Merge branch 'jk/patch-ids-no-merges' into maint
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "tempfile.h"
6 #include "quote.h"
7 #include "diff.h"
8 #include "diffcore.h"
9 #include "delta.h"
10 #include "xdiff-interface.h"
11 #include "color.h"
12 #include "attr.h"
13 #include "run-command.h"
14 #include "utf8.h"
15 #include "userdiff.h"
16 #include "submodule-config.h"
17 #include "submodule.h"
18 #include "ll-merge.h"
19 #include "string-list.h"
20 #include "argv-array.h"
21
22 #ifdef NO_FAST_WORKING_DIRECTORY
23 #define FAST_WORKING_DIRECTORY 0
24 #else
25 #define FAST_WORKING_DIRECTORY 1
26 #endif
27
28 static int diff_detect_rename_default;
29 static int diff_compaction_heuristic; /* experimental */
30 static int diff_rename_limit_default = 400;
31 static int diff_suppress_blank_empty;
32 static int diff_use_color_default = -1;
33 static int diff_context_default = 3;
34 static const char *diff_word_regex_cfg;
35 static const char *external_diff_cmd_cfg;
36 static const char *diff_order_file_cfg;
37 int diff_auto_refresh_index = 1;
38 static int diff_mnemonic_prefix;
39 static int diff_no_prefix;
40 static int diff_stat_graph_width;
41 static int diff_dirstat_permille_default = 30;
42 static struct diff_options default_diff_options;
43 static long diff_algorithm;
44
45 static char diff_colors[][COLOR_MAXLEN] = {
46         GIT_COLOR_RESET,
47         GIT_COLOR_NORMAL,       /* CONTEXT */
48         GIT_COLOR_BOLD,         /* METAINFO */
49         GIT_COLOR_CYAN,         /* FRAGINFO */
50         GIT_COLOR_RED,          /* OLD */
51         GIT_COLOR_GREEN,        /* NEW */
52         GIT_COLOR_YELLOW,       /* COMMIT */
53         GIT_COLOR_BG_RED,       /* WHITESPACE */
54         GIT_COLOR_NORMAL,       /* FUNCINFO */
55 };
56
57 static int parse_diff_color_slot(const char *var)
58 {
59         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
60                 return DIFF_CONTEXT;
61         if (!strcasecmp(var, "meta"))
62                 return DIFF_METAINFO;
63         if (!strcasecmp(var, "frag"))
64                 return DIFF_FRAGINFO;
65         if (!strcasecmp(var, "old"))
66                 return DIFF_FILE_OLD;
67         if (!strcasecmp(var, "new"))
68                 return DIFF_FILE_NEW;
69         if (!strcasecmp(var, "commit"))
70                 return DIFF_COMMIT;
71         if (!strcasecmp(var, "whitespace"))
72                 return DIFF_WHITESPACE;
73         if (!strcasecmp(var, "func"))
74                 return DIFF_FUNCINFO;
75         return -1;
76 }
77
78 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
79                                 struct strbuf *errmsg)
80 {
81         char *params_copy = xstrdup(params_string);
82         struct string_list params = STRING_LIST_INIT_NODUP;
83         int ret = 0;
84         int i;
85
86         if (*params_copy)
87                 string_list_split_in_place(&params, params_copy, ',', -1);
88         for (i = 0; i < params.nr; i++) {
89                 const char *p = params.items[i].string;
90                 if (!strcmp(p, "changes")) {
91                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
92                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
93                 } else if (!strcmp(p, "lines")) {
94                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
95                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
96                 } else if (!strcmp(p, "files")) {
97                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
98                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
99                 } else if (!strcmp(p, "noncumulative")) {
100                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
101                 } else if (!strcmp(p, "cumulative")) {
102                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
103                 } else if (isdigit(*p)) {
104                         char *end;
105                         int permille = strtoul(p, &end, 10) * 10;
106                         if (*end == '.' && isdigit(*++end)) {
107                                 /* only use first digit */
108                                 permille += *end - '0';
109                                 /* .. and ignore any further digits */
110                                 while (isdigit(*++end))
111                                         ; /* nothing */
112                         }
113                         if (!*end)
114                                 options->dirstat_permille = permille;
115                         else {
116                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
117                                             p);
118                                 ret++;
119                         }
120                 } else {
121                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
122                         ret++;
123                 }
124
125         }
126         string_list_clear(&params, 0);
127         free(params_copy);
128         return ret;
129 }
130
131 static int parse_submodule_params(struct diff_options *options, const char *value)
132 {
133         if (!strcmp(value, "log"))
134                 DIFF_OPT_SET(options, SUBMODULE_LOG);
135         else if (!strcmp(value, "short"))
136                 DIFF_OPT_CLR(options, SUBMODULE_LOG);
137         else
138                 return -1;
139         return 0;
140 }
141
142 static int git_config_rename(const char *var, const char *value)
143 {
144         if (!value)
145                 return DIFF_DETECT_RENAME;
146         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
147                 return  DIFF_DETECT_COPY;
148         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
149 }
150
151 long parse_algorithm_value(const char *value)
152 {
153         if (!value)
154                 return -1;
155         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
156                 return 0;
157         else if (!strcasecmp(value, "minimal"))
158                 return XDF_NEED_MINIMAL;
159         else if (!strcasecmp(value, "patience"))
160                 return XDF_PATIENCE_DIFF;
161         else if (!strcasecmp(value, "histogram"))
162                 return XDF_HISTOGRAM_DIFF;
163         return -1;
164 }
165
166 /*
167  * These are to give UI layer defaults.
168  * The core-level commands such as git-diff-files should
169  * never be affected by the setting of diff.renames
170  * the user happens to have in the configuration file.
171  */
172 void init_diff_ui_defaults(void)
173 {
174         diff_detect_rename_default = 1;
175 }
176
177 int git_diff_ui_config(const char *var, const char *value, void *cb)
178 {
179         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
180                 diff_use_color_default = git_config_colorbool(var, value);
181                 return 0;
182         }
183         if (!strcmp(var, "diff.context")) {
184                 diff_context_default = git_config_int(var, value);
185                 if (diff_context_default < 0)
186                         return -1;
187                 return 0;
188         }
189         if (!strcmp(var, "diff.renames")) {
190                 diff_detect_rename_default = git_config_rename(var, value);
191                 return 0;
192         }
193         if (!strcmp(var, "diff.compactionheuristic")) {
194                 diff_compaction_heuristic = git_config_bool(var, value);
195                 return 0;
196         }
197         if (!strcmp(var, "diff.autorefreshindex")) {
198                 diff_auto_refresh_index = git_config_bool(var, value);
199                 return 0;
200         }
201         if (!strcmp(var, "diff.mnemonicprefix")) {
202                 diff_mnemonic_prefix = git_config_bool(var, value);
203                 return 0;
204         }
205         if (!strcmp(var, "diff.noprefix")) {
206                 diff_no_prefix = git_config_bool(var, value);
207                 return 0;
208         }
209         if (!strcmp(var, "diff.statgraphwidth")) {
210                 diff_stat_graph_width = git_config_int(var, value);
211                 return 0;
212         }
213         if (!strcmp(var, "diff.external"))
214                 return git_config_string(&external_diff_cmd_cfg, var, value);
215         if (!strcmp(var, "diff.wordregex"))
216                 return git_config_string(&diff_word_regex_cfg, var, value);
217         if (!strcmp(var, "diff.orderfile"))
218                 return git_config_pathname(&diff_order_file_cfg, var, value);
219
220         if (!strcmp(var, "diff.ignoresubmodules"))
221                 handle_ignore_submodules_arg(&default_diff_options, value);
222
223         if (!strcmp(var, "diff.submodule")) {
224                 if (parse_submodule_params(&default_diff_options, value))
225                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
226                                 value);
227                 return 0;
228         }
229
230         if (!strcmp(var, "diff.algorithm")) {
231                 diff_algorithm = parse_algorithm_value(value);
232                 if (diff_algorithm < 0)
233                         return -1;
234                 return 0;
235         }
236
237         if (git_color_config(var, value, cb) < 0)
238                 return -1;
239
240         return git_diff_basic_config(var, value, cb);
241 }
242
243 int git_diff_basic_config(const char *var, const char *value, void *cb)
244 {
245         const char *name;
246
247         if (!strcmp(var, "diff.renamelimit")) {
248                 diff_rename_limit_default = git_config_int(var, value);
249                 return 0;
250         }
251
252         if (userdiff_config(var, value) < 0)
253                 return -1;
254
255         if (skip_prefix(var, "diff.color.", &name) ||
256             skip_prefix(var, "color.diff.", &name)) {
257                 int slot = parse_diff_color_slot(name);
258                 if (slot < 0)
259                         return 0;
260                 if (!value)
261                         return config_error_nonbool(var);
262                 return color_parse(value, diff_colors[slot]);
263         }
264
265         /* like GNU diff's --suppress-blank-empty option  */
266         if (!strcmp(var, "diff.suppressblankempty") ||
267                         /* for backwards compatibility */
268                         !strcmp(var, "diff.suppress-blank-empty")) {
269                 diff_suppress_blank_empty = git_config_bool(var, value);
270                 return 0;
271         }
272
273         if (!strcmp(var, "diff.dirstat")) {
274                 struct strbuf errmsg = STRBUF_INIT;
275                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
276                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
277                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
278                                 errmsg.buf);
279                 strbuf_release(&errmsg);
280                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
281                 return 0;
282         }
283
284         if (starts_with(var, "submodule."))
285                 return parse_submodule_config_option(var, value);
286
287         return git_default_config(var, value, cb);
288 }
289
290 static char *quote_two(const char *one, const char *two)
291 {
292         int need_one = quote_c_style(one, NULL, NULL, 1);
293         int need_two = quote_c_style(two, NULL, NULL, 1);
294         struct strbuf res = STRBUF_INIT;
295
296         if (need_one + need_two) {
297                 strbuf_addch(&res, '"');
298                 quote_c_style(one, &res, NULL, 1);
299                 quote_c_style(two, &res, NULL, 1);
300                 strbuf_addch(&res, '"');
301         } else {
302                 strbuf_addstr(&res, one);
303                 strbuf_addstr(&res, two);
304         }
305         return strbuf_detach(&res, NULL);
306 }
307
308 static const char *external_diff(void)
309 {
310         static const char *external_diff_cmd = NULL;
311         static int done_preparing = 0;
312
313         if (done_preparing)
314                 return external_diff_cmd;
315         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
316         if (!external_diff_cmd)
317                 external_diff_cmd = external_diff_cmd_cfg;
318         done_preparing = 1;
319         return external_diff_cmd;
320 }
321
322 /*
323  * Keep track of files used for diffing. Sometimes such an entry
324  * refers to a temporary file, sometimes to an existing file, and
325  * sometimes to "/dev/null".
326  */
327 static struct diff_tempfile {
328         /*
329          * filename external diff should read from, or NULL if this
330          * entry is currently not in use:
331          */
332         const char *name;
333
334         char hex[GIT_SHA1_HEXSZ + 1];
335         char mode[10];
336
337         /*
338          * If this diff_tempfile instance refers to a temporary file,
339          * this tempfile object is used to manage its lifetime.
340          */
341         struct tempfile tempfile;
342 } diff_temp[2];
343
344 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
345
346 struct emit_callback {
347         int color_diff;
348         unsigned ws_rule;
349         int blank_at_eof_in_preimage;
350         int blank_at_eof_in_postimage;
351         int lno_in_preimage;
352         int lno_in_postimage;
353         sane_truncate_fn truncate;
354         const char **label_path;
355         struct diff_words_data *diff_words;
356         struct diff_options *opt;
357         struct strbuf *header;
358 };
359
360 static int count_lines(const char *data, int size)
361 {
362         int count, ch, completely_empty = 1, nl_just_seen = 0;
363         count = 0;
364         while (0 < size--) {
365                 ch = *data++;
366                 if (ch == '\n') {
367                         count++;
368                         nl_just_seen = 1;
369                         completely_empty = 0;
370                 }
371                 else {
372                         nl_just_seen = 0;
373                         completely_empty = 0;
374                 }
375         }
376         if (completely_empty)
377                 return 0;
378         if (!nl_just_seen)
379                 count++; /* no trailing newline */
380         return count;
381 }
382
383 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
384 {
385         if (!DIFF_FILE_VALID(one)) {
386                 mf->ptr = (char *)""; /* does not matter */
387                 mf->size = 0;
388                 return 0;
389         }
390         else if (diff_populate_filespec(one, 0))
391                 return -1;
392
393         mf->ptr = one->data;
394         mf->size = one->size;
395         return 0;
396 }
397
398 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
399 static unsigned long diff_filespec_size(struct diff_filespec *one)
400 {
401         if (!DIFF_FILE_VALID(one))
402                 return 0;
403         diff_populate_filespec(one, CHECK_SIZE_ONLY);
404         return one->size;
405 }
406
407 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
408 {
409         char *ptr = mf->ptr;
410         long size = mf->size;
411         int cnt = 0;
412
413         if (!size)
414                 return cnt;
415         ptr += size - 1; /* pointing at the very end */
416         if (*ptr != '\n')
417                 ; /* incomplete line */
418         else
419                 ptr--; /* skip the last LF */
420         while (mf->ptr < ptr) {
421                 char *prev_eol;
422                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
423                         if (*prev_eol == '\n')
424                                 break;
425                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
426                         break;
427                 cnt++;
428                 ptr = prev_eol - 1;
429         }
430         return cnt;
431 }
432
433 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
434                                struct emit_callback *ecbdata)
435 {
436         int l1, l2, at;
437         unsigned ws_rule = ecbdata->ws_rule;
438         l1 = count_trailing_blank(mf1, ws_rule);
439         l2 = count_trailing_blank(mf2, ws_rule);
440         if (l2 <= l1) {
441                 ecbdata->blank_at_eof_in_preimage = 0;
442                 ecbdata->blank_at_eof_in_postimage = 0;
443                 return;
444         }
445         at = count_lines(mf1->ptr, mf1->size);
446         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
447
448         at = count_lines(mf2->ptr, mf2->size);
449         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
450 }
451
452 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
453                         int first, const char *line, int len)
454 {
455         int has_trailing_newline, has_trailing_carriage_return;
456         int nofirst;
457         FILE *file = o->file;
458
459         fputs(diff_line_prefix(o), file);
460
461         if (len == 0) {
462                 has_trailing_newline = (first == '\n');
463                 has_trailing_carriage_return = (!has_trailing_newline &&
464                                                 (first == '\r'));
465                 nofirst = has_trailing_newline || has_trailing_carriage_return;
466         } else {
467                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
468                 if (has_trailing_newline)
469                         len--;
470                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
471                 if (has_trailing_carriage_return)
472                         len--;
473                 nofirst = 0;
474         }
475
476         if (len || !nofirst) {
477                 fputs(set, file);
478                 if (!nofirst)
479                         fputc(first, file);
480                 fwrite(line, len, 1, file);
481                 fputs(reset, file);
482         }
483         if (has_trailing_carriage_return)
484                 fputc('\r', file);
485         if (has_trailing_newline)
486                 fputc('\n', file);
487 }
488
489 static void emit_line(struct diff_options *o, const char *set, const char *reset,
490                       const char *line, int len)
491 {
492         emit_line_0(o, set, reset, line[0], line+1, len-1);
493 }
494
495 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
496 {
497         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
498               ecbdata->blank_at_eof_in_preimage &&
499               ecbdata->blank_at_eof_in_postimage &&
500               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
501               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
502                 return 0;
503         return ws_blank_line(line, len, ecbdata->ws_rule);
504 }
505
506 static void emit_line_checked(const char *reset,
507                               struct emit_callback *ecbdata,
508                               const char *line, int len,
509                               enum color_diff color,
510                               unsigned ws_error_highlight,
511                               char sign)
512 {
513         const char *set = diff_get_color(ecbdata->color_diff, color);
514         const char *ws = NULL;
515
516         if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
517                 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
518                 if (!*ws)
519                         ws = NULL;
520         }
521
522         if (!ws)
523                 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
524         else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
525                 /* Blank line at EOF - paint '+' as well */
526                 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
527         else {
528                 /* Emit just the prefix, then the rest. */
529                 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
530                 ws_check_emit(line, len, ecbdata->ws_rule,
531                               ecbdata->opt->file, set, reset, ws);
532         }
533 }
534
535 static void emit_add_line(const char *reset,
536                           struct emit_callback *ecbdata,
537                           const char *line, int len)
538 {
539         emit_line_checked(reset, ecbdata, line, len,
540                           DIFF_FILE_NEW, WSEH_NEW, '+');
541 }
542
543 static void emit_del_line(const char *reset,
544                           struct emit_callback *ecbdata,
545                           const char *line, int len)
546 {
547         emit_line_checked(reset, ecbdata, line, len,
548                           DIFF_FILE_OLD, WSEH_OLD, '-');
549 }
550
551 static void emit_context_line(const char *reset,
552                               struct emit_callback *ecbdata,
553                               const char *line, int len)
554 {
555         emit_line_checked(reset, ecbdata, line, len,
556                           DIFF_CONTEXT, WSEH_CONTEXT, ' ');
557 }
558
559 static void emit_hunk_header(struct emit_callback *ecbdata,
560                              const char *line, int len)
561 {
562         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
563         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
564         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
565         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
566         static const char atat[2] = { '@', '@' };
567         const char *cp, *ep;
568         struct strbuf msgbuf = STRBUF_INIT;
569         int org_len = len;
570         int i = 1;
571
572         /*
573          * As a hunk header must begin with "@@ -<old>, +<new> @@",
574          * it always is at least 10 bytes long.
575          */
576         if (len < 10 ||
577             memcmp(line, atat, 2) ||
578             !(ep = memmem(line + 2, len - 2, atat, 2))) {
579                 emit_line(ecbdata->opt, context, reset, line, len);
580                 return;
581         }
582         ep += 2; /* skip over @@ */
583
584         /* The hunk header in fraginfo color */
585         strbuf_addstr(&msgbuf, frag);
586         strbuf_add(&msgbuf, line, ep - line);
587         strbuf_addstr(&msgbuf, reset);
588
589         /*
590          * trailing "\r\n"
591          */
592         for ( ; i < 3; i++)
593                 if (line[len - i] == '\r' || line[len - i] == '\n')
594                         len--;
595
596         /* blank before the func header */
597         for (cp = ep; ep - line < len; ep++)
598                 if (*ep != ' ' && *ep != '\t')
599                         break;
600         if (ep != cp) {
601                 strbuf_addstr(&msgbuf, context);
602                 strbuf_add(&msgbuf, cp, ep - cp);
603                 strbuf_addstr(&msgbuf, reset);
604         }
605
606         if (ep < line + len) {
607                 strbuf_addstr(&msgbuf, func);
608                 strbuf_add(&msgbuf, ep, line + len - ep);
609                 strbuf_addstr(&msgbuf, reset);
610         }
611
612         strbuf_add(&msgbuf, line + len, org_len - len);
613         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
614         strbuf_release(&msgbuf);
615 }
616
617 static struct diff_tempfile *claim_diff_tempfile(void) {
618         int i;
619         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
620                 if (!diff_temp[i].name)
621                         return diff_temp + i;
622         die("BUG: diff is failing to clean up its tempfiles");
623 }
624
625 static void remove_tempfile(void)
626 {
627         int i;
628         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
629                 if (is_tempfile_active(&diff_temp[i].tempfile))
630                         delete_tempfile(&diff_temp[i].tempfile);
631                 diff_temp[i].name = NULL;
632         }
633 }
634
635 static void print_line_count(FILE *file, int count)
636 {
637         switch (count) {
638         case 0:
639                 fprintf(file, "0,0");
640                 break;
641         case 1:
642                 fprintf(file, "1");
643                 break;
644         default:
645                 fprintf(file, "1,%d", count);
646                 break;
647         }
648 }
649
650 static void emit_rewrite_lines(struct emit_callback *ecb,
651                                int prefix, const char *data, int size)
652 {
653         const char *endp = NULL;
654         static const char *nneof = " No newline at end of file\n";
655         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
656
657         while (0 < size) {
658                 int len;
659
660                 endp = memchr(data, '\n', size);
661                 len = endp ? (endp - data + 1) : size;
662                 if (prefix != '+') {
663                         ecb->lno_in_preimage++;
664                         emit_del_line(reset, ecb, data, len);
665                 } else {
666                         ecb->lno_in_postimage++;
667                         emit_add_line(reset, ecb, data, len);
668                 }
669                 size -= len;
670                 data += len;
671         }
672         if (!endp) {
673                 const char *context = diff_get_color(ecb->color_diff,
674                                                      DIFF_CONTEXT);
675                 putc('\n', ecb->opt->file);
676                 emit_line_0(ecb->opt, context, reset, '\\',
677                             nneof, strlen(nneof));
678         }
679 }
680
681 static void emit_rewrite_diff(const char *name_a,
682                               const char *name_b,
683                               struct diff_filespec *one,
684                               struct diff_filespec *two,
685                               struct userdiff_driver *textconv_one,
686                               struct userdiff_driver *textconv_two,
687                               struct diff_options *o)
688 {
689         int lc_a, lc_b;
690         const char *name_a_tab, *name_b_tab;
691         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
692         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
693         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
694         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
695         const char *a_prefix, *b_prefix;
696         char *data_one, *data_two;
697         size_t size_one, size_two;
698         struct emit_callback ecbdata;
699         const char *line_prefix = diff_line_prefix(o);
700
701         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
702                 a_prefix = o->b_prefix;
703                 b_prefix = o->a_prefix;
704         } else {
705                 a_prefix = o->a_prefix;
706                 b_prefix = o->b_prefix;
707         }
708
709         name_a += (*name_a == '/');
710         name_b += (*name_b == '/');
711         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
712         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
713
714         strbuf_reset(&a_name);
715         strbuf_reset(&b_name);
716         quote_two_c_style(&a_name, a_prefix, name_a, 0);
717         quote_two_c_style(&b_name, b_prefix, name_b, 0);
718
719         size_one = fill_textconv(textconv_one, one, &data_one);
720         size_two = fill_textconv(textconv_two, two, &data_two);
721
722         memset(&ecbdata, 0, sizeof(ecbdata));
723         ecbdata.color_diff = want_color(o->use_color);
724         ecbdata.ws_rule = whitespace_rule(name_b);
725         ecbdata.opt = o;
726         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
727                 mmfile_t mf1, mf2;
728                 mf1.ptr = (char *)data_one;
729                 mf2.ptr = (char *)data_two;
730                 mf1.size = size_one;
731                 mf2.size = size_two;
732                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
733         }
734         ecbdata.lno_in_preimage = 1;
735         ecbdata.lno_in_postimage = 1;
736
737         lc_a = count_lines(data_one, size_one);
738         lc_b = count_lines(data_two, size_two);
739         fprintf(o->file,
740                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
741                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
742                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
743                 line_prefix, fraginfo);
744         if (!o->irreversible_delete)
745                 print_line_count(o->file, lc_a);
746         else
747                 fprintf(o->file, "?,?");
748         fprintf(o->file, " +");
749         print_line_count(o->file, lc_b);
750         fprintf(o->file, " @@%s\n", reset);
751         if (lc_a && !o->irreversible_delete)
752                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
753         if (lc_b)
754                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
755         if (textconv_one)
756                 free((char *)data_one);
757         if (textconv_two)
758                 free((char *)data_two);
759 }
760
761 struct diff_words_buffer {
762         mmfile_t text;
763         long alloc;
764         struct diff_words_orig {
765                 const char *begin, *end;
766         } *orig;
767         int orig_nr, orig_alloc;
768 };
769
770 static void diff_words_append(char *line, unsigned long len,
771                 struct diff_words_buffer *buffer)
772 {
773         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
774         line++;
775         len--;
776         memcpy(buffer->text.ptr + buffer->text.size, line, len);
777         buffer->text.size += len;
778         buffer->text.ptr[buffer->text.size] = '\0';
779 }
780
781 struct diff_words_style_elem {
782         const char *prefix;
783         const char *suffix;
784         const char *color; /* NULL; filled in by the setup code if
785                             * color is enabled */
786 };
787
788 struct diff_words_style {
789         enum diff_words_type type;
790         struct diff_words_style_elem new, old, ctx;
791         const char *newline;
792 };
793
794 static struct diff_words_style diff_words_styles[] = {
795         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
796         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
797         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
798 };
799
800 struct diff_words_data {
801         struct diff_words_buffer minus, plus;
802         const char *current_plus;
803         int last_minus;
804         struct diff_options *opt;
805         regex_t *word_regex;
806         enum diff_words_type type;
807         struct diff_words_style *style;
808 };
809
810 static int fn_out_diff_words_write_helper(FILE *fp,
811                                           struct diff_words_style_elem *st_el,
812                                           const char *newline,
813                                           size_t count, const char *buf,
814                                           const char *line_prefix)
815 {
816         int print = 0;
817
818         while (count) {
819                 char *p = memchr(buf, '\n', count);
820                 if (print)
821                         fputs(line_prefix, fp);
822                 if (p != buf) {
823                         if (st_el->color && fputs(st_el->color, fp) < 0)
824                                 return -1;
825                         if (fputs(st_el->prefix, fp) < 0 ||
826                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
827                             fputs(st_el->suffix, fp) < 0)
828                                 return -1;
829                         if (st_el->color && *st_el->color
830                             && fputs(GIT_COLOR_RESET, fp) < 0)
831                                 return -1;
832                 }
833                 if (!p)
834                         return 0;
835                 if (fputs(newline, fp) < 0)
836                         return -1;
837                 count -= p + 1 - buf;
838                 buf = p + 1;
839                 print = 1;
840         }
841         return 0;
842 }
843
844 /*
845  * '--color-words' algorithm can be described as:
846  *
847  *   1. collect a the minus/plus lines of a diff hunk, divided into
848  *      minus-lines and plus-lines;
849  *
850  *   2. break both minus-lines and plus-lines into words and
851  *      place them into two mmfile_t with one word for each line;
852  *
853  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
854  *
855  * And for the common parts of the both file, we output the plus side text.
856  * diff_words->current_plus is used to trace the current position of the plus file
857  * which printed. diff_words->last_minus is used to trace the last minus word
858  * printed.
859  *
860  * For '--graph' to work with '--color-words', we need to output the graph prefix
861  * on each line of color words output. Generally, there are two conditions on
862  * which we should output the prefix.
863  *
864  *   1. diff_words->last_minus == 0 &&
865  *      diff_words->current_plus == diff_words->plus.text.ptr
866  *
867  *      that is: the plus text must start as a new line, and if there is no minus
868  *      word printed, a graph prefix must be printed.
869  *
870  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
871  *      *(diff_words->current_plus - 1) == '\n'
872  *
873  *      that is: a graph prefix must be printed following a '\n'
874  */
875 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
876 {
877         if ((diff_words->last_minus == 0 &&
878                 diff_words->current_plus == diff_words->plus.text.ptr) ||
879                 (diff_words->current_plus > diff_words->plus.text.ptr &&
880                 *(diff_words->current_plus - 1) == '\n')) {
881                 return 1;
882         } else {
883                 return 0;
884         }
885 }
886
887 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
888 {
889         struct diff_words_data *diff_words = priv;
890         struct diff_words_style *style = diff_words->style;
891         int minus_first, minus_len, plus_first, plus_len;
892         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
893         struct diff_options *opt = diff_words->opt;
894         const char *line_prefix;
895
896         if (line[0] != '@' || parse_hunk_header(line, len,
897                         &minus_first, &minus_len, &plus_first, &plus_len))
898                 return;
899
900         assert(opt);
901         line_prefix = diff_line_prefix(opt);
902
903         /* POSIX requires that first be decremented by one if len == 0... */
904         if (minus_len) {
905                 minus_begin = diff_words->minus.orig[minus_first].begin;
906                 minus_end =
907                         diff_words->minus.orig[minus_first + minus_len - 1].end;
908         } else
909                 minus_begin = minus_end =
910                         diff_words->minus.orig[minus_first].end;
911
912         if (plus_len) {
913                 plus_begin = diff_words->plus.orig[plus_first].begin;
914                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
915         } else
916                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
917
918         if (color_words_output_graph_prefix(diff_words)) {
919                 fputs(line_prefix, diff_words->opt->file);
920         }
921         if (diff_words->current_plus != plus_begin) {
922                 fn_out_diff_words_write_helper(diff_words->opt->file,
923                                 &style->ctx, style->newline,
924                                 plus_begin - diff_words->current_plus,
925                                 diff_words->current_plus, line_prefix);
926                 if (*(plus_begin - 1) == '\n')
927                         fputs(line_prefix, diff_words->opt->file);
928         }
929         if (minus_begin != minus_end) {
930                 fn_out_diff_words_write_helper(diff_words->opt->file,
931                                 &style->old, style->newline,
932                                 minus_end - minus_begin, minus_begin,
933                                 line_prefix);
934         }
935         if (plus_begin != plus_end) {
936                 fn_out_diff_words_write_helper(diff_words->opt->file,
937                                 &style->new, style->newline,
938                                 plus_end - plus_begin, plus_begin,
939                                 line_prefix);
940         }
941
942         diff_words->current_plus = plus_end;
943         diff_words->last_minus = minus_first;
944 }
945
946 /* This function starts looking at *begin, and returns 0 iff a word was found. */
947 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
948                 int *begin, int *end)
949 {
950         if (word_regex && *begin < buffer->size) {
951                 regmatch_t match[1];
952                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
953                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
954                                         '\n', match[0].rm_eo - match[0].rm_so);
955                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
956                         *begin += match[0].rm_so;
957                         return *begin >= *end;
958                 }
959                 return -1;
960         }
961
962         /* find the next word */
963         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
964                 (*begin)++;
965         if (*begin >= buffer->size)
966                 return -1;
967
968         /* find the end of the word */
969         *end = *begin + 1;
970         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
971                 (*end)++;
972
973         return 0;
974 }
975
976 /*
977  * This function splits the words in buffer->text, stores the list with
978  * newline separator into out, and saves the offsets of the original words
979  * in buffer->orig.
980  */
981 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
982                 regex_t *word_regex)
983 {
984         int i, j;
985         long alloc = 0;
986
987         out->size = 0;
988         out->ptr = NULL;
989
990         /* fake an empty "0th" word */
991         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
992         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
993         buffer->orig_nr = 1;
994
995         for (i = 0; i < buffer->text.size; i++) {
996                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
997                         return;
998
999                 /* store original boundaries */
1000                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1001                                 buffer->orig_alloc);
1002                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1003                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1004                 buffer->orig_nr++;
1005
1006                 /* store one word */
1007                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1008                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1009                 out->ptr[out->size + j - i] = '\n';
1010                 out->size += j - i + 1;
1011
1012                 i = j - 1;
1013         }
1014 }
1015
1016 /* this executes the word diff on the accumulated buffers */
1017 static void diff_words_show(struct diff_words_data *diff_words)
1018 {
1019         xpparam_t xpp;
1020         xdemitconf_t xecfg;
1021         mmfile_t minus, plus;
1022         struct diff_words_style *style = diff_words->style;
1023
1024         struct diff_options *opt = diff_words->opt;
1025         const char *line_prefix;
1026
1027         assert(opt);
1028         line_prefix = diff_line_prefix(opt);
1029
1030         /* special case: only removal */
1031         if (!diff_words->plus.text.size) {
1032                 fputs(line_prefix, diff_words->opt->file);
1033                 fn_out_diff_words_write_helper(diff_words->opt->file,
1034                         &style->old, style->newline,
1035                         diff_words->minus.text.size,
1036                         diff_words->minus.text.ptr, line_prefix);
1037                 diff_words->minus.text.size = 0;
1038                 return;
1039         }
1040
1041         diff_words->current_plus = diff_words->plus.text.ptr;
1042         diff_words->last_minus = 0;
1043
1044         memset(&xpp, 0, sizeof(xpp));
1045         memset(&xecfg, 0, sizeof(xecfg));
1046         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1047         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1048         xpp.flags = 0;
1049         /* as only the hunk header will be parsed, we need a 0-context */
1050         xecfg.ctxlen = 0;
1051         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1052                           &xpp, &xecfg))
1053                 die("unable to generate word diff");
1054         free(minus.ptr);
1055         free(plus.ptr);
1056         if (diff_words->current_plus != diff_words->plus.text.ptr +
1057                         diff_words->plus.text.size) {
1058                 if (color_words_output_graph_prefix(diff_words))
1059                         fputs(line_prefix, diff_words->opt->file);
1060                 fn_out_diff_words_write_helper(diff_words->opt->file,
1061                         &style->ctx, style->newline,
1062                         diff_words->plus.text.ptr + diff_words->plus.text.size
1063                         - diff_words->current_plus, diff_words->current_plus,
1064                         line_prefix);
1065         }
1066         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1067 }
1068
1069 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1070 static void diff_words_flush(struct emit_callback *ecbdata)
1071 {
1072         if (ecbdata->diff_words->minus.text.size ||
1073             ecbdata->diff_words->plus.text.size)
1074                 diff_words_show(ecbdata->diff_words);
1075 }
1076
1077 static void diff_filespec_load_driver(struct diff_filespec *one)
1078 {
1079         /* Use already-loaded driver */
1080         if (one->driver)
1081                 return;
1082
1083         if (S_ISREG(one->mode))
1084                 one->driver = userdiff_find_by_path(one->path);
1085
1086         /* Fallback to default settings */
1087         if (!one->driver)
1088                 one->driver = userdiff_find_by_name("default");
1089 }
1090
1091 static const char *userdiff_word_regex(struct diff_filespec *one)
1092 {
1093         diff_filespec_load_driver(one);
1094         return one->driver->word_regex;
1095 }
1096
1097 static void init_diff_words_data(struct emit_callback *ecbdata,
1098                                  struct diff_options *orig_opts,
1099                                  struct diff_filespec *one,
1100                                  struct diff_filespec *two)
1101 {
1102         int i;
1103         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1104         memcpy(o, orig_opts, sizeof(struct diff_options));
1105
1106         ecbdata->diff_words =
1107                 xcalloc(1, sizeof(struct diff_words_data));
1108         ecbdata->diff_words->type = o->word_diff;
1109         ecbdata->diff_words->opt = o;
1110         if (!o->word_regex)
1111                 o->word_regex = userdiff_word_regex(one);
1112         if (!o->word_regex)
1113                 o->word_regex = userdiff_word_regex(two);
1114         if (!o->word_regex)
1115                 o->word_regex = diff_word_regex_cfg;
1116         if (o->word_regex) {
1117                 ecbdata->diff_words->word_regex = (regex_t *)
1118                         xmalloc(sizeof(regex_t));
1119                 if (regcomp(ecbdata->diff_words->word_regex,
1120                             o->word_regex,
1121                             REG_EXTENDED | REG_NEWLINE))
1122                         die ("Invalid regular expression: %s",
1123                              o->word_regex);
1124         }
1125         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1126                 if (o->word_diff == diff_words_styles[i].type) {
1127                         ecbdata->diff_words->style =
1128                                 &diff_words_styles[i];
1129                         break;
1130                 }
1131         }
1132         if (want_color(o->use_color)) {
1133                 struct diff_words_style *st = ecbdata->diff_words->style;
1134                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1135                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1136                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1137         }
1138 }
1139
1140 static void free_diff_words_data(struct emit_callback *ecbdata)
1141 {
1142         if (ecbdata->diff_words) {
1143                 diff_words_flush(ecbdata);
1144                 free (ecbdata->diff_words->opt);
1145                 free (ecbdata->diff_words->minus.text.ptr);
1146                 free (ecbdata->diff_words->minus.orig);
1147                 free (ecbdata->diff_words->plus.text.ptr);
1148                 free (ecbdata->diff_words->plus.orig);
1149                 if (ecbdata->diff_words->word_regex) {
1150                         regfree(ecbdata->diff_words->word_regex);
1151                         free(ecbdata->diff_words->word_regex);
1152                 }
1153                 free(ecbdata->diff_words);
1154                 ecbdata->diff_words = NULL;
1155         }
1156 }
1157
1158 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1159 {
1160         if (want_color(diff_use_color))
1161                 return diff_colors[ix];
1162         return "";
1163 }
1164
1165 const char *diff_line_prefix(struct diff_options *opt)
1166 {
1167         struct strbuf *msgbuf;
1168         if (!opt->output_prefix)
1169                 return "";
1170
1171         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1172         return msgbuf->buf;
1173 }
1174
1175 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1176 {
1177         const char *cp;
1178         unsigned long allot;
1179         size_t l = len;
1180
1181         if (ecb->truncate)
1182                 return ecb->truncate(line, len);
1183         cp = line;
1184         allot = l;
1185         while (0 < l) {
1186                 (void) utf8_width(&cp, &l);
1187                 if (!cp)
1188                         break; /* truncated in the middle? */
1189         }
1190         return allot - l;
1191 }
1192
1193 static void find_lno(const char *line, struct emit_callback *ecbdata)
1194 {
1195         const char *p;
1196         ecbdata->lno_in_preimage = 0;
1197         ecbdata->lno_in_postimage = 0;
1198         p = strchr(line, '-');
1199         if (!p)
1200                 return; /* cannot happen */
1201         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1202         p = strchr(p, '+');
1203         if (!p)
1204                 return; /* cannot happen */
1205         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1206 }
1207
1208 static void fn_out_consume(void *priv, char *line, unsigned long len)
1209 {
1210         struct emit_callback *ecbdata = priv;
1211         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1212         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1213         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1214         struct diff_options *o = ecbdata->opt;
1215         const char *line_prefix = diff_line_prefix(o);
1216
1217         o->found_changes = 1;
1218
1219         if (ecbdata->header) {
1220                 fprintf(o->file, "%s", ecbdata->header->buf);
1221                 strbuf_reset(ecbdata->header);
1222                 ecbdata->header = NULL;
1223         }
1224
1225         if (ecbdata->label_path[0]) {
1226                 const char *name_a_tab, *name_b_tab;
1227
1228                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1229                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1230
1231                 fprintf(o->file, "%s%s--- %s%s%s\n",
1232                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1233                 fprintf(o->file, "%s%s+++ %s%s%s\n",
1234                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1235                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1236         }
1237
1238         if (diff_suppress_blank_empty
1239             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1240                 line[0] = '\n';
1241                 len = 1;
1242         }
1243
1244         if (line[0] == '@') {
1245                 if (ecbdata->diff_words)
1246                         diff_words_flush(ecbdata);
1247                 len = sane_truncate_line(ecbdata, line, len);
1248                 find_lno(line, ecbdata);
1249                 emit_hunk_header(ecbdata, line, len);
1250                 if (line[len-1] != '\n')
1251                         putc('\n', o->file);
1252                 return;
1253         }
1254
1255         if (ecbdata->diff_words) {
1256                 if (line[0] == '-') {
1257                         diff_words_append(line, len,
1258                                           &ecbdata->diff_words->minus);
1259                         return;
1260                 } else if (line[0] == '+') {
1261                         diff_words_append(line, len,
1262                                           &ecbdata->diff_words->plus);
1263                         return;
1264                 } else if (starts_with(line, "\\ ")) {
1265                         /*
1266                          * Eat the "no newline at eof" marker as if we
1267                          * saw a "+" or "-" line with nothing on it,
1268                          * and return without diff_words_flush() to
1269                          * defer processing. If this is the end of
1270                          * preimage, more "+" lines may come after it.
1271                          */
1272                         return;
1273                 }
1274                 diff_words_flush(ecbdata);
1275                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1276                         emit_line(o, context, reset, line, len);
1277                         fputs("~\n", o->file);
1278                 } else {
1279                         /*
1280                          * Skip the prefix character, if any.  With
1281                          * diff_suppress_blank_empty, there may be
1282                          * none.
1283                          */
1284                         if (line[0] != '\n') {
1285                               line++;
1286                               len--;
1287                         }
1288                         emit_line(o, context, reset, line, len);
1289                 }
1290                 return;
1291         }
1292
1293         switch (line[0]) {
1294         case '+':
1295                 ecbdata->lno_in_postimage++;
1296                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1297                 break;
1298         case '-':
1299                 ecbdata->lno_in_preimage++;
1300                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1301                 break;
1302         case ' ':
1303                 ecbdata->lno_in_postimage++;
1304                 ecbdata->lno_in_preimage++;
1305                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1306                 break;
1307         default:
1308                 /* incomplete line at the end */
1309                 ecbdata->lno_in_preimage++;
1310                 emit_line(o, diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1311                           reset, line, len);
1312                 break;
1313         }
1314 }
1315
1316 static char *pprint_rename(const char *a, const char *b)
1317 {
1318         const char *old = a;
1319         const char *new = b;
1320         struct strbuf name = STRBUF_INIT;
1321         int pfx_length, sfx_length;
1322         int pfx_adjust_for_slash;
1323         int len_a = strlen(a);
1324         int len_b = strlen(b);
1325         int a_midlen, b_midlen;
1326         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1327         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1328
1329         if (qlen_a || qlen_b) {
1330                 quote_c_style(a, &name, NULL, 0);
1331                 strbuf_addstr(&name, " => ");
1332                 quote_c_style(b, &name, NULL, 0);
1333                 return strbuf_detach(&name, NULL);
1334         }
1335
1336         /* Find common prefix */
1337         pfx_length = 0;
1338         while (*old && *new && *old == *new) {
1339                 if (*old == '/')
1340                         pfx_length = old - a + 1;
1341                 old++;
1342                 new++;
1343         }
1344
1345         /* Find common suffix */
1346         old = a + len_a;
1347         new = b + len_b;
1348         sfx_length = 0;
1349         /*
1350          * If there is a common prefix, it must end in a slash.  In
1351          * that case we let this loop run 1 into the prefix to see the
1352          * same slash.
1353          *
1354          * If there is no common prefix, we cannot do this as it would
1355          * underrun the input strings.
1356          */
1357         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1358         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1359                b + pfx_length - pfx_adjust_for_slash <= new &&
1360                *old == *new) {
1361                 if (*old == '/')
1362                         sfx_length = len_a - (old - a);
1363                 old--;
1364                 new--;
1365         }
1366
1367         /*
1368          * pfx{mid-a => mid-b}sfx
1369          * {pfx-a => pfx-b}sfx
1370          * pfx{sfx-a => sfx-b}
1371          * name-a => name-b
1372          */
1373         a_midlen = len_a - pfx_length - sfx_length;
1374         b_midlen = len_b - pfx_length - sfx_length;
1375         if (a_midlen < 0)
1376                 a_midlen = 0;
1377         if (b_midlen < 0)
1378                 b_midlen = 0;
1379
1380         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1381         if (pfx_length + sfx_length) {
1382                 strbuf_add(&name, a, pfx_length);
1383                 strbuf_addch(&name, '{');
1384         }
1385         strbuf_add(&name, a + pfx_length, a_midlen);
1386         strbuf_addstr(&name, " => ");
1387         strbuf_add(&name, b + pfx_length, b_midlen);
1388         if (pfx_length + sfx_length) {
1389                 strbuf_addch(&name, '}');
1390                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1391         }
1392         return strbuf_detach(&name, NULL);
1393 }
1394
1395 struct diffstat_t {
1396         int nr;
1397         int alloc;
1398         struct diffstat_file {
1399                 char *from_name;
1400                 char *name;
1401                 char *print_name;
1402                 unsigned is_unmerged:1;
1403                 unsigned is_binary:1;
1404                 unsigned is_renamed:1;
1405                 unsigned is_interesting:1;
1406                 uintmax_t added, deleted;
1407         } **files;
1408 };
1409
1410 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1411                                           const char *name_a,
1412                                           const char *name_b)
1413 {
1414         struct diffstat_file *x;
1415         x = xcalloc(1, sizeof(*x));
1416         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1417         diffstat->files[diffstat->nr++] = x;
1418         if (name_b) {
1419                 x->from_name = xstrdup(name_a);
1420                 x->name = xstrdup(name_b);
1421                 x->is_renamed = 1;
1422         }
1423         else {
1424                 x->from_name = NULL;
1425                 x->name = xstrdup(name_a);
1426         }
1427         return x;
1428 }
1429
1430 static void diffstat_consume(void *priv, char *line, unsigned long len)
1431 {
1432         struct diffstat_t *diffstat = priv;
1433         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1434
1435         if (line[0] == '+')
1436                 x->added++;
1437         else if (line[0] == '-')
1438                 x->deleted++;
1439 }
1440
1441 const char mime_boundary_leader[] = "------------";
1442
1443 static int scale_linear(int it, int width, int max_change)
1444 {
1445         if (!it)
1446                 return 0;
1447         /*
1448          * make sure that at least one '-' or '+' is printed if
1449          * there is any change to this path. The easiest way is to
1450          * scale linearly as if the alloted width is one column shorter
1451          * than it is, and then add 1 to the result.
1452          */
1453         return 1 + (it * (width - 1) / max_change);
1454 }
1455
1456 static void show_name(FILE *file,
1457                       const char *prefix, const char *name, int len)
1458 {
1459         fprintf(file, " %s%-*s |", prefix, len, name);
1460 }
1461
1462 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1463 {
1464         if (cnt <= 0)
1465                 return;
1466         fprintf(file, "%s", set);
1467         while (cnt--)
1468                 putc(ch, file);
1469         fprintf(file, "%s", reset);
1470 }
1471
1472 static void fill_print_name(struct diffstat_file *file)
1473 {
1474         char *pname;
1475
1476         if (file->print_name)
1477                 return;
1478
1479         if (!file->is_renamed) {
1480                 struct strbuf buf = STRBUF_INIT;
1481                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1482                         pname = strbuf_detach(&buf, NULL);
1483                 } else {
1484                         pname = file->name;
1485                         strbuf_release(&buf);
1486                 }
1487         } else {
1488                 pname = pprint_rename(file->from_name, file->name);
1489         }
1490         file->print_name = pname;
1491 }
1492
1493 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1494 {
1495         struct strbuf sb = STRBUF_INIT;
1496         int ret;
1497
1498         if (!files) {
1499                 assert(insertions == 0 && deletions == 0);
1500                 return fprintf(fp, "%s\n", " 0 files changed");
1501         }
1502
1503         strbuf_addf(&sb,
1504                     (files == 1) ? " %d file changed" : " %d files changed",
1505                     files);
1506
1507         /*
1508          * For binary diff, the caller may want to print "x files
1509          * changed" with insertions == 0 && deletions == 0.
1510          *
1511          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1512          * is probably less confusing (i.e skip over "2 files changed
1513          * but nothing about added/removed lines? Is this a bug in Git?").
1514          */
1515         if (insertions || deletions == 0) {
1516                 strbuf_addf(&sb,
1517                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1518                             insertions);
1519         }
1520
1521         if (deletions || insertions == 0) {
1522                 strbuf_addf(&sb,
1523                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1524                             deletions);
1525         }
1526         strbuf_addch(&sb, '\n');
1527         ret = fputs(sb.buf, fp);
1528         strbuf_release(&sb);
1529         return ret;
1530 }
1531
1532 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1533 {
1534         int i, len, add, del, adds = 0, dels = 0;
1535         uintmax_t max_change = 0, max_len = 0;
1536         int total_files = data->nr, count;
1537         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1538         const char *reset, *add_c, *del_c;
1539         const char *line_prefix = "";
1540         int extra_shown = 0;
1541
1542         if (data->nr == 0)
1543                 return;
1544
1545         line_prefix = diff_line_prefix(options);
1546         count = options->stat_count ? options->stat_count : data->nr;
1547
1548         reset = diff_get_color_opt(options, DIFF_RESET);
1549         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1550         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1551
1552         /*
1553          * Find the longest filename and max number of changes
1554          */
1555         for (i = 0; (i < count) && (i < data->nr); i++) {
1556                 struct diffstat_file *file = data->files[i];
1557                 uintmax_t change = file->added + file->deleted;
1558
1559                 if (!file->is_interesting && (change == 0)) {
1560                         count++; /* not shown == room for one more */
1561                         continue;
1562                 }
1563                 fill_print_name(file);
1564                 len = strlen(file->print_name);
1565                 if (max_len < len)
1566                         max_len = len;
1567
1568                 if (file->is_unmerged) {
1569                         /* "Unmerged" is 8 characters */
1570                         bin_width = bin_width < 8 ? 8 : bin_width;
1571                         continue;
1572                 }
1573                 if (file->is_binary) {
1574                         /* "Bin XXX -> YYY bytes" */
1575                         int w = 14 + decimal_width(file->added)
1576                                 + decimal_width(file->deleted);
1577                         bin_width = bin_width < w ? w : bin_width;
1578                         /* Display change counts aligned with "Bin" */
1579                         number_width = 3;
1580                         continue;
1581                 }
1582
1583                 if (max_change < change)
1584                         max_change = change;
1585         }
1586         count = i; /* where we can stop scanning in data->files[] */
1587
1588         /*
1589          * We have width = stat_width or term_columns() columns total.
1590          * We want a maximum of min(max_len, stat_name_width) for the name part.
1591          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1592          * We also need 1 for " " and 4 + decimal_width(max_change)
1593          * for " | NNNN " and one the empty column at the end, altogether
1594          * 6 + decimal_width(max_change).
1595          *
1596          * If there's not enough space, we will use the smaller of
1597          * stat_name_width (if set) and 5/8*width for the filename,
1598          * and the rest for constant elements + graph part, but no more
1599          * than stat_graph_width for the graph part.
1600          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1601          * for the standard terminal size).
1602          *
1603          * In other words: stat_width limits the maximum width, and
1604          * stat_name_width fixes the maximum width of the filename,
1605          * and is also used to divide available columns if there
1606          * aren't enough.
1607          *
1608          * Binary files are displayed with "Bin XXX -> YYY bytes"
1609          * instead of the change count and graph. This part is treated
1610          * similarly to the graph part, except that it is not
1611          * "scaled". If total width is too small to accommodate the
1612          * guaranteed minimum width of the filename part and the
1613          * separators and this message, this message will "overflow"
1614          * making the line longer than the maximum width.
1615          */
1616
1617         if (options->stat_width == -1)
1618                 width = term_columns() - options->output_prefix_length;
1619         else
1620                 width = options->stat_width ? options->stat_width : 80;
1621         number_width = decimal_width(max_change) > number_width ?
1622                 decimal_width(max_change) : number_width;
1623
1624         if (options->stat_graph_width == -1)
1625                 options->stat_graph_width = diff_stat_graph_width;
1626
1627         /*
1628          * Guarantee 3/8*16==6 for the graph part
1629          * and 5/8*16==10 for the filename part
1630          */
1631         if (width < 16 + 6 + number_width)
1632                 width = 16 + 6 + number_width;
1633
1634         /*
1635          * First assign sizes that are wanted, ignoring available width.
1636          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1637          * starting from "XXX" should fit in graph_width.
1638          */
1639         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1640         if (options->stat_graph_width &&
1641             options->stat_graph_width < graph_width)
1642                 graph_width = options->stat_graph_width;
1643
1644         name_width = (options->stat_name_width > 0 &&
1645                       options->stat_name_width < max_len) ?
1646                 options->stat_name_width : max_len;
1647
1648         /*
1649          * Adjust adjustable widths not to exceed maximum width
1650          */
1651         if (name_width + number_width + 6 + graph_width > width) {
1652                 if (graph_width > width * 3/8 - number_width - 6) {
1653                         graph_width = width * 3/8 - number_width - 6;
1654                         if (graph_width < 6)
1655                                 graph_width = 6;
1656                 }
1657
1658                 if (options->stat_graph_width &&
1659                     graph_width > options->stat_graph_width)
1660                         graph_width = options->stat_graph_width;
1661                 if (name_width > width - number_width - 6 - graph_width)
1662                         name_width = width - number_width - 6 - graph_width;
1663                 else
1664                         graph_width = width - number_width - 6 - name_width;
1665         }
1666
1667         /*
1668          * From here name_width is the width of the name area,
1669          * and graph_width is the width of the graph area.
1670          * max_change is used to scale graph properly.
1671          */
1672         for (i = 0; i < count; i++) {
1673                 const char *prefix = "";
1674                 struct diffstat_file *file = data->files[i];
1675                 char *name = file->print_name;
1676                 uintmax_t added = file->added;
1677                 uintmax_t deleted = file->deleted;
1678                 int name_len;
1679
1680                 if (!file->is_interesting && (added + deleted == 0))
1681                         continue;
1682
1683                 /*
1684                  * "scale" the filename
1685                  */
1686                 len = name_width;
1687                 name_len = strlen(name);
1688                 if (name_width < name_len) {
1689                         char *slash;
1690                         prefix = "...";
1691                         len -= 3;
1692                         name += name_len - len;
1693                         slash = strchr(name, '/');
1694                         if (slash)
1695                                 name = slash;
1696                 }
1697
1698                 if (file->is_binary) {
1699                         fprintf(options->file, "%s", line_prefix);
1700                         show_name(options->file, prefix, name, len);
1701                         fprintf(options->file, " %*s", number_width, "Bin");
1702                         if (!added && !deleted) {
1703                                 putc('\n', options->file);
1704                                 continue;
1705                         }
1706                         fprintf(options->file, " %s%"PRIuMAX"%s",
1707                                 del_c, deleted, reset);
1708                         fprintf(options->file, " -> ");
1709                         fprintf(options->file, "%s%"PRIuMAX"%s",
1710                                 add_c, added, reset);
1711                         fprintf(options->file, " bytes");
1712                         fprintf(options->file, "\n");
1713                         continue;
1714                 }
1715                 else if (file->is_unmerged) {
1716                         fprintf(options->file, "%s", line_prefix);
1717                         show_name(options->file, prefix, name, len);
1718                         fprintf(options->file, " Unmerged\n");
1719                         continue;
1720                 }
1721
1722                 /*
1723                  * scale the add/delete
1724                  */
1725                 add = added;
1726                 del = deleted;
1727
1728                 if (graph_width <= max_change) {
1729                         int total = scale_linear(add + del, graph_width, max_change);
1730                         if (total < 2 && add && del)
1731                                 /* width >= 2 due to the sanity check */
1732                                 total = 2;
1733                         if (add < del) {
1734                                 add = scale_linear(add, graph_width, max_change);
1735                                 del = total - add;
1736                         } else {
1737                                 del = scale_linear(del, graph_width, max_change);
1738                                 add = total - del;
1739                         }
1740                 }
1741                 fprintf(options->file, "%s", line_prefix);
1742                 show_name(options->file, prefix, name, len);
1743                 fprintf(options->file, " %*"PRIuMAX"%s",
1744                         number_width, added + deleted,
1745                         added + deleted ? " " : "");
1746                 show_graph(options->file, '+', add, add_c, reset);
1747                 show_graph(options->file, '-', del, del_c, reset);
1748                 fprintf(options->file, "\n");
1749         }
1750
1751         for (i = 0; i < data->nr; i++) {
1752                 struct diffstat_file *file = data->files[i];
1753                 uintmax_t added = file->added;
1754                 uintmax_t deleted = file->deleted;
1755
1756                 if (file->is_unmerged ||
1757                     (!file->is_interesting && (added + deleted == 0))) {
1758                         total_files--;
1759                         continue;
1760                 }
1761
1762                 if (!file->is_binary) {
1763                         adds += added;
1764                         dels += deleted;
1765                 }
1766                 if (i < count)
1767                         continue;
1768                 if (!extra_shown)
1769                         fprintf(options->file, "%s ...\n", line_prefix);
1770                 extra_shown = 1;
1771         }
1772         fprintf(options->file, "%s", line_prefix);
1773         print_stat_summary(options->file, total_files, adds, dels);
1774 }
1775
1776 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1777 {
1778         int i, adds = 0, dels = 0, total_files = data->nr;
1779
1780         if (data->nr == 0)
1781                 return;
1782
1783         for (i = 0; i < data->nr; i++) {
1784                 int added = data->files[i]->added;
1785                 int deleted= data->files[i]->deleted;
1786
1787                 if (data->files[i]->is_unmerged ||
1788                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1789                         total_files--;
1790                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1791                         adds += added;
1792                         dels += deleted;
1793                 }
1794         }
1795         fprintf(options->file, "%s", diff_line_prefix(options));
1796         print_stat_summary(options->file, total_files, adds, dels);
1797 }
1798
1799 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1800 {
1801         int i;
1802
1803         if (data->nr == 0)
1804                 return;
1805
1806         for (i = 0; i < data->nr; i++) {
1807                 struct diffstat_file *file = data->files[i];
1808
1809                 fprintf(options->file, "%s", diff_line_prefix(options));
1810
1811                 if (file->is_binary)
1812                         fprintf(options->file, "-\t-\t");
1813                 else
1814                         fprintf(options->file,
1815                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1816                                 file->added, file->deleted);
1817                 if (options->line_termination) {
1818                         fill_print_name(file);
1819                         if (!file->is_renamed)
1820                                 write_name_quoted(file->name, options->file,
1821                                                   options->line_termination);
1822                         else {
1823                                 fputs(file->print_name, options->file);
1824                                 putc(options->line_termination, options->file);
1825                         }
1826                 } else {
1827                         if (file->is_renamed) {
1828                                 putc('\0', options->file);
1829                                 write_name_quoted(file->from_name, options->file, '\0');
1830                         }
1831                         write_name_quoted(file->name, options->file, '\0');
1832                 }
1833         }
1834 }
1835
1836 struct dirstat_file {
1837         const char *name;
1838         unsigned long changed;
1839 };
1840
1841 struct dirstat_dir {
1842         struct dirstat_file *files;
1843         int alloc, nr, permille, cumulative;
1844 };
1845
1846 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1847                 unsigned long changed, const char *base, int baselen)
1848 {
1849         unsigned long this_dir = 0;
1850         unsigned int sources = 0;
1851         const char *line_prefix = diff_line_prefix(opt);
1852
1853         while (dir->nr) {
1854                 struct dirstat_file *f = dir->files;
1855                 int namelen = strlen(f->name);
1856                 unsigned long this;
1857                 char *slash;
1858
1859                 if (namelen < baselen)
1860                         break;
1861                 if (memcmp(f->name, base, baselen))
1862                         break;
1863                 slash = strchr(f->name + baselen, '/');
1864                 if (slash) {
1865                         int newbaselen = slash + 1 - f->name;
1866                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1867                         sources++;
1868                 } else {
1869                         this = f->changed;
1870                         dir->files++;
1871                         dir->nr--;
1872                         sources += 2;
1873                 }
1874                 this_dir += this;
1875         }
1876
1877         /*
1878          * We don't report dirstat's for
1879          *  - the top level
1880          *  - or cases where everything came from a single directory
1881          *    under this directory (sources == 1).
1882          */
1883         if (baselen && sources != 1) {
1884                 if (this_dir) {
1885                         int permille = this_dir * 1000 / changed;
1886                         if (permille >= dir->permille) {
1887                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1888                                         permille / 10, permille % 10, baselen, base);
1889                                 if (!dir->cumulative)
1890                                         return 0;
1891                         }
1892                 }
1893         }
1894         return this_dir;
1895 }
1896
1897 static int dirstat_compare(const void *_a, const void *_b)
1898 {
1899         const struct dirstat_file *a = _a;
1900         const struct dirstat_file *b = _b;
1901         return strcmp(a->name, b->name);
1902 }
1903
1904 static void show_dirstat(struct diff_options *options)
1905 {
1906         int i;
1907         unsigned long changed;
1908         struct dirstat_dir dir;
1909         struct diff_queue_struct *q = &diff_queued_diff;
1910
1911         dir.files = NULL;
1912         dir.alloc = 0;
1913         dir.nr = 0;
1914         dir.permille = options->dirstat_permille;
1915         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1916
1917         changed = 0;
1918         for (i = 0; i < q->nr; i++) {
1919                 struct diff_filepair *p = q->queue[i];
1920                 const char *name;
1921                 unsigned long copied, added, damage;
1922                 int content_changed;
1923
1924                 name = p->two->path ? p->two->path : p->one->path;
1925
1926                 if (p->one->oid_valid && p->two->oid_valid)
1927                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
1928                 else
1929                         content_changed = 1;
1930
1931                 if (!content_changed) {
1932                         /*
1933                          * The SHA1 has not changed, so pre-/post-content is
1934                          * identical. We can therefore skip looking at the
1935                          * file contents altogether.
1936                          */
1937                         damage = 0;
1938                         goto found_damage;
1939                 }
1940
1941                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1942                         /*
1943                          * In --dirstat-by-file mode, we don't really need to
1944                          * look at the actual file contents at all.
1945                          * The fact that the SHA1 changed is enough for us to
1946                          * add this file to the list of results
1947                          * (with each file contributing equal damage).
1948                          */
1949                         damage = 1;
1950                         goto found_damage;
1951                 }
1952
1953                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1954                         diff_populate_filespec(p->one, 0);
1955                         diff_populate_filespec(p->two, 0);
1956                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1957                                                &copied, &added);
1958                         diff_free_filespec_data(p->one);
1959                         diff_free_filespec_data(p->two);
1960                 } else if (DIFF_FILE_VALID(p->one)) {
1961                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1962                         copied = added = 0;
1963                         diff_free_filespec_data(p->one);
1964                 } else if (DIFF_FILE_VALID(p->two)) {
1965                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1966                         copied = 0;
1967                         added = p->two->size;
1968                         diff_free_filespec_data(p->two);
1969                 } else
1970                         continue;
1971
1972                 /*
1973                  * Original minus copied is the removed material,
1974                  * added is the new material.  They are both damages
1975                  * made to the preimage.
1976                  * If the resulting damage is zero, we know that
1977                  * diffcore_count_changes() considers the two entries to
1978                  * be identical, but since content_changed is true, we
1979                  * know that there must have been _some_ kind of change,
1980                  * so we force all entries to have damage > 0.
1981                  */
1982                 damage = (p->one->size - copied) + added;
1983                 if (!damage)
1984                         damage = 1;
1985
1986 found_damage:
1987                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1988                 dir.files[dir.nr].name = name;
1989                 dir.files[dir.nr].changed = damage;
1990                 changed += damage;
1991                 dir.nr++;
1992         }
1993
1994         /* This can happen even with many files, if everything was renames */
1995         if (!changed)
1996                 return;
1997
1998         /* Show all directories with more than x% of the changes */
1999         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2000         gather_dirstat(options, &dir, changed, "", 0);
2001 }
2002
2003 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2004 {
2005         int i;
2006         unsigned long changed;
2007         struct dirstat_dir dir;
2008
2009         if (data->nr == 0)
2010                 return;
2011
2012         dir.files = NULL;
2013         dir.alloc = 0;
2014         dir.nr = 0;
2015         dir.permille = options->dirstat_permille;
2016         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2017
2018         changed = 0;
2019         for (i = 0; i < data->nr; i++) {
2020                 struct diffstat_file *file = data->files[i];
2021                 unsigned long damage = file->added + file->deleted;
2022                 if (file->is_binary)
2023                         /*
2024                          * binary files counts bytes, not lines. Must find some
2025                          * way to normalize binary bytes vs. textual lines.
2026                          * The following heuristic assumes that there are 64
2027                          * bytes per "line".
2028                          * This is stupid and ugly, but very cheap...
2029                          */
2030                         damage = (damage + 63) / 64;
2031                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2032                 dir.files[dir.nr].name = file->name;
2033                 dir.files[dir.nr].changed = damage;
2034                 changed += damage;
2035                 dir.nr++;
2036         }
2037
2038         /* This can happen even with many files, if everything was renames */
2039         if (!changed)
2040                 return;
2041
2042         /* Show all directories with more than x% of the changes */
2043         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2044         gather_dirstat(options, &dir, changed, "", 0);
2045 }
2046
2047 static void free_diffstat_info(struct diffstat_t *diffstat)
2048 {
2049         int i;
2050         for (i = 0; i < diffstat->nr; i++) {
2051                 struct diffstat_file *f = diffstat->files[i];
2052                 if (f->name != f->print_name)
2053                         free(f->print_name);
2054                 free(f->name);
2055                 free(f->from_name);
2056                 free(f);
2057         }
2058         free(diffstat->files);
2059 }
2060
2061 struct checkdiff_t {
2062         const char *filename;
2063         int lineno;
2064         int conflict_marker_size;
2065         struct diff_options *o;
2066         unsigned ws_rule;
2067         unsigned status;
2068 };
2069
2070 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2071 {
2072         char firstchar;
2073         int cnt;
2074
2075         if (len < marker_size + 1)
2076                 return 0;
2077         firstchar = line[0];
2078         switch (firstchar) {
2079         case '=': case '>': case '<': case '|':
2080                 break;
2081         default:
2082                 return 0;
2083         }
2084         for (cnt = 1; cnt < marker_size; cnt++)
2085                 if (line[cnt] != firstchar)
2086                         return 0;
2087         /* line[1] thru line[marker_size-1] are same as firstchar */
2088         if (len < marker_size + 1 || !isspace(line[marker_size]))
2089                 return 0;
2090         return 1;
2091 }
2092
2093 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2094 {
2095         struct checkdiff_t *data = priv;
2096         int marker_size = data->conflict_marker_size;
2097         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2098         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2099         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2100         char *err;
2101         const char *line_prefix;
2102
2103         assert(data->o);
2104         line_prefix = diff_line_prefix(data->o);
2105
2106         if (line[0] == '+') {
2107                 unsigned bad;
2108                 data->lineno++;
2109                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2110                         data->status |= 1;
2111                         fprintf(data->o->file,
2112                                 "%s%s:%d: leftover conflict marker\n",
2113                                 line_prefix, data->filename, data->lineno);
2114                 }
2115                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2116                 if (!bad)
2117                         return;
2118                 data->status |= bad;
2119                 err = whitespace_error_string(bad);
2120                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2121                         line_prefix, data->filename, data->lineno, err);
2122                 free(err);
2123                 emit_line(data->o, set, reset, line, 1);
2124                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2125                               data->o->file, set, reset, ws);
2126         } else if (line[0] == ' ') {
2127                 data->lineno++;
2128         } else if (line[0] == '@') {
2129                 char *plus = strchr(line, '+');
2130                 if (plus)
2131                         data->lineno = strtol(plus, NULL, 10) - 1;
2132                 else
2133                         die("invalid diff");
2134         }
2135 }
2136
2137 static unsigned char *deflate_it(char *data,
2138                                  unsigned long size,
2139                                  unsigned long *result_size)
2140 {
2141         int bound;
2142         unsigned char *deflated;
2143         git_zstream stream;
2144
2145         git_deflate_init(&stream, zlib_compression_level);
2146         bound = git_deflate_bound(&stream, size);
2147         deflated = xmalloc(bound);
2148         stream.next_out = deflated;
2149         stream.avail_out = bound;
2150
2151         stream.next_in = (unsigned char *)data;
2152         stream.avail_in = size;
2153         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2154                 ; /* nothing */
2155         git_deflate_end(&stream);
2156         *result_size = stream.total_out;
2157         return deflated;
2158 }
2159
2160 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2161                                   const char *prefix)
2162 {
2163         void *cp;
2164         void *delta;
2165         void *deflated;
2166         void *data;
2167         unsigned long orig_size;
2168         unsigned long delta_size;
2169         unsigned long deflate_size;
2170         unsigned long data_size;
2171
2172         /* We could do deflated delta, or we could do just deflated two,
2173          * whichever is smaller.
2174          */
2175         delta = NULL;
2176         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2177         if (one->size && two->size) {
2178                 delta = diff_delta(one->ptr, one->size,
2179                                    two->ptr, two->size,
2180                                    &delta_size, deflate_size);
2181                 if (delta) {
2182                         void *to_free = delta;
2183                         orig_size = delta_size;
2184                         delta = deflate_it(delta, delta_size, &delta_size);
2185                         free(to_free);
2186                 }
2187         }
2188
2189         if (delta && delta_size < deflate_size) {
2190                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2191                 free(deflated);
2192                 data = delta;
2193                 data_size = delta_size;
2194         }
2195         else {
2196                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2197                 free(delta);
2198                 data = deflated;
2199                 data_size = deflate_size;
2200         }
2201
2202         /* emit data encoded in base85 */
2203         cp = data;
2204         while (data_size) {
2205                 int bytes = (52 < data_size) ? 52 : data_size;
2206                 char line[70];
2207                 data_size -= bytes;
2208                 if (bytes <= 26)
2209                         line[0] = bytes + 'A' - 1;
2210                 else
2211                         line[0] = bytes - 26 + 'a' - 1;
2212                 encode_85(line + 1, cp, bytes);
2213                 cp = (char *) cp + bytes;
2214                 fprintf(file, "%s", prefix);
2215                 fputs(line, file);
2216                 fputc('\n', file);
2217         }
2218         fprintf(file, "%s\n", prefix);
2219         free(data);
2220 }
2221
2222 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2223                              const char *prefix)
2224 {
2225         fprintf(file, "%sGIT binary patch\n", prefix);
2226         emit_binary_diff_body(file, one, two, prefix);
2227         emit_binary_diff_body(file, two, one, prefix);
2228 }
2229
2230 int diff_filespec_is_binary(struct diff_filespec *one)
2231 {
2232         if (one->is_binary == -1) {
2233                 diff_filespec_load_driver(one);
2234                 if (one->driver->binary != -1)
2235                         one->is_binary = one->driver->binary;
2236                 else {
2237                         if (!one->data && DIFF_FILE_VALID(one))
2238                                 diff_populate_filespec(one, CHECK_BINARY);
2239                         if (one->is_binary == -1 && one->data)
2240                                 one->is_binary = buffer_is_binary(one->data,
2241                                                 one->size);
2242                         if (one->is_binary == -1)
2243                                 one->is_binary = 0;
2244                 }
2245         }
2246         return one->is_binary;
2247 }
2248
2249 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2250 {
2251         diff_filespec_load_driver(one);
2252         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2253 }
2254
2255 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2256 {
2257         if (!options->a_prefix)
2258                 options->a_prefix = a;
2259         if (!options->b_prefix)
2260                 options->b_prefix = b;
2261 }
2262
2263 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2264 {
2265         if (!DIFF_FILE_VALID(one))
2266                 return NULL;
2267
2268         diff_filespec_load_driver(one);
2269         return userdiff_get_textconv(one->driver);
2270 }
2271
2272 static void builtin_diff(const char *name_a,
2273                          const char *name_b,
2274                          struct diff_filespec *one,
2275                          struct diff_filespec *two,
2276                          const char *xfrm_msg,
2277                          int must_show_header,
2278                          struct diff_options *o,
2279                          int complete_rewrite)
2280 {
2281         mmfile_t mf1, mf2;
2282         const char *lbl[2];
2283         char *a_one, *b_two;
2284         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2285         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2286         const char *a_prefix, *b_prefix;
2287         struct userdiff_driver *textconv_one = NULL;
2288         struct userdiff_driver *textconv_two = NULL;
2289         struct strbuf header = STRBUF_INIT;
2290         const char *line_prefix = diff_line_prefix(o);
2291
2292         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2293                         (!one->mode || S_ISGITLINK(one->mode)) &&
2294                         (!two->mode || S_ISGITLINK(two->mode))) {
2295                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2296                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2297                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2298                                 line_prefix,
2299                                 one->oid.hash, two->oid.hash,
2300                                 two->dirty_submodule,
2301                                 meta, del, add, reset);
2302                 return;
2303         }
2304
2305         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2306                 textconv_one = get_textconv(one);
2307                 textconv_two = get_textconv(two);
2308         }
2309
2310         diff_set_mnemonic_prefix(o, "a/", "b/");
2311         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2312                 a_prefix = o->b_prefix;
2313                 b_prefix = o->a_prefix;
2314         } else {
2315                 a_prefix = o->a_prefix;
2316                 b_prefix = o->b_prefix;
2317         }
2318
2319         /* Never use a non-valid filename anywhere if at all possible */
2320         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2321         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2322
2323         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2324         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2325         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2326         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2327         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2328         if (lbl[0][0] == '/') {
2329                 /* /dev/null */
2330                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2331                 if (xfrm_msg)
2332                         strbuf_addstr(&header, xfrm_msg);
2333                 must_show_header = 1;
2334         }
2335         else if (lbl[1][0] == '/') {
2336                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2337                 if (xfrm_msg)
2338                         strbuf_addstr(&header, xfrm_msg);
2339                 must_show_header = 1;
2340         }
2341         else {
2342                 if (one->mode != two->mode) {
2343                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2344                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2345                         must_show_header = 1;
2346                 }
2347                 if (xfrm_msg)
2348                         strbuf_addstr(&header, xfrm_msg);
2349
2350                 /*
2351                  * we do not run diff between different kind
2352                  * of objects.
2353                  */
2354                 if ((one->mode ^ two->mode) & S_IFMT)
2355                         goto free_ab_and_return;
2356                 if (complete_rewrite &&
2357                     (textconv_one || !diff_filespec_is_binary(one)) &&
2358                     (textconv_two || !diff_filespec_is_binary(two))) {
2359                         fprintf(o->file, "%s", header.buf);
2360                         strbuf_reset(&header);
2361                         emit_rewrite_diff(name_a, name_b, one, two,
2362                                                 textconv_one, textconv_two, o);
2363                         o->found_changes = 1;
2364                         goto free_ab_and_return;
2365                 }
2366         }
2367
2368         if (o->irreversible_delete && lbl[1][0] == '/') {
2369                 fprintf(o->file, "%s", header.buf);
2370                 strbuf_reset(&header);
2371                 goto free_ab_and_return;
2372         } else if (!DIFF_OPT_TST(o, TEXT) &&
2373             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2374               (!textconv_two && diff_filespec_is_binary(two)) )) {
2375                 if (!one->data && !two->data &&
2376                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2377                     !DIFF_OPT_TST(o, BINARY)) {
2378                         if (!oidcmp(&one->oid, &two->oid)) {
2379                                 if (must_show_header)
2380                                         fprintf(o->file, "%s", header.buf);
2381                                 goto free_ab_and_return;
2382                         }
2383                         fprintf(o->file, "%s", header.buf);
2384                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2385                                 line_prefix, lbl[0], lbl[1]);
2386                         goto free_ab_and_return;
2387                 }
2388                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2389                         die("unable to read files to diff");
2390                 /* Quite common confusing case */
2391                 if (mf1.size == mf2.size &&
2392                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2393                         if (must_show_header)
2394                                 fprintf(o->file, "%s", header.buf);
2395                         goto free_ab_and_return;
2396                 }
2397                 fprintf(o->file, "%s", header.buf);
2398                 strbuf_reset(&header);
2399                 if (DIFF_OPT_TST(o, BINARY))
2400                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2401                 else
2402                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2403                                 line_prefix, lbl[0], lbl[1]);
2404                 o->found_changes = 1;
2405         } else {
2406                 /* Crazy xdl interfaces.. */
2407                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2408                 const char *v;
2409                 xpparam_t xpp;
2410                 xdemitconf_t xecfg;
2411                 struct emit_callback ecbdata;
2412                 const struct userdiff_funcname *pe;
2413
2414                 if (must_show_header) {
2415                         fprintf(o->file, "%s", header.buf);
2416                         strbuf_reset(&header);
2417                 }
2418
2419                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2420                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2421
2422                 pe = diff_funcname_pattern(one);
2423                 if (!pe)
2424                         pe = diff_funcname_pattern(two);
2425
2426                 memset(&xpp, 0, sizeof(xpp));
2427                 memset(&xecfg, 0, sizeof(xecfg));
2428                 memset(&ecbdata, 0, sizeof(ecbdata));
2429                 ecbdata.label_path = lbl;
2430                 ecbdata.color_diff = want_color(o->use_color);
2431                 ecbdata.ws_rule = whitespace_rule(name_b);
2432                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2433                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2434                 ecbdata.opt = o;
2435                 ecbdata.header = header.len ? &header : NULL;
2436                 xpp.flags = o->xdl_opts;
2437                 xecfg.ctxlen = o->context;
2438                 xecfg.interhunkctxlen = o->interhunkcontext;
2439                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2440                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2441                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2442                 if (pe)
2443                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2444                 if (!diffopts)
2445                         ;
2446                 else if (skip_prefix(diffopts, "--unified=", &v))
2447                         xecfg.ctxlen = strtoul(v, NULL, 10);
2448                 else if (skip_prefix(diffopts, "-u", &v))
2449                         xecfg.ctxlen = strtoul(v, NULL, 10);
2450                 if (o->word_diff)
2451                         init_diff_words_data(&ecbdata, o, one, two);
2452                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2453                                   &xpp, &xecfg))
2454                         die("unable to generate diff for %s", one->path);
2455                 if (o->word_diff)
2456                         free_diff_words_data(&ecbdata);
2457                 if (textconv_one)
2458                         free(mf1.ptr);
2459                 if (textconv_two)
2460                         free(mf2.ptr);
2461                 xdiff_clear_find_func(&xecfg);
2462         }
2463
2464  free_ab_and_return:
2465         strbuf_release(&header);
2466         diff_free_filespec_data(one);
2467         diff_free_filespec_data(two);
2468         free(a_one);
2469         free(b_two);
2470         return;
2471 }
2472
2473 static void builtin_diffstat(const char *name_a, const char *name_b,
2474                              struct diff_filespec *one,
2475                              struct diff_filespec *two,
2476                              struct diffstat_t *diffstat,
2477                              struct diff_options *o,
2478                              struct diff_filepair *p)
2479 {
2480         mmfile_t mf1, mf2;
2481         struct diffstat_file *data;
2482         int same_contents;
2483         int complete_rewrite = 0;
2484
2485         if (!DIFF_PAIR_UNMERGED(p)) {
2486                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2487                         complete_rewrite = 1;
2488         }
2489
2490         data = diffstat_add(diffstat, name_a, name_b);
2491         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2492
2493         if (!one || !two) {
2494                 data->is_unmerged = 1;
2495                 return;
2496         }
2497
2498         same_contents = !oidcmp(&one->oid, &two->oid);
2499
2500         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2501                 data->is_binary = 1;
2502                 if (same_contents) {
2503                         data->added = 0;
2504                         data->deleted = 0;
2505                 } else {
2506                         data->added = diff_filespec_size(two);
2507                         data->deleted = diff_filespec_size(one);
2508                 }
2509         }
2510
2511         else if (complete_rewrite) {
2512                 diff_populate_filespec(one, 0);
2513                 diff_populate_filespec(two, 0);
2514                 data->deleted = count_lines(one->data, one->size);
2515                 data->added = count_lines(two->data, two->size);
2516         }
2517
2518         else if (!same_contents) {
2519                 /* Crazy xdl interfaces.. */
2520                 xpparam_t xpp;
2521                 xdemitconf_t xecfg;
2522
2523                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2524                         die("unable to read files to diff");
2525
2526                 memset(&xpp, 0, sizeof(xpp));
2527                 memset(&xecfg, 0, sizeof(xecfg));
2528                 xpp.flags = o->xdl_opts;
2529                 xecfg.ctxlen = o->context;
2530                 xecfg.interhunkctxlen = o->interhunkcontext;
2531                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2532                                   &xpp, &xecfg))
2533                         die("unable to generate diffstat for %s", one->path);
2534         }
2535
2536         diff_free_filespec_data(one);
2537         diff_free_filespec_data(two);
2538 }
2539
2540 static void builtin_checkdiff(const char *name_a, const char *name_b,
2541                               const char *attr_path,
2542                               struct diff_filespec *one,
2543                               struct diff_filespec *two,
2544                               struct diff_options *o)
2545 {
2546         mmfile_t mf1, mf2;
2547         struct checkdiff_t data;
2548
2549         if (!two)
2550                 return;
2551
2552         memset(&data, 0, sizeof(data));
2553         data.filename = name_b ? name_b : name_a;
2554         data.lineno = 0;
2555         data.o = o;
2556         data.ws_rule = whitespace_rule(attr_path);
2557         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2558
2559         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2560                 die("unable to read files to diff");
2561
2562         /*
2563          * All the other codepaths check both sides, but not checking
2564          * the "old" side here is deliberate.  We are checking the newly
2565          * introduced changes, and as long as the "new" side is text, we
2566          * can and should check what it introduces.
2567          */
2568         if (diff_filespec_is_binary(two))
2569                 goto free_and_return;
2570         else {
2571                 /* Crazy xdl interfaces.. */
2572                 xpparam_t xpp;
2573                 xdemitconf_t xecfg;
2574
2575                 memset(&xpp, 0, sizeof(xpp));
2576                 memset(&xecfg, 0, sizeof(xecfg));
2577                 xecfg.ctxlen = 1; /* at least one context line */
2578                 xpp.flags = 0;
2579                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2580                                   &xpp, &xecfg))
2581                         die("unable to generate checkdiff for %s", one->path);
2582
2583                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2584                         struct emit_callback ecbdata;
2585                         int blank_at_eof;
2586
2587                         ecbdata.ws_rule = data.ws_rule;
2588                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2589                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2590
2591                         if (blank_at_eof) {
2592                                 static char *err;
2593                                 if (!err)
2594                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2595                                 fprintf(o->file, "%s:%d: %s.\n",
2596                                         data.filename, blank_at_eof, err);
2597                                 data.status = 1; /* report errors */
2598                         }
2599                 }
2600         }
2601  free_and_return:
2602         diff_free_filespec_data(one);
2603         diff_free_filespec_data(two);
2604         if (data.status)
2605                 DIFF_OPT_SET(o, CHECK_FAILED);
2606 }
2607
2608 struct diff_filespec *alloc_filespec(const char *path)
2609 {
2610         struct diff_filespec *spec;
2611
2612         FLEXPTR_ALLOC_STR(spec, path, path);
2613         spec->count = 1;
2614         spec->is_binary = -1;
2615         return spec;
2616 }
2617
2618 void free_filespec(struct diff_filespec *spec)
2619 {
2620         if (!--spec->count) {
2621                 diff_free_filespec_data(spec);
2622                 free(spec);
2623         }
2624 }
2625
2626 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2627                    int sha1_valid, unsigned short mode)
2628 {
2629         if (mode) {
2630                 spec->mode = canon_mode(mode);
2631                 hashcpy(spec->oid.hash, sha1);
2632                 spec->oid_valid = sha1_valid;
2633         }
2634 }
2635
2636 /*
2637  * Given a name and sha1 pair, if the index tells us the file in
2638  * the work tree has that object contents, return true, so that
2639  * prepare_temp_file() does not have to inflate and extract.
2640  */
2641 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2642 {
2643         const struct cache_entry *ce;
2644         struct stat st;
2645         int pos, len;
2646
2647         /*
2648          * We do not read the cache ourselves here, because the
2649          * benchmark with my previous version that always reads cache
2650          * shows that it makes things worse for diff-tree comparing
2651          * two linux-2.6 kernel trees in an already checked out work
2652          * tree.  This is because most diff-tree comparisons deal with
2653          * only a small number of files, while reading the cache is
2654          * expensive for a large project, and its cost outweighs the
2655          * savings we get by not inflating the object to a temporary
2656          * file.  Practically, this code only helps when we are used
2657          * by diff-cache --cached, which does read the cache before
2658          * calling us.
2659          */
2660         if (!active_cache)
2661                 return 0;
2662
2663         /* We want to avoid the working directory if our caller
2664          * doesn't need the data in a normal file, this system
2665          * is rather slow with its stat/open/mmap/close syscalls,
2666          * and the object is contained in a pack file.  The pack
2667          * is probably already open and will be faster to obtain
2668          * the data through than the working directory.  Loose
2669          * objects however would tend to be slower as they need
2670          * to be individually opened and inflated.
2671          */
2672         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2673                 return 0;
2674
2675         /*
2676          * Similarly, if we'd have to convert the file contents anyway, that
2677          * makes the optimization not worthwhile.
2678          */
2679         if (!want_file && would_convert_to_git(name))
2680                 return 0;
2681
2682         len = strlen(name);
2683         pos = cache_name_pos(name, len);
2684         if (pos < 0)
2685                 return 0;
2686         ce = active_cache[pos];
2687
2688         /*
2689          * This is not the sha1 we are looking for, or
2690          * unreusable because it is not a regular file.
2691          */
2692         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2693                 return 0;
2694
2695         /*
2696          * If ce is marked as "assume unchanged", there is no
2697          * guarantee that work tree matches what we are looking for.
2698          */
2699         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2700                 return 0;
2701
2702         /*
2703          * If ce matches the file in the work tree, we can reuse it.
2704          */
2705         if (ce_uptodate(ce) ||
2706             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2707                 return 1;
2708
2709         return 0;
2710 }
2711
2712 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2713 {
2714         struct strbuf buf = STRBUF_INIT;
2715         char *dirty = "";
2716
2717         /* Are we looking at the work tree? */
2718         if (s->dirty_submodule)
2719                 dirty = "-dirty";
2720
2721         strbuf_addf(&buf, "Subproject commit %s%s\n",
2722                     oid_to_hex(&s->oid), dirty);
2723         s->size = buf.len;
2724         if (size_only) {
2725                 s->data = NULL;
2726                 strbuf_release(&buf);
2727         } else {
2728                 s->data = strbuf_detach(&buf, NULL);
2729                 s->should_free = 1;
2730         }
2731         return 0;
2732 }
2733
2734 /*
2735  * While doing rename detection and pickaxe operation, we may need to
2736  * grab the data for the blob (or file) for our own in-core comparison.
2737  * diff_filespec has data and size fields for this purpose.
2738  */
2739 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2740 {
2741         int size_only = flags & CHECK_SIZE_ONLY;
2742         int err = 0;
2743         /*
2744          * demote FAIL to WARN to allow inspecting the situation
2745          * instead of refusing.
2746          */
2747         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2748                                     ? SAFE_CRLF_WARN
2749                                     : safe_crlf);
2750
2751         if (!DIFF_FILE_VALID(s))
2752                 die("internal error: asking to populate invalid file.");
2753         if (S_ISDIR(s->mode))
2754                 return -1;
2755
2756         if (s->data)
2757                 return 0;
2758
2759         if (size_only && 0 < s->size)
2760                 return 0;
2761
2762         if (S_ISGITLINK(s->mode))
2763                 return diff_populate_gitlink(s, size_only);
2764
2765         if (!s->oid_valid ||
2766             reuse_worktree_file(s->path, s->oid.hash, 0)) {
2767                 struct strbuf buf = STRBUF_INIT;
2768                 struct stat st;
2769                 int fd;
2770
2771                 if (lstat(s->path, &st) < 0) {
2772                         if (errno == ENOENT) {
2773                         err_empty:
2774                                 err = -1;
2775                         empty:
2776                                 s->data = (char *)"";
2777                                 s->size = 0;
2778                                 return err;
2779                         }
2780                 }
2781                 s->size = xsize_t(st.st_size);
2782                 if (!s->size)
2783                         goto empty;
2784                 if (S_ISLNK(st.st_mode)) {
2785                         struct strbuf sb = STRBUF_INIT;
2786
2787                         if (strbuf_readlink(&sb, s->path, s->size))
2788                                 goto err_empty;
2789                         s->size = sb.len;
2790                         s->data = strbuf_detach(&sb, NULL);
2791                         s->should_free = 1;
2792                         return 0;
2793                 }
2794                 if (size_only)
2795                         return 0;
2796                 if ((flags & CHECK_BINARY) &&
2797                     s->size > big_file_threshold && s->is_binary == -1) {
2798                         s->is_binary = 1;
2799                         return 0;
2800                 }
2801                 fd = open(s->path, O_RDONLY);
2802                 if (fd < 0)
2803                         goto err_empty;
2804                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2805                 close(fd);
2806                 s->should_munmap = 1;
2807
2808                 /*
2809                  * Convert from working tree format to canonical git format
2810                  */
2811                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2812                         size_t size = 0;
2813                         munmap(s->data, s->size);
2814                         s->should_munmap = 0;
2815                         s->data = strbuf_detach(&buf, &size);
2816                         s->size = size;
2817                         s->should_free = 1;
2818                 }
2819         }
2820         else {
2821                 enum object_type type;
2822                 if (size_only || (flags & CHECK_BINARY)) {
2823                         type = sha1_object_info(s->oid.hash, &s->size);
2824                         if (type < 0)
2825                                 die("unable to read %s",
2826                                     oid_to_hex(&s->oid));
2827                         if (size_only)
2828                                 return 0;
2829                         if (s->size > big_file_threshold && s->is_binary == -1) {
2830                                 s->is_binary = 1;
2831                                 return 0;
2832                         }
2833                 }
2834                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
2835                 if (!s->data)
2836                         die("unable to read %s", oid_to_hex(&s->oid));
2837                 s->should_free = 1;
2838         }
2839         return 0;
2840 }
2841
2842 void diff_free_filespec_blob(struct diff_filespec *s)
2843 {
2844         if (s->should_free)
2845                 free(s->data);
2846         else if (s->should_munmap)
2847                 munmap(s->data, s->size);
2848
2849         if (s->should_free || s->should_munmap) {
2850                 s->should_free = s->should_munmap = 0;
2851                 s->data = NULL;
2852         }
2853 }
2854
2855 void diff_free_filespec_data(struct diff_filespec *s)
2856 {
2857         diff_free_filespec_blob(s);
2858         free(s->cnt_data);
2859         s->cnt_data = NULL;
2860 }
2861
2862 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2863                            void *blob,
2864                            unsigned long size,
2865                            const struct object_id *oid,
2866                            int mode)
2867 {
2868         int fd;
2869         struct strbuf buf = STRBUF_INIT;
2870         struct strbuf template = STRBUF_INIT;
2871         char *path_dup = xstrdup(path);
2872         const char *base = basename(path_dup);
2873
2874         /* Generate "XXXXXX_basename.ext" */
2875         strbuf_addstr(&template, "XXXXXX_");
2876         strbuf_addstr(&template, base);
2877
2878         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2879         if (fd < 0)
2880                 die_errno("unable to create temp-file");
2881         if (convert_to_working_tree(path,
2882                         (const char *)blob, (size_t)size, &buf)) {
2883                 blob = buf.buf;
2884                 size = buf.len;
2885         }
2886         if (write_in_full(fd, blob, size) != size)
2887                 die_errno("unable to write temp-file");
2888         close_tempfile(&temp->tempfile);
2889         temp->name = get_tempfile_path(&temp->tempfile);
2890         oid_to_hex_r(temp->hex, oid);
2891         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
2892         strbuf_release(&buf);
2893         strbuf_release(&template);
2894         free(path_dup);
2895 }
2896
2897 static struct diff_tempfile *prepare_temp_file(const char *name,
2898                 struct diff_filespec *one)
2899 {
2900         struct diff_tempfile *temp = claim_diff_tempfile();
2901
2902         if (!DIFF_FILE_VALID(one)) {
2903         not_a_valid_file:
2904                 /* A '-' entry produces this for file-2, and
2905                  * a '+' entry produces this for file-1.
2906                  */
2907                 temp->name = "/dev/null";
2908                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
2909                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
2910                 return temp;
2911         }
2912
2913         if (!S_ISGITLINK(one->mode) &&
2914             (!one->oid_valid ||
2915              reuse_worktree_file(name, one->oid.hash, 1))) {
2916                 struct stat st;
2917                 if (lstat(name, &st) < 0) {
2918                         if (errno == ENOENT)
2919                                 goto not_a_valid_file;
2920                         die_errno("stat(%s)", name);
2921                 }
2922                 if (S_ISLNK(st.st_mode)) {
2923                         struct strbuf sb = STRBUF_INIT;
2924                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2925                                 die_errno("readlink(%s)", name);
2926                         prep_temp_blob(name, temp, sb.buf, sb.len,
2927                                        (one->oid_valid ?
2928                                         &one->oid : &null_oid),
2929                                        (one->oid_valid ?
2930                                         one->mode : S_IFLNK));
2931                         strbuf_release(&sb);
2932                 }
2933                 else {
2934                         /* we can borrow from the file in the work tree */
2935                         temp->name = name;
2936                         if (!one->oid_valid)
2937                                 sha1_to_hex_r(temp->hex, null_sha1);
2938                         else
2939                                 sha1_to_hex_r(temp->hex, one->oid.hash);
2940                         /* Even though we may sometimes borrow the
2941                          * contents from the work tree, we always want
2942                          * one->mode.  mode is trustworthy even when
2943                          * !(one->sha1_valid), as long as
2944                          * DIFF_FILE_VALID(one).
2945                          */
2946                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
2947                 }
2948                 return temp;
2949         }
2950         else {
2951                 if (diff_populate_filespec(one, 0))
2952                         die("cannot read data blob for %s", one->path);
2953                 prep_temp_blob(name, temp, one->data, one->size,
2954                                &one->oid, one->mode);
2955         }
2956         return temp;
2957 }
2958
2959 static void add_external_diff_name(struct argv_array *argv,
2960                                    const char *name,
2961                                    struct diff_filespec *df)
2962 {
2963         struct diff_tempfile *temp = prepare_temp_file(name, df);
2964         argv_array_push(argv, temp->name);
2965         argv_array_push(argv, temp->hex);
2966         argv_array_push(argv, temp->mode);
2967 }
2968
2969 /* An external diff command takes:
2970  *
2971  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2972  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2973  *
2974  */
2975 static void run_external_diff(const char *pgm,
2976                               const char *name,
2977                               const char *other,
2978                               struct diff_filespec *one,
2979                               struct diff_filespec *two,
2980                               const char *xfrm_msg,
2981                               int complete_rewrite,
2982                               struct diff_options *o)
2983 {
2984         struct argv_array argv = ARGV_ARRAY_INIT;
2985         struct argv_array env = ARGV_ARRAY_INIT;
2986         struct diff_queue_struct *q = &diff_queued_diff;
2987
2988         argv_array_push(&argv, pgm);
2989         argv_array_push(&argv, name);
2990
2991         if (one && two) {
2992                 add_external_diff_name(&argv, name, one);
2993                 if (!other)
2994                         add_external_diff_name(&argv, name, two);
2995                 else {
2996                         add_external_diff_name(&argv, other, two);
2997                         argv_array_push(&argv, other);
2998                         argv_array_push(&argv, xfrm_msg);
2999                 }
3000         }
3001
3002         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3003         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3004
3005         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3006                 die(_("external diff died, stopping at %s"), name);
3007
3008         remove_tempfile();
3009         argv_array_clear(&argv);
3010         argv_array_clear(&env);
3011 }
3012
3013 static int similarity_index(struct diff_filepair *p)
3014 {
3015         return p->score * 100 / MAX_SCORE;
3016 }
3017
3018 static void fill_metainfo(struct strbuf *msg,
3019                           const char *name,
3020                           const char *other,
3021                           struct diff_filespec *one,
3022                           struct diff_filespec *two,
3023                           struct diff_options *o,
3024                           struct diff_filepair *p,
3025                           int *must_show_header,
3026                           int use_color)
3027 {
3028         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3029         const char *reset = diff_get_color(use_color, DIFF_RESET);
3030         const char *line_prefix = diff_line_prefix(o);
3031
3032         *must_show_header = 1;
3033         strbuf_init(msg, PATH_MAX * 2 + 300);
3034         switch (p->status) {
3035         case DIFF_STATUS_COPIED:
3036                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3037                             line_prefix, set, similarity_index(p));
3038                 strbuf_addf(msg, "%s\n%s%scopy from ",
3039                             reset,  line_prefix, set);
3040                 quote_c_style(name, msg, NULL, 0);
3041                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3042                 quote_c_style(other, msg, NULL, 0);
3043                 strbuf_addf(msg, "%s\n", reset);
3044                 break;
3045         case DIFF_STATUS_RENAMED:
3046                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3047                             line_prefix, set, similarity_index(p));
3048                 strbuf_addf(msg, "%s\n%s%srename from ",
3049                             reset, line_prefix, set);
3050                 quote_c_style(name, msg, NULL, 0);
3051                 strbuf_addf(msg, "%s\n%s%srename to ",
3052                             reset, line_prefix, set);
3053                 quote_c_style(other, msg, NULL, 0);
3054                 strbuf_addf(msg, "%s\n", reset);
3055                 break;
3056         case DIFF_STATUS_MODIFIED:
3057                 if (p->score) {
3058                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3059                                     line_prefix,
3060                                     set, similarity_index(p), reset);
3061                         break;
3062                 }
3063                 /* fallthru */
3064         default:
3065                 *must_show_header = 0;
3066         }
3067         if (one && two && oidcmp(&one->oid, &two->oid)) {
3068                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3069
3070                 if (DIFF_OPT_TST(o, BINARY)) {
3071                         mmfile_t mf;
3072                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3073                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3074                                 abbrev = 40;
3075                 }
3076                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3077                             find_unique_abbrev(one->oid.hash, abbrev));
3078                 strbuf_addstr(msg, find_unique_abbrev(two->oid.hash, abbrev));
3079                 if (one->mode == two->mode)
3080                         strbuf_addf(msg, " %06o", one->mode);
3081                 strbuf_addf(msg, "%s\n", reset);
3082         }
3083 }
3084
3085 static void run_diff_cmd(const char *pgm,
3086                          const char *name,
3087                          const char *other,
3088                          const char *attr_path,
3089                          struct diff_filespec *one,
3090                          struct diff_filespec *two,
3091                          struct strbuf *msg,
3092                          struct diff_options *o,
3093                          struct diff_filepair *p)
3094 {
3095         const char *xfrm_msg = NULL;
3096         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3097         int must_show_header = 0;
3098
3099
3100         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3101                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3102                 if (drv && drv->external)
3103                         pgm = drv->external;
3104         }
3105
3106         if (msg) {
3107                 /*
3108                  * don't use colors when the header is intended for an
3109                  * external diff driver
3110                  */
3111                 fill_metainfo(msg, name, other, one, two, o, p,
3112                               &must_show_header,
3113                               want_color(o->use_color) && !pgm);
3114                 xfrm_msg = msg->len ? msg->buf : NULL;
3115         }
3116
3117         if (pgm) {
3118                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3119                                   complete_rewrite, o);
3120                 return;
3121         }
3122         if (one && two)
3123                 builtin_diff(name, other ? other : name,
3124                              one, two, xfrm_msg, must_show_header,
3125                              o, complete_rewrite);
3126         else
3127                 fprintf(o->file, "* Unmerged path %s\n", name);
3128 }
3129
3130 static void diff_fill_sha1_info(struct diff_filespec *one)
3131 {
3132         if (DIFF_FILE_VALID(one)) {
3133                 if (!one->oid_valid) {
3134                         struct stat st;
3135                         if (one->is_stdin) {
3136                                 oidclr(&one->oid);
3137                                 return;
3138                         }
3139                         if (lstat(one->path, &st) < 0)
3140                                 die_errno("stat '%s'", one->path);
3141                         if (index_path(one->oid.hash, one->path, &st, 0))
3142                                 die("cannot hash %s", one->path);
3143                 }
3144         }
3145         else
3146                 oidclr(&one->oid);
3147 }
3148
3149 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3150 {
3151         /* Strip the prefix but do not molest /dev/null and absolute paths */
3152         if (*namep && **namep != '/') {
3153                 *namep += prefix_length;
3154                 if (**namep == '/')
3155                         ++*namep;
3156         }
3157         if (*otherp && **otherp != '/') {
3158                 *otherp += prefix_length;
3159                 if (**otherp == '/')
3160                         ++*otherp;
3161         }
3162 }
3163
3164 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3165 {
3166         const char *pgm = external_diff();
3167         struct strbuf msg;
3168         struct diff_filespec *one = p->one;
3169         struct diff_filespec *two = p->two;
3170         const char *name;
3171         const char *other;
3172         const char *attr_path;
3173
3174         name  = p->one->path;
3175         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3176         attr_path = name;
3177         if (o->prefix_length)
3178                 strip_prefix(o->prefix_length, &name, &other);
3179
3180         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3181                 pgm = NULL;
3182
3183         if (DIFF_PAIR_UNMERGED(p)) {
3184                 run_diff_cmd(pgm, name, NULL, attr_path,
3185                              NULL, NULL, NULL, o, p);
3186                 return;
3187         }
3188
3189         diff_fill_sha1_info(one);
3190         diff_fill_sha1_info(two);
3191
3192         if (!pgm &&
3193             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3194             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3195                 /*
3196                  * a filepair that changes between file and symlink
3197                  * needs to be split into deletion and creation.
3198                  */
3199                 struct diff_filespec *null = alloc_filespec(two->path);
3200                 run_diff_cmd(NULL, name, other, attr_path,
3201                              one, null, &msg, o, p);
3202                 free(null);
3203                 strbuf_release(&msg);
3204
3205                 null = alloc_filespec(one->path);
3206                 run_diff_cmd(NULL, name, other, attr_path,
3207                              null, two, &msg, o, p);
3208                 free(null);
3209         }
3210         else
3211                 run_diff_cmd(pgm, name, other, attr_path,
3212                              one, two, &msg, o, p);
3213
3214         strbuf_release(&msg);
3215 }
3216
3217 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3218                          struct diffstat_t *diffstat)
3219 {
3220         const char *name;
3221         const char *other;
3222
3223         if (DIFF_PAIR_UNMERGED(p)) {
3224                 /* unmerged */
3225                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3226                 return;
3227         }
3228
3229         name = p->one->path;
3230         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3231
3232         if (o->prefix_length)
3233                 strip_prefix(o->prefix_length, &name, &other);
3234
3235         diff_fill_sha1_info(p->one);
3236         diff_fill_sha1_info(p->two);
3237
3238         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3239 }
3240
3241 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3242 {
3243         const char *name;
3244         const char *other;
3245         const char *attr_path;
3246
3247         if (DIFF_PAIR_UNMERGED(p)) {
3248                 /* unmerged */
3249                 return;
3250         }
3251
3252         name = p->one->path;
3253         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3254         attr_path = other ? other : name;
3255
3256         if (o->prefix_length)
3257                 strip_prefix(o->prefix_length, &name, &other);
3258
3259         diff_fill_sha1_info(p->one);
3260         diff_fill_sha1_info(p->two);
3261
3262         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3263 }
3264
3265 void diff_setup(struct diff_options *options)
3266 {
3267         memcpy(options, &default_diff_options, sizeof(*options));
3268
3269         options->file = stdout;
3270
3271         options->line_termination = '\n';
3272         options->break_opt = -1;
3273         options->rename_limit = -1;
3274         options->dirstat_permille = diff_dirstat_permille_default;
3275         options->context = diff_context_default;
3276         options->ws_error_highlight = WSEH_NEW;
3277         DIFF_OPT_SET(options, RENAME_EMPTY);
3278
3279         /* pathchange left =NULL by default */
3280         options->change = diff_change;
3281         options->add_remove = diff_addremove;
3282         options->use_color = diff_use_color_default;
3283         options->detect_rename = diff_detect_rename_default;
3284         options->xdl_opts |= diff_algorithm;
3285         if (diff_compaction_heuristic)
3286                 DIFF_XDL_SET(options, COMPACTION_HEURISTIC);
3287
3288         options->orderfile = diff_order_file_cfg;
3289
3290         if (diff_no_prefix) {
3291                 options->a_prefix = options->b_prefix = "";
3292         } else if (!diff_mnemonic_prefix) {
3293                 options->a_prefix = "a/";
3294                 options->b_prefix = "b/";
3295         }
3296 }
3297
3298 void diff_setup_done(struct diff_options *options)
3299 {
3300         int count = 0;
3301
3302         if (options->set_default)
3303                 options->set_default(options);
3304
3305         if (options->output_format & DIFF_FORMAT_NAME)
3306                 count++;
3307         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3308                 count++;
3309         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3310                 count++;
3311         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3312                 count++;
3313         if (count > 1)
3314                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3315
3316         /*
3317          * Most of the time we can say "there are changes"
3318          * only by checking if there are changed paths, but
3319          * --ignore-whitespace* options force us to look
3320          * inside contents.
3321          */
3322
3323         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3324             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3325             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3326                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3327         else
3328                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3329
3330         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3331                 options->detect_rename = DIFF_DETECT_COPY;
3332
3333         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3334                 options->prefix = NULL;
3335         if (options->prefix)
3336                 options->prefix_length = strlen(options->prefix);
3337         else
3338                 options->prefix_length = 0;
3339
3340         if (options->output_format & (DIFF_FORMAT_NAME |
3341                                       DIFF_FORMAT_NAME_STATUS |
3342                                       DIFF_FORMAT_CHECKDIFF |
3343                                       DIFF_FORMAT_NO_OUTPUT))
3344                 options->output_format &= ~(DIFF_FORMAT_RAW |
3345                                             DIFF_FORMAT_NUMSTAT |
3346                                             DIFF_FORMAT_DIFFSTAT |
3347                                             DIFF_FORMAT_SHORTSTAT |
3348                                             DIFF_FORMAT_DIRSTAT |
3349                                             DIFF_FORMAT_SUMMARY |
3350                                             DIFF_FORMAT_PATCH);
3351
3352         /*
3353          * These cases always need recursive; we do not drop caller-supplied
3354          * recursive bits for other formats here.
3355          */
3356         if (options->output_format & (DIFF_FORMAT_PATCH |
3357                                       DIFF_FORMAT_NUMSTAT |
3358                                       DIFF_FORMAT_DIFFSTAT |
3359                                       DIFF_FORMAT_SHORTSTAT |
3360                                       DIFF_FORMAT_DIRSTAT |
3361                                       DIFF_FORMAT_SUMMARY |
3362                                       DIFF_FORMAT_CHECKDIFF))
3363                 DIFF_OPT_SET(options, RECURSIVE);
3364         /*
3365          * Also pickaxe would not work very well if you do not say recursive
3366          */
3367         if (options->pickaxe)
3368                 DIFF_OPT_SET(options, RECURSIVE);
3369         /*
3370          * When patches are generated, submodules diffed against the work tree
3371          * must be checked for dirtiness too so it can be shown in the output
3372          */
3373         if (options->output_format & DIFF_FORMAT_PATCH)
3374                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3375
3376         if (options->detect_rename && options->rename_limit < 0)
3377                 options->rename_limit = diff_rename_limit_default;
3378         if (options->setup & DIFF_SETUP_USE_CACHE) {
3379                 if (!active_cache)
3380                         /* read-cache does not die even when it fails
3381                          * so it is safe for us to do this here.  Also
3382                          * it does not smudge active_cache or active_nr
3383                          * when it fails, so we do not have to worry about
3384                          * cleaning it up ourselves either.
3385                          */
3386                         read_cache();
3387         }
3388         if (options->abbrev <= 0 || 40 < options->abbrev)
3389                 options->abbrev = 40; /* full */
3390
3391         /*
3392          * It does not make sense to show the first hit we happened
3393          * to have found.  It does not make sense not to return with
3394          * exit code in such a case either.
3395          */
3396         if (DIFF_OPT_TST(options, QUICK)) {
3397                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3398                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3399         }
3400
3401         options->diff_path_counter = 0;
3402
3403         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3404                 die(_("--follow requires exactly one pathspec"));
3405 }
3406
3407 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3408 {
3409         char c, *eq;
3410         int len;
3411
3412         if (*arg != '-')
3413                 return 0;
3414         c = *++arg;
3415         if (!c)
3416                 return 0;
3417         if (c == arg_short) {
3418                 c = *++arg;
3419                 if (!c)
3420                         return 1;
3421                 if (val && isdigit(c)) {
3422                         char *end;
3423                         int n = strtoul(arg, &end, 10);
3424                         if (*end)
3425                                 return 0;
3426                         *val = n;
3427                         return 1;
3428                 }
3429                 return 0;
3430         }
3431         if (c != '-')
3432                 return 0;
3433         arg++;
3434         eq = strchrnul(arg, '=');
3435         len = eq - arg;
3436         if (!len || strncmp(arg, arg_long, len))
3437                 return 0;
3438         if (*eq) {
3439                 int n;
3440                 char *end;
3441                 if (!isdigit(*++eq))
3442                         return 0;
3443                 n = strtoul(eq, &end, 10);
3444                 if (*end)
3445                         return 0;
3446                 *val = n;
3447         }
3448         return 1;
3449 }
3450
3451 static int diff_scoreopt_parse(const char *opt);
3452
3453 static inline int short_opt(char opt, const char **argv,
3454                             const char **optarg)
3455 {
3456         const char *arg = argv[0];
3457         if (arg[0] != '-' || arg[1] != opt)
3458                 return 0;
3459         if (arg[2] != '\0') {
3460                 *optarg = arg + 2;
3461                 return 1;
3462         }
3463         if (!argv[1])
3464                 die("Option '%c' requires a value", opt);
3465         *optarg = argv[1];
3466         return 2;
3467 }
3468
3469 int parse_long_opt(const char *opt, const char **argv,
3470                    const char **optarg)
3471 {
3472         const char *arg = argv[0];
3473         if (!skip_prefix(arg, "--", &arg))
3474                 return 0;
3475         if (!skip_prefix(arg, opt, &arg))
3476                 return 0;
3477         if (*arg == '=') { /* stuck form: --option=value */
3478                 *optarg = arg + 1;
3479                 return 1;
3480         }
3481         if (*arg != '\0')
3482                 return 0;
3483         /* separate form: --option value */
3484         if (!argv[1])
3485                 die("Option '--%s' requires a value", opt);
3486         *optarg = argv[1];
3487         return 2;
3488 }
3489
3490 static int stat_opt(struct diff_options *options, const char **av)
3491 {
3492         const char *arg = av[0];
3493         char *end;
3494         int width = options->stat_width;
3495         int name_width = options->stat_name_width;
3496         int graph_width = options->stat_graph_width;
3497         int count = options->stat_count;
3498         int argcount = 1;
3499
3500         if (!skip_prefix(arg, "--stat", &arg))
3501                 die("BUG: stat option does not begin with --stat: %s", arg);
3502         end = (char *)arg;
3503
3504         switch (*arg) {
3505         case '-':
3506                 if (skip_prefix(arg, "-width", &arg)) {
3507                         if (*arg == '=')
3508                                 width = strtoul(arg + 1, &end, 10);
3509                         else if (!*arg && !av[1])
3510                                 die("Option '--stat-width' requires a value");
3511                         else if (!*arg) {
3512                                 width = strtoul(av[1], &end, 10);
3513                                 argcount = 2;
3514                         }
3515                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3516                         if (*arg == '=')
3517                                 name_width = strtoul(arg + 1, &end, 10);
3518                         else if (!*arg && !av[1])
3519                                 die("Option '--stat-name-width' requires a value");
3520                         else if (!*arg) {
3521                                 name_width = strtoul(av[1], &end, 10);
3522                                 argcount = 2;
3523                         }
3524                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3525                         if (*arg == '=')
3526                                 graph_width = strtoul(arg + 1, &end, 10);
3527                         else if (!*arg && !av[1])
3528                                 die("Option '--stat-graph-width' requires a value");
3529                         else if (!*arg) {
3530                                 graph_width = strtoul(av[1], &end, 10);
3531                                 argcount = 2;
3532                         }
3533                 } else if (skip_prefix(arg, "-count", &arg)) {
3534                         if (*arg == '=')
3535                                 count = strtoul(arg + 1, &end, 10);
3536                         else if (!*arg && !av[1])
3537                                 die("Option '--stat-count' requires a value");
3538                         else if (!*arg) {
3539                                 count = strtoul(av[1], &end, 10);
3540                                 argcount = 2;
3541                         }
3542                 }
3543                 break;
3544         case '=':
3545                 width = strtoul(arg+1, &end, 10);
3546                 if (*end == ',')
3547                         name_width = strtoul(end+1, &end, 10);
3548                 if (*end == ',')
3549                         count = strtoul(end+1, &end, 10);
3550         }
3551
3552         /* Important! This checks all the error cases! */
3553         if (*end)
3554                 return 0;
3555         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3556         options->stat_name_width = name_width;
3557         options->stat_graph_width = graph_width;
3558         options->stat_width = width;
3559         options->stat_count = count;
3560         return argcount;
3561 }
3562
3563 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3564 {
3565         struct strbuf errmsg = STRBUF_INIT;
3566         if (parse_dirstat_params(options, params, &errmsg))
3567                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3568                     errmsg.buf);
3569         strbuf_release(&errmsg);
3570         /*
3571          * The caller knows a dirstat-related option is given from the command
3572          * line; allow it to say "return this_function();"
3573          */
3574         options->output_format |= DIFF_FORMAT_DIRSTAT;
3575         return 1;
3576 }
3577
3578 static int parse_submodule_opt(struct diff_options *options, const char *value)
3579 {
3580         if (parse_submodule_params(options, value))
3581                 die(_("Failed to parse --submodule option parameter: '%s'"),
3582                         value);
3583         return 1;
3584 }
3585
3586 static const char diff_status_letters[] = {
3587         DIFF_STATUS_ADDED,
3588         DIFF_STATUS_COPIED,
3589         DIFF_STATUS_DELETED,
3590         DIFF_STATUS_MODIFIED,
3591         DIFF_STATUS_RENAMED,
3592         DIFF_STATUS_TYPE_CHANGED,
3593         DIFF_STATUS_UNKNOWN,
3594         DIFF_STATUS_UNMERGED,
3595         DIFF_STATUS_FILTER_AON,
3596         DIFF_STATUS_FILTER_BROKEN,
3597         '\0',
3598 };
3599
3600 static unsigned int filter_bit['Z' + 1];
3601
3602 static void prepare_filter_bits(void)
3603 {
3604         int i;
3605
3606         if (!filter_bit[DIFF_STATUS_ADDED]) {
3607                 for (i = 0; diff_status_letters[i]; i++)
3608                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3609         }
3610 }
3611
3612 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3613 {
3614         return opt->filter & filter_bit[(int) status];
3615 }
3616
3617 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3618 {
3619         int i, optch;
3620
3621         prepare_filter_bits();
3622
3623         /*
3624          * If there is a negation e.g. 'd' in the input, and we haven't
3625          * initialized the filter field with another --diff-filter, start
3626          * from full set of bits, except for AON.
3627          */
3628         if (!opt->filter) {
3629                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3630                         if (optch < 'a' || 'z' < optch)
3631                                 continue;
3632                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3633                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3634                         break;
3635                 }
3636         }
3637
3638         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3639                 unsigned int bit;
3640                 int negate;
3641
3642                 if ('a' <= optch && optch <= 'z') {
3643                         negate = 1;
3644                         optch = toupper(optch);
3645                 } else {
3646                         negate = 0;
3647                 }
3648
3649                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3650                 if (!bit)
3651                         return optarg[i];
3652                 if (negate)
3653                         opt->filter &= ~bit;
3654                 else
3655                         opt->filter |= bit;
3656         }
3657         return 0;
3658 }
3659
3660 static void enable_patch_output(int *fmt) {
3661         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3662         *fmt |= DIFF_FORMAT_PATCH;
3663 }
3664
3665 static int parse_one_token(const char **arg, const char *token)
3666 {
3667         const char *rest;
3668         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
3669                 *arg = rest;
3670                 return 1;
3671         }
3672         return 0;
3673 }
3674
3675 static int parse_ws_error_highlight(struct diff_options *opt, const char *arg)
3676 {
3677         const char *orig_arg = arg;
3678         unsigned val = 0;
3679         while (*arg) {
3680                 if (parse_one_token(&arg, "none"))
3681                         val = 0;
3682                 else if (parse_one_token(&arg, "default"))
3683                         val = WSEH_NEW;
3684                 else if (parse_one_token(&arg, "all"))
3685                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
3686                 else if (parse_one_token(&arg, "new"))
3687                         val |= WSEH_NEW;
3688                 else if (parse_one_token(&arg, "old"))
3689                         val |= WSEH_OLD;
3690                 else if (parse_one_token(&arg, "context"))
3691                         val |= WSEH_CONTEXT;
3692                 else {
3693                         error("unknown value after ws-error-highlight=%.*s",
3694                               (int)(arg - orig_arg), orig_arg);
3695                         return 0;
3696                 }
3697                 if (*arg)
3698                         arg++;
3699         }
3700         opt->ws_error_highlight = val;
3701         return 1;
3702 }
3703
3704 int diff_opt_parse(struct diff_options *options,
3705                    const char **av, int ac, const char *prefix)
3706 {
3707         const char *arg = av[0];
3708         const char *optarg;
3709         int argcount;
3710
3711         if (!prefix)
3712                 prefix = "";
3713
3714         /* Output format options */
3715         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3716             || opt_arg(arg, 'U', "unified", &options->context))
3717                 enable_patch_output(&options->output_format);
3718         else if (!strcmp(arg, "--raw"))
3719                 options->output_format |= DIFF_FORMAT_RAW;
3720         else if (!strcmp(arg, "--patch-with-raw")) {
3721                 enable_patch_output(&options->output_format);
3722                 options->output_format |= DIFF_FORMAT_RAW;
3723         } else if (!strcmp(arg, "--numstat"))
3724                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3725         else if (!strcmp(arg, "--shortstat"))
3726                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3727         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3728                 return parse_dirstat_opt(options, "");
3729         else if (skip_prefix(arg, "-X", &arg))
3730                 return parse_dirstat_opt(options, arg);
3731         else if (skip_prefix(arg, "--dirstat=", &arg))
3732                 return parse_dirstat_opt(options, arg);
3733         else if (!strcmp(arg, "--cumulative"))
3734                 return parse_dirstat_opt(options, "cumulative");
3735         else if (!strcmp(arg, "--dirstat-by-file"))
3736                 return parse_dirstat_opt(options, "files");
3737         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3738                 parse_dirstat_opt(options, "files");
3739                 return parse_dirstat_opt(options, arg);
3740         }
3741         else if (!strcmp(arg, "--check"))
3742                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3743         else if (!strcmp(arg, "--summary"))
3744                 options->output_format |= DIFF_FORMAT_SUMMARY;
3745         else if (!strcmp(arg, "--patch-with-stat")) {
3746                 enable_patch_output(&options->output_format);
3747                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3748         } else if (!strcmp(arg, "--name-only"))
3749                 options->output_format |= DIFF_FORMAT_NAME;
3750         else if (!strcmp(arg, "--name-status"))
3751                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3752         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3753                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3754         else if (starts_with(arg, "--stat"))
3755                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3756                 return stat_opt(options, av);
3757
3758         /* renames options */
3759         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3760                  !strcmp(arg, "--break-rewrites")) {
3761                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3762                         return error("invalid argument to -B: %s", arg+2);
3763         }
3764         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3765                  !strcmp(arg, "--find-renames")) {
3766                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3767                         return error("invalid argument to -M: %s", arg+2);
3768                 options->detect_rename = DIFF_DETECT_RENAME;
3769         }
3770         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3771                 options->irreversible_delete = 1;
3772         }
3773         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3774                  !strcmp(arg, "--find-copies")) {
3775                 if (options->detect_rename == DIFF_DETECT_COPY)
3776                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3777                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3778                         return error("invalid argument to -C: %s", arg+2);
3779                 options->detect_rename = DIFF_DETECT_COPY;
3780         }
3781         else if (!strcmp(arg, "--no-renames"))
3782                 options->detect_rename = 0;
3783         else if (!strcmp(arg, "--rename-empty"))
3784                 DIFF_OPT_SET(options, RENAME_EMPTY);
3785         else if (!strcmp(arg, "--no-rename-empty"))
3786                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3787         else if (!strcmp(arg, "--relative"))
3788                 DIFF_OPT_SET(options, RELATIVE_NAME);
3789         else if (skip_prefix(arg, "--relative=", &arg)) {
3790                 DIFF_OPT_SET(options, RELATIVE_NAME);
3791                 options->prefix = arg;
3792         }
3793
3794         /* xdiff options */
3795         else if (!strcmp(arg, "--minimal"))
3796                 DIFF_XDL_SET(options, NEED_MINIMAL);
3797         else if (!strcmp(arg, "--no-minimal"))
3798                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3799         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3800                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3801         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3802                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3803         else if (!strcmp(arg, "--ignore-space-at-eol"))
3804                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3805         else if (!strcmp(arg, "--ignore-blank-lines"))
3806                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3807         else if (!strcmp(arg, "--compaction-heuristic"))
3808                 DIFF_XDL_SET(options, COMPACTION_HEURISTIC);
3809         else if (!strcmp(arg, "--no-compaction-heuristic"))
3810                 DIFF_XDL_CLR(options, COMPACTION_HEURISTIC);
3811         else if (!strcmp(arg, "--patience"))
3812                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3813         else if (!strcmp(arg, "--histogram"))
3814                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3815         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3816                 long value = parse_algorithm_value(optarg);
3817                 if (value < 0)
3818                         return error("option diff-algorithm accepts \"myers\", "
3819                                      "\"minimal\", \"patience\" and \"histogram\"");
3820                 /* clear out previous settings */
3821                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3822                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3823                 options->xdl_opts |= value;
3824                 return argcount;
3825         }
3826
3827         /* flags options */
3828         else if (!strcmp(arg, "--binary")) {
3829                 enable_patch_output(&options->output_format);
3830                 DIFF_OPT_SET(options, BINARY);
3831         }
3832         else if (!strcmp(arg, "--full-index"))
3833                 DIFF_OPT_SET(options, FULL_INDEX);
3834         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3835                 DIFF_OPT_SET(options, TEXT);
3836         else if (!strcmp(arg, "-R"))
3837                 DIFF_OPT_SET(options, REVERSE_DIFF);
3838         else if (!strcmp(arg, "--find-copies-harder"))
3839                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3840         else if (!strcmp(arg, "--follow"))
3841                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3842         else if (!strcmp(arg, "--no-follow")) {
3843                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3844                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3845         } else if (!strcmp(arg, "--color"))
3846                 options->use_color = 1;
3847         else if (skip_prefix(arg, "--color=", &arg)) {
3848                 int value = git_config_colorbool(NULL, arg);
3849                 if (value < 0)
3850                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3851                 options->use_color = value;
3852         }
3853         else if (!strcmp(arg, "--no-color"))
3854                 options->use_color = 0;
3855         else if (!strcmp(arg, "--color-words")) {
3856                 options->use_color = 1;
3857                 options->word_diff = DIFF_WORDS_COLOR;
3858         }
3859         else if (skip_prefix(arg, "--color-words=", &arg)) {
3860                 options->use_color = 1;
3861                 options->word_diff = DIFF_WORDS_COLOR;
3862                 options->word_regex = arg;
3863         }
3864         else if (!strcmp(arg, "--word-diff")) {
3865                 if (options->word_diff == DIFF_WORDS_NONE)
3866                         options->word_diff = DIFF_WORDS_PLAIN;
3867         }
3868         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3869                 if (!strcmp(arg, "plain"))
3870                         options->word_diff = DIFF_WORDS_PLAIN;
3871                 else if (!strcmp(arg, "color")) {
3872                         options->use_color = 1;
3873                         options->word_diff = DIFF_WORDS_COLOR;
3874                 }
3875                 else if (!strcmp(arg, "porcelain"))
3876                         options->word_diff = DIFF_WORDS_PORCELAIN;
3877                 else if (!strcmp(arg, "none"))
3878                         options->word_diff = DIFF_WORDS_NONE;
3879                 else
3880                         die("bad --word-diff argument: %s", arg);
3881         }
3882         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3883                 if (options->word_diff == DIFF_WORDS_NONE)
3884                         options->word_diff = DIFF_WORDS_PLAIN;
3885                 options->word_regex = optarg;
3886                 return argcount;
3887         }
3888         else if (!strcmp(arg, "--exit-code"))
3889                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3890         else if (!strcmp(arg, "--quiet"))
3891                 DIFF_OPT_SET(options, QUICK);
3892         else if (!strcmp(arg, "--ext-diff"))
3893                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3894         else if (!strcmp(arg, "--no-ext-diff"))
3895                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3896         else if (!strcmp(arg, "--textconv"))
3897                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3898         else if (!strcmp(arg, "--no-textconv"))
3899                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3900         else if (!strcmp(arg, "--ignore-submodules")) {
3901                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3902                 handle_ignore_submodules_arg(options, "all");
3903         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3904                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3905                 handle_ignore_submodules_arg(options, arg);
3906         } else if (!strcmp(arg, "--submodule"))
3907                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3908         else if (skip_prefix(arg, "--submodule=", &arg))
3909                 return parse_submodule_opt(options, arg);
3910         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3911                 return parse_ws_error_highlight(options, arg);
3912
3913         /* misc options */
3914         else if (!strcmp(arg, "-z"))
3915                 options->line_termination = 0;
3916         else if ((argcount = short_opt('l', av, &optarg))) {
3917                 options->rename_limit = strtoul(optarg, NULL, 10);
3918                 return argcount;
3919         }
3920         else if ((argcount = short_opt('S', av, &optarg))) {
3921                 options->pickaxe = optarg;
3922                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3923                 return argcount;
3924         } else if ((argcount = short_opt('G', av, &optarg))) {
3925                 options->pickaxe = optarg;
3926                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3927                 return argcount;
3928         }
3929         else if (!strcmp(arg, "--pickaxe-all"))
3930                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3931         else if (!strcmp(arg, "--pickaxe-regex"))
3932                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3933         else if ((argcount = short_opt('O', av, &optarg))) {
3934                 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
3935                 options->orderfile = xstrdup(path);
3936                 return argcount;
3937         }
3938         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3939                 int offending = parse_diff_filter_opt(optarg, options);
3940                 if (offending)
3941                         die("unknown change class '%c' in --diff-filter=%s",
3942                             offending, optarg);
3943                 return argcount;
3944         }
3945         else if (!strcmp(arg, "--abbrev"))
3946                 options->abbrev = DEFAULT_ABBREV;
3947         else if (skip_prefix(arg, "--abbrev=", &arg)) {
3948                 options->abbrev = strtoul(arg, NULL, 10);
3949                 if (options->abbrev < MINIMUM_ABBREV)
3950                         options->abbrev = MINIMUM_ABBREV;
3951                 else if (40 < options->abbrev)
3952                         options->abbrev = 40;
3953         }
3954         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3955                 options->a_prefix = optarg;
3956                 return argcount;
3957         }
3958         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3959                 options->b_prefix = optarg;
3960                 return argcount;
3961         }
3962         else if (!strcmp(arg, "--no-prefix"))
3963                 options->a_prefix = options->b_prefix = "";
3964         else if (opt_arg(arg, '\0', "inter-hunk-context",
3965                          &options->interhunkcontext))
3966                 ;
3967         else if (!strcmp(arg, "-W"))
3968                 DIFF_OPT_SET(options, FUNCCONTEXT);
3969         else if (!strcmp(arg, "--function-context"))
3970                 DIFF_OPT_SET(options, FUNCCONTEXT);
3971         else if (!strcmp(arg, "--no-function-context"))
3972                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3973         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3974                 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
3975                 options->file = fopen(path, "w");
3976                 if (!options->file)
3977                         die_errno("Could not open '%s'", path);
3978                 options->close_file = 1;
3979                 if (options->use_color != GIT_COLOR_ALWAYS)
3980                         options->use_color = GIT_COLOR_NEVER;
3981                 return argcount;
3982         } else
3983                 return 0;
3984         return 1;
3985 }
3986
3987 int parse_rename_score(const char **cp_p)
3988 {
3989         unsigned long num, scale;
3990         int ch, dot;
3991         const char *cp = *cp_p;
3992
3993         num = 0;
3994         scale = 1;
3995         dot = 0;
3996         for (;;) {
3997                 ch = *cp;
3998                 if ( !dot && ch == '.' ) {
3999                         scale = 1;
4000                         dot = 1;
4001                 } else if ( ch == '%' ) {
4002                         scale = dot ? scale*100 : 100;
4003                         cp++;   /* % is always at the end */
4004                         break;
4005                 } else if ( ch >= '0' && ch <= '9' ) {
4006                         if ( scale < 100000 ) {
4007                                 scale *= 10;
4008                                 num = (num*10) + (ch-'0');
4009                         }
4010                 } else {
4011                         break;
4012                 }
4013                 cp++;
4014         }
4015         *cp_p = cp;
4016
4017         /* user says num divided by scale and we say internally that
4018          * is MAX_SCORE * num / scale.
4019          */
4020         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4021 }
4022
4023 static int diff_scoreopt_parse(const char *opt)
4024 {
4025         int opt1, opt2, cmd;
4026
4027         if (*opt++ != '-')
4028                 return -1;
4029         cmd = *opt++;
4030         if (cmd == '-') {
4031                 /* convert the long-form arguments into short-form versions */
4032                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4033                         if (*opt == 0 || *opt++ == '=')
4034                                 cmd = 'B';
4035                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4036                         if (*opt == 0 || *opt++ == '=')
4037                                 cmd = 'C';
4038                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4039                         if (*opt == 0 || *opt++ == '=')
4040                                 cmd = 'M';
4041                 }
4042         }
4043         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4044                 return -1; /* that is not a -M, -C, or -B option */
4045
4046         opt1 = parse_rename_score(&opt);
4047         if (cmd != 'B')
4048                 opt2 = 0;
4049         else {
4050                 if (*opt == 0)
4051                         opt2 = 0;
4052                 else if (*opt != '/')
4053                         return -1; /* we expect -B80/99 or -B80 */
4054                 else {
4055                         opt++;
4056                         opt2 = parse_rename_score(&opt);
4057                 }
4058         }
4059         if (*opt != 0)
4060                 return -1;
4061         return opt1 | (opt2 << 16);
4062 }
4063
4064 struct diff_queue_struct diff_queued_diff;
4065
4066 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4067 {
4068         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4069         queue->queue[queue->nr++] = dp;
4070 }
4071
4072 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4073                                  struct diff_filespec *one,
4074                                  struct diff_filespec *two)
4075 {
4076         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4077         dp->one = one;
4078         dp->two = two;
4079         if (queue)
4080                 diff_q(queue, dp);
4081         return dp;
4082 }
4083
4084 void diff_free_filepair(struct diff_filepair *p)
4085 {
4086         free_filespec(p->one);
4087         free_filespec(p->two);
4088         free(p);
4089 }
4090
4091 /* This is different from find_unique_abbrev() in that
4092  * it stuffs the result with dots for alignment.
4093  */
4094 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
4095 {
4096         int abblen;
4097         const char *abbrev;
4098         if (len == 40)
4099                 return sha1_to_hex(sha1);
4100
4101         abbrev = find_unique_abbrev(sha1, len);
4102         abblen = strlen(abbrev);
4103         if (abblen < 37) {
4104                 static char hex[41];
4105                 if (len < abblen && abblen <= len + 2)
4106                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4107                 else
4108                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4109                 return hex;
4110         }
4111         return sha1_to_hex(sha1);
4112 }
4113
4114 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4115 {
4116         int line_termination = opt->line_termination;
4117         int inter_name_termination = line_termination ? '\t' : '\0';
4118
4119         fprintf(opt->file, "%s", diff_line_prefix(opt));
4120         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4121                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4122                         diff_unique_abbrev(p->one->oid.hash, opt->abbrev));
4123                 fprintf(opt->file, "%s ",
4124                         diff_unique_abbrev(p->two->oid.hash, opt->abbrev));
4125         }
4126         if (p->score) {
4127                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4128                         inter_name_termination);
4129         } else {
4130                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4131         }
4132
4133         if (p->status == DIFF_STATUS_COPIED ||
4134             p->status == DIFF_STATUS_RENAMED) {
4135                 const char *name_a, *name_b;
4136                 name_a = p->one->path;
4137                 name_b = p->two->path;
4138                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4139                 write_name_quoted(name_a, opt->file, inter_name_termination);
4140                 write_name_quoted(name_b, opt->file, line_termination);
4141         } else {
4142                 const char *name_a, *name_b;
4143                 name_a = p->one->mode ? p->one->path : p->two->path;
4144                 name_b = NULL;
4145                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4146                 write_name_quoted(name_a, opt->file, line_termination);
4147         }
4148 }
4149
4150 int diff_unmodified_pair(struct diff_filepair *p)
4151 {
4152         /* This function is written stricter than necessary to support
4153          * the currently implemented transformers, but the idea is to
4154          * let transformers to produce diff_filepairs any way they want,
4155          * and filter and clean them up here before producing the output.
4156          */
4157         struct diff_filespec *one = p->one, *two = p->two;
4158
4159         if (DIFF_PAIR_UNMERGED(p))
4160                 return 0; /* unmerged is interesting */
4161
4162         /* deletion, addition, mode or type change
4163          * and rename are all interesting.
4164          */
4165         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4166             DIFF_PAIR_MODE_CHANGED(p) ||
4167             strcmp(one->path, two->path))
4168                 return 0;
4169
4170         /* both are valid and point at the same path.  that is, we are
4171          * dealing with a change.
4172          */
4173         if (one->oid_valid && two->oid_valid &&
4174             !oidcmp(&one->oid, &two->oid) &&
4175             !one->dirty_submodule && !two->dirty_submodule)
4176                 return 1; /* no change */
4177         if (!one->oid_valid && !two->oid_valid)
4178                 return 1; /* both look at the same file on the filesystem. */
4179         return 0;
4180 }
4181
4182 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4183 {
4184         if (diff_unmodified_pair(p))
4185                 return;
4186
4187         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4188             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4189                 return; /* no tree diffs in patch format */
4190
4191         run_diff(p, o);
4192 }
4193
4194 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4195                             struct diffstat_t *diffstat)
4196 {
4197         if (diff_unmodified_pair(p))
4198                 return;
4199
4200         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4201             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4202                 return; /* no useful stat for tree diffs */
4203
4204         run_diffstat(p, o, diffstat);
4205 }
4206
4207 static void diff_flush_checkdiff(struct diff_filepair *p,
4208                 struct diff_options *o)
4209 {
4210         if (diff_unmodified_pair(p))
4211                 return;
4212
4213         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4214             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4215                 return; /* nothing to check in tree diffs */
4216
4217         run_checkdiff(p, o);
4218 }
4219
4220 int diff_queue_is_empty(void)
4221 {
4222         struct diff_queue_struct *q = &diff_queued_diff;
4223         int i;
4224         for (i = 0; i < q->nr; i++)
4225                 if (!diff_unmodified_pair(q->queue[i]))
4226                         return 0;
4227         return 1;
4228 }
4229
4230 #if DIFF_DEBUG
4231 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4232 {
4233         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4234                 x, one ? one : "",
4235                 s->path,
4236                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4237                 s->mode,
4238                 s->oid_valid ? oid_to_hex(&s->oid) : "");
4239         fprintf(stderr, "queue[%d] %s size %lu\n",
4240                 x, one ? one : "",
4241                 s->size);
4242 }
4243
4244 void diff_debug_filepair(const struct diff_filepair *p, int i)
4245 {
4246         diff_debug_filespec(p->one, i, "one");
4247         diff_debug_filespec(p->two, i, "two");
4248         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4249                 p->score, p->status ? p->status : '?',
4250                 p->one->rename_used, p->broken_pair);
4251 }
4252
4253 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4254 {
4255         int i;
4256         if (msg)
4257                 fprintf(stderr, "%s\n", msg);
4258         fprintf(stderr, "q->nr = %d\n", q->nr);
4259         for (i = 0; i < q->nr; i++) {
4260                 struct diff_filepair *p = q->queue[i];
4261                 diff_debug_filepair(p, i);
4262         }
4263 }
4264 #endif
4265
4266 static void diff_resolve_rename_copy(void)
4267 {
4268         int i;
4269         struct diff_filepair *p;
4270         struct diff_queue_struct *q = &diff_queued_diff;
4271
4272         diff_debug_queue("resolve-rename-copy", q);
4273
4274         for (i = 0; i < q->nr; i++) {
4275                 p = q->queue[i];
4276                 p->status = 0; /* undecided */
4277                 if (DIFF_PAIR_UNMERGED(p))
4278                         p->status = DIFF_STATUS_UNMERGED;
4279                 else if (!DIFF_FILE_VALID(p->one))
4280                         p->status = DIFF_STATUS_ADDED;
4281                 else if (!DIFF_FILE_VALID(p->two))
4282                         p->status = DIFF_STATUS_DELETED;
4283                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4284                         p->status = DIFF_STATUS_TYPE_CHANGED;
4285
4286                 /* from this point on, we are dealing with a pair
4287                  * whose both sides are valid and of the same type, i.e.
4288                  * either in-place edit or rename/copy edit.
4289                  */
4290                 else if (DIFF_PAIR_RENAME(p)) {
4291                         /*
4292                          * A rename might have re-connected a broken
4293                          * pair up, causing the pathnames to be the
4294                          * same again. If so, that's not a rename at
4295                          * all, just a modification..
4296                          *
4297                          * Otherwise, see if this source was used for
4298                          * multiple renames, in which case we decrement
4299                          * the count, and call it a copy.
4300                          */
4301                         if (!strcmp(p->one->path, p->two->path))
4302                                 p->status = DIFF_STATUS_MODIFIED;
4303                         else if (--p->one->rename_used > 0)
4304                                 p->status = DIFF_STATUS_COPIED;
4305                         else
4306                                 p->status = DIFF_STATUS_RENAMED;
4307                 }
4308                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
4309                          p->one->mode != p->two->mode ||
4310                          p->one->dirty_submodule ||
4311                          p->two->dirty_submodule ||
4312                          is_null_oid(&p->one->oid))
4313                         p->status = DIFF_STATUS_MODIFIED;
4314                 else {
4315                         /* This is a "no-change" entry and should not
4316                          * happen anymore, but prepare for broken callers.
4317                          */
4318                         error("feeding unmodified %s to diffcore",
4319                               p->one->path);
4320                         p->status = DIFF_STATUS_UNKNOWN;
4321                 }
4322         }
4323         diff_debug_queue("resolve-rename-copy done", q);
4324 }
4325
4326 static int check_pair_status(struct diff_filepair *p)
4327 {
4328         switch (p->status) {
4329         case DIFF_STATUS_UNKNOWN:
4330                 return 0;
4331         case 0:
4332                 die("internal error in diff-resolve-rename-copy");
4333         default:
4334                 return 1;
4335         }
4336 }
4337
4338 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4339 {
4340         int fmt = opt->output_format;
4341
4342         if (fmt & DIFF_FORMAT_CHECKDIFF)
4343                 diff_flush_checkdiff(p, opt);
4344         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4345                 diff_flush_raw(p, opt);
4346         else if (fmt & DIFF_FORMAT_NAME) {
4347                 const char *name_a, *name_b;
4348                 name_a = p->two->path;
4349                 name_b = NULL;
4350                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4351                 write_name_quoted(name_a, opt->file, opt->line_termination);
4352         }
4353 }
4354
4355 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4356 {
4357         if (fs->mode)
4358                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4359         else
4360                 fprintf(file, " %s ", newdelete);
4361         write_name_quoted(fs->path, file, '\n');
4362 }
4363
4364
4365 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4366                 const char *line_prefix)
4367 {
4368         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4369                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4370                         p->two->mode, show_name ? ' ' : '\n');
4371                 if (show_name) {
4372                         write_name_quoted(p->two->path, file, '\n');
4373                 }
4374         }
4375 }
4376
4377 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4378                         const char *line_prefix)
4379 {
4380         char *names = pprint_rename(p->one->path, p->two->path);
4381
4382         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4383         free(names);
4384         show_mode_change(file, p, 0, line_prefix);
4385 }
4386
4387 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4388 {
4389         FILE *file = opt->file;
4390         const char *line_prefix = diff_line_prefix(opt);
4391
4392         switch(p->status) {
4393         case DIFF_STATUS_DELETED:
4394                 fputs(line_prefix, file);
4395                 show_file_mode_name(file, "delete", p->one);
4396                 break;
4397         case DIFF_STATUS_ADDED:
4398                 fputs(line_prefix, file);
4399                 show_file_mode_name(file, "create", p->two);
4400                 break;
4401         case DIFF_STATUS_COPIED:
4402                 fputs(line_prefix, file);
4403                 show_rename_copy(file, "copy", p, line_prefix);
4404                 break;
4405         case DIFF_STATUS_RENAMED:
4406                 fputs(line_prefix, file);
4407                 show_rename_copy(file, "rename", p, line_prefix);
4408                 break;
4409         default:
4410                 if (p->score) {
4411                         fprintf(file, "%s rewrite ", line_prefix);
4412                         write_name_quoted(p->two->path, file, ' ');
4413                         fprintf(file, "(%d%%)\n", similarity_index(p));
4414                 }
4415                 show_mode_change(file, p, !p->score, line_prefix);
4416                 break;
4417         }
4418 }
4419
4420 struct patch_id_t {
4421         git_SHA_CTX *ctx;
4422         int patchlen;
4423 };
4424
4425 static int remove_space(char *line, int len)
4426 {
4427         int i;
4428         char *dst = line;
4429         unsigned char c;
4430
4431         for (i = 0; i < len; i++)
4432                 if (!isspace((c = line[i])))
4433                         *dst++ = c;
4434
4435         return dst - line;
4436 }
4437
4438 static void patch_id_consume(void *priv, char *line, unsigned long len)
4439 {
4440         struct patch_id_t *data = priv;
4441         int new_len;
4442
4443         /* Ignore line numbers when computing the SHA1 of the patch */
4444         if (starts_with(line, "@@ -"))
4445                 return;
4446
4447         new_len = remove_space(line, len);
4448
4449         git_SHA1_Update(data->ctx, line, new_len);
4450         data->patchlen += new_len;
4451 }
4452
4453 /* returns 0 upon success, and writes result into sha1 */
4454 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4455 {
4456         struct diff_queue_struct *q = &diff_queued_diff;
4457         int i;
4458         git_SHA_CTX ctx;
4459         struct patch_id_t data;
4460         char buffer[PATH_MAX * 4 + 20];
4461
4462         git_SHA1_Init(&ctx);
4463         memset(&data, 0, sizeof(struct patch_id_t));
4464         data.ctx = &ctx;
4465
4466         for (i = 0; i < q->nr; i++) {
4467                 xpparam_t xpp;
4468                 xdemitconf_t xecfg;
4469                 mmfile_t mf1, mf2;
4470                 struct diff_filepair *p = q->queue[i];
4471                 int len1, len2;
4472
4473                 memset(&xpp, 0, sizeof(xpp));
4474                 memset(&xecfg, 0, sizeof(xecfg));
4475                 if (p->status == 0)
4476                         return error("internal diff status error");
4477                 if (p->status == DIFF_STATUS_UNKNOWN)
4478                         continue;
4479                 if (diff_unmodified_pair(p))
4480                         continue;
4481                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4482                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4483                         continue;
4484                 if (DIFF_PAIR_UNMERGED(p))
4485                         continue;
4486
4487                 diff_fill_sha1_info(p->one);
4488                 diff_fill_sha1_info(p->two);
4489
4490                 len1 = remove_space(p->one->path, strlen(p->one->path));
4491                 len2 = remove_space(p->two->path, strlen(p->two->path));
4492                 if (p->one->mode == 0)
4493                         len1 = snprintf(buffer, sizeof(buffer),
4494                                         "diff--gita/%.*sb/%.*s"
4495                                         "newfilemode%06o"
4496                                         "---/dev/null"
4497                                         "+++b/%.*s",
4498                                         len1, p->one->path,
4499                                         len2, p->two->path,
4500                                         p->two->mode,
4501                                         len2, p->two->path);
4502                 else if (p->two->mode == 0)
4503                         len1 = snprintf(buffer, sizeof(buffer),
4504                                         "diff--gita/%.*sb/%.*s"
4505                                         "deletedfilemode%06o"
4506                                         "---a/%.*s"
4507                                         "+++/dev/null",
4508                                         len1, p->one->path,
4509                                         len2, p->two->path,
4510                                         p->one->mode,
4511                                         len1, p->one->path);
4512                 else
4513                         len1 = snprintf(buffer, sizeof(buffer),
4514                                         "diff--gita/%.*sb/%.*s"
4515                                         "---a/%.*s"
4516                                         "+++b/%.*s",
4517                                         len1, p->one->path,
4518                                         len2, p->two->path,
4519                                         len1, p->one->path,
4520                                         len2, p->two->path);
4521                 git_SHA1_Update(&ctx, buffer, len1);
4522
4523                 if (diff_header_only)
4524                         continue;
4525
4526                 if (fill_mmfile(&mf1, p->one) < 0 ||
4527                     fill_mmfile(&mf2, p->two) < 0)
4528                         return error("unable to read files to diff");
4529
4530                 if (diff_filespec_is_binary(p->one) ||
4531                     diff_filespec_is_binary(p->two)) {
4532                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
4533                                         40);
4534                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
4535                                         40);
4536                         continue;
4537                 }
4538
4539                 xpp.flags = 0;
4540                 xecfg.ctxlen = 3;
4541                 xecfg.flags = 0;
4542                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4543                                   &xpp, &xecfg))
4544                         return error("unable to generate patch-id diff for %s",
4545                                      p->one->path);
4546         }
4547
4548         git_SHA1_Final(sha1, &ctx);
4549         return 0;
4550 }
4551
4552 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4553 {
4554         struct diff_queue_struct *q = &diff_queued_diff;
4555         int i;
4556         int result = diff_get_patch_id(options, sha1, diff_header_only);
4557
4558         for (i = 0; i < q->nr; i++)
4559                 diff_free_filepair(q->queue[i]);
4560
4561         free(q->queue);
4562         DIFF_QUEUE_CLEAR(q);
4563
4564         return result;
4565 }
4566
4567 static int is_summary_empty(const struct diff_queue_struct *q)
4568 {
4569         int i;
4570
4571         for (i = 0; i < q->nr; i++) {
4572                 const struct diff_filepair *p = q->queue[i];
4573
4574                 switch (p->status) {
4575                 case DIFF_STATUS_DELETED:
4576                 case DIFF_STATUS_ADDED:
4577                 case DIFF_STATUS_COPIED:
4578                 case DIFF_STATUS_RENAMED:
4579                         return 0;
4580                 default:
4581                         if (p->score)
4582                                 return 0;
4583                         if (p->one->mode && p->two->mode &&
4584                             p->one->mode != p->two->mode)
4585                                 return 0;
4586                         break;
4587                 }
4588         }
4589         return 1;
4590 }
4591
4592 static const char rename_limit_warning[] =
4593 "inexact rename detection was skipped due to too many files.";
4594
4595 static const char degrade_cc_to_c_warning[] =
4596 "only found copies from modified paths due to too many files.";
4597
4598 static const char rename_limit_advice[] =
4599 "you may want to set your %s variable to at least "
4600 "%d and retry the command.";
4601
4602 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4603 {
4604         if (degraded_cc)
4605                 warning(degrade_cc_to_c_warning);
4606         else if (needed)
4607                 warning(rename_limit_warning);
4608         else
4609                 return;
4610         if (0 < needed && needed < 32767)
4611                 warning(rename_limit_advice, varname, needed);
4612 }
4613
4614 void diff_flush(struct diff_options *options)
4615 {
4616         struct diff_queue_struct *q = &diff_queued_diff;
4617         int i, output_format = options->output_format;
4618         int separator = 0;
4619         int dirstat_by_line = 0;
4620
4621         /*
4622          * Order: raw, stat, summary, patch
4623          * or:    name/name-status/checkdiff (other bits clear)
4624          */
4625         if (!q->nr)
4626                 goto free_queue;
4627
4628         if (output_format & (DIFF_FORMAT_RAW |
4629                              DIFF_FORMAT_NAME |
4630                              DIFF_FORMAT_NAME_STATUS |
4631                              DIFF_FORMAT_CHECKDIFF)) {
4632                 for (i = 0; i < q->nr; i++) {
4633                         struct diff_filepair *p = q->queue[i];
4634                         if (check_pair_status(p))
4635                                 flush_one_pair(p, options);
4636                 }
4637                 separator++;
4638         }
4639
4640         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4641                 dirstat_by_line = 1;
4642
4643         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4644             dirstat_by_line) {
4645                 struct diffstat_t diffstat;
4646
4647                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4648                 for (i = 0; i < q->nr; i++) {
4649                         struct diff_filepair *p = q->queue[i];
4650                         if (check_pair_status(p))
4651                                 diff_flush_stat(p, options, &diffstat);
4652                 }
4653                 if (output_format & DIFF_FORMAT_NUMSTAT)
4654                         show_numstat(&diffstat, options);
4655                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4656                         show_stats(&diffstat, options);
4657                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4658                         show_shortstats(&diffstat, options);
4659                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4660                         show_dirstat_by_line(&diffstat, options);
4661                 free_diffstat_info(&diffstat);
4662                 separator++;
4663         }
4664         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4665                 show_dirstat(options);
4666
4667         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4668                 for (i = 0; i < q->nr; i++) {
4669                         diff_summary(options, q->queue[i]);
4670                 }
4671                 separator++;
4672         }
4673
4674         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4675             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4676             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4677                 /*
4678                  * run diff_flush_patch for the exit status. setting
4679                  * options->file to /dev/null should be safe, because we
4680                  * aren't supposed to produce any output anyway.
4681                  */
4682                 if (options->close_file)
4683                         fclose(options->file);
4684                 options->file = fopen("/dev/null", "w");
4685                 if (!options->file)
4686                         die_errno("Could not open /dev/null");
4687                 options->close_file = 1;
4688                 for (i = 0; i < q->nr; i++) {
4689                         struct diff_filepair *p = q->queue[i];
4690                         if (check_pair_status(p))
4691                                 diff_flush_patch(p, options);
4692                         if (options->found_changes)
4693                                 break;
4694                 }
4695         }
4696
4697         if (output_format & DIFF_FORMAT_PATCH) {
4698                 if (separator) {
4699                         fprintf(options->file, "%s%c",
4700                                 diff_line_prefix(options),
4701                                 options->line_termination);
4702                         if (options->stat_sep) {
4703                                 /* attach patch instead of inline */
4704                                 fputs(options->stat_sep, options->file);
4705                         }
4706                 }
4707
4708                 for (i = 0; i < q->nr; i++) {
4709                         struct diff_filepair *p = q->queue[i];
4710                         if (check_pair_status(p))
4711                                 diff_flush_patch(p, options);
4712                 }
4713         }
4714
4715         if (output_format & DIFF_FORMAT_CALLBACK)
4716                 options->format_callback(q, options, options->format_callback_data);
4717
4718         for (i = 0; i < q->nr; i++)
4719                 diff_free_filepair(q->queue[i]);
4720 free_queue:
4721         free(q->queue);
4722         DIFF_QUEUE_CLEAR(q);
4723         if (options->close_file)
4724                 fclose(options->file);
4725
4726         /*
4727          * Report the content-level differences with HAS_CHANGES;
4728          * diff_addremove/diff_change does not set the bit when
4729          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4730          */
4731         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4732                 if (options->found_changes)
4733                         DIFF_OPT_SET(options, HAS_CHANGES);
4734                 else
4735                         DIFF_OPT_CLR(options, HAS_CHANGES);
4736         }
4737 }
4738
4739 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4740 {
4741         return (((p->status == DIFF_STATUS_MODIFIED) &&
4742                  ((p->score &&
4743                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4744                   (!p->score &&
4745                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4746                 ((p->status != DIFF_STATUS_MODIFIED) &&
4747                  filter_bit_tst(p->status, options)));
4748 }
4749
4750 static void diffcore_apply_filter(struct diff_options *options)
4751 {
4752         int i;
4753         struct diff_queue_struct *q = &diff_queued_diff;
4754         struct diff_queue_struct outq;
4755
4756         DIFF_QUEUE_CLEAR(&outq);
4757
4758         if (!options->filter)
4759                 return;
4760
4761         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4762                 int found;
4763                 for (i = found = 0; !found && i < q->nr; i++) {
4764                         if (match_filter(options, q->queue[i]))
4765                                 found++;
4766                 }
4767                 if (found)
4768                         return;
4769
4770                 /* otherwise we will clear the whole queue
4771                  * by copying the empty outq at the end of this
4772                  * function, but first clear the current entries
4773                  * in the queue.
4774                  */
4775                 for (i = 0; i < q->nr; i++)
4776                         diff_free_filepair(q->queue[i]);
4777         }
4778         else {
4779                 /* Only the matching ones */
4780                 for (i = 0; i < q->nr; i++) {
4781                         struct diff_filepair *p = q->queue[i];
4782                         if (match_filter(options, p))
4783                                 diff_q(&outq, p);
4784                         else
4785                                 diff_free_filepair(p);
4786                 }
4787         }
4788         free(q->queue);
4789         *q = outq;
4790 }
4791
4792 /* Check whether two filespecs with the same mode and size are identical */
4793 static int diff_filespec_is_identical(struct diff_filespec *one,
4794                                       struct diff_filespec *two)
4795 {
4796         if (S_ISGITLINK(one->mode))
4797                 return 0;
4798         if (diff_populate_filespec(one, 0))
4799                 return 0;
4800         if (diff_populate_filespec(two, 0))
4801                 return 0;
4802         return !memcmp(one->data, two->data, one->size);
4803 }
4804
4805 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4806 {
4807         if (p->done_skip_stat_unmatch)
4808                 return p->skip_stat_unmatch_result;
4809
4810         p->done_skip_stat_unmatch = 1;
4811         p->skip_stat_unmatch_result = 0;
4812         /*
4813          * 1. Entries that come from stat info dirtiness
4814          *    always have both sides (iow, not create/delete),
4815          *    one side of the object name is unknown, with
4816          *    the same mode and size.  Keep the ones that
4817          *    do not match these criteria.  They have real
4818          *    differences.
4819          *
4820          * 2. At this point, the file is known to be modified,
4821          *    with the same mode and size, and the object
4822          *    name of one side is unknown.  Need to inspect
4823          *    the identical contents.
4824          */
4825         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4826             !DIFF_FILE_VALID(p->two) ||
4827             (p->one->oid_valid && p->two->oid_valid) ||
4828             (p->one->mode != p->two->mode) ||
4829             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4830             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4831             (p->one->size != p->two->size) ||
4832             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4833                 p->skip_stat_unmatch_result = 1;
4834         return p->skip_stat_unmatch_result;
4835 }
4836
4837 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4838 {
4839         int i;
4840         struct diff_queue_struct *q = &diff_queued_diff;
4841         struct diff_queue_struct outq;
4842         DIFF_QUEUE_CLEAR(&outq);
4843
4844         for (i = 0; i < q->nr; i++) {
4845                 struct diff_filepair *p = q->queue[i];
4846
4847                 if (diff_filespec_check_stat_unmatch(p))
4848                         diff_q(&outq, p);
4849                 else {
4850                         /*
4851                          * The caller can subtract 1 from skip_stat_unmatch
4852                          * to determine how many paths were dirty only
4853                          * due to stat info mismatch.
4854                          */
4855                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4856                                 diffopt->skip_stat_unmatch++;
4857                         diff_free_filepair(p);
4858                 }
4859         }
4860         free(q->queue);
4861         *q = outq;
4862 }
4863
4864 static int diffnamecmp(const void *a_, const void *b_)
4865 {
4866         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4867         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4868         const char *name_a, *name_b;
4869
4870         name_a = a->one ? a->one->path : a->two->path;
4871         name_b = b->one ? b->one->path : b->two->path;
4872         return strcmp(name_a, name_b);
4873 }
4874
4875 void diffcore_fix_diff_index(struct diff_options *options)
4876 {
4877         struct diff_queue_struct *q = &diff_queued_diff;
4878         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4879 }
4880
4881 void diffcore_std(struct diff_options *options)
4882 {
4883         /* NOTE please keep the following in sync with diff_tree_combined() */
4884         if (options->skip_stat_unmatch)
4885                 diffcore_skip_stat_unmatch(options);
4886         if (!options->found_follow) {
4887                 /* See try_to_follow_renames() in tree-diff.c */
4888                 if (options->break_opt != -1)
4889                         diffcore_break(options->break_opt);
4890                 if (options->detect_rename)
4891                         diffcore_rename(options);
4892                 if (options->break_opt != -1)
4893                         diffcore_merge_broken();
4894         }
4895         if (options->pickaxe)
4896                 diffcore_pickaxe(options);
4897         if (options->orderfile)
4898                 diffcore_order(options->orderfile);
4899         if (!options->found_follow)
4900                 /* See try_to_follow_renames() in tree-diff.c */
4901                 diff_resolve_rename_copy();
4902         diffcore_apply_filter(options);
4903
4904         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4905                 DIFF_OPT_SET(options, HAS_CHANGES);
4906         else
4907                 DIFF_OPT_CLR(options, HAS_CHANGES);
4908
4909         options->found_follow = 0;
4910 }
4911
4912 int diff_result_code(struct diff_options *opt, int status)
4913 {
4914         int result = 0;
4915
4916         diff_warn_rename_limit("diff.renameLimit",
4917                                opt->needed_rename_limit,
4918                                opt->degraded_cc_to_c);
4919         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4920             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4921                 return status;
4922         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4923             DIFF_OPT_TST(opt, HAS_CHANGES))
4924                 result |= 01;
4925         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4926             DIFF_OPT_TST(opt, CHECK_FAILED))
4927                 result |= 02;
4928         return result;
4929 }
4930
4931 int diff_can_quit_early(struct diff_options *opt)
4932 {
4933         return (DIFF_OPT_TST(opt, QUICK) &&
4934                 !opt->filter &&
4935                 DIFF_OPT_TST(opt, HAS_CHANGES));
4936 }
4937
4938 /*
4939  * Shall changes to this submodule be ignored?
4940  *
4941  * Submodule changes can be configured to be ignored separately for each path,
4942  * but that configuration can be overridden from the command line.
4943  */
4944 static int is_submodule_ignored(const char *path, struct diff_options *options)
4945 {
4946         int ignored = 0;
4947         unsigned orig_flags = options->flags;
4948         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4949                 set_diffopt_flags_from_submodule_config(options, path);
4950         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4951                 ignored = 1;
4952         options->flags = orig_flags;
4953         return ignored;
4954 }
4955
4956 void diff_addremove(struct diff_options *options,
4957                     int addremove, unsigned mode,
4958                     const unsigned char *sha1,
4959                     int sha1_valid,
4960                     const char *concatpath, unsigned dirty_submodule)
4961 {
4962         struct diff_filespec *one, *two;
4963
4964         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4965                 return;
4966
4967         /* This may look odd, but it is a preparation for
4968          * feeding "there are unchanged files which should
4969          * not produce diffs, but when you are doing copy
4970          * detection you would need them, so here they are"
4971          * entries to the diff-core.  They will be prefixed
4972          * with something like '=' or '*' (I haven't decided
4973          * which but should not make any difference).
4974          * Feeding the same new and old to diff_change()
4975          * also has the same effect.
4976          * Before the final output happens, they are pruned after
4977          * merged into rename/copy pairs as appropriate.
4978          */
4979         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4980                 addremove = (addremove == '+' ? '-' :
4981                              addremove == '-' ? '+' : addremove);
4982
4983         if (options->prefix &&
4984             strncmp(concatpath, options->prefix, options->prefix_length))
4985                 return;
4986
4987         one = alloc_filespec(concatpath);
4988         two = alloc_filespec(concatpath);
4989
4990         if (addremove != '+')
4991                 fill_filespec(one, sha1, sha1_valid, mode);
4992         if (addremove != '-') {
4993                 fill_filespec(two, sha1, sha1_valid, mode);
4994                 two->dirty_submodule = dirty_submodule;
4995         }
4996
4997         diff_queue(&diff_queued_diff, one, two);
4998         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4999                 DIFF_OPT_SET(options, HAS_CHANGES);
5000 }
5001
5002 void diff_change(struct diff_options *options,
5003                  unsigned old_mode, unsigned new_mode,
5004                  const unsigned char *old_sha1,
5005                  const unsigned char *new_sha1,
5006                  int old_sha1_valid, int new_sha1_valid,
5007                  const char *concatpath,
5008                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5009 {
5010         struct diff_filespec *one, *two;
5011         struct diff_filepair *p;
5012
5013         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5014             is_submodule_ignored(concatpath, options))
5015                 return;
5016
5017         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5018                 unsigned tmp;
5019                 const unsigned char *tmp_c;
5020                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
5021                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
5022                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
5023                         new_sha1_valid = tmp;
5024                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
5025                         new_dirty_submodule = tmp;
5026         }
5027
5028         if (options->prefix &&
5029             strncmp(concatpath, options->prefix, options->prefix_length))
5030                 return;
5031
5032         one = alloc_filespec(concatpath);
5033         two = alloc_filespec(concatpath);
5034         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5035         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5036         one->dirty_submodule = old_dirty_submodule;
5037         two->dirty_submodule = new_dirty_submodule;
5038         p = diff_queue(&diff_queued_diff, one, two);
5039
5040         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5041                 return;
5042
5043         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5044             !diff_filespec_check_stat_unmatch(p))
5045                 return;
5046
5047         DIFF_OPT_SET(options, HAS_CHANGES);
5048 }
5049
5050 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5051 {
5052         struct diff_filepair *pair;
5053         struct diff_filespec *one, *two;
5054
5055         if (options->prefix &&
5056             strncmp(path, options->prefix, options->prefix_length))
5057                 return NULL;
5058
5059         one = alloc_filespec(path);
5060         two = alloc_filespec(path);
5061         pair = diff_queue(&diff_queued_diff, one, two);
5062         pair->is_unmerged = 1;
5063         return pair;
5064 }
5065
5066 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5067                 size_t *outsize)
5068 {
5069         struct diff_tempfile *temp;
5070         const char *argv[3];
5071         const char **arg = argv;
5072         struct child_process child = CHILD_PROCESS_INIT;
5073         struct strbuf buf = STRBUF_INIT;
5074         int err = 0;
5075
5076         temp = prepare_temp_file(spec->path, spec);
5077         *arg++ = pgm;
5078         *arg++ = temp->name;
5079         *arg = NULL;
5080
5081         child.use_shell = 1;
5082         child.argv = argv;
5083         child.out = -1;
5084         if (start_command(&child)) {
5085                 remove_tempfile();
5086                 return NULL;
5087         }
5088
5089         if (strbuf_read(&buf, child.out, 0) < 0)
5090                 err = error("error reading from textconv command '%s'", pgm);
5091         close(child.out);
5092
5093         if (finish_command(&child) || err) {
5094                 strbuf_release(&buf);
5095                 remove_tempfile();
5096                 return NULL;
5097         }
5098         remove_tempfile();
5099
5100         return strbuf_detach(&buf, outsize);
5101 }
5102
5103 size_t fill_textconv(struct userdiff_driver *driver,
5104                      struct diff_filespec *df,
5105                      char **outbuf)
5106 {
5107         size_t size;
5108
5109         if (!driver) {
5110                 if (!DIFF_FILE_VALID(df)) {
5111                         *outbuf = "";
5112                         return 0;
5113                 }
5114                 if (diff_populate_filespec(df, 0))
5115                         die("unable to read files to diff");
5116                 *outbuf = df->data;
5117                 return df->size;
5118         }
5119
5120         if (!driver->textconv)
5121                 die("BUG: fill_textconv called with non-textconv driver");
5122
5123         if (driver->textconv_cache && df->oid_valid) {
5124                 *outbuf = notes_cache_get(driver->textconv_cache,
5125                                           df->oid.hash,
5126                                           &size);
5127                 if (*outbuf)
5128                         return size;
5129         }
5130
5131         *outbuf = run_textconv(driver->textconv, df, &size);
5132         if (!*outbuf)
5133                 die("unable to read files to diff");
5134
5135         if (driver->textconv_cache && df->oid_valid) {
5136                 /* ignore errors, as we might be in a readonly repository */
5137                 notes_cache_put(driver->textconv_cache, df->oid.hash, *outbuf,
5138                                 size);
5139                 /*
5140                  * we could save up changes and flush them all at the end,
5141                  * but we would need an extra call after all diffing is done.
5142                  * Since generating a cache entry is the slow path anyway,
5143                  * this extra overhead probably isn't a big deal.
5144                  */
5145                 notes_cache_write(driver->textconv_cache);
5146         }
5147
5148         return size;
5149 }
5150
5151 void setup_diff_pager(struct diff_options *opt)
5152 {
5153         /*
5154          * If the user asked for our exit code, then either they want --quiet
5155          * or --exit-code. We should definitely not bother with a pager in the
5156          * former case, as we will generate no output. Since we still properly
5157          * report our exit code even when a pager is run, we _could_ run a
5158          * pager with --exit-code. But since we have not done so historically,
5159          * and because it is easy to find people oneline advising "git diff
5160          * --exit-code" in hooks and other scripts, we do not do so.
5161          */
5162         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5163             check_pager_config("diff") != 0)
5164                 setup_pager();
5165 }