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