Merge branch 'jk/fix-refresh-utime'
[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,       /* CONTEXT */
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, "context") || !strcasecmp(var, "plain"))
58                 return DIFF_CONTEXT;
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                 return color_parse(value, diff_colors[slot]);
252         }
253
254         /* like GNU diff's --suppress-blank-empty option  */
255         if (!strcmp(var, "diff.suppressblankempty") ||
256                         /* for backwards compatibility */
257                         !strcmp(var, "diff.suppress-blank-empty")) {
258                 diff_suppress_blank_empty = git_config_bool(var, value);
259                 return 0;
260         }
261
262         if (!strcmp(var, "diff.dirstat")) {
263                 struct strbuf errmsg = STRBUF_INIT;
264                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
265                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
266                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
267                                 errmsg.buf);
268                 strbuf_release(&errmsg);
269                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
270                 return 0;
271         }
272
273         if (starts_with(var, "submodule."))
274                 return parse_submodule_config_option(var, value);
275
276         return git_default_config(var, value, cb);
277 }
278
279 static char *quote_two(const char *one, const char *two)
280 {
281         int need_one = quote_c_style(one, NULL, NULL, 1);
282         int need_two = quote_c_style(two, NULL, NULL, 1);
283         struct strbuf res = STRBUF_INIT;
284
285         if (need_one + need_two) {
286                 strbuf_addch(&res, '"');
287                 quote_c_style(one, &res, NULL, 1);
288                 quote_c_style(two, &res, NULL, 1);
289                 strbuf_addch(&res, '"');
290         } else {
291                 strbuf_addstr(&res, one);
292                 strbuf_addstr(&res, two);
293         }
294         return strbuf_detach(&res, NULL);
295 }
296
297 static const char *external_diff(void)
298 {
299         static const char *external_diff_cmd = NULL;
300         static int done_preparing = 0;
301
302         if (done_preparing)
303                 return external_diff_cmd;
304         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
305         if (!external_diff_cmd)
306                 external_diff_cmd = external_diff_cmd_cfg;
307         done_preparing = 1;
308         return external_diff_cmd;
309 }
310
311 static struct diff_tempfile {
312         const char *name; /* filename external diff should read from */
313         char hex[41];
314         char mode[10];
315         char tmp_path[PATH_MAX];
316 } diff_temp[2];
317
318 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
319
320 struct emit_callback {
321         int color_diff;
322         unsigned ws_rule;
323         int blank_at_eof_in_preimage;
324         int blank_at_eof_in_postimage;
325         int lno_in_preimage;
326         int lno_in_postimage;
327         sane_truncate_fn truncate;
328         const char **label_path;
329         struct diff_words_data *diff_words;
330         struct diff_options *opt;
331         int *found_changesp;
332         struct strbuf *header;
333 };
334
335 static int count_lines(const char *data, int size)
336 {
337         int count, ch, completely_empty = 1, nl_just_seen = 0;
338         count = 0;
339         while (0 < size--) {
340                 ch = *data++;
341                 if (ch == '\n') {
342                         count++;
343                         nl_just_seen = 1;
344                         completely_empty = 0;
345                 }
346                 else {
347                         nl_just_seen = 0;
348                         completely_empty = 0;
349                 }
350         }
351         if (completely_empty)
352                 return 0;
353         if (!nl_just_seen)
354                 count++; /* no trailing newline */
355         return count;
356 }
357
358 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
359 {
360         if (!DIFF_FILE_VALID(one)) {
361                 mf->ptr = (char *)""; /* does not matter */
362                 mf->size = 0;
363                 return 0;
364         }
365         else if (diff_populate_filespec(one, 0))
366                 return -1;
367
368         mf->ptr = one->data;
369         mf->size = one->size;
370         return 0;
371 }
372
373 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
374 static unsigned long diff_filespec_size(struct diff_filespec *one)
375 {
376         if (!DIFF_FILE_VALID(one))
377                 return 0;
378         diff_populate_filespec(one, CHECK_SIZE_ONLY);
379         return one->size;
380 }
381
382 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
383 {
384         char *ptr = mf->ptr;
385         long size = mf->size;
386         int cnt = 0;
387
388         if (!size)
389                 return cnt;
390         ptr += size - 1; /* pointing at the very end */
391         if (*ptr != '\n')
392                 ; /* incomplete line */
393         else
394                 ptr--; /* skip the last LF */
395         while (mf->ptr < ptr) {
396                 char *prev_eol;
397                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
398                         if (*prev_eol == '\n')
399                                 break;
400                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
401                         break;
402                 cnt++;
403                 ptr = prev_eol - 1;
404         }
405         return cnt;
406 }
407
408 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
409                                struct emit_callback *ecbdata)
410 {
411         int l1, l2, at;
412         unsigned ws_rule = ecbdata->ws_rule;
413         l1 = count_trailing_blank(mf1, ws_rule);
414         l2 = count_trailing_blank(mf2, ws_rule);
415         if (l2 <= l1) {
416                 ecbdata->blank_at_eof_in_preimage = 0;
417                 ecbdata->blank_at_eof_in_postimage = 0;
418                 return;
419         }
420         at = count_lines(mf1->ptr, mf1->size);
421         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
422
423         at = count_lines(mf2->ptr, mf2->size);
424         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
425 }
426
427 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
428                         int first, const char *line, int len)
429 {
430         int has_trailing_newline, has_trailing_carriage_return;
431         int nofirst;
432         FILE *file = o->file;
433
434         fputs(diff_line_prefix(o), file);
435
436         if (len == 0) {
437                 has_trailing_newline = (first == '\n');
438                 has_trailing_carriage_return = (!has_trailing_newline &&
439                                                 (first == '\r'));
440                 nofirst = has_trailing_newline || has_trailing_carriage_return;
441         } else {
442                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
443                 if (has_trailing_newline)
444                         len--;
445                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
446                 if (has_trailing_carriage_return)
447                         len--;
448                 nofirst = 0;
449         }
450
451         if (len || !nofirst) {
452                 fputs(set, file);
453                 if (!nofirst)
454                         fputc(first, file);
455                 fwrite(line, len, 1, file);
456                 fputs(reset, file);
457         }
458         if (has_trailing_carriage_return)
459                 fputc('\r', file);
460         if (has_trailing_newline)
461                 fputc('\n', file);
462 }
463
464 static void emit_line(struct diff_options *o, const char *set, const char *reset,
465                       const char *line, int len)
466 {
467         emit_line_0(o, set, reset, line[0], line+1, len-1);
468 }
469
470 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
471 {
472         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
473               ecbdata->blank_at_eof_in_preimage &&
474               ecbdata->blank_at_eof_in_postimage &&
475               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
476               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
477                 return 0;
478         return ws_blank_line(line, len, ecbdata->ws_rule);
479 }
480
481 static void emit_line_checked(const char *reset,
482                               struct emit_callback *ecbdata,
483                               const char *line, int len,
484                               enum color_diff color,
485                               unsigned ws_error_highlight,
486                               char sign)
487 {
488         const char *set = diff_get_color(ecbdata->color_diff, color);
489         const char *ws = NULL;
490
491         if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
492                 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
493                 if (!*ws)
494                         ws = NULL;
495         }
496
497         if (!ws)
498                 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
499         else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
500                 /* Blank line at EOF - paint '+' as well */
501                 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
502         else {
503                 /* Emit just the prefix, then the rest. */
504                 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
505                 ws_check_emit(line, len, ecbdata->ws_rule,
506                               ecbdata->opt->file, set, reset, ws);
507         }
508 }
509
510 static void emit_add_line(const char *reset,
511                           struct emit_callback *ecbdata,
512                           const char *line, int len)
513 {
514         emit_line_checked(reset, ecbdata, line, len,
515                           DIFF_FILE_NEW, WSEH_NEW, '+');
516 }
517
518 static void emit_del_line(const char *reset,
519                           struct emit_callback *ecbdata,
520                           const char *line, int len)
521 {
522         emit_line_checked(reset, ecbdata, line, len,
523                           DIFF_FILE_OLD, WSEH_OLD, '-');
524 }
525
526 static void emit_context_line(const char *reset,
527                               struct emit_callback *ecbdata,
528                               const char *line, int len)
529 {
530         emit_line_checked(reset, ecbdata, line, len,
531                           DIFF_CONTEXT, WSEH_CONTEXT, ' ');
532 }
533
534 static void emit_hunk_header(struct emit_callback *ecbdata,
535                              const char *line, int len)
536 {
537         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
538         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
539         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
540         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
541         static const char atat[2] = { '@', '@' };
542         const char *cp, *ep;
543         struct strbuf msgbuf = STRBUF_INIT;
544         int org_len = len;
545         int i = 1;
546
547         /*
548          * As a hunk header must begin with "@@ -<old>, +<new> @@",
549          * it always is at least 10 bytes long.
550          */
551         if (len < 10 ||
552             memcmp(line, atat, 2) ||
553             !(ep = memmem(line + 2, len - 2, atat, 2))) {
554                 emit_line(ecbdata->opt, context, reset, line, len);
555                 return;
556         }
557         ep += 2; /* skip over @@ */
558
559         /* The hunk header in fraginfo color */
560         strbuf_addstr(&msgbuf, frag);
561         strbuf_add(&msgbuf, line, ep - line);
562         strbuf_addstr(&msgbuf, reset);
563
564         /*
565          * trailing "\r\n"
566          */
567         for ( ; i < 3; i++)
568                 if (line[len - i] == '\r' || line[len - i] == '\n')
569                         len--;
570
571         /* blank before the func header */
572         for (cp = ep; ep - line < len; ep++)
573                 if (*ep != ' ' && *ep != '\t')
574                         break;
575         if (ep != cp) {
576                 strbuf_addstr(&msgbuf, context);
577                 strbuf_add(&msgbuf, cp, ep - cp);
578                 strbuf_addstr(&msgbuf, reset);
579         }
580
581         if (ep < line + len) {
582                 strbuf_addstr(&msgbuf, func);
583                 strbuf_add(&msgbuf, ep, line + len - ep);
584                 strbuf_addstr(&msgbuf, reset);
585         }
586
587         strbuf_add(&msgbuf, line + len, org_len - len);
588         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
589         strbuf_release(&msgbuf);
590 }
591
592 static struct diff_tempfile *claim_diff_tempfile(void) {
593         int i;
594         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
595                 if (!diff_temp[i].name)
596                         return diff_temp + i;
597         die("BUG: diff is failing to clean up its tempfiles");
598 }
599
600 static int remove_tempfile_installed;
601
602 static void remove_tempfile(void)
603 {
604         int i;
605         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
606                 if (diff_temp[i].name == diff_temp[i].tmp_path)
607                         unlink_or_warn(diff_temp[i].name);
608                 diff_temp[i].name = NULL;
609         }
610 }
611
612 static void remove_tempfile_on_signal(int signo)
613 {
614         remove_tempfile();
615         sigchain_pop(signo);
616         raise(signo);
617 }
618
619 static void print_line_count(FILE *file, int count)
620 {
621         switch (count) {
622         case 0:
623                 fprintf(file, "0,0");
624                 break;
625         case 1:
626                 fprintf(file, "1");
627                 break;
628         default:
629                 fprintf(file, "1,%d", count);
630                 break;
631         }
632 }
633
634 static void emit_rewrite_lines(struct emit_callback *ecb,
635                                int prefix, const char *data, int size)
636 {
637         const char *endp = NULL;
638         static const char *nneof = " No newline at end of file\n";
639         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
640
641         while (0 < size) {
642                 int len;
643
644                 endp = memchr(data, '\n', size);
645                 len = endp ? (endp - data + 1) : size;
646                 if (prefix != '+') {
647                         ecb->lno_in_preimage++;
648                         emit_del_line(reset, ecb, data, len);
649                 } else {
650                         ecb->lno_in_postimage++;
651                         emit_add_line(reset, ecb, data, len);
652                 }
653                 size -= len;
654                 data += len;
655         }
656         if (!endp) {
657                 const char *context = diff_get_color(ecb->color_diff,
658                                                      DIFF_CONTEXT);
659                 putc('\n', ecb->opt->file);
660                 emit_line_0(ecb->opt, context, reset, '\\',
661                             nneof, strlen(nneof));
662         }
663 }
664
665 static void emit_rewrite_diff(const char *name_a,
666                               const char *name_b,
667                               struct diff_filespec *one,
668                               struct diff_filespec *two,
669                               struct userdiff_driver *textconv_one,
670                               struct userdiff_driver *textconv_two,
671                               struct diff_options *o)
672 {
673         int lc_a, lc_b;
674         const char *name_a_tab, *name_b_tab;
675         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
676         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
677         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
678         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
679         const char *a_prefix, *b_prefix;
680         char *data_one, *data_two;
681         size_t size_one, size_two;
682         struct emit_callback ecbdata;
683         const char *line_prefix = diff_line_prefix(o);
684
685         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
686                 a_prefix = o->b_prefix;
687                 b_prefix = o->a_prefix;
688         } else {
689                 a_prefix = o->a_prefix;
690                 b_prefix = o->b_prefix;
691         }
692
693         name_a += (*name_a == '/');
694         name_b += (*name_b == '/');
695         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
696         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
697
698         strbuf_reset(&a_name);
699         strbuf_reset(&b_name);
700         quote_two_c_style(&a_name, a_prefix, name_a, 0);
701         quote_two_c_style(&b_name, b_prefix, name_b, 0);
702
703         size_one = fill_textconv(textconv_one, one, &data_one);
704         size_two = fill_textconv(textconv_two, two, &data_two);
705
706         memset(&ecbdata, 0, sizeof(ecbdata));
707         ecbdata.color_diff = want_color(o->use_color);
708         ecbdata.found_changesp = &o->found_changes;
709         ecbdata.ws_rule = whitespace_rule(name_b);
710         ecbdata.opt = o;
711         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
712                 mmfile_t mf1, mf2;
713                 mf1.ptr = (char *)data_one;
714                 mf2.ptr = (char *)data_two;
715                 mf1.size = size_one;
716                 mf2.size = size_two;
717                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
718         }
719         ecbdata.lno_in_preimage = 1;
720         ecbdata.lno_in_postimage = 1;
721
722         lc_a = count_lines(data_one, size_one);
723         lc_b = count_lines(data_two, size_two);
724         fprintf(o->file,
725                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
726                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
727                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
728                 line_prefix, fraginfo);
729         if (!o->irreversible_delete)
730                 print_line_count(o->file, lc_a);
731         else
732                 fprintf(o->file, "?,?");
733         fprintf(o->file, " +");
734         print_line_count(o->file, lc_b);
735         fprintf(o->file, " @@%s\n", reset);
736         if (lc_a && !o->irreversible_delete)
737                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
738         if (lc_b)
739                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
740         if (textconv_one)
741                 free((char *)data_one);
742         if (textconv_two)
743                 free((char *)data_two);
744 }
745
746 struct diff_words_buffer {
747         mmfile_t text;
748         long alloc;
749         struct diff_words_orig {
750                 const char *begin, *end;
751         } *orig;
752         int orig_nr, orig_alloc;
753 };
754
755 static void diff_words_append(char *line, unsigned long len,
756                 struct diff_words_buffer *buffer)
757 {
758         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
759         line++;
760         len--;
761         memcpy(buffer->text.ptr + buffer->text.size, line, len);
762         buffer->text.size += len;
763         buffer->text.ptr[buffer->text.size] = '\0';
764 }
765
766 struct diff_words_style_elem {
767         const char *prefix;
768         const char *suffix;
769         const char *color; /* NULL; filled in by the setup code if
770                             * color is enabled */
771 };
772
773 struct diff_words_style {
774         enum diff_words_type type;
775         struct diff_words_style_elem new, old, ctx;
776         const char *newline;
777 };
778
779 static struct diff_words_style diff_words_styles[] = {
780         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
781         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
782         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
783 };
784
785 struct diff_words_data {
786         struct diff_words_buffer minus, plus;
787         const char *current_plus;
788         int last_minus;
789         struct diff_options *opt;
790         regex_t *word_regex;
791         enum diff_words_type type;
792         struct diff_words_style *style;
793 };
794
795 static int fn_out_diff_words_write_helper(FILE *fp,
796                                           struct diff_words_style_elem *st_el,
797                                           const char *newline,
798                                           size_t count, const char *buf,
799                                           const char *line_prefix)
800 {
801         int print = 0;
802
803         while (count) {
804                 char *p = memchr(buf, '\n', count);
805                 if (print)
806                         fputs(line_prefix, fp);
807                 if (p != buf) {
808                         if (st_el->color && fputs(st_el->color, fp) < 0)
809                                 return -1;
810                         if (fputs(st_el->prefix, fp) < 0 ||
811                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
812                             fputs(st_el->suffix, fp) < 0)
813                                 return -1;
814                         if (st_el->color && *st_el->color
815                             && fputs(GIT_COLOR_RESET, fp) < 0)
816                                 return -1;
817                 }
818                 if (!p)
819                         return 0;
820                 if (fputs(newline, fp) < 0)
821                         return -1;
822                 count -= p + 1 - buf;
823                 buf = p + 1;
824                 print = 1;
825         }
826         return 0;
827 }
828
829 /*
830  * '--color-words' algorithm can be described as:
831  *
832  *   1. collect a the minus/plus lines of a diff hunk, divided into
833  *      minus-lines and plus-lines;
834  *
835  *   2. break both minus-lines and plus-lines into words and
836  *      place them into two mmfile_t with one word for each line;
837  *
838  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
839  *
840  * And for the common parts of the both file, we output the plus side text.
841  * diff_words->current_plus is used to trace the current position of the plus file
842  * which printed. diff_words->last_minus is used to trace the last minus word
843  * printed.
844  *
845  * For '--graph' to work with '--color-words', we need to output the graph prefix
846  * on each line of color words output. Generally, there are two conditions on
847  * which we should output the prefix.
848  *
849  *   1. diff_words->last_minus == 0 &&
850  *      diff_words->current_plus == diff_words->plus.text.ptr
851  *
852  *      that is: the plus text must start as a new line, and if there is no minus
853  *      word printed, a graph prefix must be printed.
854  *
855  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
856  *      *(diff_words->current_plus - 1) == '\n'
857  *
858  *      that is: a graph prefix must be printed following a '\n'
859  */
860 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
861 {
862         if ((diff_words->last_minus == 0 &&
863                 diff_words->current_plus == diff_words->plus.text.ptr) ||
864                 (diff_words->current_plus > diff_words->plus.text.ptr &&
865                 *(diff_words->current_plus - 1) == '\n')) {
866                 return 1;
867         } else {
868                 return 0;
869         }
870 }
871
872 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
873 {
874         struct diff_words_data *diff_words = priv;
875         struct diff_words_style *style = diff_words->style;
876         int minus_first, minus_len, plus_first, plus_len;
877         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
878         struct diff_options *opt = diff_words->opt;
879         const char *line_prefix;
880
881         if (line[0] != '@' || parse_hunk_header(line, len,
882                         &minus_first, &minus_len, &plus_first, &plus_len))
883                 return;
884
885         assert(opt);
886         line_prefix = diff_line_prefix(opt);
887
888         /* POSIX requires that first be decremented by one if len == 0... */
889         if (minus_len) {
890                 minus_begin = diff_words->minus.orig[minus_first].begin;
891                 minus_end =
892                         diff_words->minus.orig[minus_first + minus_len - 1].end;
893         } else
894                 minus_begin = minus_end =
895                         diff_words->minus.orig[minus_first].end;
896
897         if (plus_len) {
898                 plus_begin = diff_words->plus.orig[plus_first].begin;
899                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
900         } else
901                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
902
903         if (color_words_output_graph_prefix(diff_words)) {
904                 fputs(line_prefix, diff_words->opt->file);
905         }
906         if (diff_words->current_plus != plus_begin) {
907                 fn_out_diff_words_write_helper(diff_words->opt->file,
908                                 &style->ctx, style->newline,
909                                 plus_begin - diff_words->current_plus,
910                                 diff_words->current_plus, line_prefix);
911                 if (*(plus_begin - 1) == '\n')
912                         fputs(line_prefix, diff_words->opt->file);
913         }
914         if (minus_begin != minus_end) {
915                 fn_out_diff_words_write_helper(diff_words->opt->file,
916                                 &style->old, style->newline,
917                                 minus_end - minus_begin, minus_begin,
918                                 line_prefix);
919         }
920         if (plus_begin != plus_end) {
921                 fn_out_diff_words_write_helper(diff_words->opt->file,
922                                 &style->new, style->newline,
923                                 plus_end - plus_begin, plus_begin,
924                                 line_prefix);
925         }
926
927         diff_words->current_plus = plus_end;
928         diff_words->last_minus = minus_first;
929 }
930
931 /* This function starts looking at *begin, and returns 0 iff a word was found. */
932 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
933                 int *begin, int *end)
934 {
935         if (word_regex && *begin < buffer->size) {
936                 regmatch_t match[1];
937                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
938                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
939                                         '\n', match[0].rm_eo - match[0].rm_so);
940                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
941                         *begin += match[0].rm_so;
942                         return *begin >= *end;
943                 }
944                 return -1;
945         }
946
947         /* find the next word */
948         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
949                 (*begin)++;
950         if (*begin >= buffer->size)
951                 return -1;
952
953         /* find the end of the word */
954         *end = *begin + 1;
955         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
956                 (*end)++;
957
958         return 0;
959 }
960
961 /*
962  * This function splits the words in buffer->text, stores the list with
963  * newline separator into out, and saves the offsets of the original words
964  * in buffer->orig.
965  */
966 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
967                 regex_t *word_regex)
968 {
969         int i, j;
970         long alloc = 0;
971
972         out->size = 0;
973         out->ptr = NULL;
974
975         /* fake an empty "0th" word */
976         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
977         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
978         buffer->orig_nr = 1;
979
980         for (i = 0; i < buffer->text.size; i++) {
981                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
982                         return;
983
984                 /* store original boundaries */
985                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
986                                 buffer->orig_alloc);
987                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
988                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
989                 buffer->orig_nr++;
990
991                 /* store one word */
992                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
993                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
994                 out->ptr[out->size + j - i] = '\n';
995                 out->size += j - i + 1;
996
997                 i = j - 1;
998         }
999 }
1000
1001 /* this executes the word diff on the accumulated buffers */
1002 static void diff_words_show(struct diff_words_data *diff_words)
1003 {
1004         xpparam_t xpp;
1005         xdemitconf_t xecfg;
1006         mmfile_t minus, plus;
1007         struct diff_words_style *style = diff_words->style;
1008
1009         struct diff_options *opt = diff_words->opt;
1010         const char *line_prefix;
1011
1012         assert(opt);
1013         line_prefix = diff_line_prefix(opt);
1014
1015         /* special case: only removal */
1016         if (!diff_words->plus.text.size) {
1017                 fputs(line_prefix, diff_words->opt->file);
1018                 fn_out_diff_words_write_helper(diff_words->opt->file,
1019                         &style->old, style->newline,
1020                         diff_words->minus.text.size,
1021                         diff_words->minus.text.ptr, line_prefix);
1022                 diff_words->minus.text.size = 0;
1023                 return;
1024         }
1025
1026         diff_words->current_plus = diff_words->plus.text.ptr;
1027         diff_words->last_minus = 0;
1028
1029         memset(&xpp, 0, sizeof(xpp));
1030         memset(&xecfg, 0, sizeof(xecfg));
1031         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1032         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1033         xpp.flags = 0;
1034         /* as only the hunk header will be parsed, we need a 0-context */
1035         xecfg.ctxlen = 0;
1036         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1037                       &xpp, &xecfg);
1038         free(minus.ptr);
1039         free(plus.ptr);
1040         if (diff_words->current_plus != diff_words->plus.text.ptr +
1041                         diff_words->plus.text.size) {
1042                 if (color_words_output_graph_prefix(diff_words))
1043                         fputs(line_prefix, diff_words->opt->file);
1044                 fn_out_diff_words_write_helper(diff_words->opt->file,
1045                         &style->ctx, style->newline,
1046                         diff_words->plus.text.ptr + diff_words->plus.text.size
1047                         - diff_words->current_plus, diff_words->current_plus,
1048                         line_prefix);
1049         }
1050         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1051 }
1052
1053 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1054 static void diff_words_flush(struct emit_callback *ecbdata)
1055 {
1056         if (ecbdata->diff_words->minus.text.size ||
1057             ecbdata->diff_words->plus.text.size)
1058                 diff_words_show(ecbdata->diff_words);
1059 }
1060
1061 static void diff_filespec_load_driver(struct diff_filespec *one)
1062 {
1063         /* Use already-loaded driver */
1064         if (one->driver)
1065                 return;
1066
1067         if (S_ISREG(one->mode))
1068                 one->driver = userdiff_find_by_path(one->path);
1069
1070         /* Fallback to default settings */
1071         if (!one->driver)
1072                 one->driver = userdiff_find_by_name("default");
1073 }
1074
1075 static const char *userdiff_word_regex(struct diff_filespec *one)
1076 {
1077         diff_filespec_load_driver(one);
1078         return one->driver->word_regex;
1079 }
1080
1081 static void init_diff_words_data(struct emit_callback *ecbdata,
1082                                  struct diff_options *orig_opts,
1083                                  struct diff_filespec *one,
1084                                  struct diff_filespec *two)
1085 {
1086         int i;
1087         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1088         memcpy(o, orig_opts, sizeof(struct diff_options));
1089
1090         ecbdata->diff_words =
1091                 xcalloc(1, sizeof(struct diff_words_data));
1092         ecbdata->diff_words->type = o->word_diff;
1093         ecbdata->diff_words->opt = o;
1094         if (!o->word_regex)
1095                 o->word_regex = userdiff_word_regex(one);
1096         if (!o->word_regex)
1097                 o->word_regex = userdiff_word_regex(two);
1098         if (!o->word_regex)
1099                 o->word_regex = diff_word_regex_cfg;
1100         if (o->word_regex) {
1101                 ecbdata->diff_words->word_regex = (regex_t *)
1102                         xmalloc(sizeof(regex_t));
1103                 if (regcomp(ecbdata->diff_words->word_regex,
1104                             o->word_regex,
1105                             REG_EXTENDED | REG_NEWLINE))
1106                         die ("Invalid regular expression: %s",
1107                              o->word_regex);
1108         }
1109         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1110                 if (o->word_diff == diff_words_styles[i].type) {
1111                         ecbdata->diff_words->style =
1112                                 &diff_words_styles[i];
1113                         break;
1114                 }
1115         }
1116         if (want_color(o->use_color)) {
1117                 struct diff_words_style *st = ecbdata->diff_words->style;
1118                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1119                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1120                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1121         }
1122 }
1123
1124 static void free_diff_words_data(struct emit_callback *ecbdata)
1125 {
1126         if (ecbdata->diff_words) {
1127                 diff_words_flush(ecbdata);
1128                 free (ecbdata->diff_words->opt);
1129                 free (ecbdata->diff_words->minus.text.ptr);
1130                 free (ecbdata->diff_words->minus.orig);
1131                 free (ecbdata->diff_words->plus.text.ptr);
1132                 free (ecbdata->diff_words->plus.orig);
1133                 if (ecbdata->diff_words->word_regex) {
1134                         regfree(ecbdata->diff_words->word_regex);
1135                         free(ecbdata->diff_words->word_regex);
1136                 }
1137                 free(ecbdata->diff_words);
1138                 ecbdata->diff_words = NULL;
1139         }
1140 }
1141
1142 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1143 {
1144         if (want_color(diff_use_color))
1145                 return diff_colors[ix];
1146         return "";
1147 }
1148
1149 const char *diff_line_prefix(struct diff_options *opt)
1150 {
1151         struct strbuf *msgbuf;
1152         if (!opt->output_prefix)
1153                 return "";
1154
1155         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1156         return msgbuf->buf;
1157 }
1158
1159 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1160 {
1161         const char *cp;
1162         unsigned long allot;
1163         size_t l = len;
1164
1165         if (ecb->truncate)
1166                 return ecb->truncate(line, len);
1167         cp = line;
1168         allot = l;
1169         while (0 < l) {
1170                 (void) utf8_width(&cp, &l);
1171                 if (!cp)
1172                         break; /* truncated in the middle? */
1173         }
1174         return allot - l;
1175 }
1176
1177 static void find_lno(const char *line, struct emit_callback *ecbdata)
1178 {
1179         const char *p;
1180         ecbdata->lno_in_preimage = 0;
1181         ecbdata->lno_in_postimage = 0;
1182         p = strchr(line, '-');
1183         if (!p)
1184                 return; /* cannot happen */
1185         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1186         p = strchr(p, '+');
1187         if (!p)
1188                 return; /* cannot happen */
1189         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1190 }
1191
1192 static void fn_out_consume(void *priv, char *line, unsigned long len)
1193 {
1194         struct emit_callback *ecbdata = priv;
1195         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1196         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1197         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1198         struct diff_options *o = ecbdata->opt;
1199         const char *line_prefix = diff_line_prefix(o);
1200
1201         if (ecbdata->header) {
1202                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1203                 strbuf_reset(ecbdata->header);
1204                 ecbdata->header = NULL;
1205         }
1206         *(ecbdata->found_changesp) = 1;
1207
1208         if (ecbdata->label_path[0]) {
1209                 const char *name_a_tab, *name_b_tab;
1210
1211                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1212                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1213
1214                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1215                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1216                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1217                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1218                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1219         }
1220
1221         if (diff_suppress_blank_empty
1222             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1223                 line[0] = '\n';
1224                 len = 1;
1225         }
1226
1227         if (line[0] == '@') {
1228                 if (ecbdata->diff_words)
1229                         diff_words_flush(ecbdata);
1230                 len = sane_truncate_line(ecbdata, line, len);
1231                 find_lno(line, ecbdata);
1232                 emit_hunk_header(ecbdata, line, len);
1233                 if (line[len-1] != '\n')
1234                         putc('\n', ecbdata->opt->file);
1235                 return;
1236         }
1237
1238         if (len < 1) {
1239                 emit_line(ecbdata->opt, reset, reset, line, len);
1240                 if (ecbdata->diff_words
1241                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1242                         fputs("~\n", ecbdata->opt->file);
1243                 return;
1244         }
1245
1246         if (ecbdata->diff_words) {
1247                 if (line[0] == '-') {
1248                         diff_words_append(line, len,
1249                                           &ecbdata->diff_words->minus);
1250                         return;
1251                 } else if (line[0] == '+') {
1252                         diff_words_append(line, len,
1253                                           &ecbdata->diff_words->plus);
1254                         return;
1255                 } else if (starts_with(line, "\\ ")) {
1256                         /*
1257                          * Eat the "no newline at eof" marker as if we
1258                          * saw a "+" or "-" line with nothing on it,
1259                          * and return without diff_words_flush() to
1260                          * defer processing. If this is the end of
1261                          * preimage, more "+" lines may come after it.
1262                          */
1263                         return;
1264                 }
1265                 diff_words_flush(ecbdata);
1266                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1267                         emit_line(ecbdata->opt, context, reset, line, len);
1268                         fputs("~\n", ecbdata->opt->file);
1269                 } else {
1270                         /*
1271                          * Skip the prefix character, if any.  With
1272                          * diff_suppress_blank_empty, there may be
1273                          * none.
1274                          */
1275                         if (line[0] != '\n') {
1276                               line++;
1277                               len--;
1278                         }
1279                         emit_line(ecbdata->opt, context, reset, line, len);
1280                 }
1281                 return;
1282         }
1283
1284         switch (line[0]) {
1285         case '+':
1286                 ecbdata->lno_in_postimage++;
1287                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1288                 break;
1289         case '-':
1290                 ecbdata->lno_in_preimage++;
1291                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1292                 break;
1293         case ' ':
1294                 ecbdata->lno_in_postimage++;
1295                 ecbdata->lno_in_preimage++;
1296                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1297                 break;
1298         default:
1299                 /* incomplete line at the end */
1300                 ecbdata->lno_in_preimage++;
1301                 emit_line(ecbdata->opt,
1302                           diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1303                           reset, line, len);
1304                 break;
1305         }
1306 }
1307
1308 static char *pprint_rename(const char *a, const char *b)
1309 {
1310         const char *old = a;
1311         const char *new = b;
1312         struct strbuf name = STRBUF_INIT;
1313         int pfx_length, sfx_length;
1314         int pfx_adjust_for_slash;
1315         int len_a = strlen(a);
1316         int len_b = strlen(b);
1317         int a_midlen, b_midlen;
1318         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1319         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1320
1321         if (qlen_a || qlen_b) {
1322                 quote_c_style(a, &name, NULL, 0);
1323                 strbuf_addstr(&name, " => ");
1324                 quote_c_style(b, &name, NULL, 0);
1325                 return strbuf_detach(&name, NULL);
1326         }
1327
1328         /* Find common prefix */
1329         pfx_length = 0;
1330         while (*old && *new && *old == *new) {
1331                 if (*old == '/')
1332                         pfx_length = old - a + 1;
1333                 old++;
1334                 new++;
1335         }
1336
1337         /* Find common suffix */
1338         old = a + len_a;
1339         new = b + len_b;
1340         sfx_length = 0;
1341         /*
1342          * If there is a common prefix, it must end in a slash.  In
1343          * that case we let this loop run 1 into the prefix to see the
1344          * same slash.
1345          *
1346          * If there is no common prefix, we cannot do this as it would
1347          * underrun the input strings.
1348          */
1349         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1350         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1351                b + pfx_length - pfx_adjust_for_slash <= new &&
1352                *old == *new) {
1353                 if (*old == '/')
1354                         sfx_length = len_a - (old - a);
1355                 old--;
1356                 new--;
1357         }
1358
1359         /*
1360          * pfx{mid-a => mid-b}sfx
1361          * {pfx-a => pfx-b}sfx
1362          * pfx{sfx-a => sfx-b}
1363          * name-a => name-b
1364          */
1365         a_midlen = len_a - pfx_length - sfx_length;
1366         b_midlen = len_b - pfx_length - sfx_length;
1367         if (a_midlen < 0)
1368                 a_midlen = 0;
1369         if (b_midlen < 0)
1370                 b_midlen = 0;
1371
1372         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1373         if (pfx_length + sfx_length) {
1374                 strbuf_add(&name, a, pfx_length);
1375                 strbuf_addch(&name, '{');
1376         }
1377         strbuf_add(&name, a + pfx_length, a_midlen);
1378         strbuf_addstr(&name, " => ");
1379         strbuf_add(&name, b + pfx_length, b_midlen);
1380         if (pfx_length + sfx_length) {
1381                 strbuf_addch(&name, '}');
1382                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1383         }
1384         return strbuf_detach(&name, NULL);
1385 }
1386
1387 struct diffstat_t {
1388         int nr;
1389         int alloc;
1390         struct diffstat_file {
1391                 char *from_name;
1392                 char *name;
1393                 char *print_name;
1394                 unsigned is_unmerged:1;
1395                 unsigned is_binary:1;
1396                 unsigned is_renamed:1;
1397                 unsigned is_interesting:1;
1398                 uintmax_t added, deleted;
1399         } **files;
1400 };
1401
1402 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1403                                           const char *name_a,
1404                                           const char *name_b)
1405 {
1406         struct diffstat_file *x;
1407         x = xcalloc(1, sizeof(*x));
1408         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1409         diffstat->files[diffstat->nr++] = x;
1410         if (name_b) {
1411                 x->from_name = xstrdup(name_a);
1412                 x->name = xstrdup(name_b);
1413                 x->is_renamed = 1;
1414         }
1415         else {
1416                 x->from_name = NULL;
1417                 x->name = xstrdup(name_a);
1418         }
1419         return x;
1420 }
1421
1422 static void diffstat_consume(void *priv, char *line, unsigned long len)
1423 {
1424         struct diffstat_t *diffstat = priv;
1425         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1426
1427         if (line[0] == '+')
1428                 x->added++;
1429         else if (line[0] == '-')
1430                 x->deleted++;
1431 }
1432
1433 const char mime_boundary_leader[] = "------------";
1434
1435 static int scale_linear(int it, int width, int max_change)
1436 {
1437         if (!it)
1438                 return 0;
1439         /*
1440          * make sure that at least one '-' or '+' is printed if
1441          * there is any change to this path. The easiest way is to
1442          * scale linearly as if the alloted width is one column shorter
1443          * than it is, and then add 1 to the result.
1444          */
1445         return 1 + (it * (width - 1) / max_change);
1446 }
1447
1448 static void show_name(FILE *file,
1449                       const char *prefix, const char *name, int len)
1450 {
1451         fprintf(file, " %s%-*s |", prefix, len, name);
1452 }
1453
1454 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1455 {
1456         if (cnt <= 0)
1457                 return;
1458         fprintf(file, "%s", set);
1459         while (cnt--)
1460                 putc(ch, file);
1461         fprintf(file, "%s", reset);
1462 }
1463
1464 static void fill_print_name(struct diffstat_file *file)
1465 {
1466         char *pname;
1467
1468         if (file->print_name)
1469                 return;
1470
1471         if (!file->is_renamed) {
1472                 struct strbuf buf = STRBUF_INIT;
1473                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1474                         pname = strbuf_detach(&buf, NULL);
1475                 } else {
1476                         pname = file->name;
1477                         strbuf_release(&buf);
1478                 }
1479         } else {
1480                 pname = pprint_rename(file->from_name, file->name);
1481         }
1482         file->print_name = pname;
1483 }
1484
1485 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1486 {
1487         struct strbuf sb = STRBUF_INIT;
1488         int ret;
1489
1490         if (!files) {
1491                 assert(insertions == 0 && deletions == 0);
1492                 return fprintf(fp, "%s\n", " 0 files changed");
1493         }
1494
1495         strbuf_addf(&sb,
1496                     (files == 1) ? " %d file changed" : " %d files changed",
1497                     files);
1498
1499         /*
1500          * For binary diff, the caller may want to print "x files
1501          * changed" with insertions == 0 && deletions == 0.
1502          *
1503          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1504          * is probably less confusing (i.e skip over "2 files changed
1505          * but nothing about added/removed lines? Is this a bug in Git?").
1506          */
1507         if (insertions || deletions == 0) {
1508                 strbuf_addf(&sb,
1509                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1510                             insertions);
1511         }
1512
1513         if (deletions || insertions == 0) {
1514                 strbuf_addf(&sb,
1515                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1516                             deletions);
1517         }
1518         strbuf_addch(&sb, '\n');
1519         ret = fputs(sb.buf, fp);
1520         strbuf_release(&sb);
1521         return ret;
1522 }
1523
1524 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1525 {
1526         int i, len, add, del, adds = 0, dels = 0;
1527         uintmax_t max_change = 0, max_len = 0;
1528         int total_files = data->nr, count;
1529         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1530         const char *reset, *add_c, *del_c;
1531         const char *line_prefix = "";
1532         int extra_shown = 0;
1533
1534         if (data->nr == 0)
1535                 return;
1536
1537         line_prefix = diff_line_prefix(options);
1538         count = options->stat_count ? options->stat_count : data->nr;
1539
1540         reset = diff_get_color_opt(options, DIFF_RESET);
1541         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1542         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1543
1544         /*
1545          * Find the longest filename and max number of changes
1546          */
1547         for (i = 0; (i < count) && (i < data->nr); i++) {
1548                 struct diffstat_file *file = data->files[i];
1549                 uintmax_t change = file->added + file->deleted;
1550
1551                 if (!file->is_interesting && (change == 0)) {
1552                         count++; /* not shown == room for one more */
1553                         continue;
1554                 }
1555                 fill_print_name(file);
1556                 len = strlen(file->print_name);
1557                 if (max_len < len)
1558                         max_len = len;
1559
1560                 if (file->is_unmerged) {
1561                         /* "Unmerged" is 8 characters */
1562                         bin_width = bin_width < 8 ? 8 : bin_width;
1563                         continue;
1564                 }
1565                 if (file->is_binary) {
1566                         /* "Bin XXX -> YYY bytes" */
1567                         int w = 14 + decimal_width(file->added)
1568                                 + decimal_width(file->deleted);
1569                         bin_width = bin_width < w ? w : bin_width;
1570                         /* Display change counts aligned with "Bin" */
1571                         number_width = 3;
1572                         continue;
1573                 }
1574
1575                 if (max_change < change)
1576                         max_change = change;
1577         }
1578         count = i; /* where we can stop scanning in data->files[] */
1579
1580         /*
1581          * We have width = stat_width or term_columns() columns total.
1582          * We want a maximum of min(max_len, stat_name_width) for the name part.
1583          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1584          * We also need 1 for " " and 4 + decimal_width(max_change)
1585          * for " | NNNN " and one the empty column at the end, altogether
1586          * 6 + decimal_width(max_change).
1587          *
1588          * If there's not enough space, we will use the smaller of
1589          * stat_name_width (if set) and 5/8*width for the filename,
1590          * and the rest for constant elements + graph part, but no more
1591          * than stat_graph_width for the graph part.
1592          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1593          * for the standard terminal size).
1594          *
1595          * In other words: stat_width limits the maximum width, and
1596          * stat_name_width fixes the maximum width of the filename,
1597          * and is also used to divide available columns if there
1598          * aren't enough.
1599          *
1600          * Binary files are displayed with "Bin XXX -> YYY bytes"
1601          * instead of the change count and graph. This part is treated
1602          * similarly to the graph part, except that it is not
1603          * "scaled". If total width is too small to accommodate the
1604          * guaranteed minimum width of the filename part and the
1605          * separators and this message, this message will "overflow"
1606          * making the line longer than the maximum width.
1607          */
1608
1609         if (options->stat_width == -1)
1610                 width = term_columns() - options->output_prefix_length;
1611         else
1612                 width = options->stat_width ? options->stat_width : 80;
1613         number_width = decimal_width(max_change) > number_width ?
1614                 decimal_width(max_change) : number_width;
1615
1616         if (options->stat_graph_width == -1)
1617                 options->stat_graph_width = diff_stat_graph_width;
1618
1619         /*
1620          * Guarantee 3/8*16==6 for the graph part
1621          * and 5/8*16==10 for the filename part
1622          */
1623         if (width < 16 + 6 + number_width)
1624                 width = 16 + 6 + number_width;
1625
1626         /*
1627          * First assign sizes that are wanted, ignoring available width.
1628          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1629          * starting from "XXX" should fit in graph_width.
1630          */
1631         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1632         if (options->stat_graph_width &&
1633             options->stat_graph_width < graph_width)
1634                 graph_width = options->stat_graph_width;
1635
1636         name_width = (options->stat_name_width > 0 &&
1637                       options->stat_name_width < max_len) ?
1638                 options->stat_name_width : max_len;
1639
1640         /*
1641          * Adjust adjustable widths not to exceed maximum width
1642          */
1643         if (name_width + number_width + 6 + graph_width > width) {
1644                 if (graph_width > width * 3/8 - number_width - 6) {
1645                         graph_width = width * 3/8 - number_width - 6;
1646                         if (graph_width < 6)
1647                                 graph_width = 6;
1648                 }
1649
1650                 if (options->stat_graph_width &&
1651                     graph_width > options->stat_graph_width)
1652                         graph_width = options->stat_graph_width;
1653                 if (name_width > width - number_width - 6 - graph_width)
1654                         name_width = width - number_width - 6 - graph_width;
1655                 else
1656                         graph_width = width - number_width - 6 - name_width;
1657         }
1658
1659         /*
1660          * From here name_width is the width of the name area,
1661          * and graph_width is the width of the graph area.
1662          * max_change is used to scale graph properly.
1663          */
1664         for (i = 0; i < count; i++) {
1665                 const char *prefix = "";
1666                 struct diffstat_file *file = data->files[i];
1667                 char *name = file->print_name;
1668                 uintmax_t added = file->added;
1669                 uintmax_t deleted = file->deleted;
1670                 int name_len;
1671
1672                 if (!file->is_interesting && (added + deleted == 0))
1673                         continue;
1674
1675                 /*
1676                  * "scale" the filename
1677                  */
1678                 len = name_width;
1679                 name_len = strlen(name);
1680                 if (name_width < name_len) {
1681                         char *slash;
1682                         prefix = "...";
1683                         len -= 3;
1684                         name += name_len - len;
1685                         slash = strchr(name, '/');
1686                         if (slash)
1687                                 name = slash;
1688                 }
1689
1690                 if (file->is_binary) {
1691                         fprintf(options->file, "%s", line_prefix);
1692                         show_name(options->file, prefix, name, len);
1693                         fprintf(options->file, " %*s", number_width, "Bin");
1694                         if (!added && !deleted) {
1695                                 putc('\n', options->file);
1696                                 continue;
1697                         }
1698                         fprintf(options->file, " %s%"PRIuMAX"%s",
1699                                 del_c, deleted, reset);
1700                         fprintf(options->file, " -> ");
1701                         fprintf(options->file, "%s%"PRIuMAX"%s",
1702                                 add_c, added, reset);
1703                         fprintf(options->file, " bytes");
1704                         fprintf(options->file, "\n");
1705                         continue;
1706                 }
1707                 else if (file->is_unmerged) {
1708                         fprintf(options->file, "%s", line_prefix);
1709                         show_name(options->file, prefix, name, len);
1710                         fprintf(options->file, " Unmerged\n");
1711                         continue;
1712                 }
1713
1714                 /*
1715                  * scale the add/delete
1716                  */
1717                 add = added;
1718                 del = deleted;
1719
1720                 if (graph_width <= max_change) {
1721                         int total = scale_linear(add + del, graph_width, max_change);
1722                         if (total < 2 && add && del)
1723                                 /* width >= 2 due to the sanity check */
1724                                 total = 2;
1725                         if (add < del) {
1726                                 add = scale_linear(add, graph_width, max_change);
1727                                 del = total - add;
1728                         } else {
1729                                 del = scale_linear(del, graph_width, max_change);
1730                                 add = total - del;
1731                         }
1732                 }
1733                 fprintf(options->file, "%s", line_prefix);
1734                 show_name(options->file, prefix, name, len);
1735                 fprintf(options->file, " %*"PRIuMAX"%s",
1736                         number_width, added + deleted,
1737                         added + deleted ? " " : "");
1738                 show_graph(options->file, '+', add, add_c, reset);
1739                 show_graph(options->file, '-', del, del_c, reset);
1740                 fprintf(options->file, "\n");
1741         }
1742
1743         for (i = 0; i < data->nr; i++) {
1744                 struct diffstat_file *file = data->files[i];
1745                 uintmax_t added = file->added;
1746                 uintmax_t deleted = file->deleted;
1747
1748                 if (file->is_unmerged ||
1749                     (!file->is_interesting && (added + deleted == 0))) {
1750                         total_files--;
1751                         continue;
1752                 }
1753
1754                 if (!file->is_binary) {
1755                         adds += added;
1756                         dels += deleted;
1757                 }
1758                 if (i < count)
1759                         continue;
1760                 if (!extra_shown)
1761                         fprintf(options->file, "%s ...\n", line_prefix);
1762                 extra_shown = 1;
1763         }
1764         fprintf(options->file, "%s", line_prefix);
1765         print_stat_summary(options->file, total_files, adds, dels);
1766 }
1767
1768 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1769 {
1770         int i, adds = 0, dels = 0, total_files = data->nr;
1771
1772         if (data->nr == 0)
1773                 return;
1774
1775         for (i = 0; i < data->nr; i++) {
1776                 int added = data->files[i]->added;
1777                 int deleted= data->files[i]->deleted;
1778
1779                 if (data->files[i]->is_unmerged ||
1780                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1781                         total_files--;
1782                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1783                         adds += added;
1784                         dels += deleted;
1785                 }
1786         }
1787         fprintf(options->file, "%s", diff_line_prefix(options));
1788         print_stat_summary(options->file, total_files, adds, dels);
1789 }
1790
1791 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1792 {
1793         int i;
1794
1795         if (data->nr == 0)
1796                 return;
1797
1798         for (i = 0; i < data->nr; i++) {
1799                 struct diffstat_file *file = data->files[i];
1800
1801                 fprintf(options->file, "%s", diff_line_prefix(options));
1802
1803                 if (file->is_binary)
1804                         fprintf(options->file, "-\t-\t");
1805                 else
1806                         fprintf(options->file,
1807                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1808                                 file->added, file->deleted);
1809                 if (options->line_termination) {
1810                         fill_print_name(file);
1811                         if (!file->is_renamed)
1812                                 write_name_quoted(file->name, options->file,
1813                                                   options->line_termination);
1814                         else {
1815                                 fputs(file->print_name, options->file);
1816                                 putc(options->line_termination, options->file);
1817                         }
1818                 } else {
1819                         if (file->is_renamed) {
1820                                 putc('\0', options->file);
1821                                 write_name_quoted(file->from_name, options->file, '\0');
1822                         }
1823                         write_name_quoted(file->name, options->file, '\0');
1824                 }
1825         }
1826 }
1827
1828 struct dirstat_file {
1829         const char *name;
1830         unsigned long changed;
1831 };
1832
1833 struct dirstat_dir {
1834         struct dirstat_file *files;
1835         int alloc, nr, permille, cumulative;
1836 };
1837
1838 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1839                 unsigned long changed, const char *base, int baselen)
1840 {
1841         unsigned long this_dir = 0;
1842         unsigned int sources = 0;
1843         const char *line_prefix = diff_line_prefix(opt);
1844
1845         while (dir->nr) {
1846                 struct dirstat_file *f = dir->files;
1847                 int namelen = strlen(f->name);
1848                 unsigned long this;
1849                 char *slash;
1850
1851                 if (namelen < baselen)
1852                         break;
1853                 if (memcmp(f->name, base, baselen))
1854                         break;
1855                 slash = strchr(f->name + baselen, '/');
1856                 if (slash) {
1857                         int newbaselen = slash + 1 - f->name;
1858                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1859                         sources++;
1860                 } else {
1861                         this = f->changed;
1862                         dir->files++;
1863                         dir->nr--;
1864                         sources += 2;
1865                 }
1866                 this_dir += this;
1867         }
1868
1869         /*
1870          * We don't report dirstat's for
1871          *  - the top level
1872          *  - or cases where everything came from a single directory
1873          *    under this directory (sources == 1).
1874          */
1875         if (baselen && sources != 1) {
1876                 if (this_dir) {
1877                         int permille = this_dir * 1000 / changed;
1878                         if (permille >= dir->permille) {
1879                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1880                                         permille / 10, permille % 10, baselen, base);
1881                                 if (!dir->cumulative)
1882                                         return 0;
1883                         }
1884                 }
1885         }
1886         return this_dir;
1887 }
1888
1889 static int dirstat_compare(const void *_a, const void *_b)
1890 {
1891         const struct dirstat_file *a = _a;
1892         const struct dirstat_file *b = _b;
1893         return strcmp(a->name, b->name);
1894 }
1895
1896 static void show_dirstat(struct diff_options *options)
1897 {
1898         int i;
1899         unsigned long changed;
1900         struct dirstat_dir dir;
1901         struct diff_queue_struct *q = &diff_queued_diff;
1902
1903         dir.files = NULL;
1904         dir.alloc = 0;
1905         dir.nr = 0;
1906         dir.permille = options->dirstat_permille;
1907         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1908
1909         changed = 0;
1910         for (i = 0; i < q->nr; i++) {
1911                 struct diff_filepair *p = q->queue[i];
1912                 const char *name;
1913                 unsigned long copied, added, damage;
1914                 int content_changed;
1915
1916                 name = p->two->path ? p->two->path : p->one->path;
1917
1918                 if (p->one->sha1_valid && p->two->sha1_valid)
1919                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1920                 else
1921                         content_changed = 1;
1922
1923                 if (!content_changed) {
1924                         /*
1925                          * The SHA1 has not changed, so pre-/post-content is
1926                          * identical. We can therefore skip looking at the
1927                          * file contents altogether.
1928                          */
1929                         damage = 0;
1930                         goto found_damage;
1931                 }
1932
1933                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1934                         /*
1935                          * In --dirstat-by-file mode, we don't really need to
1936                          * look at the actual file contents at all.
1937                          * The fact that the SHA1 changed is enough for us to
1938                          * add this file to the list of results
1939                          * (with each file contributing equal damage).
1940                          */
1941                         damage = 1;
1942                         goto found_damage;
1943                 }
1944
1945                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1946                         diff_populate_filespec(p->one, 0);
1947                         diff_populate_filespec(p->two, 0);
1948                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1949                                                &copied, &added);
1950                         diff_free_filespec_data(p->one);
1951                         diff_free_filespec_data(p->two);
1952                 } else if (DIFF_FILE_VALID(p->one)) {
1953                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1954                         copied = added = 0;
1955                         diff_free_filespec_data(p->one);
1956                 } else if (DIFF_FILE_VALID(p->two)) {
1957                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1958                         copied = 0;
1959                         added = p->two->size;
1960                         diff_free_filespec_data(p->two);
1961                 } else
1962                         continue;
1963
1964                 /*
1965                  * Original minus copied is the removed material,
1966                  * added is the new material.  They are both damages
1967                  * made to the preimage.
1968                  * If the resulting damage is zero, we know that
1969                  * diffcore_count_changes() considers the two entries to
1970                  * be identical, but since content_changed is true, we
1971                  * know that there must have been _some_ kind of change,
1972                  * so we force all entries to have damage > 0.
1973                  */
1974                 damage = (p->one->size - copied) + added;
1975                 if (!damage)
1976                         damage = 1;
1977
1978 found_damage:
1979                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1980                 dir.files[dir.nr].name = name;
1981                 dir.files[dir.nr].changed = damage;
1982                 changed += damage;
1983                 dir.nr++;
1984         }
1985
1986         /* This can happen even with many files, if everything was renames */
1987         if (!changed)
1988                 return;
1989
1990         /* Show all directories with more than x% of the changes */
1991         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1992         gather_dirstat(options, &dir, changed, "", 0);
1993 }
1994
1995 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1996 {
1997         int i;
1998         unsigned long changed;
1999         struct dirstat_dir dir;
2000
2001         if (data->nr == 0)
2002                 return;
2003
2004         dir.files = NULL;
2005         dir.alloc = 0;
2006         dir.nr = 0;
2007         dir.permille = options->dirstat_permille;
2008         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2009
2010         changed = 0;
2011         for (i = 0; i < data->nr; i++) {
2012                 struct diffstat_file *file = data->files[i];
2013                 unsigned long damage = file->added + file->deleted;
2014                 if (file->is_binary)
2015                         /*
2016                          * binary files counts bytes, not lines. Must find some
2017                          * way to normalize binary bytes vs. textual lines.
2018                          * The following heuristic assumes that there are 64
2019                          * bytes per "line".
2020                          * This is stupid and ugly, but very cheap...
2021                          */
2022                         damage = (damage + 63) / 64;
2023                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2024                 dir.files[dir.nr].name = file->name;
2025                 dir.files[dir.nr].changed = damage;
2026                 changed += damage;
2027                 dir.nr++;
2028         }
2029
2030         /* This can happen even with many files, if everything was renames */
2031         if (!changed)
2032                 return;
2033
2034         /* Show all directories with more than x% of the changes */
2035         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2036         gather_dirstat(options, &dir, changed, "", 0);
2037 }
2038
2039 static void free_diffstat_info(struct diffstat_t *diffstat)
2040 {
2041         int i;
2042         for (i = 0; i < diffstat->nr; i++) {
2043                 struct diffstat_file *f = diffstat->files[i];
2044                 if (f->name != f->print_name)
2045                         free(f->print_name);
2046                 free(f->name);
2047                 free(f->from_name);
2048                 free(f);
2049         }
2050         free(diffstat->files);
2051 }
2052
2053 struct checkdiff_t {
2054         const char *filename;
2055         int lineno;
2056         int conflict_marker_size;
2057         struct diff_options *o;
2058         unsigned ws_rule;
2059         unsigned status;
2060 };
2061
2062 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2063 {
2064         char firstchar;
2065         int cnt;
2066
2067         if (len < marker_size + 1)
2068                 return 0;
2069         firstchar = line[0];
2070         switch (firstchar) {
2071         case '=': case '>': case '<': case '|':
2072                 break;
2073         default:
2074                 return 0;
2075         }
2076         for (cnt = 1; cnt < marker_size; cnt++)
2077                 if (line[cnt] != firstchar)
2078                         return 0;
2079         /* line[1] thru line[marker_size-1] are same as firstchar */
2080         if (len < marker_size + 1 || !isspace(line[marker_size]))
2081                 return 0;
2082         return 1;
2083 }
2084
2085 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2086 {
2087         struct checkdiff_t *data = priv;
2088         int marker_size = data->conflict_marker_size;
2089         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2090         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2091         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2092         char *err;
2093         const char *line_prefix;
2094
2095         assert(data->o);
2096         line_prefix = diff_line_prefix(data->o);
2097
2098         if (line[0] == '+') {
2099                 unsigned bad;
2100                 data->lineno++;
2101                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2102                         data->status |= 1;
2103                         fprintf(data->o->file,
2104                                 "%s%s:%d: leftover conflict marker\n",
2105                                 line_prefix, data->filename, data->lineno);
2106                 }
2107                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2108                 if (!bad)
2109                         return;
2110                 data->status |= bad;
2111                 err = whitespace_error_string(bad);
2112                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2113                         line_prefix, data->filename, data->lineno, err);
2114                 free(err);
2115                 emit_line(data->o, set, reset, line, 1);
2116                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2117                               data->o->file, set, reset, ws);
2118         } else if (line[0] == ' ') {
2119                 data->lineno++;
2120         } else if (line[0] == '@') {
2121                 char *plus = strchr(line, '+');
2122                 if (plus)
2123                         data->lineno = strtol(plus, NULL, 10) - 1;
2124                 else
2125                         die("invalid diff");
2126         }
2127 }
2128
2129 static unsigned char *deflate_it(char *data,
2130                                  unsigned long size,
2131                                  unsigned long *result_size)
2132 {
2133         int bound;
2134         unsigned char *deflated;
2135         git_zstream stream;
2136
2137         git_deflate_init(&stream, zlib_compression_level);
2138         bound = git_deflate_bound(&stream, size);
2139         deflated = xmalloc(bound);
2140         stream.next_out = deflated;
2141         stream.avail_out = bound;
2142
2143         stream.next_in = (unsigned char *)data;
2144         stream.avail_in = size;
2145         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2146                 ; /* nothing */
2147         git_deflate_end(&stream);
2148         *result_size = stream.total_out;
2149         return deflated;
2150 }
2151
2152 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2153                                   const char *prefix)
2154 {
2155         void *cp;
2156         void *delta;
2157         void *deflated;
2158         void *data;
2159         unsigned long orig_size;
2160         unsigned long delta_size;
2161         unsigned long deflate_size;
2162         unsigned long data_size;
2163
2164         /* We could do deflated delta, or we could do just deflated two,
2165          * whichever is smaller.
2166          */
2167         delta = NULL;
2168         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2169         if (one->size && two->size) {
2170                 delta = diff_delta(one->ptr, one->size,
2171                                    two->ptr, two->size,
2172                                    &delta_size, deflate_size);
2173                 if (delta) {
2174                         void *to_free = delta;
2175                         orig_size = delta_size;
2176                         delta = deflate_it(delta, delta_size, &delta_size);
2177                         free(to_free);
2178                 }
2179         }
2180
2181         if (delta && delta_size < deflate_size) {
2182                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2183                 free(deflated);
2184                 data = delta;
2185                 data_size = delta_size;
2186         }
2187         else {
2188                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2189                 free(delta);
2190                 data = deflated;
2191                 data_size = deflate_size;
2192         }
2193
2194         /* emit data encoded in base85 */
2195         cp = data;
2196         while (data_size) {
2197                 int bytes = (52 < data_size) ? 52 : data_size;
2198                 char line[70];
2199                 data_size -= bytes;
2200                 if (bytes <= 26)
2201                         line[0] = bytes + 'A' - 1;
2202                 else
2203                         line[0] = bytes - 26 + 'a' - 1;
2204                 encode_85(line + 1, cp, bytes);
2205                 cp = (char *) cp + bytes;
2206                 fprintf(file, "%s", prefix);
2207                 fputs(line, file);
2208                 fputc('\n', file);
2209         }
2210         fprintf(file, "%s\n", prefix);
2211         free(data);
2212 }
2213
2214 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2215                              const char *prefix)
2216 {
2217         fprintf(file, "%sGIT binary patch\n", prefix);
2218         emit_binary_diff_body(file, one, two, prefix);
2219         emit_binary_diff_body(file, two, one, prefix);
2220 }
2221
2222 int diff_filespec_is_binary(struct diff_filespec *one)
2223 {
2224         if (one->is_binary == -1) {
2225                 diff_filespec_load_driver(one);
2226                 if (one->driver->binary != -1)
2227                         one->is_binary = one->driver->binary;
2228                 else {
2229                         if (!one->data && DIFF_FILE_VALID(one))
2230                                 diff_populate_filespec(one, CHECK_BINARY);
2231                         if (one->is_binary == -1 && one->data)
2232                                 one->is_binary = buffer_is_binary(one->data,
2233                                                 one->size);
2234                         if (one->is_binary == -1)
2235                                 one->is_binary = 0;
2236                 }
2237         }
2238         return one->is_binary;
2239 }
2240
2241 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2242 {
2243         diff_filespec_load_driver(one);
2244         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2245 }
2246
2247 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2248 {
2249         if (!options->a_prefix)
2250                 options->a_prefix = a;
2251         if (!options->b_prefix)
2252                 options->b_prefix = b;
2253 }
2254
2255 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2256 {
2257         if (!DIFF_FILE_VALID(one))
2258                 return NULL;
2259
2260         diff_filespec_load_driver(one);
2261         return userdiff_get_textconv(one->driver);
2262 }
2263
2264 static void builtin_diff(const char *name_a,
2265                          const char *name_b,
2266                          struct diff_filespec *one,
2267                          struct diff_filespec *two,
2268                          const char *xfrm_msg,
2269                          int must_show_header,
2270                          struct diff_options *o,
2271                          int complete_rewrite)
2272 {
2273         mmfile_t mf1, mf2;
2274         const char *lbl[2];
2275         char *a_one, *b_two;
2276         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2277         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2278         const char *a_prefix, *b_prefix;
2279         struct userdiff_driver *textconv_one = NULL;
2280         struct userdiff_driver *textconv_two = NULL;
2281         struct strbuf header = STRBUF_INIT;
2282         const char *line_prefix = diff_line_prefix(o);
2283
2284         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2285                         (!one->mode || S_ISGITLINK(one->mode)) &&
2286                         (!two->mode || S_ISGITLINK(two->mode))) {
2287                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2288                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2289                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2290                                 line_prefix,
2291                                 one->sha1, two->sha1, two->dirty_submodule,
2292                                 meta, del, add, reset);
2293                 return;
2294         }
2295
2296         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2297                 textconv_one = get_textconv(one);
2298                 textconv_two = get_textconv(two);
2299         }
2300
2301         diff_set_mnemonic_prefix(o, "a/", "b/");
2302         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2303                 a_prefix = o->b_prefix;
2304                 b_prefix = o->a_prefix;
2305         } else {
2306                 a_prefix = o->a_prefix;
2307                 b_prefix = o->b_prefix;
2308         }
2309
2310         /* Never use a non-valid filename anywhere if at all possible */
2311         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2312         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2313
2314         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2315         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2316         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2317         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2318         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2319         if (lbl[0][0] == '/') {
2320                 /* /dev/null */
2321                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2322                 if (xfrm_msg)
2323                         strbuf_addstr(&header, xfrm_msg);
2324                 must_show_header = 1;
2325         }
2326         else if (lbl[1][0] == '/') {
2327                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2328                 if (xfrm_msg)
2329                         strbuf_addstr(&header, xfrm_msg);
2330                 must_show_header = 1;
2331         }
2332         else {
2333                 if (one->mode != two->mode) {
2334                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2335                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2336                         must_show_header = 1;
2337                 }
2338                 if (xfrm_msg)
2339                         strbuf_addstr(&header, xfrm_msg);
2340
2341                 /*
2342                  * we do not run diff between different kind
2343                  * of objects.
2344                  */
2345                 if ((one->mode ^ two->mode) & S_IFMT)
2346                         goto free_ab_and_return;
2347                 if (complete_rewrite &&
2348                     (textconv_one || !diff_filespec_is_binary(one)) &&
2349                     (textconv_two || !diff_filespec_is_binary(two))) {
2350                         fprintf(o->file, "%s", header.buf);
2351                         strbuf_reset(&header);
2352                         emit_rewrite_diff(name_a, name_b, one, two,
2353                                                 textconv_one, textconv_two, o);
2354                         o->found_changes = 1;
2355                         goto free_ab_and_return;
2356                 }
2357         }
2358
2359         if (o->irreversible_delete && lbl[1][0] == '/') {
2360                 fprintf(o->file, "%s", header.buf);
2361                 strbuf_reset(&header);
2362                 goto free_ab_and_return;
2363         } else if (!DIFF_OPT_TST(o, TEXT) &&
2364             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2365               (!textconv_two && diff_filespec_is_binary(two)) )) {
2366                 if (!one->data && !two->data &&
2367                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2368                     !DIFF_OPT_TST(o, BINARY)) {
2369                         if (!hashcmp(one->sha1, two->sha1)) {
2370                                 if (must_show_header)
2371                                         fprintf(o->file, "%s", header.buf);
2372                                 goto free_ab_and_return;
2373                         }
2374                         fprintf(o->file, "%s", header.buf);
2375                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2376                                 line_prefix, lbl[0], lbl[1]);
2377                         goto free_ab_and_return;
2378                 }
2379                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2380                         die("unable to read files to diff");
2381                 /* Quite common confusing case */
2382                 if (mf1.size == mf2.size &&
2383                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2384                         if (must_show_header)
2385                                 fprintf(o->file, "%s", header.buf);
2386                         goto free_ab_and_return;
2387                 }
2388                 fprintf(o->file, "%s", header.buf);
2389                 strbuf_reset(&header);
2390                 if (DIFF_OPT_TST(o, BINARY))
2391                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2392                 else
2393                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2394                                 line_prefix, lbl[0], lbl[1]);
2395                 o->found_changes = 1;
2396         } else {
2397                 /* Crazy xdl interfaces.. */
2398                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2399                 const char *v;
2400                 xpparam_t xpp;
2401                 xdemitconf_t xecfg;
2402                 struct emit_callback ecbdata;
2403                 const struct userdiff_funcname *pe;
2404
2405                 if (must_show_header) {
2406                         fprintf(o->file, "%s", header.buf);
2407                         strbuf_reset(&header);
2408                 }
2409
2410                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2411                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2412
2413                 pe = diff_funcname_pattern(one);
2414                 if (!pe)
2415                         pe = diff_funcname_pattern(two);
2416
2417                 memset(&xpp, 0, sizeof(xpp));
2418                 memset(&xecfg, 0, sizeof(xecfg));
2419                 memset(&ecbdata, 0, sizeof(ecbdata));
2420                 ecbdata.label_path = lbl;
2421                 ecbdata.color_diff = want_color(o->use_color);
2422                 ecbdata.found_changesp = &o->found_changes;
2423                 ecbdata.ws_rule = whitespace_rule(name_b);
2424                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2425                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2426                 ecbdata.opt = o;
2427                 ecbdata.header = header.len ? &header : NULL;
2428                 xpp.flags = o->xdl_opts;
2429                 xecfg.ctxlen = o->context;
2430                 xecfg.interhunkctxlen = o->interhunkcontext;
2431                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2432                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2433                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2434                 if (pe)
2435                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2436                 if (!diffopts)
2437                         ;
2438                 else if (skip_prefix(diffopts, "--unified=", &v))
2439                         xecfg.ctxlen = strtoul(v, NULL, 10);
2440                 else if (skip_prefix(diffopts, "-u", &v))
2441                         xecfg.ctxlen = strtoul(v, NULL, 10);
2442                 if (o->word_diff)
2443                         init_diff_words_data(&ecbdata, o, one, two);
2444                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2445                               &xpp, &xecfg);
2446                 if (o->word_diff)
2447                         free_diff_words_data(&ecbdata);
2448                 if (textconv_one)
2449                         free(mf1.ptr);
2450                 if (textconv_two)
2451                         free(mf2.ptr);
2452                 xdiff_clear_find_func(&xecfg);
2453         }
2454
2455  free_ab_and_return:
2456         strbuf_release(&header);
2457         diff_free_filespec_data(one);
2458         diff_free_filespec_data(two);
2459         free(a_one);
2460         free(b_two);
2461         return;
2462 }
2463
2464 static void builtin_diffstat(const char *name_a, const char *name_b,
2465                              struct diff_filespec *one,
2466                              struct diff_filespec *two,
2467                              struct diffstat_t *diffstat,
2468                              struct diff_options *o,
2469                              struct diff_filepair *p)
2470 {
2471         mmfile_t mf1, mf2;
2472         struct diffstat_file *data;
2473         int same_contents;
2474         int complete_rewrite = 0;
2475
2476         if (!DIFF_PAIR_UNMERGED(p)) {
2477                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2478                         complete_rewrite = 1;
2479         }
2480
2481         data = diffstat_add(diffstat, name_a, name_b);
2482         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2483
2484         if (!one || !two) {
2485                 data->is_unmerged = 1;
2486                 return;
2487         }
2488
2489         same_contents = !hashcmp(one->sha1, two->sha1);
2490
2491         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2492                 data->is_binary = 1;
2493                 if (same_contents) {
2494                         data->added = 0;
2495                         data->deleted = 0;
2496                 } else {
2497                         data->added = diff_filespec_size(two);
2498                         data->deleted = diff_filespec_size(one);
2499                 }
2500         }
2501
2502         else if (complete_rewrite) {
2503                 diff_populate_filespec(one, 0);
2504                 diff_populate_filespec(two, 0);
2505                 data->deleted = count_lines(one->data, one->size);
2506                 data->added = count_lines(two->data, two->size);
2507         }
2508
2509         else if (!same_contents) {
2510                 /* Crazy xdl interfaces.. */
2511                 xpparam_t xpp;
2512                 xdemitconf_t xecfg;
2513
2514                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2515                         die("unable to read files to diff");
2516
2517                 memset(&xpp, 0, sizeof(xpp));
2518                 memset(&xecfg, 0, sizeof(xecfg));
2519                 xpp.flags = o->xdl_opts;
2520                 xecfg.ctxlen = o->context;
2521                 xecfg.interhunkctxlen = o->interhunkcontext;
2522                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2523                               &xpp, &xecfg);
2524         }
2525
2526         diff_free_filespec_data(one);
2527         diff_free_filespec_data(two);
2528 }
2529
2530 static void builtin_checkdiff(const char *name_a, const char *name_b,
2531                               const char *attr_path,
2532                               struct diff_filespec *one,
2533                               struct diff_filespec *two,
2534                               struct diff_options *o)
2535 {
2536         mmfile_t mf1, mf2;
2537         struct checkdiff_t data;
2538
2539         if (!two)
2540                 return;
2541
2542         memset(&data, 0, sizeof(data));
2543         data.filename = name_b ? name_b : name_a;
2544         data.lineno = 0;
2545         data.o = o;
2546         data.ws_rule = whitespace_rule(attr_path);
2547         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2548
2549         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2550                 die("unable to read files to diff");
2551
2552         /*
2553          * All the other codepaths check both sides, but not checking
2554          * the "old" side here is deliberate.  We are checking the newly
2555          * introduced changes, and as long as the "new" side is text, we
2556          * can and should check what it introduces.
2557          */
2558         if (diff_filespec_is_binary(two))
2559                 goto free_and_return;
2560         else {
2561                 /* Crazy xdl interfaces.. */
2562                 xpparam_t xpp;
2563                 xdemitconf_t xecfg;
2564
2565                 memset(&xpp, 0, sizeof(xpp));
2566                 memset(&xecfg, 0, sizeof(xecfg));
2567                 xecfg.ctxlen = 1; /* at least one context line */
2568                 xpp.flags = 0;
2569                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2570                               &xpp, &xecfg);
2571
2572                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2573                         struct emit_callback ecbdata;
2574                         int blank_at_eof;
2575
2576                         ecbdata.ws_rule = data.ws_rule;
2577                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2578                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2579
2580                         if (blank_at_eof) {
2581                                 static char *err;
2582                                 if (!err)
2583                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2584                                 fprintf(o->file, "%s:%d: %s.\n",
2585                                         data.filename, blank_at_eof, err);
2586                                 data.status = 1; /* report errors */
2587                         }
2588                 }
2589         }
2590  free_and_return:
2591         diff_free_filespec_data(one);
2592         diff_free_filespec_data(two);
2593         if (data.status)
2594                 DIFF_OPT_SET(o, CHECK_FAILED);
2595 }
2596
2597 struct diff_filespec *alloc_filespec(const char *path)
2598 {
2599         int namelen = strlen(path);
2600         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2601
2602         memset(spec, 0, sizeof(*spec));
2603         spec->path = (char *)(spec + 1);
2604         memcpy(spec->path, path, namelen+1);
2605         spec->count = 1;
2606         spec->is_binary = -1;
2607         return spec;
2608 }
2609
2610 void free_filespec(struct diff_filespec *spec)
2611 {
2612         if (!--spec->count) {
2613                 diff_free_filespec_data(spec);
2614                 free(spec);
2615         }
2616 }
2617
2618 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2619                    int sha1_valid, unsigned short mode)
2620 {
2621         if (mode) {
2622                 spec->mode = canon_mode(mode);
2623                 hashcpy(spec->sha1, sha1);
2624                 spec->sha1_valid = sha1_valid;
2625         }
2626 }
2627
2628 /*
2629  * Given a name and sha1 pair, if the index tells us the file in
2630  * the work tree has that object contents, return true, so that
2631  * prepare_temp_file() does not have to inflate and extract.
2632  */
2633 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2634 {
2635         const struct cache_entry *ce;
2636         struct stat st;
2637         int pos, len;
2638
2639         /*
2640          * We do not read the cache ourselves here, because the
2641          * benchmark with my previous version that always reads cache
2642          * shows that it makes things worse for diff-tree comparing
2643          * two linux-2.6 kernel trees in an already checked out work
2644          * tree.  This is because most diff-tree comparisons deal with
2645          * only a small number of files, while reading the cache is
2646          * expensive for a large project, and its cost outweighs the
2647          * savings we get by not inflating the object to a temporary
2648          * file.  Practically, this code only helps when we are used
2649          * by diff-cache --cached, which does read the cache before
2650          * calling us.
2651          */
2652         if (!active_cache)
2653                 return 0;
2654
2655         /* We want to avoid the working directory if our caller
2656          * doesn't need the data in a normal file, this system
2657          * is rather slow with its stat/open/mmap/close syscalls,
2658          * and the object is contained in a pack file.  The pack
2659          * is probably already open and will be faster to obtain
2660          * the data through than the working directory.  Loose
2661          * objects however would tend to be slower as they need
2662          * to be individually opened and inflated.
2663          */
2664         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2665                 return 0;
2666
2667         len = strlen(name);
2668         pos = cache_name_pos(name, len);
2669         if (pos < 0)
2670                 return 0;
2671         ce = active_cache[pos];
2672
2673         /*
2674          * This is not the sha1 we are looking for, or
2675          * unreusable because it is not a regular file.
2676          */
2677         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2678                 return 0;
2679
2680         /*
2681          * If ce is marked as "assume unchanged", there is no
2682          * guarantee that work tree matches what we are looking for.
2683          */
2684         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2685                 return 0;
2686
2687         /*
2688          * If ce matches the file in the work tree, we can reuse it.
2689          */
2690         if (ce_uptodate(ce) ||
2691             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2692                 return 1;
2693
2694         return 0;
2695 }
2696
2697 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2698 {
2699         int len;
2700         char *data = xmalloc(100), *dirty = "";
2701
2702         /* Are we looking at the work tree? */
2703         if (s->dirty_submodule)
2704                 dirty = "-dirty";
2705
2706         len = snprintf(data, 100,
2707                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2708         s->data = data;
2709         s->size = len;
2710         s->should_free = 1;
2711         if (size_only) {
2712                 s->data = NULL;
2713                 free(data);
2714         }
2715         return 0;
2716 }
2717
2718 /*
2719  * While doing rename detection and pickaxe operation, we may need to
2720  * grab the data for the blob (or file) for our own in-core comparison.
2721  * diff_filespec has data and size fields for this purpose.
2722  */
2723 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2724 {
2725         int size_only = flags & CHECK_SIZE_ONLY;
2726         int err = 0;
2727         /*
2728          * demote FAIL to WARN to allow inspecting the situation
2729          * instead of refusing.
2730          */
2731         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2732                                     ? SAFE_CRLF_WARN
2733                                     : safe_crlf);
2734
2735         if (!DIFF_FILE_VALID(s))
2736                 die("internal error: asking to populate invalid file.");
2737         if (S_ISDIR(s->mode))
2738                 return -1;
2739
2740         if (s->data)
2741                 return 0;
2742
2743         if (size_only && 0 < s->size)
2744                 return 0;
2745
2746         if (S_ISGITLINK(s->mode))
2747                 return diff_populate_gitlink(s, size_only);
2748
2749         if (!s->sha1_valid ||
2750             reuse_worktree_file(s->path, s->sha1, 0)) {
2751                 struct strbuf buf = STRBUF_INIT;
2752                 struct stat st;
2753                 int fd;
2754
2755                 if (lstat(s->path, &st) < 0) {
2756                         if (errno == ENOENT) {
2757                         err_empty:
2758                                 err = -1;
2759                         empty:
2760                                 s->data = (char *)"";
2761                                 s->size = 0;
2762                                 return err;
2763                         }
2764                 }
2765                 s->size = xsize_t(st.st_size);
2766                 if (!s->size)
2767                         goto empty;
2768                 if (S_ISLNK(st.st_mode)) {
2769                         struct strbuf sb = STRBUF_INIT;
2770
2771                         if (strbuf_readlink(&sb, s->path, s->size))
2772                                 goto err_empty;
2773                         s->size = sb.len;
2774                         s->data = strbuf_detach(&sb, NULL);
2775                         s->should_free = 1;
2776                         return 0;
2777                 }
2778                 if (size_only)
2779                         return 0;
2780                 if ((flags & CHECK_BINARY) &&
2781                     s->size > big_file_threshold && s->is_binary == -1) {
2782                         s->is_binary = 1;
2783                         return 0;
2784                 }
2785                 fd = open(s->path, O_RDONLY);
2786                 if (fd < 0)
2787                         goto err_empty;
2788                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2789                 close(fd);
2790                 s->should_munmap = 1;
2791
2792                 /*
2793                  * Convert from working tree format to canonical git format
2794                  */
2795                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2796                         size_t size = 0;
2797                         munmap(s->data, s->size);
2798                         s->should_munmap = 0;
2799                         s->data = strbuf_detach(&buf, &size);
2800                         s->size = size;
2801                         s->should_free = 1;
2802                 }
2803         }
2804         else {
2805                 enum object_type type;
2806                 if (size_only || (flags & CHECK_BINARY)) {
2807                         type = sha1_object_info(s->sha1, &s->size);
2808                         if (type < 0)
2809                                 die("unable to read %s", sha1_to_hex(s->sha1));
2810                         if (size_only)
2811                                 return 0;
2812                         if (s->size > big_file_threshold && s->is_binary == -1) {
2813                                 s->is_binary = 1;
2814                                 return 0;
2815                         }
2816                 }
2817                 s->data = read_sha1_file(s->sha1, &type, &s->size);
2818                 if (!s->data)
2819                         die("unable to read %s", sha1_to_hex(s->sha1));
2820                 s->should_free = 1;
2821         }
2822         return 0;
2823 }
2824
2825 void diff_free_filespec_blob(struct diff_filespec *s)
2826 {
2827         if (s->should_free)
2828                 free(s->data);
2829         else if (s->should_munmap)
2830                 munmap(s->data, s->size);
2831
2832         if (s->should_free || s->should_munmap) {
2833                 s->should_free = s->should_munmap = 0;
2834                 s->data = NULL;
2835         }
2836 }
2837
2838 void diff_free_filespec_data(struct diff_filespec *s)
2839 {
2840         diff_free_filespec_blob(s);
2841         free(s->cnt_data);
2842         s->cnt_data = NULL;
2843 }
2844
2845 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2846                            void *blob,
2847                            unsigned long size,
2848                            const unsigned char *sha1,
2849                            int mode)
2850 {
2851         int fd;
2852         struct strbuf buf = STRBUF_INIT;
2853         struct strbuf template = STRBUF_INIT;
2854         char *path_dup = xstrdup(path);
2855         const char *base = basename(path_dup);
2856
2857         /* Generate "XXXXXX_basename.ext" */
2858         strbuf_addstr(&template, "XXXXXX_");
2859         strbuf_addstr(&template, base);
2860
2861         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2862                         strlen(base) + 1);
2863         if (fd < 0)
2864                 die_errno("unable to create temp-file");
2865         if (convert_to_working_tree(path,
2866                         (const char *)blob, (size_t)size, &buf)) {
2867                 blob = buf.buf;
2868                 size = buf.len;
2869         }
2870         if (write_in_full(fd, blob, size) != size)
2871                 die_errno("unable to write temp-file");
2872         close(fd);
2873         temp->name = temp->tmp_path;
2874         strcpy(temp->hex, sha1_to_hex(sha1));
2875         temp->hex[40] = 0;
2876         sprintf(temp->mode, "%06o", mode);
2877         strbuf_release(&buf);
2878         strbuf_release(&template);
2879         free(path_dup);
2880 }
2881
2882 static struct diff_tempfile *prepare_temp_file(const char *name,
2883                 struct diff_filespec *one)
2884 {
2885         struct diff_tempfile *temp = claim_diff_tempfile();
2886
2887         if (!DIFF_FILE_VALID(one)) {
2888         not_a_valid_file:
2889                 /* A '-' entry produces this for file-2, and
2890                  * a '+' entry produces this for file-1.
2891                  */
2892                 temp->name = "/dev/null";
2893                 strcpy(temp->hex, ".");
2894                 strcpy(temp->mode, ".");
2895                 return temp;
2896         }
2897
2898         if (!remove_tempfile_installed) {
2899                 atexit(remove_tempfile);
2900                 sigchain_push_common(remove_tempfile_on_signal);
2901                 remove_tempfile_installed = 1;
2902         }
2903
2904         if (!S_ISGITLINK(one->mode) &&
2905             (!one->sha1_valid ||
2906              reuse_worktree_file(name, one->sha1, 1))) {
2907                 struct stat st;
2908                 if (lstat(name, &st) < 0) {
2909                         if (errno == ENOENT)
2910                                 goto not_a_valid_file;
2911                         die_errno("stat(%s)", name);
2912                 }
2913                 if (S_ISLNK(st.st_mode)) {
2914                         struct strbuf sb = STRBUF_INIT;
2915                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2916                                 die_errno("readlink(%s)", name);
2917                         prep_temp_blob(name, temp, sb.buf, sb.len,
2918                                        (one->sha1_valid ?
2919                                         one->sha1 : null_sha1),
2920                                        (one->sha1_valid ?
2921                                         one->mode : S_IFLNK));
2922                         strbuf_release(&sb);
2923                 }
2924                 else {
2925                         /* we can borrow from the file in the work tree */
2926                         temp->name = name;
2927                         if (!one->sha1_valid)
2928                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2929                         else
2930                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2931                         /* Even though we may sometimes borrow the
2932                          * contents from the work tree, we always want
2933                          * one->mode.  mode is trustworthy even when
2934                          * !(one->sha1_valid), as long as
2935                          * DIFF_FILE_VALID(one).
2936                          */
2937                         sprintf(temp->mode, "%06o", one->mode);
2938                 }
2939                 return temp;
2940         }
2941         else {
2942                 if (diff_populate_filespec(one, 0))
2943                         die("cannot read data blob for %s", one->path);
2944                 prep_temp_blob(name, temp, one->data, one->size,
2945                                one->sha1, one->mode);
2946         }
2947         return temp;
2948 }
2949
2950 static void add_external_diff_name(struct argv_array *argv,
2951                                    const char *name,
2952                                    struct diff_filespec *df)
2953 {
2954         struct diff_tempfile *temp = prepare_temp_file(name, df);
2955         argv_array_push(argv, temp->name);
2956         argv_array_push(argv, temp->hex);
2957         argv_array_push(argv, temp->mode);
2958 }
2959
2960 /* An external diff command takes:
2961  *
2962  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2963  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2964  *
2965  */
2966 static void run_external_diff(const char *pgm,
2967                               const char *name,
2968                               const char *other,
2969                               struct diff_filespec *one,
2970                               struct diff_filespec *two,
2971                               const char *xfrm_msg,
2972                               int complete_rewrite,
2973                               struct diff_options *o)
2974 {
2975         struct argv_array argv = ARGV_ARRAY_INIT;
2976         struct argv_array env = ARGV_ARRAY_INIT;
2977         struct diff_queue_struct *q = &diff_queued_diff;
2978
2979         argv_array_push(&argv, pgm);
2980         argv_array_push(&argv, name);
2981
2982         if (one && two) {
2983                 add_external_diff_name(&argv, name, one);
2984                 if (!other)
2985                         add_external_diff_name(&argv, name, two);
2986                 else {
2987                         add_external_diff_name(&argv, other, two);
2988                         argv_array_push(&argv, other);
2989                         argv_array_push(&argv, xfrm_msg);
2990                 }
2991         }
2992
2993         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
2994         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2995
2996         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
2997                 die(_("external diff died, stopping at %s"), name);
2998
2999         remove_tempfile();
3000         argv_array_clear(&argv);
3001         argv_array_clear(&env);
3002 }
3003
3004 static int similarity_index(struct diff_filepair *p)
3005 {
3006         return p->score * 100 / MAX_SCORE;
3007 }
3008
3009 static void fill_metainfo(struct strbuf *msg,
3010                           const char *name,
3011                           const char *other,
3012                           struct diff_filespec *one,
3013                           struct diff_filespec *two,
3014                           struct diff_options *o,
3015                           struct diff_filepair *p,
3016                           int *must_show_header,
3017                           int use_color)
3018 {
3019         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3020         const char *reset = diff_get_color(use_color, DIFF_RESET);
3021         const char *line_prefix = diff_line_prefix(o);
3022
3023         *must_show_header = 1;
3024         strbuf_init(msg, PATH_MAX * 2 + 300);
3025         switch (p->status) {
3026         case DIFF_STATUS_COPIED:
3027                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3028                             line_prefix, set, similarity_index(p));
3029                 strbuf_addf(msg, "%s\n%s%scopy from ",
3030                             reset,  line_prefix, set);
3031                 quote_c_style(name, msg, NULL, 0);
3032                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3033                 quote_c_style(other, msg, NULL, 0);
3034                 strbuf_addf(msg, "%s\n", reset);
3035                 break;
3036         case DIFF_STATUS_RENAMED:
3037                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3038                             line_prefix, set, similarity_index(p));
3039                 strbuf_addf(msg, "%s\n%s%srename from ",
3040                             reset, line_prefix, set);
3041                 quote_c_style(name, msg, NULL, 0);
3042                 strbuf_addf(msg, "%s\n%s%srename to ",
3043                             reset, line_prefix, set);
3044                 quote_c_style(other, msg, NULL, 0);
3045                 strbuf_addf(msg, "%s\n", reset);
3046                 break;
3047         case DIFF_STATUS_MODIFIED:
3048                 if (p->score) {
3049                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3050                                     line_prefix,
3051                                     set, similarity_index(p), reset);
3052                         break;
3053                 }
3054                 /* fallthru */
3055         default:
3056                 *must_show_header = 0;
3057         }
3058         if (one && two && hashcmp(one->sha1, two->sha1)) {
3059                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3060
3061                 if (DIFF_OPT_TST(o, BINARY)) {
3062                         mmfile_t mf;
3063                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3064                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3065                                 abbrev = 40;
3066                 }
3067                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3068                             find_unique_abbrev(one->sha1, abbrev));
3069                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3070                 if (one->mode == two->mode)
3071                         strbuf_addf(msg, " %06o", one->mode);
3072                 strbuf_addf(msg, "%s\n", reset);
3073         }
3074 }
3075
3076 static void run_diff_cmd(const char *pgm,
3077                          const char *name,
3078                          const char *other,
3079                          const char *attr_path,
3080                          struct diff_filespec *one,
3081                          struct diff_filespec *two,
3082                          struct strbuf *msg,
3083                          struct diff_options *o,
3084                          struct diff_filepair *p)
3085 {
3086         const char *xfrm_msg = NULL;
3087         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3088         int must_show_header = 0;
3089
3090
3091         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3092                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3093                 if (drv && drv->external)
3094                         pgm = drv->external;
3095         }
3096
3097         if (msg) {
3098                 /*
3099                  * don't use colors when the header is intended for an
3100                  * external diff driver
3101                  */
3102                 fill_metainfo(msg, name, other, one, two, o, p,
3103                               &must_show_header,
3104                               want_color(o->use_color) && !pgm);
3105                 xfrm_msg = msg->len ? msg->buf : NULL;
3106         }
3107
3108         if (pgm) {
3109                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3110                                   complete_rewrite, o);
3111                 return;
3112         }
3113         if (one && two)
3114                 builtin_diff(name, other ? other : name,
3115                              one, two, xfrm_msg, must_show_header,
3116                              o, complete_rewrite);
3117         else
3118                 fprintf(o->file, "* Unmerged path %s\n", name);
3119 }
3120
3121 static void diff_fill_sha1_info(struct diff_filespec *one)
3122 {
3123         if (DIFF_FILE_VALID(one)) {
3124                 if (!one->sha1_valid) {
3125                         struct stat st;
3126                         if (one->is_stdin) {
3127                                 hashcpy(one->sha1, null_sha1);
3128                                 return;
3129                         }
3130                         if (lstat(one->path, &st) < 0)
3131                                 die_errno("stat '%s'", one->path);
3132                         if (index_path(one->sha1, one->path, &st, 0))
3133                                 die("cannot hash %s", one->path);
3134                 }
3135         }
3136         else
3137                 hashclr(one->sha1);
3138 }
3139
3140 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3141 {
3142         /* Strip the prefix but do not molest /dev/null and absolute paths */
3143         if (*namep && **namep != '/') {
3144                 *namep += prefix_length;
3145                 if (**namep == '/')
3146                         ++*namep;
3147         }
3148         if (*otherp && **otherp != '/') {
3149                 *otherp += prefix_length;
3150                 if (**otherp == '/')
3151                         ++*otherp;
3152         }
3153 }
3154
3155 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3156 {
3157         const char *pgm = external_diff();
3158         struct strbuf msg;
3159         struct diff_filespec *one = p->one;
3160         struct diff_filespec *two = p->two;
3161         const char *name;
3162         const char *other;
3163         const char *attr_path;
3164
3165         name  = p->one->path;
3166         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3167         attr_path = name;
3168         if (o->prefix_length)
3169                 strip_prefix(o->prefix_length, &name, &other);
3170
3171         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3172                 pgm = NULL;
3173
3174         if (DIFF_PAIR_UNMERGED(p)) {
3175                 run_diff_cmd(pgm, name, NULL, attr_path,
3176                              NULL, NULL, NULL, o, p);
3177                 return;
3178         }
3179
3180         diff_fill_sha1_info(one);
3181         diff_fill_sha1_info(two);
3182
3183         if (!pgm &&
3184             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3185             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3186                 /*
3187                  * a filepair that changes between file and symlink
3188                  * needs to be split into deletion and creation.
3189                  */
3190                 struct diff_filespec *null = alloc_filespec(two->path);
3191                 run_diff_cmd(NULL, name, other, attr_path,
3192                              one, null, &msg, o, p);
3193                 free(null);
3194                 strbuf_release(&msg);
3195
3196                 null = alloc_filespec(one->path);
3197                 run_diff_cmd(NULL, name, other, attr_path,
3198                              null, two, &msg, o, p);
3199                 free(null);
3200         }
3201         else
3202                 run_diff_cmd(pgm, name, other, attr_path,
3203                              one, two, &msg, o, p);
3204
3205         strbuf_release(&msg);
3206 }
3207
3208 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3209                          struct diffstat_t *diffstat)
3210 {
3211         const char *name;
3212         const char *other;
3213
3214         if (DIFF_PAIR_UNMERGED(p)) {
3215                 /* unmerged */
3216                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3217                 return;
3218         }
3219
3220         name = p->one->path;
3221         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3222
3223         if (o->prefix_length)
3224                 strip_prefix(o->prefix_length, &name, &other);
3225
3226         diff_fill_sha1_info(p->one);
3227         diff_fill_sha1_info(p->two);
3228
3229         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3230 }
3231
3232 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3233 {
3234         const char *name;
3235         const char *other;
3236         const char *attr_path;
3237
3238         if (DIFF_PAIR_UNMERGED(p)) {
3239                 /* unmerged */
3240                 return;
3241         }
3242
3243         name = p->one->path;
3244         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3245         attr_path = other ? other : name;
3246
3247         if (o->prefix_length)
3248                 strip_prefix(o->prefix_length, &name, &other);
3249
3250         diff_fill_sha1_info(p->one);
3251         diff_fill_sha1_info(p->two);
3252
3253         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3254 }
3255
3256 void diff_setup(struct diff_options *options)
3257 {
3258         memcpy(options, &default_diff_options, sizeof(*options));
3259
3260         options->file = stdout;
3261
3262         options->line_termination = '\n';
3263         options->break_opt = -1;
3264         options->rename_limit = -1;
3265         options->dirstat_permille = diff_dirstat_permille_default;
3266         options->context = diff_context_default;
3267         options->ws_error_highlight = WSEH_NEW;
3268         DIFF_OPT_SET(options, RENAME_EMPTY);
3269
3270         /* pathchange left =NULL by default */
3271         options->change = diff_change;
3272         options->add_remove = diff_addremove;
3273         options->use_color = diff_use_color_default;
3274         options->detect_rename = diff_detect_rename_default;
3275         options->xdl_opts |= diff_algorithm;
3276
3277         options->orderfile = diff_order_file_cfg;
3278
3279         if (diff_no_prefix) {
3280                 options->a_prefix = options->b_prefix = "";
3281         } else if (!diff_mnemonic_prefix) {
3282                 options->a_prefix = "a/";
3283                 options->b_prefix = "b/";
3284         }
3285 }
3286
3287 void diff_setup_done(struct diff_options *options)
3288 {
3289         int count = 0;
3290
3291         if (options->set_default)
3292                 options->set_default(options);
3293
3294         if (options->output_format & DIFF_FORMAT_NAME)
3295                 count++;
3296         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3297                 count++;
3298         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3299                 count++;
3300         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3301                 count++;
3302         if (count > 1)
3303                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3304
3305         /*
3306          * Most of the time we can say "there are changes"
3307          * only by checking if there are changed paths, but
3308          * --ignore-whitespace* options force us to look
3309          * inside contents.
3310          */
3311
3312         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3313             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3314             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3315                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3316         else
3317                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3318
3319         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3320                 options->detect_rename = DIFF_DETECT_COPY;
3321
3322         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3323                 options->prefix = NULL;
3324         if (options->prefix)
3325                 options->prefix_length = strlen(options->prefix);
3326         else
3327                 options->prefix_length = 0;
3328
3329         if (options->output_format & (DIFF_FORMAT_NAME |
3330                                       DIFF_FORMAT_NAME_STATUS |
3331                                       DIFF_FORMAT_CHECKDIFF |
3332                                       DIFF_FORMAT_NO_OUTPUT))
3333                 options->output_format &= ~(DIFF_FORMAT_RAW |
3334                                             DIFF_FORMAT_NUMSTAT |
3335                                             DIFF_FORMAT_DIFFSTAT |
3336                                             DIFF_FORMAT_SHORTSTAT |
3337                                             DIFF_FORMAT_DIRSTAT |
3338                                             DIFF_FORMAT_SUMMARY |
3339                                             DIFF_FORMAT_PATCH);
3340
3341         /*
3342          * These cases always need recursive; we do not drop caller-supplied
3343          * recursive bits for other formats here.
3344          */
3345         if (options->output_format & (DIFF_FORMAT_PATCH |
3346                                       DIFF_FORMAT_NUMSTAT |
3347                                       DIFF_FORMAT_DIFFSTAT |
3348                                       DIFF_FORMAT_SHORTSTAT |
3349                                       DIFF_FORMAT_DIRSTAT |
3350                                       DIFF_FORMAT_SUMMARY |
3351                                       DIFF_FORMAT_CHECKDIFF))
3352                 DIFF_OPT_SET(options, RECURSIVE);
3353         /*
3354          * Also pickaxe would not work very well if you do not say recursive
3355          */
3356         if (options->pickaxe)
3357                 DIFF_OPT_SET(options, RECURSIVE);
3358         /*
3359          * When patches are generated, submodules diffed against the work tree
3360          * must be checked for dirtiness too so it can be shown in the output
3361          */
3362         if (options->output_format & DIFF_FORMAT_PATCH)
3363                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3364
3365         if (options->detect_rename && options->rename_limit < 0)
3366                 options->rename_limit = diff_rename_limit_default;
3367         if (options->setup & DIFF_SETUP_USE_CACHE) {
3368                 if (!active_cache)
3369                         /* read-cache does not die even when it fails
3370                          * so it is safe for us to do this here.  Also
3371                          * it does not smudge active_cache or active_nr
3372                          * when it fails, so we do not have to worry about
3373                          * cleaning it up ourselves either.
3374                          */
3375                         read_cache();
3376         }
3377         if (options->abbrev <= 0 || 40 < options->abbrev)
3378                 options->abbrev = 40; /* full */
3379
3380         /*
3381          * It does not make sense to show the first hit we happened
3382          * to have found.  It does not make sense not to return with
3383          * exit code in such a case either.
3384          */
3385         if (DIFF_OPT_TST(options, QUICK)) {
3386                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3387                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3388         }
3389
3390         options->diff_path_counter = 0;
3391
3392         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3393                 die(_("--follow requires exactly one pathspec"));
3394 }
3395
3396 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3397 {
3398         char c, *eq;
3399         int len;
3400
3401         if (*arg != '-')
3402                 return 0;
3403         c = *++arg;
3404         if (!c)
3405                 return 0;
3406         if (c == arg_short) {
3407                 c = *++arg;
3408                 if (!c)
3409                         return 1;
3410                 if (val && isdigit(c)) {
3411                         char *end;
3412                         int n = strtoul(arg, &end, 10);
3413                         if (*end)
3414                                 return 0;
3415                         *val = n;
3416                         return 1;
3417                 }
3418                 return 0;
3419         }
3420         if (c != '-')
3421                 return 0;
3422         arg++;
3423         eq = strchrnul(arg, '=');
3424         len = eq - arg;
3425         if (!len || strncmp(arg, arg_long, len))
3426                 return 0;
3427         if (*eq) {
3428                 int n;
3429                 char *end;
3430                 if (!isdigit(*++eq))
3431                         return 0;
3432                 n = strtoul(eq, &end, 10);
3433                 if (*end)
3434                         return 0;
3435                 *val = n;
3436         }
3437         return 1;
3438 }
3439
3440 static int diff_scoreopt_parse(const char *opt);
3441
3442 static inline int short_opt(char opt, const char **argv,
3443                             const char **optarg)
3444 {
3445         const char *arg = argv[0];
3446         if (arg[0] != '-' || arg[1] != opt)
3447                 return 0;
3448         if (arg[2] != '\0') {
3449                 *optarg = arg + 2;
3450                 return 1;
3451         }
3452         if (!argv[1])
3453                 die("Option '%c' requires a value", opt);
3454         *optarg = argv[1];
3455         return 2;
3456 }
3457
3458 int parse_long_opt(const char *opt, const char **argv,
3459                    const char **optarg)
3460 {
3461         const char *arg = argv[0];
3462         if (!skip_prefix(arg, "--", &arg))
3463                 return 0;
3464         if (!skip_prefix(arg, opt, &arg))
3465                 return 0;
3466         if (*arg == '=') { /* stuck form: --option=value */
3467                 *optarg = arg + 1;
3468                 return 1;
3469         }
3470         if (*arg != '\0')
3471                 return 0;
3472         /* separate form: --option value */
3473         if (!argv[1])
3474                 die("Option '--%s' requires a value", opt);
3475         *optarg = argv[1];
3476         return 2;
3477 }
3478
3479 static int stat_opt(struct diff_options *options, const char **av)
3480 {
3481         const char *arg = av[0];
3482         char *end;
3483         int width = options->stat_width;
3484         int name_width = options->stat_name_width;
3485         int graph_width = options->stat_graph_width;
3486         int count = options->stat_count;
3487         int argcount = 1;
3488
3489         if (!skip_prefix(arg, "--stat", &arg))
3490                 die("BUG: stat option does not begin with --stat: %s", arg);
3491         end = (char *)arg;
3492
3493         switch (*arg) {
3494         case '-':
3495                 if (skip_prefix(arg, "-width", &arg)) {
3496                         if (*arg == '=')
3497                                 width = strtoul(arg + 1, &end, 10);
3498                         else if (!*arg && !av[1])
3499                                 die("Option '--stat-width' requires a value");
3500                         else if (!*arg) {
3501                                 width = strtoul(av[1], &end, 10);
3502                                 argcount = 2;
3503                         }
3504                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3505                         if (*arg == '=')
3506                                 name_width = strtoul(arg + 1, &end, 10);
3507                         else if (!*arg && !av[1])
3508                                 die("Option '--stat-name-width' requires a value");
3509                         else if (!*arg) {
3510                                 name_width = strtoul(av[1], &end, 10);
3511                                 argcount = 2;
3512                         }
3513                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3514                         if (*arg == '=')
3515                                 graph_width = strtoul(arg + 1, &end, 10);
3516                         else if (!*arg && !av[1])
3517                                 die("Option '--stat-graph-width' requires a value");
3518                         else if (!*arg) {
3519                                 graph_width = strtoul(av[1], &end, 10);
3520                                 argcount = 2;
3521                         }
3522                 } else if (skip_prefix(arg, "-count", &arg)) {
3523                         if (*arg == '=')
3524                                 count = strtoul(arg + 1, &end, 10);
3525                         else if (!*arg && !av[1])
3526                                 die("Option '--stat-count' requires a value");
3527                         else if (!*arg) {
3528                                 count = strtoul(av[1], &end, 10);
3529                                 argcount = 2;
3530                         }
3531                 }
3532                 break;
3533         case '=':
3534                 width = strtoul(arg+1, &end, 10);
3535                 if (*end == ',')
3536                         name_width = strtoul(end+1, &end, 10);
3537                 if (*end == ',')
3538                         count = strtoul(end+1, &end, 10);
3539         }
3540
3541         /* Important! This checks all the error cases! */
3542         if (*end)
3543                 return 0;
3544         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3545         options->stat_name_width = name_width;
3546         options->stat_graph_width = graph_width;
3547         options->stat_width = width;
3548         options->stat_count = count;
3549         return argcount;
3550 }
3551
3552 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3553 {
3554         struct strbuf errmsg = STRBUF_INIT;
3555         if (parse_dirstat_params(options, params, &errmsg))
3556                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3557                     errmsg.buf);
3558         strbuf_release(&errmsg);
3559         /*
3560          * The caller knows a dirstat-related option is given from the command
3561          * line; allow it to say "return this_function();"
3562          */
3563         options->output_format |= DIFF_FORMAT_DIRSTAT;
3564         return 1;
3565 }
3566
3567 static int parse_submodule_opt(struct diff_options *options, const char *value)
3568 {
3569         if (parse_submodule_params(options, value))
3570                 die(_("Failed to parse --submodule option parameter: '%s'"),
3571                         value);
3572         return 1;
3573 }
3574
3575 static const char diff_status_letters[] = {
3576         DIFF_STATUS_ADDED,
3577         DIFF_STATUS_COPIED,
3578         DIFF_STATUS_DELETED,
3579         DIFF_STATUS_MODIFIED,
3580         DIFF_STATUS_RENAMED,
3581         DIFF_STATUS_TYPE_CHANGED,
3582         DIFF_STATUS_UNKNOWN,
3583         DIFF_STATUS_UNMERGED,
3584         DIFF_STATUS_FILTER_AON,
3585         DIFF_STATUS_FILTER_BROKEN,
3586         '\0',
3587 };
3588
3589 static unsigned int filter_bit['Z' + 1];
3590
3591 static void prepare_filter_bits(void)
3592 {
3593         int i;
3594
3595         if (!filter_bit[DIFF_STATUS_ADDED]) {
3596                 for (i = 0; diff_status_letters[i]; i++)
3597                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3598         }
3599 }
3600
3601 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3602 {
3603         return opt->filter & filter_bit[(int) status];
3604 }
3605
3606 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3607 {
3608         int i, optch;
3609
3610         prepare_filter_bits();
3611
3612         /*
3613          * If there is a negation e.g. 'd' in the input, and we haven't
3614          * initialized the filter field with another --diff-filter, start
3615          * from full set of bits, except for AON.
3616          */
3617         if (!opt->filter) {
3618                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3619                         if (optch < 'a' || 'z' < optch)
3620                                 continue;
3621                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3622                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3623                         break;
3624                 }
3625         }
3626
3627         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3628                 unsigned int bit;
3629                 int negate;
3630
3631                 if ('a' <= optch && optch <= 'z') {
3632                         negate = 1;
3633                         optch = toupper(optch);
3634                 } else {
3635                         negate = 0;
3636                 }
3637
3638                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3639                 if (!bit)
3640                         return optarg[i];
3641                 if (negate)
3642                         opt->filter &= ~bit;
3643                 else
3644                         opt->filter |= bit;
3645         }
3646         return 0;
3647 }
3648
3649 static void enable_patch_output(int *fmt) {
3650         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3651         *fmt |= DIFF_FORMAT_PATCH;
3652 }
3653
3654 static int parse_one_token(const char **arg, const char *token)
3655 {
3656         return skip_prefix(*arg, token, arg) && (!**arg || **arg == ',');
3657 }
3658
3659 static int parse_ws_error_highlight(struct diff_options *opt, const char *arg)
3660 {
3661         const char *orig_arg = arg;
3662         unsigned val = 0;
3663         while (*arg) {
3664                 if (parse_one_token(&arg, "none"))
3665                         val = 0;
3666                 else if (parse_one_token(&arg, "default"))
3667                         val = WSEH_NEW;
3668                 else if (parse_one_token(&arg, "all"))
3669                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
3670                 else if (parse_one_token(&arg, "new"))
3671                         val |= WSEH_NEW;
3672                 else if (parse_one_token(&arg, "old"))
3673                         val |= WSEH_OLD;
3674                 else if (parse_one_token(&arg, "context"))
3675                         val |= WSEH_CONTEXT;
3676                 else {
3677                         error("unknown value after ws-error-highlight=%.*s",
3678                               (int)(arg - orig_arg), orig_arg);
3679                         return 0;
3680                 }
3681                 if (*arg)
3682                         arg++;
3683         }
3684         opt->ws_error_highlight = val;
3685         return 1;
3686 }
3687
3688 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3689 {
3690         const char *arg = av[0];
3691         const char *optarg;
3692         int argcount;
3693
3694         /* Output format options */
3695         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3696             || opt_arg(arg, 'U', "unified", &options->context))
3697                 enable_patch_output(&options->output_format);
3698         else if (!strcmp(arg, "--raw"))
3699                 options->output_format |= DIFF_FORMAT_RAW;
3700         else if (!strcmp(arg, "--patch-with-raw")) {
3701                 enable_patch_output(&options->output_format);
3702                 options->output_format |= DIFF_FORMAT_RAW;
3703         } else if (!strcmp(arg, "--numstat"))
3704                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3705         else if (!strcmp(arg, "--shortstat"))
3706                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3707         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3708                 return parse_dirstat_opt(options, "");
3709         else if (skip_prefix(arg, "-X", &arg))
3710                 return parse_dirstat_opt(options, arg);
3711         else if (skip_prefix(arg, "--dirstat=", &arg))
3712                 return parse_dirstat_opt(options, arg);
3713         else if (!strcmp(arg, "--cumulative"))
3714                 return parse_dirstat_opt(options, "cumulative");
3715         else if (!strcmp(arg, "--dirstat-by-file"))
3716                 return parse_dirstat_opt(options, "files");
3717         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3718                 parse_dirstat_opt(options, "files");
3719                 return parse_dirstat_opt(options, arg);
3720         }
3721         else if (!strcmp(arg, "--check"))
3722                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3723         else if (!strcmp(arg, "--summary"))
3724                 options->output_format |= DIFF_FORMAT_SUMMARY;
3725         else if (!strcmp(arg, "--patch-with-stat")) {
3726                 enable_patch_output(&options->output_format);
3727                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3728         } else if (!strcmp(arg, "--name-only"))
3729                 options->output_format |= DIFF_FORMAT_NAME;
3730         else if (!strcmp(arg, "--name-status"))
3731                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3732         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3733                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3734         else if (starts_with(arg, "--stat"))
3735                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3736                 return stat_opt(options, av);
3737
3738         /* renames options */
3739         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3740                  !strcmp(arg, "--break-rewrites")) {
3741                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3742                         return error("invalid argument to -B: %s", arg+2);
3743         }
3744         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3745                  !strcmp(arg, "--find-renames")) {
3746                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3747                         return error("invalid argument to -M: %s", arg+2);
3748                 options->detect_rename = DIFF_DETECT_RENAME;
3749         }
3750         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3751                 options->irreversible_delete = 1;
3752         }
3753         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3754                  !strcmp(arg, "--find-copies")) {
3755                 if (options->detect_rename == DIFF_DETECT_COPY)
3756                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3757                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3758                         return error("invalid argument to -C: %s", arg+2);
3759                 options->detect_rename = DIFF_DETECT_COPY;
3760         }
3761         else if (!strcmp(arg, "--no-renames"))
3762                 options->detect_rename = 0;
3763         else if (!strcmp(arg, "--rename-empty"))
3764                 DIFF_OPT_SET(options, RENAME_EMPTY);
3765         else if (!strcmp(arg, "--no-rename-empty"))
3766                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3767         else if (!strcmp(arg, "--relative"))
3768                 DIFF_OPT_SET(options, RELATIVE_NAME);
3769         else if (skip_prefix(arg, "--relative=", &arg)) {
3770                 DIFF_OPT_SET(options, RELATIVE_NAME);
3771                 options->prefix = arg;
3772         }
3773
3774         /* xdiff options */
3775         else if (!strcmp(arg, "--minimal"))
3776                 DIFF_XDL_SET(options, NEED_MINIMAL);
3777         else if (!strcmp(arg, "--no-minimal"))
3778                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3779         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3780                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3781         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3782                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3783         else if (!strcmp(arg, "--ignore-space-at-eol"))
3784                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3785         else if (!strcmp(arg, "--ignore-blank-lines"))
3786                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3787         else if (!strcmp(arg, "--patience"))
3788                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3789         else if (!strcmp(arg, "--histogram"))
3790                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3791         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3792                 long value = parse_algorithm_value(optarg);
3793                 if (value < 0)
3794                         return error("option diff-algorithm accepts \"myers\", "
3795                                      "\"minimal\", \"patience\" and \"histogram\"");
3796                 /* clear out previous settings */
3797                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3798                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3799                 options->xdl_opts |= value;
3800                 return argcount;
3801         }
3802
3803         /* flags options */
3804         else if (!strcmp(arg, "--binary")) {
3805                 enable_patch_output(&options->output_format);
3806                 DIFF_OPT_SET(options, BINARY);
3807         }
3808         else if (!strcmp(arg, "--full-index"))
3809                 DIFF_OPT_SET(options, FULL_INDEX);
3810         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3811                 DIFF_OPT_SET(options, TEXT);
3812         else if (!strcmp(arg, "-R"))
3813                 DIFF_OPT_SET(options, REVERSE_DIFF);
3814         else if (!strcmp(arg, "--find-copies-harder"))
3815                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3816         else if (!strcmp(arg, "--follow"))
3817                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3818         else if (!strcmp(arg, "--no-follow"))
3819                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3820         else if (!strcmp(arg, "--color"))
3821                 options->use_color = 1;
3822         else if (skip_prefix(arg, "--color=", &arg)) {
3823                 int value = git_config_colorbool(NULL, arg);
3824                 if (value < 0)
3825                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3826                 options->use_color = value;
3827         }
3828         else if (!strcmp(arg, "--no-color"))
3829                 options->use_color = 0;
3830         else if (!strcmp(arg, "--color-words")) {
3831                 options->use_color = 1;
3832                 options->word_diff = DIFF_WORDS_COLOR;
3833         }
3834         else if (skip_prefix(arg, "--color-words=", &arg)) {
3835                 options->use_color = 1;
3836                 options->word_diff = DIFF_WORDS_COLOR;
3837                 options->word_regex = arg;
3838         }
3839         else if (!strcmp(arg, "--word-diff")) {
3840                 if (options->word_diff == DIFF_WORDS_NONE)
3841                         options->word_diff = DIFF_WORDS_PLAIN;
3842         }
3843         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3844                 if (!strcmp(arg, "plain"))
3845                         options->word_diff = DIFF_WORDS_PLAIN;
3846                 else if (!strcmp(arg, "color")) {
3847                         options->use_color = 1;
3848                         options->word_diff = DIFF_WORDS_COLOR;
3849                 }
3850                 else if (!strcmp(arg, "porcelain"))
3851                         options->word_diff = DIFF_WORDS_PORCELAIN;
3852                 else if (!strcmp(arg, "none"))
3853                         options->word_diff = DIFF_WORDS_NONE;
3854                 else
3855                         die("bad --word-diff argument: %s", arg);
3856         }
3857         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3858                 if (options->word_diff == DIFF_WORDS_NONE)
3859                         options->word_diff = DIFF_WORDS_PLAIN;
3860                 options->word_regex = optarg;
3861                 return argcount;
3862         }
3863         else if (!strcmp(arg, "--exit-code"))
3864                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3865         else if (!strcmp(arg, "--quiet"))
3866                 DIFF_OPT_SET(options, QUICK);
3867         else if (!strcmp(arg, "--ext-diff"))
3868                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3869         else if (!strcmp(arg, "--no-ext-diff"))
3870                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3871         else if (!strcmp(arg, "--textconv"))
3872                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3873         else if (!strcmp(arg, "--no-textconv"))
3874                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3875         else if (!strcmp(arg, "--ignore-submodules")) {
3876                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3877                 handle_ignore_submodules_arg(options, "all");
3878         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3879                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3880                 handle_ignore_submodules_arg(options, arg);
3881         } else if (!strcmp(arg, "--submodule"))
3882                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3883         else if (skip_prefix(arg, "--submodule=", &arg))
3884                 return parse_submodule_opt(options, arg);
3885         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3886                 return parse_ws_error_highlight(options, arg);
3887
3888         /* misc options */
3889         else if (!strcmp(arg, "-z"))
3890                 options->line_termination = 0;
3891         else if ((argcount = short_opt('l', av, &optarg))) {
3892                 options->rename_limit = strtoul(optarg, NULL, 10);
3893                 return argcount;
3894         }
3895         else if ((argcount = short_opt('S', av, &optarg))) {
3896                 options->pickaxe = optarg;
3897                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3898                 return argcount;
3899         } else if ((argcount = short_opt('G', av, &optarg))) {
3900                 options->pickaxe = optarg;
3901                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3902                 return argcount;
3903         }
3904         else if (!strcmp(arg, "--pickaxe-all"))
3905                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3906         else if (!strcmp(arg, "--pickaxe-regex"))
3907                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3908         else if ((argcount = short_opt('O', av, &optarg))) {
3909                 options->orderfile = optarg;
3910                 return argcount;
3911         }
3912         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3913                 int offending = parse_diff_filter_opt(optarg, options);
3914                 if (offending)
3915                         die("unknown change class '%c' in --diff-filter=%s",
3916                             offending, optarg);
3917                 return argcount;
3918         }
3919         else if (!strcmp(arg, "--abbrev"))
3920                 options->abbrev = DEFAULT_ABBREV;
3921         else if (skip_prefix(arg, "--abbrev=", &arg)) {
3922                 options->abbrev = strtoul(arg, NULL, 10);
3923                 if (options->abbrev < MINIMUM_ABBREV)
3924                         options->abbrev = MINIMUM_ABBREV;
3925                 else if (40 < options->abbrev)
3926                         options->abbrev = 40;
3927         }
3928         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3929                 options->a_prefix = optarg;
3930                 return argcount;
3931         }
3932         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3933                 options->b_prefix = optarg;
3934                 return argcount;
3935         }
3936         else if (!strcmp(arg, "--no-prefix"))
3937                 options->a_prefix = options->b_prefix = "";
3938         else if (opt_arg(arg, '\0', "inter-hunk-context",
3939                          &options->interhunkcontext))
3940                 ;
3941         else if (!strcmp(arg, "-W"))
3942                 DIFF_OPT_SET(options, FUNCCONTEXT);
3943         else if (!strcmp(arg, "--function-context"))
3944                 DIFF_OPT_SET(options, FUNCCONTEXT);
3945         else if (!strcmp(arg, "--no-function-context"))
3946                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3947         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3948                 options->file = fopen(optarg, "w");
3949                 if (!options->file)
3950                         die_errno("Could not open '%s'", optarg);
3951                 options->close_file = 1;
3952                 return argcount;
3953         } else
3954                 return 0;
3955         return 1;
3956 }
3957
3958 int parse_rename_score(const char **cp_p)
3959 {
3960         unsigned long num, scale;
3961         int ch, dot;
3962         const char *cp = *cp_p;
3963
3964         num = 0;
3965         scale = 1;
3966         dot = 0;
3967         for (;;) {
3968                 ch = *cp;
3969                 if ( !dot && ch == '.' ) {
3970                         scale = 1;
3971                         dot = 1;
3972                 } else if ( ch == '%' ) {
3973                         scale = dot ? scale*100 : 100;
3974                         cp++;   /* % is always at the end */
3975                         break;
3976                 } else if ( ch >= '0' && ch <= '9' ) {
3977                         if ( scale < 100000 ) {
3978                                 scale *= 10;
3979                                 num = (num*10) + (ch-'0');
3980                         }
3981                 } else {
3982                         break;
3983                 }
3984                 cp++;
3985         }
3986         *cp_p = cp;
3987
3988         /* user says num divided by scale and we say internally that
3989          * is MAX_SCORE * num / scale.
3990          */
3991         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3992 }
3993
3994 static int diff_scoreopt_parse(const char *opt)
3995 {
3996         int opt1, opt2, cmd;
3997
3998         if (*opt++ != '-')
3999                 return -1;
4000         cmd = *opt++;
4001         if (cmd == '-') {
4002                 /* convert the long-form arguments into short-form versions */
4003                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4004                         if (*opt == 0 || *opt++ == '=')
4005                                 cmd = 'B';
4006                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4007                         if (*opt == 0 || *opt++ == '=')
4008                                 cmd = 'C';
4009                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4010                         if (*opt == 0 || *opt++ == '=')
4011                                 cmd = 'M';
4012                 }
4013         }
4014         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4015                 return -1; /* that is not a -M, -C, or -B option */
4016
4017         opt1 = parse_rename_score(&opt);
4018         if (cmd != 'B')
4019                 opt2 = 0;
4020         else {
4021                 if (*opt == 0)
4022                         opt2 = 0;
4023                 else if (*opt != '/')
4024                         return -1; /* we expect -B80/99 or -B80 */
4025                 else {
4026                         opt++;
4027                         opt2 = parse_rename_score(&opt);
4028                 }
4029         }
4030         if (*opt != 0)
4031                 return -1;
4032         return opt1 | (opt2 << 16);
4033 }
4034
4035 struct diff_queue_struct diff_queued_diff;
4036
4037 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4038 {
4039         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4040         queue->queue[queue->nr++] = dp;
4041 }
4042
4043 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4044                                  struct diff_filespec *one,
4045                                  struct diff_filespec *two)
4046 {
4047         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4048         dp->one = one;
4049         dp->two = two;
4050         if (queue)
4051                 diff_q(queue, dp);
4052         return dp;
4053 }
4054
4055 void diff_free_filepair(struct diff_filepair *p)
4056 {
4057         free_filespec(p->one);
4058         free_filespec(p->two);
4059         free(p);
4060 }
4061
4062 /* This is different from find_unique_abbrev() in that
4063  * it stuffs the result with dots for alignment.
4064  */
4065 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
4066 {
4067         int abblen;
4068         const char *abbrev;
4069         if (len == 40)
4070                 return sha1_to_hex(sha1);
4071
4072         abbrev = find_unique_abbrev(sha1, len);
4073         abblen = strlen(abbrev);
4074         if (abblen < 37) {
4075                 static char hex[41];
4076                 if (len < abblen && abblen <= len + 2)
4077                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
4078                 else
4079                         sprintf(hex, "%s...", abbrev);
4080                 return hex;
4081         }
4082         return sha1_to_hex(sha1);
4083 }
4084
4085 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4086 {
4087         int line_termination = opt->line_termination;
4088         int inter_name_termination = line_termination ? '\t' : '\0';
4089
4090         fprintf(opt->file, "%s", diff_line_prefix(opt));
4091         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4092                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4093                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
4094                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4095         }
4096         if (p->score) {
4097                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4098                         inter_name_termination);
4099         } else {
4100                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4101         }
4102
4103         if (p->status == DIFF_STATUS_COPIED ||
4104             p->status == DIFF_STATUS_RENAMED) {
4105                 const char *name_a, *name_b;
4106                 name_a = p->one->path;
4107                 name_b = p->two->path;
4108                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4109                 write_name_quoted(name_a, opt->file, inter_name_termination);
4110                 write_name_quoted(name_b, opt->file, line_termination);
4111         } else {
4112                 const char *name_a, *name_b;
4113                 name_a = p->one->mode ? p->one->path : p->two->path;
4114                 name_b = NULL;
4115                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4116                 write_name_quoted(name_a, opt->file, line_termination);
4117         }
4118 }
4119
4120 int diff_unmodified_pair(struct diff_filepair *p)
4121 {
4122         /* This function is written stricter than necessary to support
4123          * the currently implemented transformers, but the idea is to
4124          * let transformers to produce diff_filepairs any way they want,
4125          * and filter and clean them up here before producing the output.
4126          */
4127         struct diff_filespec *one = p->one, *two = p->two;
4128
4129         if (DIFF_PAIR_UNMERGED(p))
4130                 return 0; /* unmerged is interesting */
4131
4132         /* deletion, addition, mode or type change
4133          * and rename are all interesting.
4134          */
4135         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4136             DIFF_PAIR_MODE_CHANGED(p) ||
4137             strcmp(one->path, two->path))
4138                 return 0;
4139
4140         /* both are valid and point at the same path.  that is, we are
4141          * dealing with a change.
4142          */
4143         if (one->sha1_valid && two->sha1_valid &&
4144             !hashcmp(one->sha1, two->sha1) &&
4145             !one->dirty_submodule && !two->dirty_submodule)
4146                 return 1; /* no change */
4147         if (!one->sha1_valid && !two->sha1_valid)
4148                 return 1; /* both look at the same file on the filesystem. */
4149         return 0;
4150 }
4151
4152 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4153 {
4154         if (diff_unmodified_pair(p))
4155                 return;
4156
4157         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4158             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4159                 return; /* no tree diffs in patch format */
4160
4161         run_diff(p, o);
4162 }
4163
4164 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4165                             struct diffstat_t *diffstat)
4166 {
4167         if (diff_unmodified_pair(p))
4168                 return;
4169
4170         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4171             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4172                 return; /* no useful stat for tree diffs */
4173
4174         run_diffstat(p, o, diffstat);
4175 }
4176
4177 static void diff_flush_checkdiff(struct diff_filepair *p,
4178                 struct diff_options *o)
4179 {
4180         if (diff_unmodified_pair(p))
4181                 return;
4182
4183         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4184             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4185                 return; /* nothing to check in tree diffs */
4186
4187         run_checkdiff(p, o);
4188 }
4189
4190 int diff_queue_is_empty(void)
4191 {
4192         struct diff_queue_struct *q = &diff_queued_diff;
4193         int i;
4194         for (i = 0; i < q->nr; i++)
4195                 if (!diff_unmodified_pair(q->queue[i]))
4196                         return 0;
4197         return 1;
4198 }
4199
4200 #if DIFF_DEBUG
4201 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4202 {
4203         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4204                 x, one ? one : "",
4205                 s->path,
4206                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4207                 s->mode,
4208                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4209         fprintf(stderr, "queue[%d] %s size %lu\n",
4210                 x, one ? one : "",
4211                 s->size);
4212 }
4213
4214 void diff_debug_filepair(const struct diff_filepair *p, int i)
4215 {
4216         diff_debug_filespec(p->one, i, "one");
4217         diff_debug_filespec(p->two, i, "two");
4218         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4219                 p->score, p->status ? p->status : '?',
4220                 p->one->rename_used, p->broken_pair);
4221 }
4222
4223 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4224 {
4225         int i;
4226         if (msg)
4227                 fprintf(stderr, "%s\n", msg);
4228         fprintf(stderr, "q->nr = %d\n", q->nr);
4229         for (i = 0; i < q->nr; i++) {
4230                 struct diff_filepair *p = q->queue[i];
4231                 diff_debug_filepair(p, i);
4232         }
4233 }
4234 #endif
4235
4236 static void diff_resolve_rename_copy(void)
4237 {
4238         int i;
4239         struct diff_filepair *p;
4240         struct diff_queue_struct *q = &diff_queued_diff;
4241
4242         diff_debug_queue("resolve-rename-copy", q);
4243
4244         for (i = 0; i < q->nr; i++) {
4245                 p = q->queue[i];
4246                 p->status = 0; /* undecided */
4247                 if (DIFF_PAIR_UNMERGED(p))
4248                         p->status = DIFF_STATUS_UNMERGED;
4249                 else if (!DIFF_FILE_VALID(p->one))
4250                         p->status = DIFF_STATUS_ADDED;
4251                 else if (!DIFF_FILE_VALID(p->two))
4252                         p->status = DIFF_STATUS_DELETED;
4253                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4254                         p->status = DIFF_STATUS_TYPE_CHANGED;
4255
4256                 /* from this point on, we are dealing with a pair
4257                  * whose both sides are valid and of the same type, i.e.
4258                  * either in-place edit or rename/copy edit.
4259                  */
4260                 else if (DIFF_PAIR_RENAME(p)) {
4261                         /*
4262                          * A rename might have re-connected a broken
4263                          * pair up, causing the pathnames to be the
4264                          * same again. If so, that's not a rename at
4265                          * all, just a modification..
4266                          *
4267                          * Otherwise, see if this source was used for
4268                          * multiple renames, in which case we decrement
4269                          * the count, and call it a copy.
4270                          */
4271                         if (!strcmp(p->one->path, p->two->path))
4272                                 p->status = DIFF_STATUS_MODIFIED;
4273                         else if (--p->one->rename_used > 0)
4274                                 p->status = DIFF_STATUS_COPIED;
4275                         else
4276                                 p->status = DIFF_STATUS_RENAMED;
4277                 }
4278                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4279                          p->one->mode != p->two->mode ||
4280                          p->one->dirty_submodule ||
4281                          p->two->dirty_submodule ||
4282                          is_null_sha1(p->one->sha1))
4283                         p->status = DIFF_STATUS_MODIFIED;
4284                 else {
4285                         /* This is a "no-change" entry and should not
4286                          * happen anymore, but prepare for broken callers.
4287                          */
4288                         error("feeding unmodified %s to diffcore",
4289                               p->one->path);
4290                         p->status = DIFF_STATUS_UNKNOWN;
4291                 }
4292         }
4293         diff_debug_queue("resolve-rename-copy done", q);
4294 }
4295
4296 static int check_pair_status(struct diff_filepair *p)
4297 {
4298         switch (p->status) {
4299         case DIFF_STATUS_UNKNOWN:
4300                 return 0;
4301         case 0:
4302                 die("internal error in diff-resolve-rename-copy");
4303         default:
4304                 return 1;
4305         }
4306 }
4307
4308 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4309 {
4310         int fmt = opt->output_format;
4311
4312         if (fmt & DIFF_FORMAT_CHECKDIFF)
4313                 diff_flush_checkdiff(p, opt);
4314         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4315                 diff_flush_raw(p, opt);
4316         else if (fmt & DIFF_FORMAT_NAME) {
4317                 const char *name_a, *name_b;
4318                 name_a = p->two->path;
4319                 name_b = NULL;
4320                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4321                 write_name_quoted(name_a, opt->file, opt->line_termination);
4322         }
4323 }
4324
4325 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4326 {
4327         if (fs->mode)
4328                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4329         else
4330                 fprintf(file, " %s ", newdelete);
4331         write_name_quoted(fs->path, file, '\n');
4332 }
4333
4334
4335 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4336                 const char *line_prefix)
4337 {
4338         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4339                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4340                         p->two->mode, show_name ? ' ' : '\n');
4341                 if (show_name) {
4342                         write_name_quoted(p->two->path, file, '\n');
4343                 }
4344         }
4345 }
4346
4347 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4348                         const char *line_prefix)
4349 {
4350         char *names = pprint_rename(p->one->path, p->two->path);
4351
4352         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4353         free(names);
4354         show_mode_change(file, p, 0, line_prefix);
4355 }
4356
4357 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4358 {
4359         FILE *file = opt->file;
4360         const char *line_prefix = diff_line_prefix(opt);
4361
4362         switch(p->status) {
4363         case DIFF_STATUS_DELETED:
4364                 fputs(line_prefix, file);
4365                 show_file_mode_name(file, "delete", p->one);
4366                 break;
4367         case DIFF_STATUS_ADDED:
4368                 fputs(line_prefix, file);
4369                 show_file_mode_name(file, "create", p->two);
4370                 break;
4371         case DIFF_STATUS_COPIED:
4372                 fputs(line_prefix, file);
4373                 show_rename_copy(file, "copy", p, line_prefix);
4374                 break;
4375         case DIFF_STATUS_RENAMED:
4376                 fputs(line_prefix, file);
4377                 show_rename_copy(file, "rename", p, line_prefix);
4378                 break;
4379         default:
4380                 if (p->score) {
4381                         fprintf(file, "%s rewrite ", line_prefix);
4382                         write_name_quoted(p->two->path, file, ' ');
4383                         fprintf(file, "(%d%%)\n", similarity_index(p));
4384                 }
4385                 show_mode_change(file, p, !p->score, line_prefix);
4386                 break;
4387         }
4388 }
4389
4390 struct patch_id_t {
4391         git_SHA_CTX *ctx;
4392         int patchlen;
4393 };
4394
4395 static int remove_space(char *line, int len)
4396 {
4397         int i;
4398         char *dst = line;
4399         unsigned char c;
4400
4401         for (i = 0; i < len; i++)
4402                 if (!isspace((c = line[i])))
4403                         *dst++ = c;
4404
4405         return dst - line;
4406 }
4407
4408 static void patch_id_consume(void *priv, char *line, unsigned long len)
4409 {
4410         struct patch_id_t *data = priv;
4411         int new_len;
4412
4413         /* Ignore line numbers when computing the SHA1 of the patch */
4414         if (starts_with(line, "@@ -"))
4415                 return;
4416
4417         new_len = remove_space(line, len);
4418
4419         git_SHA1_Update(data->ctx, line, new_len);
4420         data->patchlen += new_len;
4421 }
4422
4423 /* returns 0 upon success, and writes result into sha1 */
4424 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4425 {
4426         struct diff_queue_struct *q = &diff_queued_diff;
4427         int i;
4428         git_SHA_CTX ctx;
4429         struct patch_id_t data;
4430         char buffer[PATH_MAX * 4 + 20];
4431
4432         git_SHA1_Init(&ctx);
4433         memset(&data, 0, sizeof(struct patch_id_t));
4434         data.ctx = &ctx;
4435
4436         for (i = 0; i < q->nr; i++) {
4437                 xpparam_t xpp;
4438                 xdemitconf_t xecfg;
4439                 mmfile_t mf1, mf2;
4440                 struct diff_filepair *p = q->queue[i];
4441                 int len1, len2;
4442
4443                 memset(&xpp, 0, sizeof(xpp));
4444                 memset(&xecfg, 0, sizeof(xecfg));
4445                 if (p->status == 0)
4446                         return error("internal diff status error");
4447                 if (p->status == DIFF_STATUS_UNKNOWN)
4448                         continue;
4449                 if (diff_unmodified_pair(p))
4450                         continue;
4451                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4452                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4453                         continue;
4454                 if (DIFF_PAIR_UNMERGED(p))
4455                         continue;
4456
4457                 diff_fill_sha1_info(p->one);
4458                 diff_fill_sha1_info(p->two);
4459                 if (fill_mmfile(&mf1, p->one) < 0 ||
4460                                 fill_mmfile(&mf2, p->two) < 0)
4461                         return error("unable to read files to diff");
4462
4463                 len1 = remove_space(p->one->path, strlen(p->one->path));
4464                 len2 = remove_space(p->two->path, strlen(p->two->path));
4465                 if (p->one->mode == 0)
4466                         len1 = snprintf(buffer, sizeof(buffer),
4467                                         "diff--gita/%.*sb/%.*s"
4468                                         "newfilemode%06o"
4469                                         "---/dev/null"
4470                                         "+++b/%.*s",
4471                                         len1, p->one->path,
4472                                         len2, p->two->path,
4473                                         p->two->mode,
4474                                         len2, p->two->path);
4475                 else if (p->two->mode == 0)
4476                         len1 = snprintf(buffer, sizeof(buffer),
4477                                         "diff--gita/%.*sb/%.*s"
4478                                         "deletedfilemode%06o"
4479                                         "---a/%.*s"
4480                                         "+++/dev/null",
4481                                         len1, p->one->path,
4482                                         len2, p->two->path,
4483                                         p->one->mode,
4484                                         len1, p->one->path);
4485                 else
4486                         len1 = snprintf(buffer, sizeof(buffer),
4487                                         "diff--gita/%.*sb/%.*s"
4488                                         "---a/%.*s"
4489                                         "+++b/%.*s",
4490                                         len1, p->one->path,
4491                                         len2, p->two->path,
4492                                         len1, p->one->path,
4493                                         len2, p->two->path);
4494                 git_SHA1_Update(&ctx, buffer, len1);
4495
4496                 if (diff_filespec_is_binary(p->one) ||
4497                     diff_filespec_is_binary(p->two)) {
4498                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4499                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4500                         continue;
4501                 }
4502
4503                 xpp.flags = 0;
4504                 xecfg.ctxlen = 3;
4505                 xecfg.flags = 0;
4506                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4507                               &xpp, &xecfg);
4508         }
4509
4510         git_SHA1_Final(sha1, &ctx);
4511         return 0;
4512 }
4513
4514 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4515 {
4516         struct diff_queue_struct *q = &diff_queued_diff;
4517         int i;
4518         int result = diff_get_patch_id(options, sha1);
4519
4520         for (i = 0; i < q->nr; i++)
4521                 diff_free_filepair(q->queue[i]);
4522
4523         free(q->queue);
4524         DIFF_QUEUE_CLEAR(q);
4525
4526         return result;
4527 }
4528
4529 static int is_summary_empty(const struct diff_queue_struct *q)
4530 {
4531         int i;
4532
4533         for (i = 0; i < q->nr; i++) {
4534                 const struct diff_filepair *p = q->queue[i];
4535
4536                 switch (p->status) {
4537                 case DIFF_STATUS_DELETED:
4538                 case DIFF_STATUS_ADDED:
4539                 case DIFF_STATUS_COPIED:
4540                 case DIFF_STATUS_RENAMED:
4541                         return 0;
4542                 default:
4543                         if (p->score)
4544                                 return 0;
4545                         if (p->one->mode && p->two->mode &&
4546                             p->one->mode != p->two->mode)
4547                                 return 0;
4548                         break;
4549                 }
4550         }
4551         return 1;
4552 }
4553
4554 static const char rename_limit_warning[] =
4555 "inexact rename detection was skipped due to too many files.";
4556
4557 static const char degrade_cc_to_c_warning[] =
4558 "only found copies from modified paths due to too many files.";
4559
4560 static const char rename_limit_advice[] =
4561 "you may want to set your %s variable to at least "
4562 "%d and retry the command.";
4563
4564 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4565 {
4566         if (degraded_cc)
4567                 warning(degrade_cc_to_c_warning);
4568         else if (needed)
4569                 warning(rename_limit_warning);
4570         else
4571                 return;
4572         if (0 < needed && needed < 32767)
4573                 warning(rename_limit_advice, varname, needed);
4574 }
4575
4576 void diff_flush(struct diff_options *options)
4577 {
4578         struct diff_queue_struct *q = &diff_queued_diff;
4579         int i, output_format = options->output_format;
4580         int separator = 0;
4581         int dirstat_by_line = 0;
4582
4583         /*
4584          * Order: raw, stat, summary, patch
4585          * or:    name/name-status/checkdiff (other bits clear)
4586          */
4587         if (!q->nr)
4588                 goto free_queue;
4589
4590         if (output_format & (DIFF_FORMAT_RAW |
4591                              DIFF_FORMAT_NAME |
4592                              DIFF_FORMAT_NAME_STATUS |
4593                              DIFF_FORMAT_CHECKDIFF)) {
4594                 for (i = 0; i < q->nr; i++) {
4595                         struct diff_filepair *p = q->queue[i];
4596                         if (check_pair_status(p))
4597                                 flush_one_pair(p, options);
4598                 }
4599                 separator++;
4600         }
4601
4602         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4603                 dirstat_by_line = 1;
4604
4605         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4606             dirstat_by_line) {
4607                 struct diffstat_t diffstat;
4608
4609                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4610                 for (i = 0; i < q->nr; i++) {
4611                         struct diff_filepair *p = q->queue[i];
4612                         if (check_pair_status(p))
4613                                 diff_flush_stat(p, options, &diffstat);
4614                 }
4615                 if (output_format & DIFF_FORMAT_NUMSTAT)
4616                         show_numstat(&diffstat, options);
4617                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4618                         show_stats(&diffstat, options);
4619                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4620                         show_shortstats(&diffstat, options);
4621                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4622                         show_dirstat_by_line(&diffstat, options);
4623                 free_diffstat_info(&diffstat);
4624                 separator++;
4625         }
4626         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4627                 show_dirstat(options);
4628
4629         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4630                 for (i = 0; i < q->nr; i++) {
4631                         diff_summary(options, q->queue[i]);
4632                 }
4633                 separator++;
4634         }
4635
4636         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4637             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4638             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4639                 /*
4640                  * run diff_flush_patch for the exit status. setting
4641                  * options->file to /dev/null should be safe, because we
4642                  * aren't supposed to produce any output anyway.
4643                  */
4644                 if (options->close_file)
4645                         fclose(options->file);
4646                 options->file = fopen("/dev/null", "w");
4647                 if (!options->file)
4648                         die_errno("Could not open /dev/null");
4649                 options->close_file = 1;
4650                 for (i = 0; i < q->nr; i++) {
4651                         struct diff_filepair *p = q->queue[i];
4652                         if (check_pair_status(p))
4653                                 diff_flush_patch(p, options);
4654                         if (options->found_changes)
4655                                 break;
4656                 }
4657         }
4658
4659         if (output_format & DIFF_FORMAT_PATCH) {
4660                 if (separator) {
4661                         fprintf(options->file, "%s%c",
4662                                 diff_line_prefix(options),
4663                                 options->line_termination);
4664                         if (options->stat_sep) {
4665                                 /* attach patch instead of inline */
4666                                 fputs(options->stat_sep, options->file);
4667                         }
4668                 }
4669
4670                 for (i = 0; i < q->nr; i++) {
4671                         struct diff_filepair *p = q->queue[i];
4672                         if (check_pair_status(p))
4673                                 diff_flush_patch(p, options);
4674                 }
4675         }
4676
4677         if (output_format & DIFF_FORMAT_CALLBACK)
4678                 options->format_callback(q, options, options->format_callback_data);
4679
4680         for (i = 0; i < q->nr; i++)
4681                 diff_free_filepair(q->queue[i]);
4682 free_queue:
4683         free(q->queue);
4684         DIFF_QUEUE_CLEAR(q);
4685         if (options->close_file)
4686                 fclose(options->file);
4687
4688         /*
4689          * Report the content-level differences with HAS_CHANGES;
4690          * diff_addremove/diff_change does not set the bit when
4691          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4692          */
4693         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4694                 if (options->found_changes)
4695                         DIFF_OPT_SET(options, HAS_CHANGES);
4696                 else
4697                         DIFF_OPT_CLR(options, HAS_CHANGES);
4698         }
4699 }
4700
4701 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4702 {
4703         return (((p->status == DIFF_STATUS_MODIFIED) &&
4704                  ((p->score &&
4705                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4706                   (!p->score &&
4707                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4708                 ((p->status != DIFF_STATUS_MODIFIED) &&
4709                  filter_bit_tst(p->status, options)));
4710 }
4711
4712 static void diffcore_apply_filter(struct diff_options *options)
4713 {
4714         int i;
4715         struct diff_queue_struct *q = &diff_queued_diff;
4716         struct diff_queue_struct outq;
4717
4718         DIFF_QUEUE_CLEAR(&outq);
4719
4720         if (!options->filter)
4721                 return;
4722
4723         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4724                 int found;
4725                 for (i = found = 0; !found && i < q->nr; i++) {
4726                         if (match_filter(options, q->queue[i]))
4727                                 found++;
4728                 }
4729                 if (found)
4730                         return;
4731
4732                 /* otherwise we will clear the whole queue
4733                  * by copying the empty outq at the end of this
4734                  * function, but first clear the current entries
4735                  * in the queue.
4736                  */
4737                 for (i = 0; i < q->nr; i++)
4738                         diff_free_filepair(q->queue[i]);
4739         }
4740         else {
4741                 /* Only the matching ones */
4742                 for (i = 0; i < q->nr; i++) {
4743                         struct diff_filepair *p = q->queue[i];
4744                         if (match_filter(options, p))
4745                                 diff_q(&outq, p);
4746                         else
4747                                 diff_free_filepair(p);
4748                 }
4749         }
4750         free(q->queue);
4751         *q = outq;
4752 }
4753
4754 /* Check whether two filespecs with the same mode and size are identical */
4755 static int diff_filespec_is_identical(struct diff_filespec *one,
4756                                       struct diff_filespec *two)
4757 {
4758         if (S_ISGITLINK(one->mode))
4759                 return 0;
4760         if (diff_populate_filespec(one, 0))
4761                 return 0;
4762         if (diff_populate_filespec(two, 0))
4763                 return 0;
4764         return !memcmp(one->data, two->data, one->size);
4765 }
4766
4767 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4768 {
4769         if (p->done_skip_stat_unmatch)
4770                 return p->skip_stat_unmatch_result;
4771
4772         p->done_skip_stat_unmatch = 1;
4773         p->skip_stat_unmatch_result = 0;
4774         /*
4775          * 1. Entries that come from stat info dirtiness
4776          *    always have both sides (iow, not create/delete),
4777          *    one side of the object name is unknown, with
4778          *    the same mode and size.  Keep the ones that
4779          *    do not match these criteria.  They have real
4780          *    differences.
4781          *
4782          * 2. At this point, the file is known to be modified,
4783          *    with the same mode and size, and the object
4784          *    name of one side is unknown.  Need to inspect
4785          *    the identical contents.
4786          */
4787         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4788             !DIFF_FILE_VALID(p->two) ||
4789             (p->one->sha1_valid && p->two->sha1_valid) ||
4790             (p->one->mode != p->two->mode) ||
4791             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4792             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4793             (p->one->size != p->two->size) ||
4794             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4795                 p->skip_stat_unmatch_result = 1;
4796         return p->skip_stat_unmatch_result;
4797 }
4798
4799 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4800 {
4801         int i;
4802         struct diff_queue_struct *q = &diff_queued_diff;
4803         struct diff_queue_struct outq;
4804         DIFF_QUEUE_CLEAR(&outq);
4805
4806         for (i = 0; i < q->nr; i++) {
4807                 struct diff_filepair *p = q->queue[i];
4808
4809                 if (diff_filespec_check_stat_unmatch(p))
4810                         diff_q(&outq, p);
4811                 else {
4812                         /*
4813                          * The caller can subtract 1 from skip_stat_unmatch
4814                          * to determine how many paths were dirty only
4815                          * due to stat info mismatch.
4816                          */
4817                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4818                                 diffopt->skip_stat_unmatch++;
4819                         diff_free_filepair(p);
4820                 }
4821         }
4822         free(q->queue);
4823         *q = outq;
4824 }
4825
4826 static int diffnamecmp(const void *a_, const void *b_)
4827 {
4828         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4829         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4830         const char *name_a, *name_b;
4831
4832         name_a = a->one ? a->one->path : a->two->path;
4833         name_b = b->one ? b->one->path : b->two->path;
4834         return strcmp(name_a, name_b);
4835 }
4836
4837 void diffcore_fix_diff_index(struct diff_options *options)
4838 {
4839         struct diff_queue_struct *q = &diff_queued_diff;
4840         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4841 }
4842
4843 void diffcore_std(struct diff_options *options)
4844 {
4845         /* NOTE please keep the following in sync with diff_tree_combined() */
4846         if (options->skip_stat_unmatch)
4847                 diffcore_skip_stat_unmatch(options);
4848         if (!options->found_follow) {
4849                 /* See try_to_follow_renames() in tree-diff.c */
4850                 if (options->break_opt != -1)
4851                         diffcore_break(options->break_opt);
4852                 if (options->detect_rename)
4853                         diffcore_rename(options);
4854                 if (options->break_opt != -1)
4855                         diffcore_merge_broken();
4856         }
4857         if (options->pickaxe)
4858                 diffcore_pickaxe(options);
4859         if (options->orderfile)
4860                 diffcore_order(options->orderfile);
4861         if (!options->found_follow)
4862                 /* See try_to_follow_renames() in tree-diff.c */
4863                 diff_resolve_rename_copy();
4864         diffcore_apply_filter(options);
4865
4866         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4867                 DIFF_OPT_SET(options, HAS_CHANGES);
4868         else
4869                 DIFF_OPT_CLR(options, HAS_CHANGES);
4870
4871         options->found_follow = 0;
4872 }
4873
4874 int diff_result_code(struct diff_options *opt, int status)
4875 {
4876         int result = 0;
4877
4878         diff_warn_rename_limit("diff.renameLimit",
4879                                opt->needed_rename_limit,
4880                                opt->degraded_cc_to_c);
4881         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4882             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4883                 return status;
4884         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4885             DIFF_OPT_TST(opt, HAS_CHANGES))
4886                 result |= 01;
4887         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4888             DIFF_OPT_TST(opt, CHECK_FAILED))
4889                 result |= 02;
4890         return result;
4891 }
4892
4893 int diff_can_quit_early(struct diff_options *opt)
4894 {
4895         return (DIFF_OPT_TST(opt, QUICK) &&
4896                 !opt->filter &&
4897                 DIFF_OPT_TST(opt, HAS_CHANGES));
4898 }
4899
4900 /*
4901  * Shall changes to this submodule be ignored?
4902  *
4903  * Submodule changes can be configured to be ignored separately for each path,
4904  * but that configuration can be overridden from the command line.
4905  */
4906 static int is_submodule_ignored(const char *path, struct diff_options *options)
4907 {
4908         int ignored = 0;
4909         unsigned orig_flags = options->flags;
4910         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4911                 set_diffopt_flags_from_submodule_config(options, path);
4912         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4913                 ignored = 1;
4914         options->flags = orig_flags;
4915         return ignored;
4916 }
4917
4918 void diff_addremove(struct diff_options *options,
4919                     int addremove, unsigned mode,
4920                     const unsigned char *sha1,
4921                     int sha1_valid,
4922                     const char *concatpath, unsigned dirty_submodule)
4923 {
4924         struct diff_filespec *one, *two;
4925
4926         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4927                 return;
4928
4929         /* This may look odd, but it is a preparation for
4930          * feeding "there are unchanged files which should
4931          * not produce diffs, but when you are doing copy
4932          * detection you would need them, so here they are"
4933          * entries to the diff-core.  They will be prefixed
4934          * with something like '=' or '*' (I haven't decided
4935          * which but should not make any difference).
4936          * Feeding the same new and old to diff_change()
4937          * also has the same effect.
4938          * Before the final output happens, they are pruned after
4939          * merged into rename/copy pairs as appropriate.
4940          */
4941         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4942                 addremove = (addremove == '+' ? '-' :
4943                              addremove == '-' ? '+' : addremove);
4944
4945         if (options->prefix &&
4946             strncmp(concatpath, options->prefix, options->prefix_length))
4947                 return;
4948
4949         one = alloc_filespec(concatpath);
4950         two = alloc_filespec(concatpath);
4951
4952         if (addremove != '+')
4953                 fill_filespec(one, sha1, sha1_valid, mode);
4954         if (addremove != '-') {
4955                 fill_filespec(two, sha1, sha1_valid, mode);
4956                 two->dirty_submodule = dirty_submodule;
4957         }
4958
4959         diff_queue(&diff_queued_diff, one, two);
4960         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4961                 DIFF_OPT_SET(options, HAS_CHANGES);
4962 }
4963
4964 void diff_change(struct diff_options *options,
4965                  unsigned old_mode, unsigned new_mode,
4966                  const unsigned char *old_sha1,
4967                  const unsigned char *new_sha1,
4968                  int old_sha1_valid, int new_sha1_valid,
4969                  const char *concatpath,
4970                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4971 {
4972         struct diff_filespec *one, *two;
4973         struct diff_filepair *p;
4974
4975         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4976             is_submodule_ignored(concatpath, options))
4977                 return;
4978
4979         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4980                 unsigned tmp;
4981                 const unsigned char *tmp_c;
4982                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4983                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4984                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4985                         new_sha1_valid = tmp;
4986                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4987                         new_dirty_submodule = tmp;
4988         }
4989
4990         if (options->prefix &&
4991             strncmp(concatpath, options->prefix, options->prefix_length))
4992                 return;
4993
4994         one = alloc_filespec(concatpath);
4995         two = alloc_filespec(concatpath);
4996         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4997         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4998         one->dirty_submodule = old_dirty_submodule;
4999         two->dirty_submodule = new_dirty_submodule;
5000         p = diff_queue(&diff_queued_diff, one, two);
5001
5002         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5003                 return;
5004
5005         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5006             !diff_filespec_check_stat_unmatch(p))
5007                 return;
5008
5009         DIFF_OPT_SET(options, HAS_CHANGES);
5010 }
5011
5012 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5013 {
5014         struct diff_filepair *pair;
5015         struct diff_filespec *one, *two;
5016
5017         if (options->prefix &&
5018             strncmp(path, options->prefix, options->prefix_length))
5019                 return NULL;
5020
5021         one = alloc_filespec(path);
5022         two = alloc_filespec(path);
5023         pair = diff_queue(&diff_queued_diff, one, two);
5024         pair->is_unmerged = 1;
5025         return pair;
5026 }
5027
5028 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5029                 size_t *outsize)
5030 {
5031         struct diff_tempfile *temp;
5032         const char *argv[3];
5033         const char **arg = argv;
5034         struct child_process child = CHILD_PROCESS_INIT;
5035         struct strbuf buf = STRBUF_INIT;
5036         int err = 0;
5037
5038         temp = prepare_temp_file(spec->path, spec);
5039         *arg++ = pgm;
5040         *arg++ = temp->name;
5041         *arg = NULL;
5042
5043         child.use_shell = 1;
5044         child.argv = argv;
5045         child.out = -1;
5046         if (start_command(&child)) {
5047                 remove_tempfile();
5048                 return NULL;
5049         }
5050
5051         if (strbuf_read(&buf, child.out, 0) < 0)
5052                 err = error("error reading from textconv command '%s'", pgm);
5053         close(child.out);
5054
5055         if (finish_command(&child) || err) {
5056                 strbuf_release(&buf);
5057                 remove_tempfile();
5058                 return NULL;
5059         }
5060         remove_tempfile();
5061
5062         return strbuf_detach(&buf, outsize);
5063 }
5064
5065 size_t fill_textconv(struct userdiff_driver *driver,
5066                      struct diff_filespec *df,
5067                      char **outbuf)
5068 {
5069         size_t size;
5070
5071         if (!driver || !driver->textconv) {
5072                 if (!DIFF_FILE_VALID(df)) {
5073                         *outbuf = "";
5074                         return 0;
5075                 }
5076                 if (diff_populate_filespec(df, 0))
5077                         die("unable to read files to diff");
5078                 *outbuf = df->data;
5079                 return df->size;
5080         }
5081
5082         if (driver->textconv_cache && df->sha1_valid) {
5083                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5084                                           &size);
5085                 if (*outbuf)
5086                         return size;
5087         }
5088
5089         *outbuf = run_textconv(driver->textconv, df, &size);
5090         if (!*outbuf)
5091                 die("unable to read files to diff");
5092
5093         if (driver->textconv_cache && df->sha1_valid) {
5094                 /* ignore errors, as we might be in a readonly repository */
5095                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5096                                 size);
5097                 /*
5098                  * we could save up changes and flush them all at the end,
5099                  * but we would need an extra call after all diffing is done.
5100                  * Since generating a cache entry is the slow path anyway,
5101                  * this extra overhead probably isn't a big deal.
5102                  */
5103                 notes_cache_write(driver->textconv_cache);
5104         }
5105
5106         return size;
5107 }
5108
5109 void setup_diff_pager(struct diff_options *opt)
5110 {
5111         /*
5112          * If the user asked for our exit code, then either they want --quiet
5113          * or --exit-code. We should definitely not bother with a pager in the
5114          * former case, as we will generate no output. Since we still properly
5115          * report our exit code even when a pager is run, we _could_ run a
5116          * pager with --exit-code. But since we have not done so historically,
5117          * and because it is easy to find people oneline advising "git diff
5118          * --exit-code" in hooks and other scripts, we do not do so.
5119          */
5120         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5121             check_pager_config("diff") != 0)
5122                 setup_pager();
5123 }