submodule = old_dirty_submodule;
4931         two->dirty_submodule = new_dirty_submodule;
4932         p = diff_queue(&diff_queued_diff, one, two);
4933
4934         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4935                 return;
4936
4937         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
4938             !diff_filespec_check_stat_unmatch(p))
4939                 return;
4940
4941         DIFF_OPT_SET(options, HAS_CHANGES);
4942 }
4943
4944 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4945 {
4946         struct diff_filepair *pair;
4947         struct diff_filespec *one, *two;
4948
4949         if (options->prefix &&
4950             strncmp(path, options->prefix, options->prefix_length))
4951                 return NULL;
4952
4953         one = alloc_filespec(path);
4954         two = alloc_filespec(path);
4955         pair = diff_queue(&diff_queued_diff, one, two);
4956         pair->is_unmerged = 1;
4957         return pair;
4958 }
4959
4960 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4961                 size_t *outsize)
4962 {
4963         struct diff_tempfile *temp;
4964         const char *argv[3];
4965         const char **arg = argv;
4966         struct child_process child;
4967         struct strbuf buf = STRBUF_INIT;
4968         int err = 0;
4969
4970         temp = prepare_temp_file(spec->path, spec);
4971         *arg++ = pgm;
4972         *arg++ = temp->name;
4973         *arg = NULL;
4974
4975         memset(&child, 0, sizeof(child));
4976         child.use_shell = 1;
4977         child.argv = argv;
4978         child.out = -1;
4979         if (start_command(&child)) {
4980                 remove_tempfile();
4981                 return NULL;
4982         }
4983
4984         if (strbuf_read(&buf, child.out, 0) < 0)
4985                 err = error("error reading from textconv command '%s'", pgm);
4986         close(child.out);
4987
4988         if (finish_command(&child) || err) {
4989                 strbuf_release(&buf);
4990                 remove_tempfile();
4991                 return NULL;
4992         }
4993         remove_tempfile();
4994
4995         return strbuf_detach(&buf, outsize);
4996 }
4997
4998 size_t fill_textconv(struct userdiff_driver *driver,
4999                      struct diff_filespec *df,
5000                      char **outbuf)
5001 {
5002         size_t size;
5003
5004         if (!driver || !driver->textconv) {
5005                 if (!DIFF_FILE_VALID(df)) {
5006                         *outbuf = "";
5007                         return 0;
5008                 }
5009                 if (diff_populate_filespec(df, 0))
5010                         die("unable to read files to diff");
5011                 *outbuf = df->data;
5012                 return df->size;
5013         }
5014
5015         if (driver->textconv_cache && df->sha1_valid) {
5016                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5017                                           &size);
5018                 if (*outbuf)
5019                         return size;
5020         }
5021
5022         *outbuf = run_textconv(driver->textconv, df, &size);
5023         if (!*outbuf)
5024                 die("unable to read files to diff");
5025
5026         if (driver->textconv_cache && df->sha1_valid) {
5027                 /* ignore errors, as we might be in a readonly repository */
5028                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5029                                 size);
5030                 /*
5031                  * we could save up changes and flush them all at the end,
5032                  * but we would need an extra call after all diffing is done.
5033                  * Since generating a cache entry is the slow path anyway,
5034                  * this extra overhead probably isn't a big deal.
5035                  */
5036                 notes_cache_write(driver->textconv_cache);
5037         }
5038
5039         return size;
5040 }
5041
5042 void setup_diff_pager(struct diff_options *opt)
5043 {
5044         /*
5045          * If the user asked for our exit code, then either they want --quiet
5046          * or --exit-code. We should definitely not bother with a pager in the
5047          * former case, as we will generate no output. Since we still properly
5048          * report our exit code even when a pager is run, we _could_ run a
5049          * pager with --exit-code. But since we have not done so historically,
5050          * and because it is easy to find people oneline advising "git diff
5051          * --exit-code" in hooks and other scripts, we do not do so.
5052          */
5053         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5054             check_pager_config("diff") != 0)
5055                 setup_pager();
5056 }