Merge branch 'sb/t9300-typofix'
[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 #include "argv-array.h"
20
21 #ifdef NO_FAST_WORKING_DIRECTORY
22 #define FAST_WORKING_DIRECTORY 0
23 #else
24 #define FAST_WORKING_DIRECTORY 1
25 #endif
26
27 static int diff_detect_rename_default;
28 static int diff_rename_limit_default = 400;
29 static int diff_suppress_blank_empty;
30 static int diff_use_color_default = -1;
31 static int diff_context_default = 3;
32 static const char *diff_word_regex_cfg;
33 static const char *external_diff_cmd_cfg;
34 static const char *diff_order_file_cfg;
35 int diff_auto_refresh_index = 1;
36 static int diff_mnemonic_prefix;
37 static int diff_no_prefix;
38 static int diff_stat_graph_width;
39 static int diff_dirstat_permille_default = 30;
40 static struct diff_options default_diff_options;
41 static long diff_algorithm;
42
43 static char diff_colors[][COLOR_MAXLEN] = {
44         GIT_COLOR_RESET,
45         GIT_COLOR_NORMAL,       /* PLAIN */
46         GIT_COLOR_BOLD,         /* METAINFO */
47         GIT_COLOR_CYAN,         /* FRAGINFO */
48         GIT_COLOR_RED,          /* OLD */
49         GIT_COLOR_GREEN,        /* NEW */
50         GIT_COLOR_YELLOW,       /* COMMIT */
51         GIT_COLOR_BG_RED,       /* WHITESPACE */
52         GIT_COLOR_NORMAL,       /* FUNCINFO */
53 };
54
55 static int parse_diff_color_slot(const char *var)
56 {
57         if (!strcasecmp(var, "plain"))
58                 return DIFF_PLAIN;
59         if (!strcasecmp(var, "meta"))
60                 return DIFF_METAINFO;
61         if (!strcasecmp(var, "frag"))
62                 return DIFF_FRAGINFO;
63         if (!strcasecmp(var, "old"))
64                 return DIFF_FILE_OLD;
65         if (!strcasecmp(var, "new"))
66                 return DIFF_FILE_NEW;
67         if (!strcasecmp(var, "commit"))
68                 return DIFF_COMMIT;
69         if (!strcasecmp(var, "whitespace"))
70                 return DIFF_WHITESPACE;
71         if (!strcasecmp(var, "func"))
72                 return DIFF_FUNCINFO;
73         return -1;
74 }
75
76 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
77                                 struct strbuf *errmsg)
78 {
79         char *params_copy = xstrdup(params_string);
80         struct string_list params = STRING_LIST_INIT_NODUP;
81         int ret = 0;
82         int i;
83
84         if (*params_copy)
85                 string_list_split_in_place(&params, params_copy, ',', -1);
86         for (i = 0; i < params.nr; i++) {
87                 const char *p = params.items[i].string;
88                 if (!strcmp(p, "changes")) {
89                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
90                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
91                 } else if (!strcmp(p, "lines")) {
92                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
93                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
94                 } else if (!strcmp(p, "files")) {
95                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
96                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
97                 } else if (!strcmp(p, "noncumulative")) {
98                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
99                 } else if (!strcmp(p, "cumulative")) {
100                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
101                 } else if (isdigit(*p)) {
102                         char *end;
103                         int permille = strtoul(p, &end, 10) * 10;
104                         if (*end == '.' && isdigit(*++end)) {
105                                 /* only use first digit */
106                                 permille += *end - '0';
107                                 /* .. and ignore any further digits */
108                                 while (isdigit(*++end))
109                                         ; /* nothing */
110                         }
111                         if (!*end)
112                                 options->dirstat_permille = permille;
113                         else {
114                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
115                                             p);
116                                 ret++;
117                         }
118                 } else {
119                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
120                         ret++;
121                 }
122
123         }
124         string_list_clear(&params, 0);
125         free(params_copy);
126         return ret;
127 }
128
129 static int parse_submodule_params(struct diff_options *options, const char *value)
130 {
131         if (!strcmp(value, "log"))
132                 DIFF_OPT_SET(options, SUBMODULE_LOG);
133         else if (!strcmp(value, "short"))
134                 DIFF_OPT_CLR(options, SUBMODULE_LOG);
135         else
136                 return -1;
137         return 0;
138 }
139
140 static int git_config_rename(const char *var, const char *value)
141 {
142         if (!value)
143                 return DIFF_DETECT_RENAME;
144         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
145                 return  DIFF_DETECT_COPY;
146         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
147 }
148
149 long parse_algorithm_value(const char *value)
150 {
151         if (!value)
152                 return -1;
153         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
154                 return 0;
155         else if (!strcasecmp(value, "minimal"))
156                 return XDF_NEED_MINIMAL;
157         else if (!strcasecmp(value, "patience"))
158                 return XDF_PATIENCE_DIFF;
159         else if (!strcasecmp(value, "histogram"))
160                 return XDF_HISTOGRAM_DIFF;
161         return -1;
162 }
163
164 /*
165  * These are to give UI layer defaults.
166  * The core-level commands such as git-diff-files should
167  * never be affected by the setting of diff.renames
168  * the user happens to have in the configuration file.
169  */
170 int git_diff_ui_config(const char *var, const char *value, void *cb)
171 {
172         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
173                 diff_use_color_default = git_config_colorbool(var, value);
174                 return 0;
175         }
176         if (!strcmp(var, "diff.context")) {
177                 diff_context_default = git_config_int(var, value);
178                 if (diff_context_default < 0)
179                         return -1;
180                 return 0;
181         }
182         if (!strcmp(var, "diff.renames")) {
183                 diff_detect_rename_default = git_config_rename(var, value);
184                 return 0;
185         }
186         if (!strcmp(var, "diff.autorefreshindex")) {
187                 diff_auto_refresh_index = git_config_bool(var, value);
188                 return 0;
189         }
190         if (!strcmp(var, "diff.mnemonicprefix")) {
191                 diff_mnemonic_prefix = git_config_bool(var, value);
192                 return 0;
193         }
194         if (!strcmp(var, "diff.noprefix")) {
195                 diff_no_prefix = git_config_bool(var, value);
196                 return 0;
197         }
198         if (!strcmp(var, "diff.statgraphwidth")) {
199                 diff_stat_graph_width = git_config_int(var, value);
200                 return 0;
201         }
202         if (!strcmp(var, "diff.external"))
203                 return git_config_string(&external_diff_cmd_cfg, var, value);
204         if (!strcmp(var, "diff.wordregex"))
205                 return git_config_string(&diff_word_regex_cfg, var, value);
206         if (!strcmp(var, "diff.orderfile"))
207                 return git_config_pathname(&diff_order_file_cfg, var, value);
208
209         if (!strcmp(var, "diff.ignoresubmodules"))
210                 handle_ignore_submodules_arg(&default_diff_options, value);
211
212         if (!strcmp(var, "diff.submodule")) {
213                 if (parse_submodule_params(&default_diff_options, value))
214                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
215                                 value);
216                 return 0;
217         }
218
219         if (!strcmp(var, "diff.algorithm")) {
220                 diff_algorithm = parse_algorithm_value(value);
221                 if (diff_algorithm < 0)
222                         return -1;
223                 return 0;
224         }
225
226         if (git_color_config(var, value, cb) < 0)
227                 return -1;
228
229         return git_diff_basic_config(var, value, cb);
230 }
231
232 int git_diff_basic_config(const char *var, const char *value, void *cb)
233 {
234         const char *name;
235
236         if (!strcmp(var, "diff.renamelimit")) {
237                 diff_rename_limit_default = git_config_int(var, value);
238                 return 0;
239         }
240
241         if (userdiff_config(var, value) < 0)
242                 return -1;
243
244         if (skip_prefix(var, "diff.color.", &name) ||
245             skip_prefix(var, "color.diff.", &name)) {
246                 int slot = parse_diff_color_slot(name);
247                 if (slot < 0)
248                         return 0;
249                 if (!value)
250                         return config_error_nonbool(var);
251                 color_parse(value, var, diff_colors[slot]);
252                 return 0;
253         }
254
255         /* like GNU diff's --suppress-blank-empty option  */
256         if (!strcmp(var, "diff.suppressblankempty") ||
257                         /* for backwards compatibility */
258                         !strcmp(var, "diff.suppress-blank-empty")) {
259                 diff_suppress_blank_empty = git_config_bool(var, value);
260                 return 0;
261         }
262
263         if (!strcmp(var, "diff.dirstat")) {
264                 struct strbuf errmsg = STRBUF_INIT;
265                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
266                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
267                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
268                                 errmsg.buf);
269                 strbuf_release(&errmsg);
270                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
271                 return 0;
272         }
273
274         if (starts_with(var, "submodule."))
275                 return parse_submodule_config_option(var, value);
276
277         return git_default_config(var, value, cb);
278 }
279
280 static char *quote_two(const char *one, const char *two)
281 {
282         int need_one = quote_c_style(one, NULL, NULL, 1);
283         int need_two = quote_c_style(two, NULL, NULL, 1);
284         struct strbuf res = STRBUF_INIT;
285
286         if (need_one + need_two) {
287                 strbuf_addch(&res, '"');
288                 quote_c_style(one, &res, NULL, 1);
289                 quote_c_style(two, &res, NULL, 1);
290                 strbuf_addch(&res, '"');
291         } else {
292                 strbuf_addstr(&res, one);
293                 strbuf_addstr(&res, two);
294         }
295         return strbuf_detach(&res, NULL);
296 }
297
298 static const char *external_diff(void)
299 {
300         static const char *external_diff_cmd = NULL;
301         static int done_preparing = 0;
302
303         if (done_preparing)
304                 return external_diff_cmd;
305         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
306         if (!external_diff_cmd)
307                 external_diff_cmd = external_diff_cmd_cfg;
308         done_preparing = 1;
309         return external_diff_cmd;
310 }
311
312 static struct diff_tempfile {
313         const char *name; /* filename external diff should read from */
314         char hex[41];
315         char mode[10];
316         char tmp_path[PATH_MAX];
317 } diff_temp[2];
318
319 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
320
321 struct emit_callback {
322         int color_diff;
323         unsigned ws_rule;
324         int blank_at_eof_in_preimage;
325         int blank_at_eof_in_postimage;
326         int lno_in_preimage;
327         int lno_in_postimage;
328         sane_truncate_fn truncate;
329         const char **label_path;
330         struct diff_words_data *diff_words;
331         struct diff_options *opt;
332         int *found_changesp;
333         struct strbuf *header;
334 };
335
336 static int count_lines(const char *data, int size)
337 {
338         int count, ch, completely_empty = 1, nl_just_seen = 0;
339         count = 0;
340         while (0 < size--) {
341                 ch = *data++;
342                 if (ch == '\n') {
343                         count++;
344                         nl_just_seen = 1;
345                         completely_empty = 0;
346                 }
347                 else {
348                         nl_just_seen = 0;
349                         completely_empty = 0;
350                 }
351         }
352         if (completely_empty)
353                 return 0;
354         if (!nl_just_seen)
355                 count++; /* no trailing newline */
356         return count;
357 }
358
359 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
360 {
361         if (!DIFF_FILE_VALID(one)) {
362                 mf->ptr = (char *)""; /* does not matter */
363                 mf->size = 0;
364                 return 0;
365         }
366         else if (diff_populate_filespec(one, 0))
367                 return -1;
368
369         mf->ptr = one->data;
370         mf->size = one->size;
371         return 0;
372 }
373
374 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
375 static unsigned long diff_filespec_size(struct diff_filespec *one)
376 {
377         if (!DIFF_FILE_VALID(one))
378                 return 0;
379         diff_populate_filespec(one, CHECK_SIZE_ONLY);
380         return one->size;
381 }
382
383 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
384 {
385         char *ptr = mf->ptr;
386         long size = mf->size;
387         int cnt = 0;
388
389         if (!size)
390                 return cnt;
391         ptr += size - 1; /* pointing at the very end */
392         if (*ptr != '\n')
393                 ; /* incomplete line */
394         else
395                 ptr--; /* skip the last LF */
396         while (mf->ptr < ptr) {
397                 char *prev_eol;
398                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
399                         if (*prev_eol == '\n')
400                                 break;
401                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
402                         break;
403                 cnt++;
404                 ptr = prev_eol - 1;
405         }
406         return cnt;
407 }
408
409 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
410                                struct emit_callback *ecbdata)
411 {
412         int l1, l2, at;
413         unsigned ws_rule = ecbdata->ws_rule;
414         l1 = count_trailing_blank(mf1, ws_rule);
415         l2 = count_trailing_blank(mf2, ws_rule);
416         if (l2 <= l1) {
417                 ecbdata->blank_at_eof_in_preimage = 0;
418                 ecbdata->blank_at_eof_in_postimage = 0;
419                 return;
420         }
421         at = count_lines(mf1->ptr, mf1->size);
422         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
423
424         at = count_lines(mf2->ptr, mf2->size);
425         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
426 }
427
428 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
429                         int first, const char *line, int len)
430 {
431         int has_trailing_newline, has_trailing_carriage_return;
432         int nofirst;
433         FILE *file = o->file;
434
435         fputs(diff_line_prefix(o), file);
436
437         if (len == 0) {
438                 has_trailing_newline = (first == '\n');
439                 has_trailing_carriage_return = (!has_trailing_newline &&
440                                                 (first == '\r'));
441                 nofirst = has_trailing_newline || has_trailing_carriage_return;
442         } else {
443                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
444                 if (has_trailing_newline)
445                         len--;
446                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
447                 if (has_trailing_carriage_return)
448                         len--;
449                 nofirst = 0;
450         }
451
452         if (len || !nofirst) {
453                 fputs(set, file);
454                 if (!nofirst)
455                         fputc(first, file);
456                 fwrite(line, len, 1, file);
457                 fputs(reset, file);
458         }
459         if (has_trailing_carriage_return)
460                 fputc('\r', file);
461         if (has_trailing_newline)
462                 fputc('\n', file);
463 }
464
465 static void emit_line(struct diff_options *o, const char *set, const char *reset,
466                       const char *line, int len)
467 {
468         emit_line_0(o, set, reset, line[0], line+1, len-1);
469 }
470
471 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
472 {
473         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
474               ecbdata->blank_at_eof_in_preimage &&
475               ecbdata->blank_at_eof_in_postimage &&
476               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
477               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
478                 return 0;
479         return ws_blank_line(line, len, ecbdata->ws_rule);
480 }
481
482 static void emit_add_line(const char *reset,
483                           struct emit_callback *ecbdata,
484                           const char *line, int len)
485 {
486         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
487         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
488
489         if (!*ws)
490                 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
491         else if (new_blank_line_at_eof(ecbdata, line, len))
492                 /* Blank line at EOF - paint '+' as well */
493                 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
494         else {
495                 /* Emit just the prefix, then the rest. */
496                 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
497                 ws_check_emit(line, len, ecbdata->ws_rule,
498                               ecbdata->opt->file, set, reset, ws);
499         }
500 }
501
502 static void emit_hunk_header(struct emit_callback *ecbdata,
503                              const char *line, int len)
504 {
505         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
506         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
507         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
508         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
509         static const char atat[2] = { '@', '@' };
510         const char *cp, *ep;
511         struct strbuf msgbuf = STRBUF_INIT;
512         int org_len = len;
513         int i = 1;
514
515         /*
516          * As a hunk header must begin with "@@ -<old>, +<new> @@",
517          * it always is at least 10 bytes long.
518          */
519         if (len < 10 ||
520             memcmp(line, atat, 2) ||
521             !(ep = memmem(line + 2, len - 2, atat, 2))) {
522                 emit_line(ecbdata->opt, plain, reset, line, len);
523                 return;
524         }
525         ep += 2; /* skip over @@ */
526
527         /* The hunk header in fraginfo color */
528         strbuf_addstr(&msgbuf, frag);
529         strbuf_add(&msgbuf, line, ep - line);
530         strbuf_addstr(&msgbuf, reset);
531
532         /*
533          * trailing "\r\n"
534          */
535         for ( ; i < 3; i++)
536                 if (line[len - i] == '\r' || line[len - i] == '\n')
537                         len--;
538
539         /* blank before the func header */
540         for (cp = ep; ep - line < len; ep++)
541                 if (*ep != ' ' && *ep != '\t')
542                         break;
543         if (ep != cp) {
544                 strbuf_addstr(&msgbuf, plain);
545                 strbuf_add(&msgbuf, cp, ep - cp);
546                 strbuf_addstr(&msgbuf, reset);
547         }
548
549         if (ep < line + len) {
550                 strbuf_addstr(&msgbuf, func);
551                 strbuf_add(&msgbuf, ep, line + len - ep);
552                 strbuf_addstr(&msgbuf, reset);
553         }
554
555         strbuf_add(&msgbuf, line + len, org_len - len);
556         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
557         strbuf_release(&msgbuf);
558 }
559
560 static struct diff_tempfile *claim_diff_tempfile(void) {
561         int i;
562         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
563                 if (!diff_temp[i].name)
564                         return diff_temp + i;
565         die("BUG: diff is failing to clean up its tempfiles");
566 }
567
568 static int remove_tempfile_installed;
569
570 static void remove_tempfile(void)
571 {
572         int i;
573         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
574                 if (diff_temp[i].name == diff_temp[i].tmp_path)
575                         unlink_or_warn(diff_temp[i].name);
576                 diff_temp[i].name = NULL;
577         }
578 }
579
580 static void remove_tempfile_on_signal(int signo)
581 {
582         remove_tempfile();
583         sigchain_pop(signo);
584         raise(signo);
585 }
586
587 static void print_line_count(FILE *file, int count)
588 {
589         switch (count) {
590         case 0:
591                 fprintf(file, "0,0");
592                 break;
593         case 1:
594                 fprintf(file, "1");
595                 break;
596         default:
597                 fprintf(file, "1,%d", count);
598                 break;
599         }
600 }
601
602 static void emit_rewrite_lines(struct emit_callback *ecb,
603                                int prefix, const char *data, int size)
604 {
605         const char *endp = NULL;
606         static const char *nneof = " No newline at end of file\n";
607         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
608         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
609
610         while (0 < size) {
611                 int len;
612
613                 endp = memchr(data, '\n', size);
614                 len = endp ? (endp - data + 1) : size;
615                 if (prefix != '+') {
616                         ecb->lno_in_preimage++;
617                         emit_line_0(ecb->opt, old, reset, '-',
618                                     data, len);
619                 } else {
620                         ecb->lno_in_postimage++;
621                         emit_add_line(reset, ecb, data, len);
622                 }
623                 size -= len;
624                 data += len;
625         }
626         if (!endp) {
627                 const char *plain = diff_get_color(ecb->color_diff,
628                                                    DIFF_PLAIN);
629                 putc('\n', ecb->opt->file);
630                 emit_line_0(ecb->opt, plain, reset, '\\',
631                             nneof, strlen(nneof));
632         }
633 }
634
635 static void emit_rewrite_diff(const char *name_a,
636                               const char *name_b,
637                               struct diff_filespec *one,
638                               struct diff_filespec *two,
639                               struct userdiff_driver *textconv_one,
640                               struct userdiff_driver *textconv_two,
641                               struct diff_options *o)
642 {
643         int lc_a, lc_b;
644         const char *name_a_tab, *name_b_tab;
645         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
646         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
647         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
648         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
649         const char *a_prefix, *b_prefix;
650         char *data_one, *data_two;
651         size_t size_one, size_two;
652         struct emit_callback ecbdata;
653         const char *line_prefix = diff_line_prefix(o);
654
655         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
656                 a_prefix = o->b_prefix;
657                 b_prefix = o->a_prefix;
658         } else {
659                 a_prefix = o->a_prefix;
660                 b_prefix = o->b_prefix;
661         }
662
663         name_a += (*name_a == '/');
664         name_b += (*name_b == '/');
665         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
666         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
667
668         strbuf_reset(&a_name);
669         strbuf_reset(&b_name);
670         quote_two_c_style(&a_name, a_prefix, name_a, 0);
671         quote_two_c_style(&b_name, b_prefix, name_b, 0);
672
673         size_one = fill_textconv(textconv_one, one, &data_one);
674         size_two = fill_textconv(textconv_two, two, &data_two);
675
676         memset(&ecbdata, 0, sizeof(ecbdata));
677         ecbdata.color_diff = want_color(o->use_color);
678         ecbdata.found_changesp = &o->found_changes;
679         ecbdata.ws_rule = whitespace_rule(name_b);
680         ecbdata.opt = o;
681         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
682                 mmfile_t mf1, mf2;
683                 mf1.ptr = (char *)data_one;
684                 mf2.ptr = (char *)data_two;
685                 mf1.size = size_one;
686                 mf2.size = size_two;
687                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
688         }
689         ecbdata.lno_in_preimage = 1;
690         ecbdata.lno_in_postimage = 1;
691
692         lc_a = count_lines(data_one, size_one);
693         lc_b = count_lines(data_two, size_two);
694         fprintf(o->file,
695                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
696                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
697                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
698                 line_prefix, fraginfo);
699         if (!o->irreversible_delete)
700                 print_line_count(o->file, lc_a);
701         else
702                 fprintf(o->file, "?,?");
703         fprintf(o->file, " +");
704         print_line_count(o->file, lc_b);
705         fprintf(o->file, " @@%s\n", reset);
706         if (lc_a && !o->irreversible_delete)
707                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
708         if (lc_b)
709                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
710         if (textconv_one)
711                 free((char *)data_one);
712         if (textconv_two)
713                 free((char *)data_two);
714 }
715
716 struct diff_words_buffer {
717         mmfile_t text;
718         long alloc;
719         struct diff_words_orig {
720                 const char *begin, *end;
721         } *orig;
722         int orig_nr, orig_alloc;
723 };
724
725 static void diff_words_append(char *line, unsigned long len,
726                 struct diff_words_buffer *buffer)
727 {
728         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
729         line++;
730         len--;
731         memcpy(buffer->text.ptr + buffer->text.size, line, len);
732         buffer->text.size += len;
733         buffer->text.ptr[buffer->text.size] = '\0';
734 }
735
736 struct diff_words_style_elem {
737         const char *prefix;
738         const char *suffix;
739         const char *color; /* NULL; filled in by the setup code if
740                             * color is enabled */
741 };
742
743 struct diff_words_style {
744         enum diff_words_type type;
745         struct diff_words_style_elem new, old, ctx;
746         const char *newline;
747 };
748
749 static struct diff_words_style diff_words_styles[] = {
750         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
751         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
752         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
753 };
754
755 struct diff_words_data {
756         struct diff_words_buffer minus, plus;
757         const char *current_plus;
758         int last_minus;
759         struct diff_options *opt;
760         regex_t *word_regex;
761         enum diff_words_type type;
762         struct diff_words_style *style;
763 };
764
765 static int fn_out_diff_words_write_helper(FILE *fp,
766                                           struct diff_words_style_elem *st_el,
767                                           const char *newline,
768                                           size_t count, const char *buf,
769                                           const char *line_prefix)
770 {
771         int print = 0;
772
773         while (count) {
774                 char *p = memchr(buf, '\n', count);
775                 if (print)
776                         fputs(line_prefix, fp);
777                 if (p != buf) {
778                         if (st_el->color && fputs(st_el->color, fp) < 0)
779                                 return -1;
780                         if (fputs(st_el->prefix, fp) < 0 ||
781                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
782                             fputs(st_el->suffix, fp) < 0)
783                                 return -1;
784                         if (st_el->color && *st_el->color
785                             && fputs(GIT_COLOR_RESET, fp) < 0)
786                                 return -1;
787                 }
788                 if (!p)
789                         return 0;
790                 if (fputs(newline, fp) < 0)
791                         return -1;
792                 count -= p + 1 - buf;
793                 buf = p + 1;
794                 print = 1;
795         }
796         return 0;
797 }
798
799 /*
800  * '--color-words' algorithm can be described as:
801  *
802  *   1. collect a the minus/plus lines of a diff hunk, divided into
803  *      minus-lines and plus-lines;
804  *
805  *   2. break both minus-lines and plus-lines into words and
806  *      place them into two mmfile_t with one word for each line;
807  *
808  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
809  *
810  * And for the common parts of the both file, we output the plus side text.
811  * diff_words->current_plus is used to trace the current position of the plus file
812  * which printed. diff_words->last_minus is used to trace the last minus word
813  * printed.
814  *
815  * For '--graph' to work with '--color-words', we need to output the graph prefix
816  * on each line of color words output. Generally, there are two conditions on
817  * which we should output the prefix.
818  *
819  *   1. diff_words->last_minus == 0 &&
820  *      diff_words->current_plus == diff_words->plus.text.ptr
821  *
822  *      that is: the plus text must start as a new line, and if there is no minus
823  *      word printed, a graph prefix must be printed.
824  *
825  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
826  *      *(diff_words->current_plus - 1) == '\n'
827  *
828  *      that is: a graph prefix must be printed following a '\n'
829  */
830 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
831 {
832         if ((diff_words->last_minus == 0 &&
833                 diff_words->current_plus == diff_words->plus.text.ptr) ||
834                 (diff_words->current_plus > diff_words->plus.text.ptr &&
835                 *(diff_words->current_plus - 1) == '\n')) {
836                 return 1;
837         } else {
838                 return 0;
839         }
840 }
841
842 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
843 {
844         struct diff_words_data *diff_words = priv;
845         struct diff_words_style *style = diff_words->style;
846         int minus_first, minus_len, plus_first, plus_len;
847         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
848         struct diff_options *opt = diff_words->opt;
849         const char *line_prefix;
850
851         if (line[0] != '@' || parse_hunk_header(line, len,
852                         &minus_first, &minus_len, &plus_first, &plus_len))
853                 return;
854
855         assert(opt);
856         line_prefix = diff_line_prefix(opt);
857
858         /* POSIX requires that first be decremented by one if len == 0... */
859         if (minus_len) {
860                 minus_begin = diff_words->minus.orig[minus_first].begin;
861                 minus_end =
862                         diff_words->minus.orig[minus_first + minus_len - 1].end;
863         } else
864                 minus_begin = minus_end =
865                         diff_words->minus.orig[minus_first].end;
866
867         if (plus_len) {
868                 plus_begin = diff_words->plus.orig[plus_first].begin;
869                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
870         } else
871                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
872
873         if (color_words_output_graph_prefix(diff_words)) {
874                 fputs(line_prefix, diff_words->opt->file);
875         }
876         if (diff_words->current_plus != plus_begin) {
877                 fn_out_diff_words_write_helper(diff_words->opt->file,
878                                 &style->ctx, style->newline,
879                                 plus_begin - diff_words->current_plus,
880                                 diff_words->current_plus, line_prefix);
881                 if (*(plus_begin - 1) == '\n')
882                         fputs(line_prefix, diff_words->opt->file);
883         }
884         if (minus_begin != minus_end) {
885                 fn_out_diff_words_write_helper(diff_words->opt->file,
886                                 &style->old, style->newline,
887                                 minus_end - minus_begin, minus_begin,
888                                 line_prefix);
889         }
890         if (plus_begin != plus_end) {
891                 fn_out_diff_words_write_helper(diff_words->opt->file,
892                                 &style->new, style->newline,
893                                 plus_end - plus_begin, plus_begin,
894                                 line_prefix);
895         }
896
897         diff_words->current_plus = plus_end;
898         diff_words->last_minus = minus_first;
899 }
900
901 /* This function starts looking at *begin, and returns 0 iff a word was found. */
902 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
903                 int *begin, int *end)
904 {
905         if (word_regex && *begin < buffer->size) {
906                 regmatch_t match[1];
907                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
908                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
909                                         '\n', match[0].rm_eo - match[0].rm_so);
910                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
911                         *begin += match[0].rm_so;
912                         return *begin >= *end;
913                 }
914                 return -1;
915         }
916
917         /* find the next word */
918         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
919                 (*begin)++;
920         if (*begin >= buffer->size)
921                 return -1;
922
923         /* find the end of the word */
924         *end = *begin + 1;
925         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
926                 (*end)++;
927
928         return 0;
929 }
930
931 /*
932  * This function splits the words in buffer->text, stores the list with
933  * newline separator into out, and saves the offsets of the original words
934  * in buffer->orig.
935  */
936 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
937                 regex_t *word_regex)
938 {
939         int i, j;
940         long alloc = 0;
941
942         out->size = 0;
943         out->ptr = NULL;
944
945         /* fake an empty "0th" word */
946         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
947         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
948         buffer->orig_nr = 1;
949
950         for (i = 0; i < buffer->text.size; i++) {
951                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
952                         return;
953
954                 /* store original boundaries */
955                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
956                                 buffer->orig_alloc);
957                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
958                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
959                 buffer->orig_nr++;
960
961                 /* store one word */
962                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
963                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
964                 out->ptr[out->size + j - i] = '\n';
965                 out->size += j - i + 1;
966
967                 i = j - 1;
968         }
969 }
970
971 /* this executes the word diff on the accumulated buffers */
972 static void diff_words_show(struct diff_words_data *diff_words)
973 {
974         xpparam_t xpp;
975         xdemitconf_t xecfg;
976         mmfile_t minus, plus;
977         struct diff_words_style *style = diff_words->style;
978
979         struct diff_options *opt = diff_words->opt;
980         const char *line_prefix;
981
982         assert(opt);
983         line_prefix = diff_line_prefix(opt);
984
985         /* special case: only removal */
986         if (!diff_words->plus.text.size) {
987                 fputs(line_prefix, diff_words->opt->file);
988                 fn_out_diff_words_write_helper(diff_words->opt->file,
989                         &style->old, style->newline,
990                         diff_words->minus.text.size,
991                         diff_words->minus.text.ptr, line_prefix);
992                 diff_words->minus.text.size = 0;
993                 return;
994         }
995
996         diff_words->current_plus = diff_words->plus.text.ptr;
997         diff_words->last_minus = 0;
998
999         memset(&xpp, 0, sizeof(xpp));
1000         memset(&xecfg, 0, sizeof(xecfg));
1001         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1002         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1003         xpp.flags = 0;
1004         /* as only the hunk header will be parsed, we need a 0-context */
1005         xecfg.ctxlen = 0;
1006         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1007                       &xpp, &xecfg);
1008         free(minus.ptr);
1009         free(plus.ptr);
1010         if (diff_words->current_plus != diff_words->plus.text.ptr +
1011                         diff_words->plus.text.size) {
1012                 if (color_words_output_graph_prefix(diff_words))
1013                         fputs(line_prefix, diff_words->opt->file);
1014                 fn_out_diff_words_write_helper(diff_words->opt->file,
1015                         &style->ctx, style->newline,
1016                         diff_words->plus.text.ptr + diff_words->plus.text.size
1017                         - diff_words->current_plus, diff_words->current_plus,
1018                         line_prefix);
1019         }
1020         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1021 }
1022
1023 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1024 static void diff_words_flush(struct emit_callback *ecbdata)
1025 {
1026         if (ecbdata->diff_words->minus.text.size ||
1027             ecbdata->diff_words->plus.text.size)
1028                 diff_words_show(ecbdata->diff_words);
1029 }
1030
1031 static void diff_filespec_load_driver(struct diff_filespec *one)
1032 {
1033         /* Use already-loaded driver */
1034         if (one->driver)
1035                 return;
1036
1037         if (S_ISREG(one->mode))
1038                 one->driver = userdiff_find_by_path(one->path);
1039
1040         /* Fallback to default settings */
1041         if (!one->driver)
1042                 one->driver = userdiff_find_by_name("default");
1043 }
1044
1045 static const char *userdiff_word_regex(struct diff_filespec *one)
1046 {
1047         diff_filespec_load_driver(one);
1048         return one->driver->word_regex;
1049 }
1050
1051 static void init_diff_words_data(struct emit_callback *ecbdata,
1052                                  struct diff_options *orig_opts,
1053                                  struct diff_filespec *one,
1054                                  struct diff_filespec *two)
1055 {
1056         int i;
1057         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1058         memcpy(o, orig_opts, sizeof(struct diff_options));
1059
1060         ecbdata->diff_words =
1061                 xcalloc(1, sizeof(struct diff_words_data));
1062         ecbdata->diff_words->type = o->word_diff;
1063         ecbdata->diff_words->opt = o;
1064         if (!o->word_regex)
1065                 o->word_regex = userdiff_word_regex(one);
1066         if (!o->word_regex)
1067                 o->word_regex = userdiff_word_regex(two);
1068         if (!o->word_regex)
1069                 o->word_regex = diff_word_regex_cfg;
1070         if (o->word_regex) {
1071                 ecbdata->diff_words->word_regex = (regex_t *)
1072                         xmalloc(sizeof(regex_t));
1073                 if (regcomp(ecbdata->diff_words->word_regex,
1074                             o->word_regex,
1075                             REG_EXTENDED | REG_NEWLINE))
1076                         die ("Invalid regular expression: %s",
1077                              o->word_regex);
1078         }
1079         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1080                 if (o->word_diff == diff_words_styles[i].type) {
1081                         ecbdata->diff_words->style =
1082                                 &diff_words_styles[i];
1083                         break;
1084                 }
1085         }
1086         if (want_color(o->use_color)) {
1087                 struct diff_words_style *st = ecbdata->diff_words->style;
1088                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1089                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1090                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1091         }
1092 }
1093
1094 static void free_diff_words_data(struct emit_callback *ecbdata)
1095 {
1096         if (ecbdata->diff_words) {
1097                 diff_words_flush(ecbdata);
1098                 free (ecbdata->diff_words->opt);
1099                 free (ecbdata->diff_words->minus.text.ptr);
1100                 free (ecbdata->diff_words->minus.orig);
1101                 free (ecbdata->diff_words->plus.text.ptr);
1102                 free (ecbdata->diff_words->plus.orig);
1103                 if (ecbdata->diff_words->word_regex) {
1104                         regfree(ecbdata->diff_words->word_regex);
1105                         free(ecbdata->diff_words->word_regex);
1106                 }
1107                 free(ecbdata->diff_words);
1108                 ecbdata->diff_words = NULL;
1109         }
1110 }
1111
1112 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1113 {
1114         if (want_color(diff_use_color))
1115                 return diff_colors[ix];
1116         return "";
1117 }
1118
1119 const char *diff_line_prefix(struct diff_options *opt)
1120 {
1121         struct strbuf *msgbuf;
1122         if (!opt->output_prefix)
1123                 return "";
1124
1125         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1126         return msgbuf->buf;
1127 }
1128
1129 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1130 {
1131         const char *cp;
1132         unsigned long allot;
1133         size_t l = len;
1134
1135         if (ecb->truncate)
1136                 return ecb->truncate(line, len);
1137         cp = line;
1138         allot = l;
1139         while (0 < l) {
1140                 (void) utf8_width(&cp, &l);
1141                 if (!cp)
1142                         break; /* truncated in the middle? */
1143         }
1144         return allot - l;
1145 }
1146
1147 static void find_lno(const char *line, struct emit_callback *ecbdata)
1148 {
1149         const char *p;
1150         ecbdata->lno_in_preimage = 0;
1151         ecbdata->lno_in_postimage = 0;
1152         p = strchr(line, '-');
1153         if (!p)
1154                 return; /* cannot happen */
1155         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1156         p = strchr(p, '+');
1157         if (!p)
1158                 return; /* cannot happen */
1159         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1160 }
1161
1162 static void fn_out_consume(void *priv, char *line, unsigned long len)
1163 {
1164         struct emit_callback *ecbdata = priv;
1165         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1166         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1167         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1168         struct diff_options *o = ecbdata->opt;
1169         const char *line_prefix = diff_line_prefix(o);
1170
1171         if (ecbdata->header) {
1172                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1173                 strbuf_reset(ecbdata->header);
1174                 ecbdata->header = NULL;
1175         }
1176         *(ecbdata->found_changesp) = 1;
1177
1178         if (ecbdata->label_path[0]) {
1179                 const char *name_a_tab, *name_b_tab;
1180
1181                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1182                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1183
1184                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1185                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1186                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1187                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1188                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1189         }
1190
1191         if (diff_suppress_blank_empty
1192             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1193                 line[0] = '\n';
1194                 len = 1;
1195         }
1196
1197         if (line[0] == '@') {
1198                 if (ecbdata->diff_words)
1199                         diff_words_flush(ecbdata);
1200                 len = sane_truncate_line(ecbdata, line, len);
1201                 find_lno(line, ecbdata);
1202                 emit_hunk_header(ecbdata, line, len);
1203                 if (line[len-1] != '\n')
1204                         putc('\n', ecbdata->opt->file);
1205                 return;
1206         }
1207
1208         if (len < 1) {
1209                 emit_line(ecbdata->opt, reset, reset, line, len);
1210                 if (ecbdata->diff_words
1211                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1212                         fputs("~\n", ecbdata->opt->file);
1213                 return;
1214         }
1215
1216         if (ecbdata->diff_words) {
1217                 if (line[0] == '-') {
1218                         diff_words_append(line, len,
1219                                           &ecbdata->diff_words->minus);
1220                         return;
1221                 } else if (line[0] == '+') {
1222                         diff_words_append(line, len,
1223                                           &ecbdata->diff_words->plus);
1224                         return;
1225                 } else if (starts_with(line, "\\ ")) {
1226                         /*
1227                          * Eat the "no newline at eof" marker as if we
1228                          * saw a "+" or "-" line with nothing on it,
1229                          * and return without diff_words_flush() to
1230                          * defer processing. If this is the end of
1231                          * preimage, more "+" lines may come after it.
1232                          */
1233                         return;
1234                 }
1235                 diff_words_flush(ecbdata);
1236                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1237                         emit_line(ecbdata->opt, plain, reset, line, len);
1238                         fputs("~\n", ecbdata->opt->file);
1239                 } else {
1240                         /*
1241                          * Skip the prefix character, if any.  With
1242                          * diff_suppress_blank_empty, there may be
1243                          * none.
1244                          */
1245                         if (line[0] != '\n') {
1246                               line++;
1247                               len--;
1248                         }
1249                         emit_line(ecbdata->opt, plain, reset, line, len);
1250                 }
1251                 return;
1252         }
1253
1254         if (line[0] != '+') {
1255                 const char *color =
1256                         diff_get_color(ecbdata->color_diff,
1257                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1258                 ecbdata->lno_in_preimage++;
1259                 if (line[0] == ' ')
1260                         ecbdata->lno_in_postimage++;
1261                 emit_line(ecbdata->opt, color, reset, line, len);
1262         } else {
1263                 ecbdata->lno_in_postimage++;
1264                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1265         }
1266 }
1267
1268 static char *pprint_rename(const char *a, const char *b)
1269 {
1270         const char *old = a;
1271         const char *new = b;
1272         struct strbuf name = STRBUF_INIT;
1273         int pfx_length, sfx_length;
1274         int pfx_adjust_for_slash;
1275         int len_a = strlen(a);
1276         int len_b = strlen(b);
1277         int a_midlen, b_midlen;
1278         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1279         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1280
1281         if (qlen_a || qlen_b) {
1282                 quote_c_style(a, &name, NULL, 0);
1283                 strbuf_addstr(&name, " => ");
1284                 quote_c_style(b, &name, NULL, 0);
1285                 return strbuf_detach(&name, NULL);
1286         }
1287
1288         /* Find common prefix */
1289         pfx_length = 0;
1290         while (*old && *new && *old == *new) {
1291                 if (*old == '/')
1292                         pfx_length = old - a + 1;
1293                 old++;
1294                 new++;
1295         }
1296
1297         /* Find common suffix */
1298         old = a + len_a;
1299         new = b + len_b;
1300         sfx_length = 0;
1301         /*
1302          * If there is a common prefix, it must end in a slash.  In
1303          * that case we let this loop run 1 into the prefix to see the
1304          * same slash.
1305          *
1306          * If there is no common prefix, we cannot do this as it would
1307          * underrun the input strings.
1308          */
1309         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1310         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1311                b + pfx_length - pfx_adjust_for_slash <= new &&
1312                *old == *new) {
1313                 if (*old == '/')
1314                         sfx_length = len_a - (old - a);
1315                 old--;
1316                 new--;
1317         }
1318
1319         /*
1320          * pfx{mid-a => mid-b}sfx
1321          * {pfx-a => pfx-b}sfx
1322          * pfx{sfx-a => sfx-b}
1323          * name-a => name-b
1324          */
1325         a_midlen = len_a - pfx_length - sfx_length;
1326         b_midlen = len_b - pfx_length - sfx_length;
1327         if (a_midlen < 0)
1328                 a_midlen = 0;
1329         if (b_midlen < 0)
1330                 b_midlen = 0;
1331
1332         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1333         if (pfx_length + sfx_length) {
1334                 strbuf_add(&name, a, pfx_length);
1335                 strbuf_addch(&name, '{');
1336         }
1337         strbuf_add(&name, a + pfx_length, a_midlen);
1338         strbuf_addstr(&name, " => ");
1339         strbuf_add(&name, b + pfx_length, b_midlen);
1340         if (pfx_length + sfx_length) {
1341                 strbuf_addch(&name, '}');
1342                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1343         }
1344         return strbuf_detach(&name, NULL);
1345 }
1346
1347 struct diffstat_t {
1348         int nr;
1349         int alloc;
1350         struct diffstat_file {
1351                 char *from_name;
1352                 char *name;
1353                 char *print_name;
1354                 unsigned is_unmerged:1;
1355                 unsigned is_binary:1;
1356                 unsigned is_renamed:1;
1357                 unsigned is_interesting:1;
1358                 uintmax_t added, deleted;
1359         } **files;
1360 };
1361
1362 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1363                                           const char *name_a,
1364                                           const char *name_b)
1365 {
1366         struct diffstat_file *x;
1367         x = xcalloc(1, sizeof(*x));
1368         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1369         diffstat->files[diffstat->nr++] = x;
1370         if (name_b) {
1371                 x->from_name = xstrdup(name_a);
1372                 x->name = xstrdup(name_b);
1373                 x->is_renamed = 1;
1374         }
1375         else {
1376                 x->from_name = NULL;
1377                 x->name = xstrdup(name_a);
1378         }
1379         return x;
1380 }
1381
1382 static void diffstat_consume(void *priv, char *line, unsigned long len)
1383 {
1384         struct diffstat_t *diffstat = priv;
1385         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1386
1387         if (line[0] == '+')
1388                 x->added++;
1389         else if (line[0] == '-')
1390                 x->deleted++;
1391 }
1392
1393 const char mime_boundary_leader[] = "------------";
1394
1395 static int scale_linear(int it, int width, int max_change)
1396 {
1397         if (!it)
1398                 return 0;
1399         /*
1400          * make sure that at least one '-' or '+' is printed if
1401          * there is any change to this path. The easiest way is to
1402          * scale linearly as if the alloted width is one column shorter
1403          * than it is, and then add 1 to the result.
1404          */
1405         return 1 + (it * (width - 1) / max_change);
1406 }
1407
1408 static void show_name(FILE *file,
1409                       const char *prefix, const char *name, int len)
1410 {
1411         fprintf(file, " %s%-*s |", prefix, len, name);
1412 }
1413
1414 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1415 {
1416         if (cnt <= 0)
1417                 return;
1418         fprintf(file, "%s", set);
1419         while (cnt--)
1420                 putc(ch, file);
1421         fprintf(file, "%s", reset);
1422 }
1423
1424 static void fill_print_name(struct diffstat_file *file)
1425 {
1426         char *pname;
1427
1428         if (file->print_name)
1429                 return;
1430
1431         if (!file->is_renamed) {
1432                 struct strbuf buf = STRBUF_INIT;
1433                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1434                         pname = strbuf_detach(&buf, NULL);
1435                 } else {
1436                         pname = file->name;
1437                         strbuf_release(&buf);
1438                 }
1439         } else {
1440                 pname = pprint_rename(file->from_name, file->name);
1441         }
1442         file->print_name = pname;
1443 }
1444
1445 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1446 {
1447         struct strbuf sb = STRBUF_INIT;
1448         int ret;
1449
1450         if (!files) {
1451                 assert(insertions == 0 && deletions == 0);
1452                 return fprintf(fp, "%s\n", " 0 files changed");
1453         }
1454
1455         strbuf_addf(&sb,
1456                     (files == 1) ? " %d file changed" : " %d files changed",
1457                     files);
1458
1459         /*
1460          * For binary diff, the caller may want to print "x files
1461          * changed" with insertions == 0 && deletions == 0.
1462          *
1463          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1464          * is probably less confusing (i.e skip over "2 files changed
1465          * but nothing about added/removed lines? Is this a bug in Git?").
1466          */
1467         if (insertions || deletions == 0) {
1468                 strbuf_addf(&sb,
1469                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1470                             insertions);
1471         }
1472
1473         if (deletions || insertions == 0) {
1474                 strbuf_addf(&sb,
1475                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1476                             deletions);
1477         }
1478         strbuf_addch(&sb, '\n');
1479         ret = fputs(sb.buf, fp);
1480         strbuf_release(&sb);
1481         return ret;
1482 }
1483
1484 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1485 {
1486         int i, len, add, del, adds = 0, dels = 0;
1487         uintmax_t max_change = 0, max_len = 0;
1488         int total_files = data->nr, count;
1489         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1490         const char *reset, *add_c, *del_c;
1491         const char *line_prefix = "";
1492         int extra_shown = 0;
1493
1494         if (data->nr == 0)
1495                 return;
1496
1497         line_prefix = diff_line_prefix(options);
1498         count = options->stat_count ? options->stat_count : data->nr;
1499
1500         reset = diff_get_color_opt(options, DIFF_RESET);
1501         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1502         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1503
1504         /*
1505          * Find the longest filename and max number of changes
1506          */
1507         for (i = 0; (i < count) && (i < data->nr); i++) {
1508                 struct diffstat_file *file = data->files[i];
1509                 uintmax_t change = file->added + file->deleted;
1510
1511                 if (!file->is_interesting && (change == 0)) {
1512                         count++; /* not shown == room for one more */
1513                         continue;
1514                 }
1515                 fill_print_name(file);
1516                 len = strlen(file->print_name);
1517                 if (max_len < len)
1518                         max_len = len;
1519
1520                 if (file->is_unmerged) {
1521                         /* "Unmerged" is 8 characters */
1522                         bin_width = bin_width < 8 ? 8 : bin_width;
1523                         continue;
1524                 }
1525                 if (file->is_binary) {
1526                         /* "Bin XXX -> YYY bytes" */
1527                         int w = 14 + decimal_width(file->added)
1528                                 + decimal_width(file->deleted);
1529                         bin_width = bin_width < w ? w : bin_width;
1530                         /* Display change counts aligned with "Bin" */
1531                         number_width = 3;
1532                         continue;
1533                 }
1534
1535                 if (max_change < change)
1536                         max_change = change;
1537         }
1538         count = i; /* where we can stop scanning in data->files[] */
1539
1540         /*
1541          * We have width = stat_width or term_columns() columns total.
1542          * We want a maximum of min(max_len, stat_name_width) for the name part.
1543          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1544          * We also need 1 for " " and 4 + decimal_width(max_change)
1545          * for " | NNNN " and one the empty column at the end, altogether
1546          * 6 + decimal_width(max_change).
1547          *
1548          * If there's not enough space, we will use the smaller of
1549          * stat_name_width (if set) and 5/8*width for the filename,
1550          * and the rest for constant elements + graph part, but no more
1551          * than stat_graph_width for the graph part.
1552          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1553          * for the standard terminal size).
1554          *
1555          * In other words: stat_width limits the maximum width, and
1556          * stat_name_width fixes the maximum width of the filename,
1557          * and is also used to divide available columns if there
1558          * aren't enough.
1559          *
1560          * Binary files are displayed with "Bin XXX -> YYY bytes"
1561          * instead of the change count and graph. This part is treated
1562          * similarly to the graph part, except that it is not
1563          * "scaled". If total width is too small to accommodate the
1564          * guaranteed minimum width of the filename part and the
1565          * separators and this message, this message will "overflow"
1566          * making the line longer than the maximum width.
1567          */
1568
1569         if (options->stat_width == -1)
1570                 width = term_columns() - options->output_prefix_length;
1571         else
1572                 width = options->stat_width ? options->stat_width : 80;
1573         number_width = decimal_width(max_change) > number_width ?
1574                 decimal_width(max_change) : number_width;
1575
1576         if (options->stat_graph_width == -1)
1577                 options->stat_graph_width = diff_stat_graph_width;
1578
1579         /*
1580          * Guarantee 3/8*16==6 for the graph part
1581          * and 5/8*16==10 for the filename part
1582          */
1583         if (width < 16 + 6 + number_width)
1584                 width = 16 + 6 + number_width;
1585
1586         /*
1587          * First assign sizes that are wanted, ignoring available width.
1588          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1589          * starting from "XXX" should fit in graph_width.
1590          */
1591         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1592         if (options->stat_graph_width &&
1593             options->stat_graph_width < graph_width)
1594                 graph_width = options->stat_graph_width;
1595
1596         name_width = (options->stat_name_width > 0 &&
1597                       options->stat_name_width < max_len) ?
1598                 options->stat_name_width : max_len;
1599
1600         /*
1601          * Adjust adjustable widths not to exceed maximum width
1602          */
1603         if (name_width + number_width + 6 + graph_width > width) {
1604                 if (graph_width > width * 3/8 - number_width - 6) {
1605                         graph_width = width * 3/8 - number_width - 6;
1606                         if (graph_width < 6)
1607                                 graph_width = 6;
1608                 }
1609
1610                 if (options->stat_graph_width &&
1611                     graph_width > options->stat_graph_width)
1612                         graph_width = options->stat_graph_width;
1613                 if (name_width > width - number_width - 6 - graph_width)
1614                         name_width = width - number_width - 6 - graph_width;
1615                 else
1616                         graph_width = width - number_width - 6 - name_width;
1617         }
1618
1619         /*
1620          * From here name_width is the width of the name area,
1621          * and graph_width is the width of the graph area.
1622          * max_change is used to scale graph properly.
1623          */
1624         for (i = 0; i < count; i++) {
1625                 const char *prefix = "";
1626                 struct diffstat_file *file = data->files[i];
1627                 char *name = file->print_name;
1628                 uintmax_t added = file->added;
1629                 uintmax_t deleted = file->deleted;
1630                 int name_len;
1631
1632                 if (!file->is_interesting && (added + deleted == 0))
1633                         continue;
1634
1635                 /*
1636                  * "scale" the filename
1637                  */
1638                 len = name_width;
1639                 name_len = strlen(name);
1640                 if (name_width < name_len) {
1641                         char *slash;
1642                         prefix = "...";
1643                         len -= 3;
1644                         name += name_len - len;
1645                         slash = strchr(name, '/');
1646                         if (slash)
1647                                 name = slash;
1648                 }
1649
1650                 if (file->is_binary) {
1651                         fprintf(options->file, "%s", line_prefix);
1652                         show_name(options->file, prefix, name, len);
1653                         fprintf(options->file, " %*s", number_width, "Bin");
1654                         if (!added && !deleted) {
1655                                 putc('\n', options->file);
1656                                 continue;
1657                         }
1658                         fprintf(options->file, " %s%"PRIuMAX"%s",
1659                                 del_c, deleted, reset);
1660                         fprintf(options->file, " -> ");
1661                         fprintf(options->file, "%s%"PRIuMAX"%s",
1662                                 add_c, added, reset);
1663                         fprintf(options->file, " bytes");
1664                         fprintf(options->file, "\n");
1665                         continue;
1666                 }
1667                 else if (file->is_unmerged) {
1668                         fprintf(options->file, "%s", line_prefix);
1669                         show_name(options->file, prefix, name, len);
1670                         fprintf(options->file, " Unmerged\n");
1671                         continue;
1672                 }
1673
1674                 /*
1675                  * scale the add/delete
1676                  */
1677                 add = added;
1678                 del = deleted;
1679
1680                 if (graph_width <= max_change) {
1681                         int total = scale_linear(add + del, graph_width, max_change);
1682                         if (total < 2 && add && del)
1683                                 /* width >= 2 due to the sanity check */
1684                                 total = 2;
1685                         if (add < del) {
1686                                 add = scale_linear(add, graph_width, max_change);
1687                                 del = total - add;
1688                         } else {
1689                                 del = scale_linear(del, graph_width, max_change);
1690                                 add = total - del;
1691                         }
1692                 }
1693                 fprintf(options->file, "%s", line_prefix);
1694                 show_name(options->file, prefix, name, len);
1695                 fprintf(options->file, " %*"PRIuMAX"%s",
1696                         number_width, added + deleted,
1697                         added + deleted ? " " : "");
1698                 show_graph(options->file, '+', add, add_c, reset);
1699                 show_graph(options->file, '-', del, del_c, reset);
1700                 fprintf(options->file, "\n");
1701         }
1702
1703         for (i = 0; i < data->nr; i++) {
1704                 struct diffstat_file *file = data->files[i];
1705                 uintmax_t added = file->added;
1706                 uintmax_t deleted = file->deleted;
1707
1708                 if (file->is_unmerged ||
1709                     (!file->is_interesting && (added + deleted == 0))) {
1710                         total_files--;
1711                         continue;
1712                 }
1713
1714                 if (!file->is_binary) {
1715                         adds += added;
1716                         dels += deleted;
1717                 }
1718                 if (i < count)
1719                         continue;
1720                 if (!extra_shown)
1721                         fprintf(options->file, "%s ...\n", line_prefix);
1722                 extra_shown = 1;
1723         }
1724         fprintf(options->file, "%s", line_prefix);
1725         print_stat_summary(options->file, total_files, adds, dels);
1726 }
1727
1728 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1729 {
1730         int i, adds = 0, dels = 0, total_files = data->nr;
1731
1732         if (data->nr == 0)
1733                 return;
1734
1735         for (i = 0; i < data->nr; i++) {
1736                 int added = data->files[i]->added;
1737                 int deleted= data->files[i]->deleted;
1738
1739                 if (data->files[i]->is_unmerged ||
1740                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1741                         total_files--;
1742                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1743                         adds += added;
1744                         dels += deleted;
1745                 }
1746         }
1747         fprintf(options->file, "%s", diff_line_prefix(options));
1748         print_stat_summary(options->file, total_files, adds, dels);
1749 }
1750
1751 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1752 {
1753         int i;
1754
1755         if (data->nr == 0)
1756                 return;
1757
1758         for (i = 0; i < data->nr; i++) {
1759                 struct diffstat_file *file = data->files[i];
1760
1761                 fprintf(options->file, "%s", diff_line_prefix(options));
1762
1763                 if (file->is_binary)
1764                         fprintf(options->file, "-\t-\t");
1765                 else
1766                         fprintf(options->file,
1767                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1768                                 file->added, file->deleted);
1769                 if (options->line_termination) {
1770                         fill_print_name(file);
1771                         if (!file->is_renamed)
1772                                 write_name_quoted(file->name, options->file,
1773                                                   options->line_termination);
1774                         else {
1775                                 fputs(file->print_name, options->file);
1776                                 putc(options->line_termination, options->file);
1777                         }
1778                 } else {
1779                         if (file->is_renamed) {
1780                                 putc('\0', options->file);
1781                                 write_name_quoted(file->from_name, options->file, '\0');
1782                         }
1783                         write_name_quoted(file->name, options->file, '\0');
1784                 }
1785         }
1786 }
1787
1788 struct dirstat_file {
1789         const char *name;
1790         unsigned long changed;
1791 };
1792
1793 struct dirstat_dir {
1794         struct dirstat_file *files;
1795         int alloc, nr, permille, cumulative;
1796 };
1797
1798 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1799                 unsigned long changed, const char *base, int baselen)
1800 {
1801         unsigned long this_dir = 0;
1802         unsigned int sources = 0;
1803         const char *line_prefix = diff_line_prefix(opt);
1804
1805         while (dir->nr) {
1806                 struct dirstat_file *f = dir->files;
1807                 int namelen = strlen(f->name);
1808                 unsigned long this;
1809                 char *slash;
1810
1811                 if (namelen < baselen)
1812                         break;
1813                 if (memcmp(f->name, base, baselen))
1814                         break;
1815                 slash = strchr(f->name + baselen, '/');
1816                 if (slash) {
1817                         int newbaselen = slash + 1 - f->name;
1818                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1819                         sources++;
1820                 } else {
1821                         this = f->changed;
1822                         dir->files++;
1823                         dir->nr--;
1824                         sources += 2;
1825                 }
1826                 this_dir += this;
1827         }
1828
1829         /*
1830          * We don't report dirstat's for
1831          *  - the top level
1832          *  - or cases where everything came from a single directory
1833          *    under this directory (sources == 1).
1834          */
1835         if (baselen && sources != 1) {
1836                 if (this_dir) {
1837                         int permille = this_dir * 1000 / changed;
1838                         if (permille >= dir->permille) {
1839                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1840                                         permille / 10, permille % 10, baselen, base);
1841                                 if (!dir->cumulative)
1842                                         return 0;
1843                         }
1844                 }
1845         }
1846         return this_dir;
1847 }
1848
1849 static int dirstat_compare(const void *_a, const void *_b)
1850 {
1851         const struct dirstat_file *a = _a;
1852         const struct dirstat_file *b = _b;
1853         return strcmp(a->name, b->name);
1854 }
1855
1856 static void show_dirstat(struct diff_options *options)
1857 {
1858         int i;
1859         unsigned long changed;
1860         struct dirstat_dir dir;
1861         struct diff_queue_struct *q = &diff_queued_diff;
1862
1863         dir.files = NULL;
1864         dir.alloc = 0;
1865         dir.nr = 0;
1866         dir.permille = options->dirstat_permille;
1867         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1868
1869         changed = 0;
1870         for (i = 0; i < q->nr; i++) {
1871                 struct diff_filepair *p = q->queue[i];
1872                 const char *name;
1873                 unsigned long copied, added, damage;
1874                 int content_changed;
1875
1876                 name = p->two->path ? p->two->path : p->one->path;
1877
1878                 if (p->one->sha1_valid && p->two->sha1_valid)
1879                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1880                 else
1881                         content_changed = 1;
1882
1883                 if (!content_changed) {
1884                         /*
1885                          * The SHA1 has not changed, so pre-/post-content is
1886                          * identical. We can therefore skip looking at the
1887                          * file contents altogether.
1888                          */
1889                         damage = 0;
1890                         goto found_damage;
1891                 }
1892
1893                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1894                         /*
1895                          * In --dirstat-by-file mode, we don't really need to
1896                          * look at the actual file contents at all.
1897                          * The fact that the SHA1 changed is enough for us to
1898                          * add this file to the list of results
1899                          * (with each file contributing equal damage).
1900                          */
1901                         damage = 1;
1902                         goto found_damage;
1903                 }
1904
1905                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1906                         diff_populate_filespec(p->one, 0);
1907                         diff_populate_filespec(p->two, 0);
1908                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1909                                                &copied, &added);
1910                         diff_free_filespec_data(p->one);
1911                         diff_free_filespec_data(p->two);
1912                 } else if (DIFF_FILE_VALID(p->one)) {
1913                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1914                         copied = added = 0;
1915                         diff_free_filespec_data(p->one);
1916                 } else if (DIFF_FILE_VALID(p->two)) {
1917                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1918                         copied = 0;
1919                         added = p->two->size;
1920                         diff_free_filespec_data(p->two);
1921                 } else
1922                         continue;
1923
1924                 /*
1925                  * Original minus copied is the removed material,
1926                  * added is the new material.  They are both damages
1927                  * made to the preimage.
1928                  * If the resulting damage is zero, we know that
1929                  * diffcore_count_changes() considers the two entries to
1930                  * be identical, but since content_changed is true, we
1931                  * know that there must have been _some_ kind of change,
1932                  * so we force all entries to have damage > 0.
1933                  */
1934                 damage = (p->one->size - copied) + added;
1935                 if (!damage)
1936                         damage = 1;
1937
1938 found_damage:
1939                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1940                 dir.files[dir.nr].name = name;
1941                 dir.files[dir.nr].changed = damage;
1942                 changed += damage;
1943                 dir.nr++;
1944         }
1945
1946         /* This can happen even with many files, if everything was renames */
1947         if (!changed)
1948                 return;
1949
1950         /* Show all directories with more than x% of the changes */
1951         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1952         gather_dirstat(options, &dir, changed, "", 0);
1953 }
1954
1955 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1956 {
1957         int i;
1958         unsigned long changed;
1959         struct dirstat_dir dir;
1960
1961         if (data->nr == 0)
1962                 return;
1963
1964         dir.files = NULL;
1965         dir.alloc = 0;
1966         dir.nr = 0;
1967         dir.permille = options->dirstat_permille;
1968         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1969
1970         changed = 0;
1971         for (i = 0; i < data->nr; i++) {
1972                 struct diffstat_file *file = data->files[i];
1973                 unsigned long damage = file->added + file->deleted;
1974                 if (file->is_binary)
1975                         /*
1976                          * binary files counts bytes, not lines. Must find some
1977                          * way to normalize binary bytes vs. textual lines.
1978                          * The following heuristic assumes that there are 64
1979                          * bytes per "line".
1980                          * This is stupid and ugly, but very cheap...
1981                          */
1982                         damage = (damage + 63) / 64;
1983                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1984                 dir.files[dir.nr].name = file->name;
1985                 dir.files[dir.nr].changed = damage;
1986                 changed += damage;
1987                 dir.nr++;
1988         }
1989
1990         /* This can happen even with many files, if everything was renames */
1991         if (!changed)
1992                 return;
1993
1994         /* Show all directories with more than x% of the changes */
1995         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1996         gather_dirstat(options, &dir, changed, "", 0);
1997 }
1998
1999 static void free_diffstat_info(struct diffstat_t *diffstat)
2000 {
2001         int i;
2002         for (i = 0; i < diffstat->nr; i++) {
2003                 struct diffstat_file *f = diffstat->files[i];
2004                 if (f->name != f->print_name)
2005                         free(f->print_name);
2006                 free(f->name);
2007                 free(f->from_name);
2008                 free(f);
2009         }
2010         free(diffstat->files);
2011 }
2012
2013 struct checkdiff_t {
2014         const char *filename;
2015         int lineno;
2016         int conflict_marker_size;
2017         struct diff_options *o;
2018         unsigned ws_rule;
2019         unsigned status;
2020 };
2021
2022 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2023 {
2024         char firstchar;
2025         int cnt;
2026
2027         if (len < marker_size + 1)
2028                 return 0;
2029         firstchar = line[0];
2030         switch (firstchar) {
2031         case '=': case '>': case '<': case '|':
2032                 break;
2033         default:
2034                 return 0;
2035         }
2036         for (cnt = 1; cnt < marker_size; cnt++)
2037                 if (line[cnt] != firstchar)
2038                         return 0;
2039         /* line[1] thru line[marker_size-1] are same as firstchar */
2040         if (len < marker_size + 1 || !isspace(line[marker_size]))
2041                 return 0;
2042         return 1;
2043 }
2044
2045 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2046 {
2047         struct checkdiff_t *data = priv;
2048         int marker_size = data->conflict_marker_size;
2049         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2050         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2051         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2052         char *err;
2053         const char *line_prefix;
2054
2055         assert(data->o);
2056         line_prefix = diff_line_prefix(data->o);
2057
2058         if (line[0] == '+') {
2059                 unsigned bad;
2060                 data->lineno++;
2061                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2062                         data->status |= 1;
2063                         fprintf(data->o->file,
2064                                 "%s%s:%d: leftover conflict marker\n",
2065                                 line_prefix, data->filename, data->lineno);
2066                 }
2067                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2068                 if (!bad)
2069                         return;
2070                 data->status |= bad;
2071                 err = whitespace_error_string(bad);
2072                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2073                         line_prefix, data->filename, data->lineno, err);
2074                 free(err);
2075                 emit_line(data->o, set, reset, line, 1);
2076                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2077                               data->o->file, set, reset, ws);
2078         } else if (line[0] == ' ') {
2079                 data->lineno++;
2080         } else if (line[0] == '@') {
2081                 char *plus = strchr(line, '+');
2082                 if (plus)
2083                         data->lineno = strtol(plus, NULL, 10) - 1;
2084                 else
2085                         die("invalid diff");
2086         }
2087 }
2088
2089 static unsigned char *deflate_it(char *data,
2090                                  unsigned long size,
2091                                  unsigned long *result_size)
2092 {
2093         int bound;
2094         unsigned char *deflated;
2095         git_zstream stream;
2096
2097         memset(&stream, 0, sizeof(stream));
2098         git_deflate_init(&stream, zlib_compression_level);
2099         bound = git_deflate_bound(&stream, size);
2100         deflated = xmalloc(bound);
2101         stream.next_out = deflated;
2102         stream.avail_out = bound;
2103
2104         stream.next_in = (unsigned char *)data;
2105         stream.avail_in = size;
2106         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2107                 ; /* nothing */
2108         git_deflate_end(&stream);
2109         *result_size = stream.total_out;
2110         return deflated;
2111 }
2112
2113 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2114                                   const char *prefix)
2115 {
2116         void *cp;
2117         void *delta;
2118         void *deflated;
2119         void *data;
2120         unsigned long orig_size;
2121         unsigned long delta_size;
2122         unsigned long deflate_size;
2123         unsigned long data_size;
2124
2125         /* We could do deflated delta, or we could do just deflated two,
2126          * whichever is smaller.
2127          */
2128         delta = NULL;
2129         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2130         if (one->size && two->size) {
2131                 delta = diff_delta(one->ptr, one->size,
2132                                    two->ptr, two->size,
2133                                    &delta_size, deflate_size);
2134                 if (delta) {
2135                         void *to_free = delta;
2136                         orig_size = delta_size;
2137                         delta = deflate_it(delta, delta_size, &delta_size);
2138                         free(to_free);
2139                 }
2140         }
2141
2142         if (delta && delta_size < deflate_size) {
2143                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2144                 free(deflated);
2145                 data = delta;
2146                 data_size = delta_size;
2147         }
2148         else {
2149                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2150                 free(delta);
2151                 data = deflated;
2152                 data_size = deflate_size;
2153         }
2154
2155         /* emit data encoded in base85 */
2156         cp = data;
2157         while (data_size) {
2158                 int bytes = (52 < data_size) ? 52 : data_size;
2159                 char line[70];
2160                 data_size -= bytes;
2161                 if (bytes <= 26)
2162                         line[0] = bytes + 'A' - 1;
2163                 else
2164                         line[0] = bytes - 26 + 'a' - 1;
2165                 encode_85(line + 1, cp, bytes);
2166                 cp = (char *) cp + bytes;
2167                 fprintf(file, "%s", prefix);
2168                 fputs(line, file);
2169                 fputc('\n', file);
2170         }
2171         fprintf(file, "%s\n", prefix);
2172         free(data);
2173 }
2174
2175 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2176                              const char *prefix)
2177 {
2178         fprintf(file, "%sGIT binary patch\n", prefix);
2179         emit_binary_diff_body(file, one, two, prefix);
2180         emit_binary_diff_body(file, two, one, prefix);
2181 }
2182
2183 int diff_filespec_is_binary(struct diff_filespec *one)
2184 {
2185         if (one->is_binary == -1) {
2186                 diff_filespec_load_driver(one);
2187                 if (one->driver->binary != -1)
2188                         one->is_binary = one->driver->binary;
2189                 else {
2190                         if (!one->data && DIFF_FILE_VALID(one))
2191                                 diff_populate_filespec(one, CHECK_BINARY);
2192                         if (one->is_binary == -1 && one->data)
2193                                 one->is_binary = buffer_is_binary(one->data,
2194                                                 one->size);
2195                         if (one->is_binary == -1)
2196                                 one->is_binary = 0;
2197                 }
2198         }
2199         return one->is_binary;
2200 }
2201
2202 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2203 {
2204         diff_filespec_load_driver(one);
2205         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2206 }
2207
2208 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2209 {
2210         if (!options->a_prefix)
2211                 options->a_prefix = a;
2212         if (!options->b_prefix)
2213                 options->b_prefix = b;
2214 }
2215
2216 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2217 {
2218         if (!DIFF_FILE_VALID(one))
2219                 return NULL;
2220
2221         diff_filespec_load_driver(one);
2222         return userdiff_get_textconv(one->driver);
2223 }
2224
2225 static void builtin_diff(const char *name_a,
2226                          const char *name_b,
2227                          struct diff_filespec *one,
2228                          struct diff_filespec *two,
2229                          const char *xfrm_msg,
2230                          int must_show_header,
2231                          struct diff_options *o,
2232                          int complete_rewrite)
2233 {
2234         mmfile_t mf1, mf2;
2235         const char *lbl[2];
2236         char *a_one, *b_two;
2237         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2238         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2239         const char *a_prefix, *b_prefix;
2240         struct userdiff_driver *textconv_one = NULL;
2241         struct userdiff_driver *textconv_two = NULL;
2242         struct strbuf header = STRBUF_INIT;
2243         const char *line_prefix = diff_line_prefix(o);
2244
2245         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2246                         (!one->mode || S_ISGITLINK(one->mode)) &&
2247                         (!two->mode || S_ISGITLINK(two->mode))) {
2248                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2249                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2250                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2251                                 line_prefix,
2252                                 one->sha1, two->sha1, two->dirty_submodule,
2253                                 meta, del, add, reset);
2254                 return;
2255         }
2256
2257         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2258                 textconv_one = get_textconv(one);
2259                 textconv_two = get_textconv(two);
2260         }
2261
2262         diff_set_mnemonic_prefix(o, "a/", "b/");
2263         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2264                 a_prefix = o->b_prefix;
2265                 b_prefix = o->a_prefix;
2266         } else {
2267                 a_prefix = o->a_prefix;
2268                 b_prefix = o->b_prefix;
2269         }
2270
2271         /* Never use a non-valid filename anywhere if at all possible */
2272         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2273         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2274
2275         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2276         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2277         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2278         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2279         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2280         if (lbl[0][0] == '/') {
2281                 /* /dev/null */
2282                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2283                 if (xfrm_msg)
2284                         strbuf_addstr(&header, xfrm_msg);
2285                 must_show_header = 1;
2286         }
2287         else if (lbl[1][0] == '/') {
2288                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2289                 if (xfrm_msg)
2290                         strbuf_addstr(&header, xfrm_msg);
2291                 must_show_header = 1;
2292         }
2293         else {
2294                 if (one->mode != two->mode) {
2295                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2296                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2297                         must_show_header = 1;
2298                 }
2299                 if (xfrm_msg)
2300                         strbuf_addstr(&header, xfrm_msg);
2301
2302                 /*
2303                  * we do not run diff between different kind
2304                  * of objects.
2305                  */
2306                 if ((one->mode ^ two->mode) & S_IFMT)
2307                         goto free_ab_and_return;
2308                 if (complete_rewrite &&
2309                     (textconv_one || !diff_filespec_is_binary(one)) &&
2310                     (textconv_two || !diff_filespec_is_binary(two))) {
2311                         fprintf(o->file, "%s", header.buf);
2312                         strbuf_reset(&header);
2313                         emit_rewrite_diff(name_a, name_b, one, two,
2314                                                 textconv_one, textconv_two, o);
2315                         o->found_changes = 1;
2316                         goto free_ab_and_return;
2317                 }
2318         }
2319
2320         if (o->irreversible_delete && lbl[1][0] == '/') {
2321                 fprintf(o->file, "%s", header.buf);
2322                 strbuf_reset(&header);
2323                 goto free_ab_and_return;
2324         } else if (!DIFF_OPT_TST(o, TEXT) &&
2325             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2326               (!textconv_two && diff_filespec_is_binary(two)) )) {
2327                 if (!one->data && !two->data &&
2328                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2329                     !DIFF_OPT_TST(o, BINARY)) {
2330                         if (!hashcmp(one->sha1, two->sha1)) {
2331                                 if (must_show_header)
2332                                         fprintf(o->file, "%s", header.buf);
2333                                 goto free_ab_and_return;
2334                         }
2335                         fprintf(o->file, "%s", header.buf);
2336                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2337                                 line_prefix, lbl[0], lbl[1]);
2338                         goto free_ab_and_return;
2339                 }
2340                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2341                         die("unable to read files to diff");
2342                 /* Quite common confusing case */
2343                 if (mf1.size == mf2.size &&
2344                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2345                         if (must_show_header)
2346                                 fprintf(o->file, "%s", header.buf);
2347                         goto free_ab_and_return;
2348                 }
2349                 fprintf(o->file, "%s", header.buf);
2350                 strbuf_reset(&header);
2351                 if (DIFF_OPT_TST(o, BINARY))
2352                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2353                 else
2354                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2355                                 line_prefix, lbl[0], lbl[1]);
2356                 o->found_changes = 1;
2357         } else {
2358                 /* Crazy xdl interfaces.. */
2359                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2360                 const char *v;
2361                 xpparam_t xpp;
2362                 xdemitconf_t xecfg;
2363                 struct emit_callback ecbdata;
2364                 const struct userdiff_funcname *pe;
2365
2366                 if (must_show_header) {
2367                         fprintf(o->file, "%s", header.buf);
2368                         strbuf_reset(&header);
2369                 }
2370
2371                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2372                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2373
2374                 pe = diff_funcname_pattern(one);
2375                 if (!pe)
2376                         pe = diff_funcname_pattern(two);
2377
2378                 memset(&xpp, 0, sizeof(xpp));
2379                 memset(&xecfg, 0, sizeof(xecfg));
2380                 memset(&ecbdata, 0, sizeof(ecbdata));
2381                 ecbdata.label_path = lbl;
2382                 ecbdata.color_diff = want_color(o->use_color);
2383                 ecbdata.found_changesp = &o->found_changes;
2384                 ecbdata.ws_rule = whitespace_rule(name_b);
2385                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2386                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2387                 ecbdata.opt = o;
2388                 ecbdata.header = header.len ? &header : NULL;
2389                 xpp.flags = o->xdl_opts;
2390                 xecfg.ctxlen = o->context;
2391                 xecfg.interhunkctxlen = o->interhunkcontext;
2392                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2393                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2394                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2395                 if (pe)
2396                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2397                 if (!diffopts)
2398                         ;
2399                 else if (skip_prefix(diffopts, "--unified=", &v))
2400                         xecfg.ctxlen = strtoul(v, NULL, 10);
2401                 else if (skip_prefix(diffopts, "-u", &v))
2402                         xecfg.ctxlen = strtoul(v, NULL, 10);
2403                 if (o->word_diff)
2404                         init_diff_words_data(&ecbdata, o, one, two);
2405                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2406                               &xpp, &xecfg);
2407                 if (o->word_diff)
2408                         free_diff_words_data(&ecbdata);
2409                 if (textconv_one)
2410                         free(mf1.ptr);
2411                 if (textconv_two)
2412                         free(mf2.ptr);
2413                 xdiff_clear_find_func(&xecfg);
2414         }
2415
2416  free_ab_and_return:
2417         strbuf_release(&header);
2418         diff_free_filespec_data(one);
2419         diff_free_filespec_data(two);
2420         free(a_one);
2421         free(b_two);
2422         return;
2423 }
2424
2425 static void builtin_diffstat(const char *name_a, const char *name_b,
2426                              struct diff_filespec *one,
2427                              struct diff_filespec *two,
2428                              struct diffstat_t *diffstat,
2429                              struct diff_options *o,
2430                              struct diff_filepair *p)
2431 {
2432         mmfile_t mf1, mf2;
2433         struct diffstat_file *data;
2434         int same_contents;
2435         int complete_rewrite = 0;
2436
2437         if (!DIFF_PAIR_UNMERGED(p)) {
2438                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2439                         complete_rewrite = 1;
2440         }
2441
2442         data = diffstat_add(diffstat, name_a, name_b);
2443         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2444
2445         if (!one || !two) {
2446                 data->is_unmerged = 1;
2447                 return;
2448         }
2449
2450         same_contents = !hashcmp(one->sha1, two->sha1);
2451
2452         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2453                 data->is_binary = 1;
2454                 if (same_contents) {
2455                         data->added = 0;
2456                         data->deleted = 0;
2457                 } else {
2458                         data->added = diff_filespec_size(two);
2459                         data->deleted = diff_filespec_size(one);
2460                 }
2461         }
2462
2463         else if (complete_rewrite) {
2464                 diff_populate_filespec(one, 0);
2465                 diff_populate_filespec(two, 0);
2466                 data->deleted = count_lines(one->data, one->size);
2467                 data->added = count_lines(two->data, two->size);
2468         }
2469
2470         else if (!same_contents) {
2471                 /* Crazy xdl interfaces.. */
2472                 xpparam_t xpp;
2473                 xdemitconf_t xecfg;
2474
2475                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2476                         die("unable to read files to diff");
2477
2478                 memset(&xpp, 0, sizeof(xpp));
2479                 memset(&xecfg, 0, sizeof(xecfg));
2480                 xpp.flags = o->xdl_opts;
2481                 xecfg.ctxlen = o->context;
2482                 xecfg.interhunkctxlen = o->interhunkcontext;
2483                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2484                               &xpp, &xecfg);
2485         }
2486
2487         diff_free_filespec_data(one);
2488         diff_free_filespec_data(two);
2489 }
2490
2491 static void builtin_checkdiff(const char *name_a, const char *name_b,
2492                               const char *attr_path,
2493                               struct diff_filespec *one,
2494                               struct diff_filespec *two,
2495                               struct diff_options *o)
2496 {
2497         mmfile_t mf1, mf2;
2498         struct checkdiff_t data;
2499
2500         if (!two)
2501                 return;
2502
2503         memset(&data, 0, sizeof(data));
2504         data.filename = name_b ? name_b : name_a;
2505         data.lineno = 0;
2506         data.o = o;
2507         data.ws_rule = whitespace_rule(attr_path);
2508         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2509
2510         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2511                 die("unable to read files to diff");
2512
2513         /*
2514          * All the other codepaths check both sides, but not checking
2515          * the "old" side here is deliberate.  We are checking the newly
2516          * introduced changes, and as long as the "new" side is text, we
2517          * can and should check what it introduces.
2518          */
2519         if (diff_filespec_is_binary(two))
2520                 goto free_and_return;
2521         else {
2522                 /* Crazy xdl interfaces.. */
2523                 xpparam_t xpp;
2524                 xdemitconf_t xecfg;
2525
2526                 memset(&xpp, 0, sizeof(xpp));
2527                 memset(&xecfg, 0, sizeof(xecfg));
2528                 xecfg.ctxlen = 1; /* at least one context line */
2529                 xpp.flags = 0;
2530                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2531                               &xpp, &xecfg);
2532
2533                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2534                         struct emit_callback ecbdata;
2535                         int blank_at_eof;
2536
2537                         ecbdata.ws_rule = data.ws_rule;
2538                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2539                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2540
2541                         if (blank_at_eof) {
2542                                 static char *err;
2543                                 if (!err)
2544                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2545                                 fprintf(o->file, "%s:%d: %s.\n",
2546                                         data.filename, blank_at_eof, err);
2547                                 data.status = 1; /* report errors */
2548                         }
2549                 }
2550         }
2551  free_and_return:
2552         diff_free_filespec_data(one);
2553         diff_free_filespec_data(two);
2554         if (data.status)
2555                 DIFF_OPT_SET(o, CHECK_FAILED);
2556 }
2557
2558 struct diff_filespec *alloc_filespec(const char *path)
2559 {
2560         int namelen = strlen(path);
2561         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2562
2563         memset(spec, 0, sizeof(*spec));
2564         spec->path = (char *)(spec + 1);
2565         memcpy(spec->path, path, namelen+1);
2566         spec->count = 1;
2567         spec->is_binary = -1;
2568         return spec;
2569 }
2570
2571 void free_filespec(struct diff_filespec *spec)
2572 {
2573         if (!--spec->count) {
2574                 diff_free_filespec_data(spec);
2575                 free(spec);
2576         }
2577 }
2578
2579 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2580                    int sha1_valid, unsigned short mode)
2581 {
2582         if (mode) {
2583                 spec->mode = canon_mode(mode);
2584                 hashcpy(spec->sha1, sha1);
2585                 spec->sha1_valid = sha1_valid;
2586         }
2587 }
2588
2589 /*
2590  * Given a name and sha1 pair, if the index tells us the file in
2591  * the work tree has that object contents, return true, so that
2592  * prepare_temp_file() does not have to inflate and extract.
2593  */
2594 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2595 {
2596         const struct cache_entry *ce;
2597         struct stat st;
2598         int pos, len;
2599
2600         /*
2601          * We do not read the cache ourselves here, because the
2602          * benchmark with my previous version that always reads cache
2603          * shows that it makes things worse for diff-tree comparing
2604          * two linux-2.6 kernel trees in an already checked out work
2605          * tree.  This is because most diff-tree comparisons deal with
2606          * only a small number of files, while reading the cache is
2607          * expensive for a large project, and its cost outweighs the
2608          * savings we get by not inflating the object to a temporary
2609          * file.  Practically, this code only helps when we are used
2610          * by diff-cache --cached, which does read the cache before
2611          * calling us.
2612          */
2613         if (!active_cache)
2614                 return 0;
2615
2616         /* We want to avoid the working directory if our caller
2617          * doesn't need the data in a normal file, this system
2618          * is rather slow with its stat/open/mmap/close syscalls,
2619          * and the object is contained in a pack file.  The pack
2620          * is probably already open and will be faster to obtain
2621          * the data through than the working directory.  Loose
2622          * objects however would tend to be slower as they need
2623          * to be individually opened and inflated.
2624          */
2625         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2626                 return 0;
2627
2628         len = strlen(name);
2629         pos = cache_name_pos(name, len);
2630         if (pos < 0)
2631                 return 0;
2632         ce = active_cache[pos];
2633
2634         /*
2635          * This is not the sha1 we are looking for, or
2636          * unreusable because it is not a regular file.
2637          */
2638         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2639                 return 0;
2640
2641         /*
2642          * If ce is marked as "assume unchanged", there is no
2643          * guarantee that work tree matches what we are looking for.
2644          */
2645         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2646                 return 0;
2647
2648         /*
2649          * If ce matches the file in the work tree, we can reuse it.
2650          */
2651         if (ce_uptodate(ce) ||
2652             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2653                 return 1;
2654
2655         return 0;
2656 }
2657
2658 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2659 {
2660         int len;
2661         char *data = xmalloc(100), *dirty = "";
2662
2663         /* Are we looking at the work tree? */
2664         if (s->dirty_submodule)
2665                 dirty = "-dirty";
2666
2667         len = snprintf(data, 100,
2668                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2669         s->data = data;
2670         s->size = len;
2671         s->should_free = 1;
2672         if (size_only) {
2673                 s->data = NULL;
2674                 free(data);
2675         }
2676         return 0;
2677 }
2678
2679 /*
2680  * While doing rename detection and pickaxe operation, we may need to
2681  * grab the data for the blob (or file) for our own in-core comparison.
2682  * diff_filespec has data and size fields for this purpose.
2683  */
2684 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2685 {
2686         int size_only = flags & CHECK_SIZE_ONLY;
2687         int err = 0;
2688         /*
2689          * demote FAIL to WARN to allow inspecting the situation
2690          * instead of refusing.
2691          */
2692         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2693                                     ? SAFE_CRLF_WARN
2694                                     : safe_crlf);
2695
2696         if (!DIFF_FILE_VALID(s))
2697                 die("internal error: asking to populate invalid file.");
2698         if (S_ISDIR(s->mode))
2699                 return -1;
2700
2701         if (s->data)
2702                 return 0;
2703
2704         if (size_only && 0 < s->size)
2705                 return 0;
2706
2707         if (S_ISGITLINK(s->mode))
2708                 return diff_populate_gitlink(s, size_only);
2709
2710         if (!s->sha1_valid ||
2711             reuse_worktree_file(s->path, s->sha1, 0)) {
2712                 struct strbuf buf = STRBUF_INIT;
2713                 struct stat st;
2714                 int fd;
2715
2716                 if (lstat(s->path, &st) < 0) {
2717                         if (errno == ENOENT) {
2718                         err_empty:
2719                                 err = -1;
2720                         empty:
2721                                 s->data = (char *)"";
2722                                 s->size = 0;
2723                                 return err;
2724                         }
2725                 }
2726                 s->size = xsize_t(st.st_size);
2727                 if (!s->size)
2728                         goto empty;
2729                 if (S_ISLNK(st.st_mode)) {
2730                         struct strbuf sb = STRBUF_INIT;
2731
2732                         if (strbuf_readlink(&sb, s->path, s->size))
2733                                 goto err_empty;
2734                         s->size = sb.len;
2735                         s->data = strbuf_detach(&sb, NULL);
2736                         s->should_free = 1;
2737                         return 0;
2738                 }
2739                 if (size_only)
2740                         return 0;
2741                 if ((flags & CHECK_BINARY) &&
2742                     s->size > big_file_threshold && s->is_binary == -1) {
2743                         s->is_binary = 1;
2744                         return 0;
2745                 }
2746                 fd = open(s->path, O_RDONLY);
2747                 if (fd < 0)
2748                         goto err_empty;
2749                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2750                 close(fd);
2751                 s->should_munmap = 1;
2752
2753                 /*
2754                  * Convert from working tree format to canonical git format
2755                  */
2756                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2757                         size_t size = 0;
2758                         munmap(s->data, s->size);
2759                         s->should_munmap = 0;
2760                         s->data = strbuf_detach(&buf, &size);
2761                         s->size = size;
2762                         s->should_free = 1;
2763                 }
2764         }
2765         else {
2766                 enum object_type type;
2767                 if (size_only || (flags & CHECK_BINARY)) {
2768                         type = sha1_object_info(s->sha1, &s->size);
2769                         if (type < 0)
2770                                 die("unable to read %s", sha1_to_hex(s->sha1));
2771                         if (size_only)
2772                                 return 0;
2773                         if (s->size > big_file_threshold && s->is_binary == -1) {
2774                                 s->is_binary = 1;
2775                                 return 0;
2776                         }
2777                 }
2778                 s->data = read_sha1_file(s->sha1, &type, &s->size);
2779                 if (!s->data)
2780                         die("unable to read %s", sha1_to_hex(s->sha1));
2781                 s->should_free = 1;
2782         }
2783         return 0;
2784 }
2785
2786 void diff_free_filespec_blob(struct diff_filespec *s)
2787 {
2788         if (s->should_free)
2789                 free(s->data);
2790         else if (s->should_munmap)
2791                 munmap(s->data, s->size);
2792
2793         if (s->should_free || s->should_munmap) {
2794                 s->should_free = s->should_munmap = 0;
2795                 s->data = NULL;
2796         }
2797 }
2798
2799 void diff_free_filespec_data(struct diff_filespec *s)
2800 {
2801         diff_free_filespec_blob(s);
2802         free(s->cnt_data);
2803         s->cnt_data = NULL;
2804 }
2805
2806 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2807                            void *blob,
2808                            unsigned long size,
2809                            const unsigned char *sha1,
2810                            int mode)
2811 {
2812         int fd;
2813         struct strbuf buf = STRBUF_INIT;
2814         struct strbuf template = STRBUF_INIT;
2815         char *path_dup = xstrdup(path);
2816         const char *base = basename(path_dup);
2817
2818         /* Generate "XXXXXX_basename.ext" */
2819         strbuf_addstr(&template, "XXXXXX_");
2820         strbuf_addstr(&template, base);
2821
2822         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2823                         strlen(base) + 1);
2824         if (fd < 0)
2825                 die_errno("unable to create temp-file");
2826         if (convert_to_working_tree(path,
2827                         (const char *)blob, (size_t)size, &buf)) {
2828                 blob = buf.buf;
2829                 size = buf.len;
2830         }
2831         if (write_in_full(fd, blob, size) != size)
2832                 die_errno("unable to write temp-file");
2833         close(fd);
2834         temp->name = temp->tmp_path;
2835         strcpy(temp->hex, sha1_to_hex(sha1));
2836         temp->hex[40] = 0;
2837         sprintf(temp->mode, "%06o", mode);
2838         strbuf_release(&buf);
2839         strbuf_release(&template);
2840         free(path_dup);
2841 }
2842
2843 static struct diff_tempfile *prepare_temp_file(const char *name,
2844                 struct diff_filespec *one)
2845 {
2846         struct diff_tempfile *temp = claim_diff_tempfile();
2847
2848         if (!DIFF_FILE_VALID(one)) {
2849         not_a_valid_file:
2850                 /* A '-' entry produces this for file-2, and
2851                  * a '+' entry produces this for file-1.
2852                  */
2853                 temp->name = "/dev/null";
2854                 strcpy(temp->hex, ".");
2855                 strcpy(temp->mode, ".");
2856                 return temp;
2857         }
2858
2859         if (!remove_tempfile_installed) {
2860                 atexit(remove_tempfile);
2861                 sigchain_push_common(remove_tempfile_on_signal);
2862                 remove_tempfile_installed = 1;
2863         }
2864
2865         if (!S_ISGITLINK(one->mode) &&
2866             (!one->sha1_valid ||
2867              reuse_worktree_file(name, one->sha1, 1))) {
2868                 struct stat st;
2869                 if (lstat(name, &st) < 0) {
2870                         if (errno == ENOENT)
2871                                 goto not_a_valid_file;
2872                         die_errno("stat(%s)", name);
2873                 }
2874                 if (S_ISLNK(st.st_mode)) {
2875                         struct strbuf sb = STRBUF_INIT;
2876                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2877                                 die_errno("readlink(%s)", name);
2878                         prep_temp_blob(name, temp, sb.buf, sb.len,
2879                                        (one->sha1_valid ?
2880                                         one->sha1 : null_sha1),
2881                                        (one->sha1_valid ?
2882                                         one->mode : S_IFLNK));
2883                         strbuf_release(&sb);
2884                 }
2885                 else {
2886                         /* we can borrow from the file in the work tree */
2887                         temp->name = name;
2888                         if (!one->sha1_valid)
2889                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2890                         else
2891                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2892                         /* Even though we may sometimes borrow the
2893                          * contents from the work tree, we always want
2894                          * one->mode.  mode is trustworthy even when
2895                          * !(one->sha1_valid), as long as
2896                          * DIFF_FILE_VALID(one).
2897                          */
2898                         sprintf(temp->mode, "%06o", one->mode);
2899                 }
2900                 return temp;
2901         }
2902         else {
2903                 if (diff_populate_filespec(one, 0))
2904                         die("cannot read data blob for %s", one->path);
2905                 prep_temp_blob(name, temp, one->data, one->size,
2906                                one->sha1, one->mode);
2907         }
2908         return temp;
2909 }
2910
2911 static void add_external_diff_name(struct argv_array *argv,
2912                                    const char *name,
2913                                    struct diff_filespec *df)
2914 {
2915         struct diff_tempfile *temp = prepare_temp_file(name, df);
2916         argv_array_push(argv, temp->name);
2917         argv_array_push(argv, temp->hex);
2918         argv_array_push(argv, temp->mode);
2919 }
2920
2921 /* An external diff command takes:
2922  *
2923  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2924  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2925  *
2926  */
2927 static void run_external_diff(const char *pgm,
2928                               const char *name,
2929                               const char *other,
2930                               struct diff_filespec *one,
2931                               struct diff_filespec *two,
2932                               const char *xfrm_msg,
2933                               int complete_rewrite,
2934                               struct diff_options *o)
2935 {
2936         struct argv_array argv = ARGV_ARRAY_INIT;
2937         struct argv_array env = ARGV_ARRAY_INIT;
2938         struct diff_queue_struct *q = &diff_queued_diff;
2939
2940         argv_array_push(&argv, pgm);
2941         argv_array_push(&argv, name);
2942
2943         if (one && two) {
2944                 add_external_diff_name(&argv, name, one);
2945                 if (!other)
2946                         add_external_diff_name(&argv, name, two);
2947                 else {
2948                         add_external_diff_name(&argv, other, two);
2949                         argv_array_push(&argv, other);
2950                         argv_array_push(&argv, xfrm_msg);
2951                 }
2952         }
2953
2954         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
2955         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2956
2957         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
2958                 die(_("external diff died, stopping at %s"), name);
2959
2960         remove_tempfile();
2961         argv_array_clear(&argv);
2962         argv_array_clear(&env);
2963 }
2964
2965 static int similarity_index(struct diff_filepair *p)
2966 {
2967         return p->score * 100 / MAX_SCORE;
2968 }
2969
2970 static void fill_metainfo(struct strbuf *msg,
2971                           const char *name,
2972                           const char *other,
2973                           struct diff_filespec *one,
2974                           struct diff_filespec *two,
2975                           struct diff_options *o,
2976                           struct diff_filepair *p,
2977                           int *must_show_header,
2978                           int use_color)
2979 {
2980         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2981         const char *reset = diff_get_color(use_color, DIFF_RESET);
2982         const char *line_prefix = diff_line_prefix(o);
2983
2984         *must_show_header = 1;
2985         strbuf_init(msg, PATH_MAX * 2 + 300);
2986         switch (p->status) {
2987         case DIFF_STATUS_COPIED:
2988                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2989                             line_prefix, set, similarity_index(p));
2990                 strbuf_addf(msg, "%s\n%s%scopy from ",
2991                             reset,  line_prefix, set);
2992                 quote_c_style(name, msg, NULL, 0);
2993                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2994                 quote_c_style(other, msg, NULL, 0);
2995                 strbuf_addf(msg, "%s\n", reset);
2996                 break;
2997         case DIFF_STATUS_RENAMED:
2998                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2999                             line_prefix, set, similarity_index(p));
3000                 strbuf_addf(msg, "%s\n%s%srename from ",
3001                             reset, line_prefix, set);
3002                 quote_c_style(name, msg, NULL, 0);
3003                 strbuf_addf(msg, "%s\n%s%srename to ",
3004                             reset, line_prefix, set);
3005                 quote_c_style(other, msg, NULL, 0);
3006                 strbuf_addf(msg, "%s\n", reset);
3007                 break;
3008         case DIFF_STATUS_MODIFIED:
3009                 if (p->score) {
3010                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3011                                     line_prefix,
3012                                     set, similarity_index(p), reset);
3013                         break;
3014                 }
3015                 /* fallthru */
3016         default:
3017                 *must_show_header = 0;
3018         }
3019         if (one && two && hashcmp(one->sha1, two->sha1)) {
3020                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3021
3022                 if (DIFF_OPT_TST(o, BINARY)) {
3023                         mmfile_t mf;
3024                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3025                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3026                                 abbrev = 40;
3027                 }
3028                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3029                             find_unique_abbrev(one->sha1, abbrev));
3030                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3031                 if (one->mode == two->mode)
3032                         strbuf_addf(msg, " %06o", one->mode);
3033                 strbuf_addf(msg, "%s\n", reset);
3034         }
3035 }
3036
3037 static void run_diff_cmd(const char *pgm,
3038                          const char *name,
3039                          const char *other,
3040                          const char *attr_path,
3041                          struct diff_filespec *one,
3042                          struct diff_filespec *two,
3043                          struct strbuf *msg,
3044                          struct diff_options *o,
3045                          struct diff_filepair *p)
3046 {
3047         const char *xfrm_msg = NULL;
3048         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3049         int must_show_header = 0;
3050
3051
3052         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3053                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3054                 if (drv && drv->external)
3055                         pgm = drv->external;
3056         }
3057
3058         if (msg) {
3059                 /*
3060                  * don't use colors when the header is intended for an
3061                  * external diff driver
3062                  */
3063                 fill_metainfo(msg, name, other, one, two, o, p,
3064                               &must_show_header,
3065                               want_color(o->use_color) && !pgm);
3066                 xfrm_msg = msg->len ? msg->buf : NULL;
3067         }
3068
3069         if (pgm) {
3070                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3071                                   complete_rewrite, o);
3072                 return;
3073         }
3074         if (one && two)
3075                 builtin_diff(name, other ? other : name,
3076                              one, two, xfrm_msg, must_show_header,
3077                              o, complete_rewrite);
3078         else
3079                 fprintf(o->file, "* Unmerged path %s\n", name);
3080 }
3081
3082 static void diff_fill_sha1_info(struct diff_filespec *one)
3083 {
3084         if (DIFF_FILE_VALID(one)) {
3085                 if (!one->sha1_valid) {
3086                         struct stat st;
3087                         if (one->is_stdin) {
3088                                 hashcpy(one->sha1, null_sha1);
3089                                 return;
3090                         }
3091                         if (lstat(one->path, &st) < 0)
3092                                 die_errno("stat '%s'", one->path);
3093                         if (index_path(one->sha1, one->path, &st, 0))
3094                                 die("cannot hash %s", one->path);
3095                 }
3096         }
3097         else
3098                 hashclr(one->sha1);
3099 }
3100
3101 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3102 {
3103         /* Strip the prefix but do not molest /dev/null and absolute paths */
3104         if (*namep && **namep != '/') {
3105                 *namep += prefix_length;
3106                 if (**namep == '/')
3107                         ++*namep;
3108         }
3109         if (*otherp && **otherp != '/') {
3110                 *otherp += prefix_length;
3111                 if (**otherp == '/')
3112                         ++*otherp;
3113         }
3114 }
3115
3116 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3117 {
3118         const char *pgm = external_diff();
3119         struct strbuf msg;
3120         struct diff_filespec *one = p->one;
3121         struct diff_filespec *two = p->two;
3122         const char *name;
3123         const char *other;
3124         const char *attr_path;
3125
3126         name  = p->one->path;
3127         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3128         attr_path = name;
3129         if (o->prefix_length)
3130                 strip_prefix(o->prefix_length, &name, &other);
3131
3132         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3133                 pgm = NULL;
3134
3135         if (DIFF_PAIR_UNMERGED(p)) {
3136                 run_diff_cmd(pgm, name, NULL, attr_path,
3137                              NULL, NULL, NULL, o, p);
3138                 return;
3139         }
3140
3141         diff_fill_sha1_info(one);
3142         diff_fill_sha1_info(two);
3143
3144         if (!pgm &&
3145             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3146             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3147                 /*
3148                  * a filepair that changes between file and symlink
3149                  * needs to be split into deletion and creation.
3150                  */
3151                 struct diff_filespec *null = alloc_filespec(two->path);
3152                 run_diff_cmd(NULL, name, other, attr_path,
3153                              one, null, &msg, o, p);
3154                 free(null);
3155                 strbuf_release(&msg);
3156
3157                 null = alloc_filespec(one->path);
3158                 run_diff_cmd(NULL, name, other, attr_path,
3159                              null, two, &msg, o, p);
3160                 free(null);
3161         }
3162         else
3163                 run_diff_cmd(pgm, name, other, attr_path,
3164                              one, two, &msg, o, p);
3165
3166         strbuf_release(&msg);
3167 }
3168
3169 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3170                          struct diffstat_t *diffstat)
3171 {
3172         const char *name;
3173         const char *other;
3174
3175         if (DIFF_PAIR_UNMERGED(p)) {
3176                 /* unmerged */
3177                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3178                 return;
3179         }
3180
3181         name = p->one->path;
3182         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3183
3184         if (o->prefix_length)
3185                 strip_prefix(o->prefix_length, &name, &other);
3186
3187         diff_fill_sha1_info(p->one);
3188         diff_fill_sha1_info(p->two);
3189
3190         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3191 }
3192
3193 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3194 {
3195         const char *name;
3196         const char *other;
3197         const char *attr_path;
3198
3199         if (DIFF_PAIR_UNMERGED(p)) {
3200                 /* unmerged */
3201                 return;
3202         }
3203
3204         name = p->one->path;
3205         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3206         attr_path = other ? other : name;
3207
3208         if (o->prefix_length)
3209                 strip_prefix(o->prefix_length, &name, &other);
3210
3211         diff_fill_sha1_info(p->one);
3212         diff_fill_sha1_info(p->two);
3213
3214         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3215 }
3216
3217 void diff_setup(struct diff_options *options)
3218 {
3219         memcpy(options, &default_diff_options, sizeof(*options));
3220
3221         options->file = stdout;
3222
3223         options->line_termination = '\n';
3224         options->break_opt = -1;
3225         options->rename_limit = -1;
3226         options->dirstat_permille = diff_dirstat_permille_default;
3227         options->context = diff_context_default;
3228         DIFF_OPT_SET(options, RENAME_EMPTY);
3229
3230         /* pathchange left =NULL by default */
3231         options->change = diff_change;
3232         options->add_remove = diff_addremove;
3233         options->use_color = diff_use_color_default;
3234         options->detect_rename = diff_detect_rename_default;
3235         options->xdl_opts |= diff_algorithm;
3236
3237         options->orderfile = diff_order_file_cfg;
3238
3239         if (diff_no_prefix) {
3240                 options->a_prefix = options->b_prefix = "";
3241         } else if (!diff_mnemonic_prefix) {
3242                 options->a_prefix = "a/";
3243                 options->b_prefix = "b/";
3244         }
3245 }
3246
3247 void diff_setup_done(struct diff_options *options)
3248 {
3249         int count = 0;
3250
3251         if (options->set_default)
3252                 options->set_default(options);
3253
3254         if (options->output_format & DIFF_FORMAT_NAME)
3255                 count++;
3256         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3257                 count++;
3258         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3259                 count++;
3260         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3261                 count++;
3262         if (count > 1)
3263                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3264
3265         /*
3266          * Most of the time we can say "there are changes"
3267          * only by checking if there are changed paths, but
3268          * --ignore-whitespace* options force us to look
3269          * inside contents.
3270          */
3271
3272         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3273             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3274             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3275                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3276         else
3277                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3278
3279         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3280                 options->detect_rename = DIFF_DETECT_COPY;
3281
3282         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3283                 options->prefix = NULL;
3284         if (options->prefix)
3285                 options->prefix_length = strlen(options->prefix);
3286         else
3287                 options->prefix_length = 0;
3288
3289         if (options->output_format & (DIFF_FORMAT_NAME |
3290                                       DIFF_FORMAT_NAME_STATUS |
3291                                       DIFF_FORMAT_CHECKDIFF |
3292                                       DIFF_FORMAT_NO_OUTPUT))
3293                 options->output_format &= ~(DIFF_FORMAT_RAW |
3294                                             DIFF_FORMAT_NUMSTAT |
3295                                             DIFF_FORMAT_DIFFSTAT |
3296                                             DIFF_FORMAT_SHORTSTAT |
3297                                             DIFF_FORMAT_DIRSTAT |
3298                                             DIFF_FORMAT_SUMMARY |
3299                                             DIFF_FORMAT_PATCH);
3300
3301         /*
3302          * These cases always need recursive; we do not drop caller-supplied
3303          * recursive bits for other formats here.
3304          */
3305         if (options->output_format & (DIFF_FORMAT_PATCH |
3306                                       DIFF_FORMAT_NUMSTAT |
3307                                       DIFF_FORMAT_DIFFSTAT |
3308                                       DIFF_FORMAT_SHORTSTAT |
3309                                       DIFF_FORMAT_DIRSTAT |
3310                                       DIFF_FORMAT_SUMMARY |
3311                                       DIFF_FORMAT_CHECKDIFF))
3312                 DIFF_OPT_SET(options, RECURSIVE);
3313         /*
3314          * Also pickaxe would not work very well if you do not say recursive
3315          */
3316         if (options->pickaxe)
3317                 DIFF_OPT_SET(options, RECURSIVE);
3318         /*
3319          * When patches are generated, submodules diffed against the work tree
3320          * must be checked for dirtiness too so it can be shown in the output
3321          */
3322         if (options->output_format & DIFF_FORMAT_PATCH)
3323                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3324
3325         if (options->detect_rename && options->rename_limit < 0)
3326                 options->rename_limit = diff_rename_limit_default;
3327         if (options->setup & DIFF_SETUP_USE_CACHE) {
3328                 if (!active_cache)
3329                         /* read-cache does not die even when it fails
3330                          * so it is safe for us to do this here.  Also
3331                          * it does not smudge active_cache or active_nr
3332                          * when it fails, so we do not have to worry about
3333                          * cleaning it up ourselves either.
3334                          */
3335                         read_cache();
3336         }
3337         if (options->abbrev <= 0 || 40 < options->abbrev)
3338                 options->abbrev = 40; /* full */
3339
3340         /*
3341          * It does not make sense to show the first hit we happened
3342          * to have found.  It does not make sense not to return with
3343          * exit code in such a case either.
3344          */
3345         if (DIFF_OPT_TST(options, QUICK)) {
3346                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3347                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3348         }
3349
3350         options->diff_path_counter = 0;
3351
3352         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3353                 die(_("--follow requires exactly one pathspec"));
3354 }
3355
3356 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3357 {
3358         char c, *eq;
3359         int len;
3360
3361         if (*arg != '-')
3362                 return 0;
3363         c = *++arg;
3364         if (!c)
3365                 return 0;
3366         if (c == arg_short) {
3367                 c = *++arg;
3368                 if (!c)
3369                         return 1;
3370                 if (val && isdigit(c)) {
3371                         char *end;
3372                         int n = strtoul(arg, &end, 10);
3373                         if (*end)
3374                                 return 0;
3375                         *val = n;
3376                         return 1;
3377                 }
3378                 return 0;
3379         }
3380         if (c != '-')
3381                 return 0;
3382         arg++;
3383         eq = strchrnul(arg, '=');
3384         len = eq - arg;
3385         if (!len || strncmp(arg, arg_long, len))
3386                 return 0;
3387         if (*eq) {
3388                 int n;
3389                 char *end;
3390                 if (!isdigit(*++eq))
3391                         return 0;
3392                 n = strtoul(eq, &end, 10);
3393                 if (*end)
3394                         return 0;
3395                 *val = n;
3396         }
3397         return 1;
3398 }
3399
3400 static int diff_scoreopt_parse(const char *opt);
3401
3402 static inline int short_opt(char opt, const char **argv,
3403                             const char **optarg)
3404 {
3405         const char *arg = argv[0];
3406         if (arg[0] != '-' || arg[1] != opt)
3407                 return 0;
3408         if (arg[2] != '\0') {
3409                 *optarg = arg + 2;
3410                 return 1;
3411         }
3412         if (!argv[1])
3413                 die("Option '%c' requires a value", opt);
3414         *optarg = argv[1];
3415         return 2;
3416 }
3417
3418 int parse_long_opt(const char *opt, const char **argv,
3419                    const char **optarg)
3420 {
3421         const char *arg = argv[0];
3422         if (!skip_prefix(arg, "--", &arg))
3423                 return 0;
3424         if (!skip_prefix(arg, opt, &arg))
3425                 return 0;
3426         if (*arg == '=') { /* stuck form: --option=value */
3427                 *optarg = arg + 1;
3428                 return 1;
3429         }
3430         if (*arg != '\0')
3431                 return 0;
3432         /* separate form: --option value */
3433         if (!argv[1])
3434                 die("Option '--%s' requires a value", opt);
3435         *optarg = argv[1];
3436         return 2;
3437 }
3438
3439 static int stat_opt(struct diff_options *options, const char **av)
3440 {
3441         const char *arg = av[0];
3442         char *end;
3443         int width = options->stat_width;
3444         int name_width = options->stat_name_width;
3445         int graph_width = options->stat_graph_width;
3446         int count = options->stat_count;
3447         int argcount = 1;
3448
3449         if (!skip_prefix(arg, "--stat", &arg))
3450                 die("BUG: stat option does not begin with --stat: %s", arg);
3451         end = (char *)arg;
3452
3453         switch (*arg) {
3454         case '-':
3455                 if (skip_prefix(arg, "-width", &arg)) {
3456                         if (*arg == '=')
3457                                 width = strtoul(arg + 1, &end, 10);
3458                         else if (!*arg && !av[1])
3459                                 die("Option '--stat-width' requires a value");
3460                         else if (!*arg) {
3461                                 width = strtoul(av[1], &end, 10);
3462                                 argcount = 2;
3463                         }
3464                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3465                         if (*arg == '=')
3466                                 name_width = strtoul(arg + 1, &end, 10);
3467                         else if (!*arg && !av[1])
3468                                 die("Option '--stat-name-width' requires a value");
3469                         else if (!*arg) {
3470                                 name_width = strtoul(av[1], &end, 10);
3471                                 argcount = 2;
3472                         }
3473                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3474                         if (*arg == '=')
3475                                 graph_width = strtoul(arg + 1, &end, 10);
3476                         else if (!*arg && !av[1])
3477                                 die("Option '--stat-graph-width' requires a value");
3478                         else if (!*arg) {
3479                                 graph_width = strtoul(av[1], &end, 10);
3480                                 argcount = 2;
3481                         }
3482                 } else if (skip_prefix(arg, "-count", &arg)) {
3483                         if (*arg == '=')
3484                                 count = strtoul(arg + 1, &end, 10);
3485                         else if (!*arg && !av[1])
3486                                 die("Option '--stat-count' requires a value");
3487                         else if (!*arg) {
3488                                 count = strtoul(av[1], &end, 10);
3489                                 argcount = 2;
3490                         }
3491                 }
3492                 break;
3493         case '=':
3494                 width = strtoul(arg+1, &end, 10);
3495                 if (*end == ',')
3496                         name_width = strtoul(end+1, &end, 10);
3497                 if (*end == ',')
3498                         count = strtoul(end+1, &end, 10);
3499         }
3500
3501         /* Important! This checks all the error cases! */
3502         if (*end)
3503                 return 0;
3504         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3505         options->stat_name_width = name_width;
3506         options->stat_graph_width = graph_width;
3507         options->stat_width = width;
3508         options->stat_count = count;
3509         return argcount;
3510 }
3511
3512 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3513 {
3514         struct strbuf errmsg = STRBUF_INIT;
3515         if (parse_dirstat_params(options, params, &errmsg))
3516                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3517                     errmsg.buf);
3518         strbuf_release(&errmsg);
3519         /*
3520          * The caller knows a dirstat-related option is given from the command
3521          * line; allow it to say "return this_function();"
3522          */
3523         options->output_format |= DIFF_FORMAT_DIRSTAT;
3524         return 1;
3525 }
3526
3527 static int parse_submodule_opt(struct diff_options *options, const char *value)
3528 {
3529         if (parse_submodule_params(options, value))
3530                 die(_("Failed to parse --submodule option parameter: '%s'"),
3531                         value);
3532         return 1;
3533 }
3534
3535 static const char diff_status_letters[] = {
3536         DIFF_STATUS_ADDED,
3537         DIFF_STATUS_COPIED,
3538         DIFF_STATUS_DELETED,
3539         DIFF_STATUS_MODIFIED,
3540         DIFF_STATUS_RENAMED,
3541         DIFF_STATUS_TYPE_CHANGED,
3542         DIFF_STATUS_UNKNOWN,
3543         DIFF_STATUS_UNMERGED,
3544         DIFF_STATUS_FILTER_AON,
3545         DIFF_STATUS_FILTER_BROKEN,
3546         '\0',
3547 };
3548
3549 static unsigned int filter_bit['Z' + 1];
3550
3551 static void prepare_filter_bits(void)
3552 {
3553         int i;
3554
3555         if (!filter_bit[DIFF_STATUS_ADDED]) {
3556                 for (i = 0; diff_status_letters[i]; i++)
3557                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3558         }
3559 }
3560
3561 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3562 {
3563         return opt->filter & filter_bit[(int) status];
3564 }
3565
3566 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3567 {
3568         int i, optch;
3569
3570         prepare_filter_bits();
3571
3572         /*
3573          * If there is a negation e.g. 'd' in the input, and we haven't
3574          * initialized the filter field with another --diff-filter, start
3575          * from full set of bits, except for AON.
3576          */
3577         if (!opt->filter) {
3578                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3579                         if (optch < 'a' || 'z' < optch)
3580                                 continue;
3581                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3582                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3583                         break;
3584                 }
3585         }
3586
3587         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3588                 unsigned int bit;
3589                 int negate;
3590
3591                 if ('a' <= optch && optch <= 'z') {
3592                         negate = 1;
3593                         optch = toupper(optch);
3594                 } else {
3595                         negate = 0;
3596                 }
3597
3598                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3599                 if (!bit)
3600                         return optarg[i];
3601                 if (negate)
3602                         opt->filter &= ~bit;
3603                 else
3604                         opt->filter |= bit;
3605         }
3606         return 0;
3607 }
3608
3609 static void enable_patch_output(int *fmt) {
3610         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3611         *fmt |= DIFF_FORMAT_PATCH;
3612 }
3613
3614 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3615 {
3616         const char *arg = av[0];
3617         const char *optarg;
3618         int argcount;
3619
3620         /* Output format options */
3621         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3622             || opt_arg(arg, 'U', "unified", &options->context))
3623                 enable_patch_output(&options->output_format);
3624         else if (!strcmp(arg, "--raw"))
3625                 options->output_format |= DIFF_FORMAT_RAW;
3626         else if (!strcmp(arg, "--patch-with-raw")) {
3627                 enable_patch_output(&options->output_format);
3628                 options->output_format |= DIFF_FORMAT_RAW;
3629         } else if (!strcmp(arg, "--numstat"))
3630                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3631         else if (!strcmp(arg, "--shortstat"))
3632                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3633         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3634                 return parse_dirstat_opt(options, "");
3635         else if (skip_prefix(arg, "-X", &arg))
3636                 return parse_dirstat_opt(options, arg);
3637         else if (skip_prefix(arg, "--dirstat=", &arg))
3638                 return parse_dirstat_opt(options, arg);
3639         else if (!strcmp(arg, "--cumulative"))
3640                 return parse_dirstat_opt(options, "cumulative");
3641         else if (!strcmp(arg, "--dirstat-by-file"))
3642                 return parse_dirstat_opt(options, "files");
3643         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3644                 parse_dirstat_opt(options, "files");
3645                 return parse_dirstat_opt(options, arg);
3646         }
3647         else if (!strcmp(arg, "--check"))
3648                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3649         else if (!strcmp(arg, "--summary"))
3650                 options->output_format |= DIFF_FORMAT_SUMMARY;
3651         else if (!strcmp(arg, "--patch-with-stat")) {
3652                 enable_patch_output(&options->output_format);
3653                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3654         } else if (!strcmp(arg, "--name-only"))
3655                 options->output_format |= DIFF_FORMAT_NAME;
3656         else if (!strcmp(arg, "--name-status"))
3657                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3658         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3659                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3660         else if (starts_with(arg, "--stat"))
3661                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3662                 return stat_opt(options, av);
3663
3664         /* renames options */
3665         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3666                  !strcmp(arg, "--break-rewrites")) {
3667                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3668                         return error("invalid argument to -B: %s", arg+2);
3669         }
3670         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3671                  !strcmp(arg, "--find-renames")) {
3672                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3673                         return error("invalid argument to -M: %s", arg+2);
3674                 options->detect_rename = DIFF_DETECT_RENAME;
3675         }
3676         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3677                 options->irreversible_delete = 1;
3678         }
3679         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3680                  !strcmp(arg, "--find-copies")) {
3681                 if (options->detect_rename == DIFF_DETECT_COPY)
3682                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3683                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3684                         return error("invalid argument to -C: %s", arg+2);
3685                 options->detect_rename = DIFF_DETECT_COPY;
3686         }
3687         else if (!strcmp(arg, "--no-renames"))
3688                 options->detect_rename = 0;
3689         else if (!strcmp(arg, "--rename-empty"))
3690                 DIFF_OPT_SET(options, RENAME_EMPTY);
3691         else if (!strcmp(arg, "--no-rename-empty"))
3692                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3693         else if (!strcmp(arg, "--relative"))
3694                 DIFF_OPT_SET(options, RELATIVE_NAME);
3695         else if (skip_prefix(arg, "--relative=", &arg)) {
3696                 DIFF_OPT_SET(options, RELATIVE_NAME);
3697                 options->prefix = arg;
3698         }
3699
3700         /* xdiff options */
3701         else if (!strcmp(arg, "--minimal"))
3702                 DIFF_XDL_SET(options, NEED_MINIMAL);
3703         else if (!strcmp(arg, "--no-minimal"))
3704                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3705         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3706                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3707         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3708                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3709         else if (!strcmp(arg, "--ignore-space-at-eol"))
3710                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3711         else if (!strcmp(arg, "--ignore-blank-lines"))
3712                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3713         else if (!strcmp(arg, "--patience"))
3714                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3715         else if (!strcmp(arg, "--histogram"))
3716                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3717         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3718                 long value = parse_algorithm_value(optarg);
3719                 if (value < 0)
3720                         return error("option diff-algorithm accepts \"myers\", "
3721                                      "\"minimal\", \"patience\" and \"histogram\"");
3722                 /* clear out previous settings */
3723                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3724                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3725                 options->xdl_opts |= value;
3726                 return argcount;
3727         }
3728
3729         /* flags options */
3730         else if (!strcmp(arg, "--binary")) {
3731                 enable_patch_output(&options->output_format);
3732                 DIFF_OPT_SET(options, BINARY);
3733         }
3734         else if (!strcmp(arg, "--full-index"))
3735                 DIFF_OPT_SET(options, FULL_INDEX);
3736         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3737                 DIFF_OPT_SET(options, TEXT);
3738         else if (!strcmp(arg, "-R"))
3739                 DIFF_OPT_SET(options, REVERSE_DIFF);
3740         else if (!strcmp(arg, "--find-copies-harder"))
3741                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3742         else if (!strcmp(arg, "--follow"))
3743                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3744         else if (!strcmp(arg, "--no-follow"))
3745                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3746         else if (!strcmp(arg, "--color"))
3747                 options->use_color = 1;
3748         else if (skip_prefix(arg, "--color=", &arg)) {
3749                 int value = git_config_colorbool(NULL, arg);
3750                 if (value < 0)
3751                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3752                 options->use_color = value;
3753         }
3754         else if (!strcmp(arg, "--no-color"))
3755                 options->use_color = 0;
3756         else if (!strcmp(arg, "--color-words")) {
3757                 options->use_color = 1;
3758                 options->word_diff = DIFF_WORDS_COLOR;
3759         }
3760         else if (skip_prefix(arg, "--color-words=", &arg)) {
3761                 options->use_color = 1;
3762                 options->word_diff = DIFF_WORDS_COLOR;
3763                 options->word_regex = arg;
3764         }
3765         else if (!strcmp(arg, "--word-diff")) {
3766                 if (options->word_diff == DIFF_WORDS_NONE)
3767                         options->word_diff = DIFF_WORDS_PLAIN;
3768         }
3769         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3770                 if (!strcmp(arg, "plain"))
3771                         options->word_diff = DIFF_WORDS_PLAIN;
3772                 else if (!strcmp(arg, "color")) {
3773                         options->use_color = 1;
3774                         options->word_diff = DIFF_WORDS_COLOR;
3775                 }
3776                 else if (!strcmp(arg, "porcelain"))
3777                         options->word_diff = DIFF_WORDS_PORCELAIN;
3778                 else if (!strcmp(arg, "none"))
3779                         options->word_diff = DIFF_WORDS_NONE;
3780                 else
3781                         die("bad --word-diff argument: %s", arg);
3782         }
3783         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3784                 if (options->word_diff == DIFF_WORDS_NONE)
3785                         options->word_diff = DIFF_WORDS_PLAIN;
3786                 options->word_regex = optarg;
3787                 return argcount;
3788         }
3789         else if (!strcmp(arg, "--exit-code"))
3790                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3791         else if (!strcmp(arg, "--quiet"))
3792                 DIFF_OPT_SET(options, QUICK);
3793         else if (!strcmp(arg, "--ext-diff"))
3794                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3795         else if (!strcmp(arg, "--no-ext-diff"))
3796                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3797         else if (!strcmp(arg, "--textconv"))
3798                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3799         else if (!strcmp(arg, "--no-textconv"))
3800                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3801         else if (!strcmp(arg, "--ignore-submodules")) {
3802                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3803                 handle_ignore_submodules_arg(options, "all");
3804         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3805                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3806                 handle_ignore_submodules_arg(options, arg);
3807         } else if (!strcmp(arg, "--submodule"))
3808                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3809         else if (skip_prefix(arg, "--submodule=", &arg))
3810                 return parse_submodule_opt(options, arg);
3811
3812         /* misc options */
3813         else if (!strcmp(arg, "-z"))
3814                 options->line_termination = 0;
3815         else if ((argcount = short_opt('l', av, &optarg))) {
3816                 options->rename_limit = strtoul(optarg, NULL, 10);
3817                 return argcount;
3818         }
3819         else if ((argcount = short_opt('S', av, &optarg))) {
3820                 options->pickaxe = optarg;
3821                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3822                 return argcount;
3823         } else if ((argcount = short_opt('G', av, &optarg))) {
3824                 options->pickaxe = optarg;
3825                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3826                 return argcount;
3827         }
3828         else if (!strcmp(arg, "--pickaxe-all"))
3829                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3830         else if (!strcmp(arg, "--pickaxe-regex"))
3831                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3832         else if ((argcount = short_opt('O', av, &optarg))) {
3833                 options->orderfile = optarg;
3834                 return argcount;
3835         }
3836         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3837                 int offending = parse_diff_filter_opt(optarg, options);
3838                 if (offending)
3839                         die("unknown change class '%c' in --diff-filter=%s",
3840                             offending, optarg);
3841                 return argcount;
3842         }
3843         else if (!strcmp(arg, "--abbrev"))
3844                 options->abbrev = DEFAULT_ABBREV;
3845         else if (skip_prefix(arg, "--abbrev=", &arg)) {
3846                 options->abbrev = strtoul(arg, NULL, 10);
3847                 if (options->abbrev < MINIMUM_ABBREV)
3848                         options->abbrev = MINIMUM_ABBREV;
3849                 else if (40 < options->abbrev)
3850                         options->abbrev = 40;
3851         }
3852         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3853                 options->a_prefix = optarg;
3854                 return argcount;
3855         }
3856         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3857                 options->b_prefix = optarg;
3858                 return argcount;
3859         }
3860         else if (!strcmp(arg, "--no-prefix"))
3861                 options->a_prefix = options->b_prefix = "";
3862         else if (opt_arg(arg, '\0', "inter-hunk-context",
3863                          &options->interhunkcontext))
3864                 ;
3865         else if (!strcmp(arg, "-W"))
3866                 DIFF_OPT_SET(options, FUNCCONTEXT);
3867         else if (!strcmp(arg, "--function-context"))
3868                 DIFF_OPT_SET(options, FUNCCONTEXT);
3869         else if (!strcmp(arg, "--no-function-context"))
3870                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3871         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3872                 options->file = fopen(optarg, "w");
3873                 if (!options->file)
3874                         die_errno("Could not open '%s'", optarg);
3875                 options->close_file = 1;
3876                 return argcount;
3877         } else
3878                 return 0;
3879         return 1;
3880 }
3881
3882 int parse_rename_score(const char **cp_p)
3883 {
3884         unsigned long num, scale;
3885         int ch, dot;
3886         const char *cp = *cp_p;
3887
3888         num = 0;
3889         scale = 1;
3890         dot = 0;
3891         for (;;) {
3892                 ch = *cp;
3893                 if ( !dot && ch == '.' ) {
3894                         scale = 1;
3895                         dot = 1;
3896                 } else if ( ch == '%' ) {
3897                         scale = dot ? scale*100 : 100;
3898                         cp++;   /* % is always at the end */
3899                         break;
3900                 } else if ( ch >= '0' && ch <= '9' ) {
3901                         if ( scale < 100000 ) {
3902                                 scale *= 10;
3903                                 num = (num*10) + (ch-'0');
3904                         }
3905                 } else {
3906                         break;
3907                 }
3908                 cp++;
3909         }
3910         *cp_p = cp;
3911
3912         /* user says num divided by scale and we say internally that
3913          * is MAX_SCORE * num / scale.
3914          */
3915         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3916 }
3917
3918 static int diff_scoreopt_parse(const char *opt)
3919 {
3920         int opt1, opt2, cmd;
3921
3922         if (*opt++ != '-')
3923                 return -1;
3924         cmd = *opt++;
3925         if (cmd == '-') {
3926                 /* convert the long-form arguments into short-form versions */
3927                 if (skip_prefix(opt, "break-rewrites", &opt)) {
3928                         if (*opt == 0 || *opt++ == '=')
3929                                 cmd = 'B';
3930                 } else if (skip_prefix(opt, "find-copies", &opt)) {
3931                         if (*opt == 0 || *opt++ == '=')
3932                                 cmd = 'C';
3933                 } else if (skip_prefix(opt, "find-renames", &opt)) {
3934                         if (*opt == 0 || *opt++ == '=')
3935                                 cmd = 'M';
3936                 }
3937         }
3938         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3939                 return -1; /* that is not a -M, -C, or -B option */
3940
3941         opt1 = parse_rename_score(&opt);
3942         if (cmd != 'B')
3943                 opt2 = 0;
3944         else {
3945                 if (*opt == 0)
3946                         opt2 = 0;
3947                 else if (*opt != '/')
3948                         return -1; /* we expect -B80/99 or -B80 */
3949                 else {
3950                         opt++;
3951                         opt2 = parse_rename_score(&opt);
3952                 }
3953         }
3954         if (*opt != 0)
3955                 return -1;
3956         return opt1 | (opt2 << 16);
3957 }
3958
3959 struct diff_queue_struct diff_queued_diff;
3960
3961 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3962 {
3963         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
3964         queue->queue[queue->nr++] = dp;
3965 }
3966
3967 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3968                                  struct diff_filespec *one,
3969                                  struct diff_filespec *two)
3970 {
3971         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3972         dp->one = one;
3973         dp->two = two;
3974         if (queue)
3975                 diff_q(queue, dp);
3976         return dp;
3977 }
3978
3979 void diff_free_filepair(struct diff_filepair *p)
3980 {
3981         free_filespec(p->one);
3982         free_filespec(p->two);
3983         free(p);
3984 }
3985
3986 /* This is different from find_unique_abbrev() in that
3987  * it stuffs the result with dots for alignment.
3988  */
3989 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3990 {
3991         int abblen;
3992         const char *abbrev;
3993         if (len == 40)
3994                 return sha1_to_hex(sha1);
3995
3996         abbrev = find_unique_abbrev(sha1, len);
3997         abblen = strlen(abbrev);
3998         if (abblen < 37) {
3999                 static char hex[41];
4000                 if (len < abblen && abblen <= len + 2)
4001                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
4002                 else
4003                         sprintf(hex, "%s...", abbrev);
4004                 return hex;
4005         }
4006         return sha1_to_hex(sha1);
4007 }
4008
4009 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4010 {
4011         int line_termination = opt->line_termination;
4012         int inter_name_termination = line_termination ? '\t' : '\0';
4013
4014         fprintf(opt->file, "%s", diff_line_prefix(opt));
4015         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4016                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4017                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
4018                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4019         }
4020         if (p->score) {
4021                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4022                         inter_name_termination);
4023         } else {
4024                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4025         }
4026
4027         if (p->status == DIFF_STATUS_COPIED ||
4028             p->status == DIFF_STATUS_RENAMED) {
4029                 const char *name_a, *name_b;
4030                 name_a = p->one->path;
4031                 name_b = p->two->path;
4032                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4033                 write_name_quoted(name_a, opt->file, inter_name_termination);
4034                 write_name_quoted(name_b, opt->file, line_termination);
4035         } else {
4036                 const char *name_a, *name_b;
4037                 name_a = p->one->mode ? p->one->path : p->two->path;
4038                 name_b = NULL;
4039                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4040                 write_name_quoted(name_a, opt->file, line_termination);
4041         }
4042 }
4043
4044 int diff_unmodified_pair(struct diff_filepair *p)
4045 {
4046         /* This function is written stricter than necessary to support
4047          * the currently implemented transformers, but the idea is to
4048          * let transformers to produce diff_filepairs any way they want,
4049          * and filter and clean them up here before producing the output.
4050          */
4051         struct diff_filespec *one = p->one, *two = p->two;
4052
4053         if (DIFF_PAIR_UNMERGED(p))
4054                 return 0; /* unmerged is interesting */
4055
4056         /* deletion, addition, mode or type change
4057          * and rename are all interesting.
4058          */
4059         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4060             DIFF_PAIR_MODE_CHANGED(p) ||
4061             strcmp(one->path, two->path))
4062                 return 0;
4063
4064         /* both are valid and point at the same path.  that is, we are
4065          * dealing with a change.
4066          */
4067         if (one->sha1_valid && two->sha1_valid &&
4068             !hashcmp(one->sha1, two->sha1) &&
4069             !one->dirty_submodule && !two->dirty_submodule)
4070                 return 1; /* no change */
4071         if (!one->sha1_valid && !two->sha1_valid)
4072                 return 1; /* both look at the same file on the filesystem. */
4073         return 0;
4074 }
4075
4076 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4077 {
4078         if (diff_unmodified_pair(p))
4079                 return;
4080
4081         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4082             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4083                 return; /* no tree diffs in patch format */
4084
4085         run_diff(p, o);
4086 }
4087
4088 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4089                             struct diffstat_t *diffstat)
4090 {
4091         if (diff_unmodified_pair(p))
4092                 return;
4093
4094         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4095             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4096                 return; /* no useful stat for tree diffs */
4097
4098         run_diffstat(p, o, diffstat);
4099 }
4100
4101 static void diff_flush_checkdiff(struct diff_filepair *p,
4102                 struct diff_options *o)
4103 {
4104         if (diff_unmodified_pair(p))
4105                 return;
4106
4107         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4108             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4109                 return; /* nothing to check in tree diffs */
4110
4111         run_checkdiff(p, o);
4112 }
4113
4114 int diff_queue_is_empty(void)
4115 {
4116         struct diff_queue_struct *q = &diff_queued_diff;
4117         int i;
4118         for (i = 0; i < q->nr; i++)
4119                 if (!diff_unmodified_pair(q->queue[i]))
4120                         return 0;
4121         return 1;
4122 }
4123
4124 #if DIFF_DEBUG
4125 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4126 {
4127         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4128                 x, one ? one : "",
4129                 s->path,
4130                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4131                 s->mode,
4132                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4133         fprintf(stderr, "queue[%d] %s size %lu\n",
4134                 x, one ? one : "",
4135                 s->size);
4136 }
4137
4138 void diff_debug_filepair(const struct diff_filepair *p, int i)
4139 {
4140         diff_debug_filespec(p->one, i, "one");
4141         diff_debug_filespec(p->two, i, "two");
4142         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4143                 p->score, p->status ? p->status : '?',
4144                 p->one->rename_used, p->broken_pair);
4145 }
4146
4147 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4148 {
4149         int i;
4150         if (msg)
4151                 fprintf(stderr, "%s\n", msg);
4152         fprintf(stderr, "q->nr = %d\n", q->nr);
4153         for (i = 0; i < q->nr; i++) {
4154                 struct diff_filepair *p = q->queue[i];
4155                 diff_debug_filepair(p, i);
4156         }
4157 }
4158 #endif
4159
4160 static void diff_resolve_rename_copy(void)
4161 {
4162         int i;
4163         struct diff_filepair *p;
4164         struct diff_queue_struct *q = &diff_queued_diff;
4165
4166         diff_debug_queue("resolve-rename-copy", q);
4167
4168         for (i = 0; i < q->nr; i++) {
4169                 p = q->queue[i];
4170                 p->status = 0; /* undecided */
4171                 if (DIFF_PAIR_UNMERGED(p))
4172                         p->status = DIFF_STATUS_UNMERGED;
4173                 else if (!DIFF_FILE_VALID(p->one))
4174                         p->status = DIFF_STATUS_ADDED;
4175                 else if (!DIFF_FILE_VALID(p->two))
4176                         p->status = DIFF_STATUS_DELETED;
4177                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4178                         p->status = DIFF_STATUS_TYPE_CHANGED;
4179
4180                 /* from this point on, we are dealing with a pair
4181                  * whose both sides are valid and of the same type, i.e.
4182                  * either in-place edit or rename/copy edit.
4183                  */
4184                 else if (DIFF_PAIR_RENAME(p)) {
4185                         /*
4186                          * A rename might have re-connected a broken
4187                          * pair up, causing the pathnames to be the
4188                          * same again. If so, that's not a rename at
4189                          * all, just a modification..
4190                          *
4191                          * Otherwise, see if this source was used for
4192                          * multiple renames, in which case we decrement
4193                          * the count, and call it a copy.
4194                          */
4195                         if (!strcmp(p->one->path, p->two->path))
4196                                 p->status = DIFF_STATUS_MODIFIED;
4197                         else if (--p->one->rename_used > 0)
4198                                 p->status = DIFF_STATUS_COPIED;
4199                         else
4200                                 p->status = DIFF_STATUS_RENAMED;
4201                 }
4202                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4203                          p->one->mode != p->two->mode ||
4204                          p->one->dirty_submodule ||
4205                          p->two->dirty_submodule ||
4206                          is_null_sha1(p->one->sha1))
4207                         p->status = DIFF_STATUS_MODIFIED;
4208                 else {
4209                         /* This is a "no-change" entry and should not
4210                          * happen anymore, but prepare for broken callers.
4211                          */
4212                         error("feeding unmodified %s to diffcore",
4213                               p->one->path);
4214                         p->status = DIFF_STATUS_UNKNOWN;
4215                 }
4216         }
4217         diff_debug_queue("resolve-rename-copy done", q);
4218 }
4219
4220 static int check_pair_status(struct diff_filepair *p)
4221 {
4222         switch (p->status) {
4223         case DIFF_STATUS_UNKNOWN:
4224                 return 0;
4225         case 0:
4226                 die("internal error in diff-resolve-rename-copy");
4227         default:
4228                 return 1;
4229         }
4230 }
4231
4232 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4233 {
4234         int fmt = opt->output_format;
4235
4236         if (fmt & DIFF_FORMAT_CHECKDIFF)
4237                 diff_flush_checkdiff(p, opt);
4238         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4239                 diff_flush_raw(p, opt);
4240         else if (fmt & DIFF_FORMAT_NAME) {
4241                 const char *name_a, *name_b;
4242                 name_a = p->two->path;
4243                 name_b = NULL;
4244                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4245                 write_name_quoted(name_a, opt->file, opt->line_termination);
4246         }
4247 }
4248
4249 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4250 {
4251         if (fs->mode)
4252                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4253         else
4254                 fprintf(file, " %s ", newdelete);
4255         write_name_quoted(fs->path, file, '\n');
4256 }
4257
4258
4259 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4260                 const char *line_prefix)
4261 {
4262         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4263                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4264                         p->two->mode, show_name ? ' ' : '\n');
4265                 if (show_name) {
4266                         write_name_quoted(p->two->path, file, '\n');
4267                 }
4268         }
4269 }
4270
4271 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4272                         const char *line_prefix)
4273 {
4274         char *names = pprint_rename(p->one->path, p->two->path);
4275
4276         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4277         free(names);
4278         show_mode_change(file, p, 0, line_prefix);
4279 }
4280
4281 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4282 {
4283         FILE *file = opt->file;
4284         const char *line_prefix = diff_line_prefix(opt);
4285
4286         switch(p->status) {
4287         case DIFF_STATUS_DELETED:
4288                 fputs(line_prefix, file);
4289                 show_file_mode_name(file, "delete", p->one);
4290                 break;
4291         case DIFF_STATUS_ADDED:
4292                 fputs(line_prefix, file);
4293                 show_file_mode_name(file, "create", p->two);
4294                 break;
4295         case DIFF_STATUS_COPIED:
4296                 fputs(line_prefix, file);
4297                 show_rename_copy(file, "copy", p, line_prefix);
4298                 break;
4299         case DIFF_STATUS_RENAMED:
4300                 fputs(line_prefix, file);
4301                 show_rename_copy(file, "rename", p, line_prefix);
4302                 break;
4303         default:
4304                 if (p->score) {
4305                         fprintf(file, "%s rewrite ", line_prefix);
4306                         write_name_quoted(p->two->path, file, ' ');
4307                         fprintf(file, "(%d%%)\n", similarity_index(p));
4308                 }
4309                 show_mode_change(file, p, !p->score, line_prefix);
4310                 break;
4311         }
4312 }
4313
4314 struct patch_id_t {
4315         git_SHA_CTX *ctx;
4316         int patchlen;
4317 };
4318
4319 static int remove_space(char *line, int len)
4320 {
4321         int i;
4322         char *dst = line;
4323         unsigned char c;
4324
4325         for (i = 0; i < len; i++)
4326                 if (!isspace((c = line[i])))
4327                         *dst++ = c;
4328
4329         return dst - line;
4330 }
4331
4332 static void patch_id_consume(void *priv, char *line, unsigned long len)
4333 {
4334         struct patch_id_t *data = priv;
4335         int new_len;
4336
4337         /* Ignore line numbers when computing the SHA1 of the patch */
4338         if (starts_with(line, "@@ -"))
4339                 return;
4340
4341         new_len = remove_space(line, len);
4342
4343         git_SHA1_Update(data->ctx, line, new_len);
4344         data->patchlen += new_len;
4345 }
4346
4347 /* returns 0 upon success, and writes result into sha1 */
4348 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4349 {
4350         struct diff_queue_struct *q = &diff_queued_diff;
4351         int i;
4352         git_SHA_CTX ctx;
4353         struct patch_id_t data;
4354         char buffer[PATH_MAX * 4 + 20];
4355
4356         git_SHA1_Init(&ctx);
4357         memset(&data, 0, sizeof(struct patch_id_t));
4358         data.ctx = &ctx;
4359
4360         for (i = 0; i < q->nr; i++) {
4361                 xpparam_t xpp;
4362                 xdemitconf_t xecfg;
4363                 mmfile_t mf1, mf2;
4364                 struct diff_filepair *p = q->queue[i];
4365                 int len1, len2;
4366
4367                 memset(&xpp, 0, sizeof(xpp));
4368                 memset(&xecfg, 0, sizeof(xecfg));
4369                 if (p->status == 0)
4370                         return error("internal diff status error");
4371                 if (p->status == DIFF_STATUS_UNKNOWN)
4372                         continue;
4373                 if (diff_unmodified_pair(p))
4374                         continue;
4375                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4376                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4377                         continue;
4378                 if (DIFF_PAIR_UNMERGED(p))
4379                         continue;
4380
4381                 diff_fill_sha1_info(p->one);
4382                 diff_fill_sha1_info(p->two);
4383                 if (fill_mmfile(&mf1, p->one) < 0 ||
4384                                 fill_mmfile(&mf2, p->two) < 0)
4385                         return error("unable to read files to diff");
4386
4387                 len1 = remove_space(p->one->path, strlen(p->one->path));
4388                 len2 = remove_space(p->two->path, strlen(p->two->path));
4389                 if (p->one->mode == 0)
4390                         len1 = snprintf(buffer, sizeof(buffer),
4391                                         "diff--gita/%.*sb/%.*s"
4392                                         "newfilemode%06o"
4393                                         "---/dev/null"
4394                                         "+++b/%.*s",
4395                                         len1, p->one->path,
4396                                         len2, p->two->path,
4397                                         p->two->mode,
4398                                         len2, p->two->path);
4399                 else if (p->two->mode == 0)
4400                         len1 = snprintf(buffer, sizeof(buffer),
4401                                         "diff--gita/%.*sb/%.*s"
4402                                         "deletedfilemode%06o"
4403                                         "---a/%.*s"
4404                                         "+++/dev/null",
4405                                         len1, p->one->path,
4406                                         len2, p->two->path,
4407                                         p->one->mode,
4408                                         len1, p->one->path);
4409                 else
4410                         len1 = snprintf(buffer, sizeof(buffer),
4411                                         "diff--gita/%.*sb/%.*s"
4412                                         "---a/%.*s"
4413                                         "+++b/%.*s",
4414                                         len1, p->one->path,
4415                                         len2, p->two->path,
4416                                         len1, p->one->path,
4417                                         len2, p->two->path);
4418                 git_SHA1_Update(&ctx, buffer, len1);
4419
4420                 if (diff_filespec_is_binary(p->one) ||
4421                     diff_filespec_is_binary(p->two)) {
4422                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4423                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4424                         continue;
4425                 }
4426
4427                 xpp.flags = 0;
4428                 xecfg.ctxlen = 3;
4429                 xecfg.flags = 0;
4430                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4431                               &xpp, &xecfg);
4432         }
4433
4434         git_SHA1_Final(sha1, &ctx);
4435         return 0;
4436 }
4437
4438 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4439 {
4440         struct diff_queue_struct *q = &diff_queued_diff;
4441         int i;
4442         int result = diff_get_patch_id(options, sha1);
4443
4444         for (i = 0; i < q->nr; i++)
4445                 diff_free_filepair(q->queue[i]);
4446
4447         free(q->queue);
4448         DIFF_QUEUE_CLEAR(q);
4449
4450         return result;
4451 }
4452
4453 static int is_summary_empty(const struct diff_queue_struct *q)
4454 {
4455         int i;
4456
4457         for (i = 0; i < q->nr; i++) {
4458                 const struct diff_filepair *p = q->queue[i];
4459
4460                 switch (p->status) {
4461                 case DIFF_STATUS_DELETED:
4462                 case DIFF_STATUS_ADDED:
4463                 case DIFF_STATUS_COPIED:
4464                 case DIFF_STATUS_RENAMED:
4465                         return 0;
4466                 default:
4467                         if (p->score)
4468                                 return 0;
4469                         if (p->one->mode && p->two->mode &&
4470                             p->one->mode != p->two->mode)
4471                                 return 0;
4472                         break;
4473                 }
4474         }
4475         return 1;
4476 }
4477
4478 static const char rename_limit_warning[] =
4479 "inexact rename detection was skipped due to too many files.";
4480
4481 static const char degrade_cc_to_c_warning[] =
4482 "only found copies from modified paths due to too many files.";
4483
4484 static const char rename_limit_advice[] =
4485 "you may want to set your %s variable to at least "
4486 "%d and retry the command.";
4487
4488 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4489 {
4490         if (degraded_cc)
4491                 warning(degrade_cc_to_c_warning);
4492         else if (needed)
4493                 warning(rename_limit_warning);
4494         else
4495                 return;
4496         if (0 < needed && needed < 32767)
4497                 warning(rename_limit_advice, varname, needed);
4498 }
4499
4500 void diff_flush(struct diff_options *options)
4501 {
4502         struct diff_queue_struct *q = &diff_queued_diff;
4503         int i, output_format = options->output_format;
4504         int separator = 0;
4505         int dirstat_by_line = 0;
4506
4507         /*
4508          * Order: raw, stat, summary, patch
4509          * or:    name/name-status/checkdiff (other bits clear)
4510          */
4511         if (!q->nr)
4512                 goto free_queue;
4513
4514         if (output_format & (DIFF_FORMAT_RAW |
4515                              DIFF_FORMAT_NAME |
4516                              DIFF_FORMAT_NAME_STATUS |
4517                              DIFF_FORMAT_CHECKDIFF)) {
4518                 for (i = 0; i < q->nr; i++) {
4519                         struct diff_filepair *p = q->queue[i];
4520                         if (check_pair_status(p))
4521                                 flush_one_pair(p, options);
4522                 }
4523                 separator++;
4524         }
4525
4526         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4527                 dirstat_by_line = 1;
4528
4529         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4530             dirstat_by_line) {
4531                 struct diffstat_t diffstat;
4532
4533                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4534                 for (i = 0; i < q->nr; i++) {
4535                         struct diff_filepair *p = q->queue[i];
4536                         if (check_pair_status(p))
4537                                 diff_flush_stat(p, options, &diffstat);
4538                 }
4539                 if (output_format & DIFF_FORMAT_NUMSTAT)
4540                         show_numstat(&diffstat, options);
4541                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4542                         show_stats(&diffstat, options);
4543                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4544                         show_shortstats(&diffstat, options);
4545                 if (output_format & DIFF_FORMAT_DIRSTAT)
4546                         show_dirstat_by_line(&diffstat, options);
4547                 free_diffstat_info(&diffstat);
4548                 separator++;
4549         }
4550         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4551                 show_dirstat(options);
4552
4553         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4554                 for (i = 0; i < q->nr; i++) {
4555                         diff_summary(options, q->queue[i]);
4556                 }
4557                 separator++;
4558         }
4559
4560         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4561             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4562             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4563                 /*
4564                  * run diff_flush_patch for the exit status. setting
4565                  * options->file to /dev/null should be safe, because we
4566                  * aren't supposed to produce any output anyway.
4567                  */
4568                 if (options->close_file)
4569                         fclose(options->file);
4570                 options->file = fopen("/dev/null", "w");
4571                 if (!options->file)
4572                         die_errno("Could not open /dev/null");
4573                 options->close_file = 1;
4574                 for (i = 0; i < q->nr; i++) {
4575                         struct diff_filepair *p = q->queue[i];
4576                         if (check_pair_status(p))
4577                                 diff_flush_patch(p, options);
4578                         if (options->found_changes)
4579                                 break;
4580                 }
4581         }
4582
4583         if (output_format & DIFF_FORMAT_PATCH) {
4584                 if (separator) {
4585                         fprintf(options->file, "%s%c",
4586                                 diff_line_prefix(options),
4587                                 options->line_termination);
4588                         if (options->stat_sep) {
4589                                 /* attach patch instead of inline */
4590                                 fputs(options->stat_sep, options->file);
4591                         }
4592                 }
4593
4594                 for (i = 0; i < q->nr; i++) {
4595                         struct diff_filepair *p = q->queue[i];
4596                         if (check_pair_status(p))
4597                                 diff_flush_patch(p, options);
4598                 }
4599         }
4600
4601         if (output_format & DIFF_FORMAT_CALLBACK)
4602                 options->format_callback(q, options, options->format_callback_data);
4603
4604         for (i = 0; i < q->nr; i++)
4605                 diff_free_filepair(q->queue[i]);
4606 free_queue:
4607         free(q->queue);
4608         DIFF_QUEUE_CLEAR(q);
4609         if (options->close_file)
4610                 fclose(options->file);
4611
4612         /*
4613          * Report the content-level differences with HAS_CHANGES;
4614          * diff_addremove/diff_change does not set the bit when
4615          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4616          */
4617         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4618                 if (options->found_changes)
4619                         DIFF_OPT_SET(options, HAS_CHANGES);
4620                 else
4621                         DIFF_OPT_CLR(options, HAS_CHANGES);
4622         }
4623 }
4624
4625 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4626 {
4627         return (((p->status == DIFF_STATUS_MODIFIED) &&
4628                  ((p->score &&
4629                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4630                   (!p->score &&
4631                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4632                 ((p->status != DIFF_STATUS_MODIFIED) &&
4633                  filter_bit_tst(p->status, options)));
4634 }
4635
4636 static void diffcore_apply_filter(struct diff_options *options)
4637 {
4638         int i;
4639         struct diff_queue_struct *q = &diff_queued_diff;
4640         struct diff_queue_struct outq;
4641
4642         DIFF_QUEUE_CLEAR(&outq);
4643
4644         if (!options->filter)
4645                 return;
4646
4647         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4648                 int found;
4649                 for (i = found = 0; !found && i < q->nr; i++) {
4650                         if (match_filter(options, q->queue[i]))
4651                                 found++;
4652                 }
4653                 if (found)
4654                         return;
4655
4656                 /* otherwise we will clear the whole queue
4657                  * by copying the empty outq at the end of this
4658                  * function, but first clear the current entries
4659                  * in the queue.
4660                  */
4661                 for (i = 0; i < q->nr; i++)
4662                         diff_free_filepair(q->queue[i]);
4663         }
4664         else {
4665                 /* Only the matching ones */
4666                 for (i = 0; i < q->nr; i++) {
4667                         struct diff_filepair *p = q->queue[i];
4668                         if (match_filter(options, p))
4669                                 diff_q(&outq, p);
4670                         else
4671                                 diff_free_filepair(p);
4672                 }
4673         }
4674         free(q->queue);
4675         *q = outq;
4676 }
4677
4678 /* Check whether two filespecs with the same mode and size are identical */
4679 static int diff_filespec_is_identical(struct diff_filespec *one,
4680                                       struct diff_filespec *two)
4681 {
4682         if (S_ISGITLINK(one->mode))
4683                 return 0;
4684         if (diff_populate_filespec(one, 0))
4685                 return 0;
4686         if (diff_populate_filespec(two, 0))
4687                 return 0;
4688         return !memcmp(one->data, two->data, one->size);
4689 }
4690
4691 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4692 {
4693         if (p->done_skip_stat_unmatch)
4694                 return p->skip_stat_unmatch_result;
4695
4696         p->done_skip_stat_unmatch = 1;
4697         p->skip_stat_unmatch_result = 0;
4698         /*
4699          * 1. Entries that come from stat info dirtiness
4700          *    always have both sides (iow, not create/delete),
4701          *    one side of the object name is unknown, with
4702          *    the same mode and size.  Keep the ones that
4703          *    do not match these criteria.  They have real
4704          *    differences.
4705          *
4706          * 2. At this point, the file is known to be modified,
4707          *    with the same mode and size, and the object
4708          *    name of one side is unknown.  Need to inspect
4709          *    the identical contents.
4710          */
4711         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4712             !DIFF_FILE_VALID(p->two) ||
4713             (p->one->sha1_valid && p->two->sha1_valid) ||
4714             (p->one->mode != p->two->mode) ||
4715             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4716             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4717             (p->one->size != p->two->size) ||
4718             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4719                 p->skip_stat_unmatch_result = 1;
4720         return p->skip_stat_unmatch_result;
4721 }
4722
4723 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4724 {
4725         int i;
4726         struct diff_queue_struct *q = &diff_queued_diff;
4727         struct diff_queue_struct outq;
4728         DIFF_QUEUE_CLEAR(&outq);
4729
4730         for (i = 0; i < q->nr; i++) {
4731                 struct diff_filepair *p = q->queue[i];
4732
4733                 if (diff_filespec_check_stat_unmatch(p))
4734                         diff_q(&outq, p);
4735                 else {
4736                         /*
4737                          * The caller can subtract 1 from skip_stat_unmatch
4738                          * to determine how many paths were dirty only
4739                          * due to stat info mismatch.
4740                          */
4741                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4742                                 diffopt->skip_stat_unmatch++;
4743                         diff_free_filepair(p);
4744                 }
4745         }
4746         free(q->queue);
4747         *q = outq;
4748 }
4749
4750 static int diffnamecmp(const void *a_, const void *b_)
4751 {
4752         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4753         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4754         const char *name_a, *name_b;
4755
4756         name_a = a->one ? a->one->path : a->two->path;
4757         name_b = b->one ? b->one->path : b->two->path;
4758         return strcmp(name_a, name_b);
4759 }
4760
4761 void diffcore_fix_diff_index(struct diff_options *options)
4762 {
4763         struct diff_queue_struct *q = &diff_queued_diff;
4764         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4765 }
4766
4767 void diffcore_std(struct diff_options *options)
4768 {
4769         /* NOTE please keep the following in sync with diff_tree_combined() */
4770         if (options->skip_stat_unmatch)
4771                 diffcore_skip_stat_unmatch(options);
4772         if (!options->found_follow) {
4773                 /* See try_to_follow_renames() in tree-diff.c */
4774                 if (options->break_opt != -1)
4775                         diffcore_break(options->break_opt);
4776                 if (options->detect_rename)
4777                         diffcore_rename(options);
4778                 if (options->break_opt != -1)
4779                         diffcore_merge_broken();
4780         }
4781         if (options->pickaxe)
4782                 diffcore_pickaxe(options);
4783         if (options->orderfile)
4784                 diffcore_order(options->orderfile);
4785         if (!options->found_follow)
4786                 /* See try_to_follow_renames() in tree-diff.c */
4787                 diff_resolve_rename_copy();
4788         diffcore_apply_filter(options);
4789
4790         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4791                 DIFF_OPT_SET(options, HAS_CHANGES);
4792         else
4793                 DIFF_OPT_CLR(options, HAS_CHANGES);
4794
4795         options->found_follow = 0;
4796 }
4797
4798 int diff_result_code(struct diff_options *opt, int status)
4799 {
4800         int result = 0;
4801
4802         diff_warn_rename_limit("diff.renameLimit",
4803                                opt->needed_rename_limit,
4804                                opt->degraded_cc_to_c);
4805         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4806             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4807                 return status;
4808         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4809             DIFF_OPT_TST(opt, HAS_CHANGES))
4810                 result |= 01;
4811         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4812             DIFF_OPT_TST(opt, CHECK_FAILED))
4813                 result |= 02;
4814         return result;
4815 }
4816
4817 int diff_can_quit_early(struct diff_options *opt)
4818 {
4819         return (DIFF_OPT_TST(opt, QUICK) &&
4820                 !opt->filter &&
4821                 DIFF_OPT_TST(opt, HAS_CHANGES));
4822 }
4823
4824 /*
4825  * Shall changes to this submodule be ignored?
4826  *
4827  * Submodule changes can be configured to be ignored separately for each path,
4828  * but that configuration can be overridden from the command line.
4829  */
4830 static int is_submodule_ignored(const char *path, struct diff_options *options)
4831 {
4832         int ignored = 0;
4833         unsigned orig_flags = options->flags;
4834         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4835                 set_diffopt_flags_from_submodule_config(options, path);
4836         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4837                 ignored = 1;
4838         options->flags = orig_flags;
4839         return ignored;
4840 }
4841
4842 void diff_addremove(struct diff_options *options,
4843                     int addremove, unsigned mode,
4844                     const unsigned char *sha1,
4845                     int sha1_valid,
4846                     const char *concatpath, unsigned dirty_submodule)
4847 {
4848         struct diff_filespec *one, *two;
4849
4850         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4851                 return;
4852
4853         /* This may look odd, but it is a preparation for
4854          * feeding "there are unchanged files which should
4855          * not produce diffs, but when you are doing copy
4856          * detection you would need them, so here they are"
4857          * entries to the diff-core.  They will be prefixed
4858          * with something like '=' or '*' (I haven't decided
4859          * which but should not make any difference).
4860          * Feeding the same new and old to diff_change()
4861          * also has the same effect.
4862          * Before the final output happens, they are pruned after
4863          * merged into rename/copy pairs as appropriate.
4864          */
4865         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4866                 addremove = (addremove == '+' ? '-' :
4867                              addremove == '-' ? '+' : addremove);
4868
4869         if (options->prefix &&
4870             strncmp(concatpath, options->prefix, options->prefix_length))
4871                 return;
4872
4873         one = alloc_filespec(concatpath);
4874         two = alloc_filespec(concatpath);
4875
4876         if (addremove != '+')
4877                 fill_filespec(one, sha1, sha1_valid, mode);
4878         if (addremove != '-') {
4879                 fill_filespec(two, sha1, sha1_valid, mode);
4880                 two->dirty_submodule = dirty_submodule;
4881         }
4882
4883         diff_queue(&diff_queued_diff, one, two);
4884         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4885                 DIFF_OPT_SET(options, HAS_CHANGES);
4886 }
4887
4888 void diff_change(struct diff_options *options,
4889                  unsigned old_mode, unsigned new_mode,
4890                  const unsigned char *old_sha1,
4891                  const unsigned char *new_sha1,
4892                  int old_sha1_valid, int new_sha1_valid,
4893                  const char *concatpath,
4894                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4895 {
4896         struct diff_filespec *one, *two;
4897         struct diff_filepair *p;
4898
4899         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4900             is_submodule_ignored(concatpath, options))
4901                 return;
4902
4903         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4904                 unsigned tmp;
4905                 const unsigned char *tmp_c;
4906                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4907                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4908                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4909                         new_sha1_valid = tmp;
4910                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4911                         new_dirty_submodule = tmp;
4912         }
4913
4914         if (options->prefix &&
4915             strncmp(concatpath, options->prefix, options->prefix_length))
4916                 return;
4917
4918         one = alloc_filespec(concatpath);
4919         two = alloc_filespec(concatpath);
4920         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4921         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4922         one->dirty_submodule = old_dirty_submodule;
4923         two->dirty_submodule = new_dirty_submodule;
4924         p = diff_queue(&diff_queued_diff, one, two);
4925
4926         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4927                 return;
4928
4929         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
4930             !diff_filespec_check_stat_unmatch(p))
4931                 return;
4932
4933         DIFF_OPT_SET(options, HAS_CHANGES);
4934 }
4935
4936 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4937 {
4938         struct diff_filepair *pair;
4939         struct diff_filespec *one, *two;
4940
4941         if (options->prefix &&
4942             strncmp(path, options->prefix, options->prefix_length))
4943                 return NULL;
4944
4945         one = alloc_filespec(path);
4946         two = alloc_filespec(path);
4947         pair = diff_queue(&diff_queued_diff, one, two);
4948         pair->is_unmerged = 1;
4949         return pair;
4950 }
4951
4952 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4953                 size_t *outsize)
4954 {
4955         struct diff_tempfile *temp;
4956         const char *argv[3];
4957         const char **arg = argv;
4958         struct child_process child = CHILD_PROCESS_INIT;
4959         struct strbuf buf = STRBUF_INIT;
4960         int err = 0;
4961
4962         temp = prepare_temp_file(spec->path, spec);
4963         *arg++ = pgm;
4964         *arg++ = temp->name;
4965         *arg = NULL;
4966
4967         child.use_shell = 1;
4968         child.argv = argv;
4969         child.out = -1;
4970         if (start_command(&child)) {
4971                 remove_tempfile();
4972                 return NULL;
4973         }
4974
4975         if (strbuf_read(&buf, child.out, 0) < 0)
4976                 err = error("error reading from textconv command '%s'", pgm);
4977         close(child.out);
4978
4979         if (finish_command(&child) || err) {
4980                 strbuf_release(&buf);
4981                 remove_tempfile();
4982                 return NULL;
4983         }
4984         remove_tempfile();
4985
4986         return strbuf_detach(&buf, outsize);
4987 }
4988
4989 size_t fill_textconv(struct userdiff_driver *driver,
4990                      struct diff_filespec *df,
4991                      char **outbuf)
4992 {
4993         size_t size;
4994
4995         if (!driver || !driver->textconv) {
4996                 if (!DIFF_FILE_VALID(df)) {
4997                         *outbuf = "";
4998                         return 0;
4999                 }
5000                 if (diff_populate_filespec(df, 0))
5001                         die("unable to read files to diff");
5002                 *outbuf = df->data;
5003                 return df->size;
5004         }
5005
5006         if (driver->textconv_cache && df->sha1_valid) {
5007                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5008                                           &size);
5009                 if (*outbuf)
5010                         return size;
5011         }
5012
5013         *outbuf = run_textconv(driver->textconv, df, &size);
5014         if (!*outbuf)
5015                 die("unable to read files to diff");
5016
5017         if (driver->textconv_cache && df->sha1_valid) {
5018                 /* ignore errors, as we might be in a readonly repository */
5019                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5020                                 size);
5021                 /*
5022                  * we could save up changes and flush them all at the end,
5023                  * but we would need an extra call after all diffing is done.
5024                  * Since generating a cache entry is the slow path anyway,
5025                  * this extra overhead probably isn't a big deal.
5026                  */
5027                 notes_cache_write(driver->textconv_cache);
5028         }
5029
5030         return size;
5031 }
5032
5033 void setup_diff_pager(struct diff_options *opt)
5034 {
5035         /*
5036          * If the user asked for our exit code, then either they want --quiet
5037          * or --exit-code. We should definitely not bother with a pager in the
5038          * former case, as we will generate no output. Since we still properly
5039          * report our exit code even when a pager is run, we _could_ run a
5040          * pager with --exit-code. But since we have not done so historically,
5041          * and because it is easy to find people oneline advising "git diff
5042          * --exit-code" in hooks and other scripts, we do not do so.
5043          */
5044         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5045             check_pager_config("diff") != 0)
5046                 setup_pager();
5047 }