builtin/apply: avoid parameter shadowing 'linenr' global
[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 = 1;
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         int *found_changesp;
358         struct strbuf *header;
359 };
360
361 static int count_lines(const char *data, int size)
362 {
363         int count, ch, completely_empty = 1, nl_just_seen = 0;
364         count = 0;
365         while (0 < size--) {
366                 ch = *data++;
367                 if (ch == '\n') {
368                         count++;
369                         nl_just_seen = 1;
370                         completely_empty = 0;
371                 }
372                 else {
373                         nl_just_seen = 0;
374                         completely_empty = 0;
375                 }
376         }
377         if (completely_empty)
378                 return 0;
379         if (!nl_just_seen)
380                 count++; /* no trailing newline */
381         return count;
382 }
383
384 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
385 {
386         if (!DIFF_FILE_VALID(one)) {
387                 mf->ptr = (char *)""; /* does not matter */
388                 mf->size = 0;
389                 return 0;
390         }
391         else if (diff_populate_filespec(one, 0))
392                 return -1;
393
394         mf->ptr = one->data;
395         mf->size = one->size;
396         return 0;
397 }
398
399 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
400 static unsigned long diff_filespec_size(struct diff_filespec *one)
401 {
402         if (!DIFF_FILE_VALID(one))
403                 return 0;
404         diff_populate_filespec(one, CHECK_SIZE_ONLY);
405         return one->size;
406 }
407
408 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
409 {
410         char *ptr = mf->ptr;
411         long size = mf->size;
412         int cnt = 0;
413
414         if (!size)
415                 return cnt;
416         ptr += size - 1; /* pointing at the very end */
417         if (*ptr != '\n')
418                 ; /* incomplete line */
419         else
420                 ptr--; /* skip the last LF */
421         while (mf->ptr < ptr) {
422                 char *prev_eol;
423                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
424                         if (*prev_eol == '\n')
425                                 break;
426                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
427                         break;
428                 cnt++;
429                 ptr = prev_eol - 1;
430         }
431         return cnt;
432 }
433
434 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
435                                struct emit_callback *ecbdata)
436 {
437         int l1, l2, at;
438         unsigned ws_rule = ecbdata->ws_rule;
439         l1 = count_trailing_blank(mf1, ws_rule);
440         l2 = count_trailing_blank(mf2, ws_rule);
441         if (l2 <= l1) {
442                 ecbdata->blank_at_eof_in_preimage = 0;
443                 ecbdata->blank_at_eof_in_postimage = 0;
444                 return;
445         }
446         at = count_lines(mf1->ptr, mf1->size);
447         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
448
449         at = count_lines(mf2->ptr, mf2->size);
450         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
451 }
452
453 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
454                         int first, const char *line, int len)
455 {
456         int has_trailing_newline, has_trailing_carriage_return;
457         int nofirst;
458         FILE *file = o->file;
459
460         fputs(diff_line_prefix(o), file);
461
462         if (len == 0) {
463                 has_trailing_newline = (first == '\n');
464                 has_trailing_carriage_return = (!has_trailing_newline &&
465                                                 (first == '\r'));
466                 nofirst = has_trailing_newline || has_trailing_carriage_return;
467         } else {
468                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
469                 if (has_trailing_newline)
470                         len--;
471                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
472                 if (has_trailing_carriage_return)
473                         len--;
474                 nofirst = 0;
475         }
476
477         if (len || !nofirst) {
478                 fputs(set, file);
479                 if (!nofirst)
480                         fputc(first, file);
481                 fwrite(line, len, 1, file);
482                 fputs(reset, file);
483         }
484         if (has_trailing_carriage_return)
485                 fputc('\r', file);
486         if (has_trailing_newline)
487                 fputc('\n', file);
488 }
489
490 static void emit_line(struct diff_options *o, const char *set, const char *reset,
491                       const char *line, int len)
492 {
493         emit_line_0(o, set, reset, line[0], line+1, len-1);
494 }
495
496 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
497 {
498         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
499               ecbdata->blank_at_eof_in_preimage &&
500               ecbdata->blank_at_eof_in_postimage &&
501               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
502               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
503                 return 0;
504         return ws_blank_line(line, len, ecbdata->ws_rule);
505 }
506
507 static void emit_line_checked(const char *reset,
508                               struct emit_callback *ecbdata,
509                               const char *line, int len,
510                               enum color_diff color,
511                               unsigned ws_error_highlight,
512                               char sign)
513 {
514         const char *set = diff_get_color(ecbdata->color_diff, color);
515         const char *ws = NULL;
516
517         if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
518                 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
519                 if (!*ws)
520                         ws = NULL;
521         }
522
523         if (!ws)
524                 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
525         else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
526                 /* Blank line at EOF - paint '+' as well */
527                 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
528         else {
529                 /* Emit just the prefix, then the rest. */
530                 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
531                 ws_check_emit(line, len, ecbdata->ws_rule,
532                               ecbdata->opt->file, set, reset, ws);
533         }
534 }
535
536 static void emit_add_line(const char *reset,
537                           struct emit_callback *ecbdata,
538                           const char *line, int len)
539 {
540         emit_line_checked(reset, ecbdata, line, len,
541                           DIFF_FILE_NEW, WSEH_NEW, '+');
542 }
543
544 static void emit_del_line(const char *reset,
545                           struct emit_callback *ecbdata,
546                           const char *line, int len)
547 {
548         emit_line_checked(reset, ecbdata, line, len,
549                           DIFF_FILE_OLD, WSEH_OLD, '-');
550 }
551
552 static void emit_context_line(const char *reset,
553                               struct emit_callback *ecbdata,
554                               const char *line, int len)
555 {
556         emit_line_checked(reset, ecbdata, line, len,
557                           DIFF_CONTEXT, WSEH_CONTEXT, ' ');
558 }
559
560 static void emit_hunk_header(struct emit_callback *ecbdata,
561                              const char *line, int len)
562 {
563         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
564         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
565         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
566         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
567         static const char atat[2] = { '@', '@' };
568         const char *cp, *ep;
569         struct strbuf msgbuf = STRBUF_INIT;
570         int org_len = len;
571         int i = 1;
572
573         /*
574          * As a hunk header must begin with "@@ -<old>, +<new> @@",
575          * it always is at least 10 bytes long.
576          */
577         if (len < 10 ||
578             memcmp(line, atat, 2) ||
579             !(ep = memmem(line + 2, len - 2, atat, 2))) {
580                 emit_line(ecbdata->opt, context, reset, line, len);
581                 return;
582         }
583         ep += 2; /* skip over @@ */
584
585         /* The hunk header in fraginfo color */
586         strbuf_addstr(&msgbuf, frag);
587         strbuf_add(&msgbuf, line, ep - line);
588         strbuf_addstr(&msgbuf, reset);
589
590         /*
591          * trailing "\r\n"
592          */
593         for ( ; i < 3; i++)
594                 if (line[len - i] == '\r' || line[len - i] == '\n')
595                         len--;
596
597         /* blank before the func header */
598         for (cp = ep; ep - line < len; ep++)
599                 if (*ep != ' ' && *ep != '\t')
600                         break;
601         if (ep != cp) {
602                 strbuf_addstr(&msgbuf, context);
603                 strbuf_add(&msgbuf, cp, ep - cp);
604                 strbuf_addstr(&msgbuf, reset);
605         }
606
607         if (ep < line + len) {
608                 strbuf_addstr(&msgbuf, func);
609                 strbuf_add(&msgbuf, ep, line + len - ep);
610                 strbuf_addstr(&msgbuf, reset);
611         }
612
613         strbuf_add(&msgbuf, line + len, org_len - len);
614         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
615         strbuf_release(&msgbuf);
616 }
617
618 static struct diff_tempfile *claim_diff_tempfile(void) {
619         int i;
620         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
621                 if (!diff_temp[i].name)
622                         return diff_temp + i;
623         die("BUG: diff is failing to clean up its tempfiles");
624 }
625
626 static void remove_tempfile(void)
627 {
628         int i;
629         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
630                 if (is_tempfile_active(&diff_temp[i].tempfile))
631                         delete_tempfile(&diff_temp[i].tempfile);
632                 diff_temp[i].name = NULL;
633         }
634 }
635
636 static void print_line_count(FILE *file, int count)
637 {
638         switch (count) {
639         case 0:
640                 fprintf(file, "0,0");
641                 break;
642         case 1:
643                 fprintf(file, "1");
644                 break;
645         default:
646                 fprintf(file, "1,%d", count);
647                 break;
648         }
649 }
650
651 static void emit_rewrite_lines(struct emit_callback *ecb,
652                                int prefix, const char *data, int size)
653 {
654         const char *endp = NULL;
655         static const char *nneof = " No newline at end of file\n";
656         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
657
658         while (0 < size) {
659                 int len;
660
661                 endp = memchr(data, '\n', size);
662                 len = endp ? (endp - data + 1) : size;
663                 if (prefix != '+') {
664                         ecb->lno_in_preimage++;
665                         emit_del_line(reset, ecb, data, len);
666                 } else {
667                         ecb->lno_in_postimage++;
668                         emit_add_line(reset, ecb, data, len);
669                 }
670                 size -= len;
671                 data += len;
672         }
673         if (!endp) {
674                 const char *context = diff_get_color(ecb->color_diff,
675                                                      DIFF_CONTEXT);
676                 putc('\n', ecb->opt->file);
677                 emit_line_0(ecb->opt, context, reset, '\\',
678                             nneof, strlen(nneof));
679         }
680 }
681
682 static void emit_rewrite_diff(const char *name_a,
683                               const char *name_b,
684                               struct diff_filespec *one,
685                               struct diff_filespec *two,
686                               struct userdiff_driver *textconv_one,
687                               struct userdiff_driver *textconv_two,
688                               struct diff_options *o)
689 {
690         int lc_a, lc_b;
691         const char *name_a_tab, *name_b_tab;
692         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
693         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
694         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
695         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
696         const char *a_prefix, *b_prefix;
697         char *data_one, *data_two;
698         size_t size_one, size_two;
699         struct emit_callback ecbdata;
700         const char *line_prefix = diff_line_prefix(o);
701
702         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
703                 a_prefix = o->b_prefix;
704                 b_prefix = o->a_prefix;
705         } else {
706                 a_prefix = o->a_prefix;
707                 b_prefix = o->b_prefix;
708         }
709
710         name_a += (*name_a == '/');
711         name_b += (*name_b == '/');
712         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
713         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
714
715         strbuf_reset(&a_name);
716         strbuf_reset(&b_name);
717         quote_two_c_style(&a_name, a_prefix, name_a, 0);
718         quote_two_c_style(&b_name, b_prefix, name_b, 0);
719
720         size_one = fill_textconv(textconv_one, one, &data_one);
721         size_two = fill_textconv(textconv_two, two, &data_two);
722
723         memset(&ecbdata, 0, sizeof(ecbdata));
724         ecbdata.color_diff = want_color(o->use_color);
725         ecbdata.found_changesp = &o->found_changes;
726         ecbdata.ws_rule = whitespace_rule(name_b);
727         ecbdata.opt = o;
728         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
729                 mmfile_t mf1, mf2;
730                 mf1.ptr = (char *)data_one;
731                 mf2.ptr = (char *)data_two;
732                 mf1.size = size_one;
733                 mf2.size = size_two;
734                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
735         }
736         ecbdata.lno_in_preimage = 1;
737         ecbdata.lno_in_postimage = 1;
738
739         lc_a = count_lines(data_one, size_one);
740         lc_b = count_lines(data_two, size_two);
741         fprintf(o->file,
742                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
743                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
744                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
745                 line_prefix, fraginfo);
746         if (!o->irreversible_delete)
747                 print_line_count(o->file, lc_a);
748         else
749                 fprintf(o->file, "?,?");
750         fprintf(o->file, " +");
751         print_line_count(o->file, lc_b);
752         fprintf(o->file, " @@%s\n", reset);
753         if (lc_a && !o->irreversible_delete)
754                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
755         if (lc_b)
756                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
757         if (textconv_one)
758                 free((char *)data_one);
759         if (textconv_two)
760                 free((char *)data_two);
761 }
762
763 struct diff_words_buffer {
764         mmfile_t text;
765         long alloc;
766         struct diff_words_orig {
767                 const char *begin, *end;
768         } *orig;
769         int orig_nr, orig_alloc;
770 };
771
772 static void diff_words_append(char *line, unsigned long len,
773                 struct diff_words_buffer *buffer)
774 {
775         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
776         line++;
777         len--;
778         memcpy(buffer->text.ptr + buffer->text.size, line, len);
779         buffer->text.size += len;
780         buffer->text.ptr[buffer->text.size] = '\0';
781 }
782
783 struct diff_words_style_elem {
784         const char *prefix;
785         const char *suffix;
786         const char *color; /* NULL; filled in by the setup code if
787                             * color is enabled */
788 };
789
790 struct diff_words_style {
791         enum diff_words_type type;
792         struct diff_words_style_elem new, old, ctx;
793         const char *newline;
794 };
795
796 static struct diff_words_style diff_words_styles[] = {
797         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
798         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
799         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
800 };
801
802 struct diff_words_data {
803         struct diff_words_buffer minus, plus;
804         const char *current_plus;
805         int last_minus;
806         struct diff_options *opt;
807         regex_t *word_regex;
808         enum diff_words_type type;
809         struct diff_words_style *style;
810 };
811
812 static int fn_out_diff_words_write_helper(FILE *fp,
813                                           struct diff_words_style_elem *st_el,
814                                           const char *newline,
815                                           size_t count, const char *buf,
816                                           const char *line_prefix)
817 {
818         int print = 0;
819
820         while (count) {
821                 char *p = memchr(buf, '\n', count);
822                 if (print)
823                         fputs(line_prefix, fp);
824                 if (p != buf) {
825                         if (st_el->color && fputs(st_el->color, fp) < 0)
826                                 return -1;
827                         if (fputs(st_el->prefix, fp) < 0 ||
828                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
829                             fputs(st_el->suffix, fp) < 0)
830                                 return -1;
831                         if (st_el->color && *st_el->color
832                             && fputs(GIT_COLOR_RESET, fp) < 0)
833                                 return -1;
834                 }
835                 if (!p)
836                         return 0;
837                 if (fputs(newline, fp) < 0)
838                         return -1;
839                 count -= p + 1 - buf;
840                 buf = p + 1;
841                 print = 1;
842         }
843         return 0;
844 }
845
846 /*
847  * '--color-words' algorithm can be described as:
848  *
849  *   1. collect a the minus/plus lines of a diff hunk, divided into
850  *      minus-lines and plus-lines;
851  *
852  *   2. break both minus-lines and plus-lines into words and
853  *      place them into two mmfile_t with one word for each line;
854  *
855  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
856  *
857  * And for the common parts of the both file, we output the plus side text.
858  * diff_words->current_plus is used to trace the current position of the plus file
859  * which printed. diff_words->last_minus is used to trace the last minus word
860  * printed.
861  *
862  * For '--graph' to work with '--color-words', we need to output the graph prefix
863  * on each line of color words output. Generally, there are two conditions on
864  * which we should output the prefix.
865  *
866  *   1. diff_words->last_minus == 0 &&
867  *      diff_words->current_plus == diff_words->plus.text.ptr
868  *
869  *      that is: the plus text must start as a new line, and if there is no minus
870  *      word printed, a graph prefix must be printed.
871  *
872  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
873  *      *(diff_words->current_plus - 1) == '\n'
874  *
875  *      that is: a graph prefix must be printed following a '\n'
876  */
877 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
878 {
879         if ((diff_words->last_minus == 0 &&
880                 diff_words->current_plus == diff_words->plus.text.ptr) ||
881                 (diff_words->current_plus > diff_words->plus.text.ptr &&
882                 *(diff_words->current_plus - 1) == '\n')) {
883                 return 1;
884         } else {
885                 return 0;
886         }
887 }
888
889 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
890 {
891         struct diff_words_data *diff_words = priv;
892         struct diff_words_style *style = diff_words->style;
893         int minus_first, minus_len, plus_first, plus_len;
894         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
895         struct diff_options *opt = diff_words->opt;
896         const char *line_prefix;
897
898         if (line[0] != '@' || parse_hunk_header(line, len,
899                         &minus_first, &minus_len, &plus_first, &plus_len))
900                 return;
901
902         assert(opt);
903         line_prefix = diff_line_prefix(opt);
904
905         /* POSIX requires that first be decremented by one if len == 0... */
906         if (minus_len) {
907                 minus_begin = diff_words->minus.orig[minus_first].begin;
908                 minus_end =
909                         diff_words->minus.orig[minus_first + minus_len - 1].end;
910         } else
911                 minus_begin = minus_end =
912                         diff_words->minus.orig[minus_first].end;
913
914         if (plus_len) {
915                 plus_begin = diff_words->plus.orig[plus_first].begin;
916                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
917         } else
918                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
919
920         if (color_words_output_graph_prefix(diff_words)) {
921                 fputs(line_prefix, diff_words->opt->file);
922         }
923         if (diff_words->current_plus != plus_begin) {
924                 fn_out_diff_words_write_helper(diff_words->opt->file,
925                                 &style->ctx, style->newline,
926                                 plus_begin - diff_words->current_plus,
927                                 diff_words->current_plus, line_prefix);
928                 if (*(plus_begin - 1) == '\n')
929                         fputs(line_prefix, diff_words->opt->file);
930         }
931         if (minus_begin != minus_end) {
932                 fn_out_diff_words_write_helper(diff_words->opt->file,
933                                 &style->old, style->newline,
934                                 minus_end - minus_begin, minus_begin,
935                                 line_prefix);
936         }
937         if (plus_begin != plus_end) {
938                 fn_out_diff_words_write_helper(diff_words->opt->file,
939                                 &style->new, style->newline,
940                                 plus_end - plus_begin, plus_begin,
941                                 line_prefix);
942         }
943
944         diff_words->current_plus = plus_end;
945         diff_words->last_minus = minus_first;
946 }
947
948 /* This function starts looking at *begin, and returns 0 iff a word was found. */
949 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
950                 int *begin, int *end)
951 {
952         if (word_regex && *begin < buffer->size) {
953                 regmatch_t match[1];
954                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
955                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
956                                         '\n', match[0].rm_eo - match[0].rm_so);
957                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
958                         *begin += match[0].rm_so;
959                         return *begin >= *end;
960                 }
961                 return -1;
962         }
963
964         /* find the next word */
965         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
966                 (*begin)++;
967         if (*begin >= buffer->size)
968                 return -1;
969
970         /* find the end of the word */
971         *end = *begin + 1;
972         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
973                 (*end)++;
974
975         return 0;
976 }
977
978 /*
979  * This function splits the words in buffer->text, stores the list with
980  * newline separator into out, and saves the offsets of the original words
981  * in buffer->orig.
982  */
983 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
984                 regex_t *word_regex)
985 {
986         int i, j;
987         long alloc = 0;
988
989         out->size = 0;
990         out->ptr = NULL;
991
992         /* fake an empty "0th" word */
993         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
994         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
995         buffer->orig_nr = 1;
996
997         for (i = 0; i < buffer->text.size; i++) {
998                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
999                         return;
1000
1001                 /* store original boundaries */
1002                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1003                                 buffer->orig_alloc);
1004                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1005                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1006                 buffer->orig_nr++;
1007
1008                 /* store one word */
1009                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1010                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1011                 out->ptr[out->size + j - i] = '\n';
1012                 out->size += j - i + 1;
1013
1014                 i = j - 1;
1015         }
1016 }
1017
1018 /* this executes the word diff on the accumulated buffers */
1019 static void diff_words_show(struct diff_words_data *diff_words)
1020 {
1021         xpparam_t xpp;
1022         xdemitconf_t xecfg;
1023         mmfile_t minus, plus;
1024         struct diff_words_style *style = diff_words->style;
1025
1026         struct diff_options *opt = diff_words->opt;
1027         const char *line_prefix;
1028
1029         assert(opt);
1030         line_prefix = diff_line_prefix(opt);
1031
1032         /* special case: only removal */
1033         if (!diff_words->plus.text.size) {
1034                 fputs(line_prefix, diff_words->opt->file);
1035                 fn_out_diff_words_write_helper(diff_words->opt->file,
1036                         &style->old, style->newline,
1037                         diff_words->minus.text.size,
1038                         diff_words->minus.text.ptr, line_prefix);
1039                 diff_words->minus.text.size = 0;
1040                 return;
1041         }
1042
1043         diff_words->current_plus = diff_words->plus.text.ptr;
1044         diff_words->last_minus = 0;
1045
1046         memset(&xpp, 0, sizeof(xpp));
1047         memset(&xecfg, 0, sizeof(xecfg));
1048         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1049         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1050         xpp.flags = 0;
1051         /* as only the hunk header will be parsed, we need a 0-context */
1052         xecfg.ctxlen = 0;
1053         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1054                           &xpp, &xecfg))
1055                 die("unable to generate word diff");
1056         free(minus.ptr);
1057         free(plus.ptr);
1058         if (diff_words->current_plus != diff_words->plus.text.ptr +
1059                         diff_words->plus.text.size) {
1060                 if (color_words_output_graph_prefix(diff_words))
1061                         fputs(line_prefix, diff_words->opt->file);
1062                 fn_out_diff_words_write_helper(diff_words->opt->file,
1063                         &style->ctx, style->newline,
1064                         diff_words->plus.text.ptr + diff_words->plus.text.size
1065                         - diff_words->current_plus, diff_words->current_plus,
1066                         line_prefix);
1067         }
1068         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1069 }
1070
1071 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1072 static void diff_words_flush(struct emit_callback *ecbdata)
1073 {
1074         if (ecbdata->diff_words->minus.text.size ||
1075             ecbdata->diff_words->plus.text.size)
1076                 diff_words_show(ecbdata->diff_words);
1077 }
1078
1079 static void diff_filespec_load_driver(struct diff_filespec *one)
1080 {
1081         /* Use already-loaded driver */
1082         if (one->driver)
1083                 return;
1084
1085         if (S_ISREG(one->mode))
1086                 one->driver = userdiff_find_by_path(one->path);
1087
1088         /* Fallback to default settings */
1089         if (!one->driver)
1090                 one->driver = userdiff_find_by_name("default");
1091 }
1092
1093 static const char *userdiff_word_regex(struct diff_filespec *one)
1094 {
1095         diff_filespec_load_driver(one);
1096         return one->driver->word_regex;
1097 }
1098
1099 static void init_diff_words_data(struct emit_callback *ecbdata,
1100                                  struct diff_options *orig_opts,
1101                                  struct diff_filespec *one,
1102                                  struct diff_filespec *two)
1103 {
1104         int i;
1105         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1106         memcpy(o, orig_opts, sizeof(struct diff_options));
1107
1108         ecbdata->diff_words =
1109                 xcalloc(1, sizeof(struct diff_words_data));
1110         ecbdata->diff_words->type = o->word_diff;
1111         ecbdata->diff_words->opt = o;
1112         if (!o->word_regex)
1113                 o->word_regex = userdiff_word_regex(one);
1114         if (!o->word_regex)
1115                 o->word_regex = userdiff_word_regex(two);
1116         if (!o->word_regex)
1117                 o->word_regex = diff_word_regex_cfg;
1118         if (o->word_regex) {
1119                 ecbdata->diff_words->word_regex = (regex_t *)
1120                         xmalloc(sizeof(regex_t));
1121                 if (regcomp(ecbdata->diff_words->word_regex,
1122                             o->word_regex,
1123                             REG_EXTENDED | REG_NEWLINE))
1124                         die ("Invalid regular expression: %s",
1125                              o->word_regex);
1126         }
1127         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1128                 if (o->word_diff == diff_words_styles[i].type) {
1129                         ecbdata->diff_words->style =
1130                                 &diff_words_styles[i];
1131                         break;
1132                 }
1133         }
1134         if (want_color(o->use_color)) {
1135                 struct diff_words_style *st = ecbdata->diff_words->style;
1136                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1137                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1138                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1139         }
1140 }
1141
1142 static void free_diff_words_data(struct emit_callback *ecbdata)
1143 {
1144         if (ecbdata->diff_words) {
1145                 diff_words_flush(ecbdata);
1146                 free (ecbdata->diff_words->opt);
1147                 free (ecbdata->diff_words->minus.text.ptr);
1148                 free (ecbdata->diff_words->minus.orig);
1149                 free (ecbdata->diff_words->plus.text.ptr);
1150                 free (ecbdata->diff_words->plus.orig);
1151                 if (ecbdata->diff_words->word_regex) {
1152                         regfree(ecbdata->diff_words->word_regex);
1153                         free(ecbdata->diff_words->word_regex);
1154                 }
1155                 free(ecbdata->diff_words);
1156                 ecbdata->diff_words = NULL;
1157         }
1158 }
1159
1160 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1161 {
1162         if (want_color(diff_use_color))
1163                 return diff_colors[ix];
1164         return "";
1165 }
1166
1167 const char *diff_line_prefix(struct diff_options *opt)
1168 {
1169         struct strbuf *msgbuf;
1170         if (!opt->output_prefix)
1171                 return "";
1172
1173         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1174         return msgbuf->buf;
1175 }
1176
1177 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1178 {
1179         const char *cp;
1180         unsigned long allot;
1181         size_t l = len;
1182
1183         if (ecb->truncate)
1184                 return ecb->truncate(line, len);
1185         cp = line;
1186         allot = l;
1187         while (0 < l) {
1188                 (void) utf8_width(&cp, &l);
1189                 if (!cp)
1190                         break; /* truncated in the middle? */
1191         }
1192         return allot - l;
1193 }
1194
1195 static void find_lno(const char *line, struct emit_callback *ecbdata)
1196 {
1197         const char *p;
1198         ecbdata->lno_in_preimage = 0;
1199         ecbdata->lno_in_postimage = 0;
1200         p = strchr(line, '-');
1201         if (!p)
1202                 return; /* cannot happen */
1203         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1204         p = strchr(p, '+');
1205         if (!p)
1206                 return; /* cannot happen */
1207         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1208 }
1209
1210 static void fn_out_consume(void *priv, char *line, unsigned long len)
1211 {
1212         struct emit_callback *ecbdata = priv;
1213         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1214         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1215         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1216         struct diff_options *o = ecbdata->opt;
1217         const char *line_prefix = diff_line_prefix(o);
1218
1219         if (ecbdata->header) {
1220                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1221                 strbuf_reset(ecbdata->header);
1222                 ecbdata->header = NULL;
1223         }
1224         *(ecbdata->found_changesp) = 1;
1225
1226         if (ecbdata->label_path[0]) {
1227                 const char *name_a_tab, *name_b_tab;
1228
1229                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1230                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1231
1232                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1233                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1234                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1235                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1236                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1237         }
1238
1239         if (diff_suppress_blank_empty
1240             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1241                 line[0] = '\n';
1242                 len = 1;
1243         }
1244
1245         if (line[0] == '@') {
1246                 if (ecbdata->diff_words)
1247                         diff_words_flush(ecbdata);
1248                 len = sane_truncate_line(ecbdata, line, len);
1249                 find_lno(line, ecbdata);
1250                 emit_hunk_header(ecbdata, line, len);
1251                 if (line[len-1] != '\n')
1252                         putc('\n', ecbdata->opt->file);
1253                 return;
1254         }
1255
1256         if (len < 1) {
1257                 emit_line(ecbdata->opt, reset, reset, line, len);
1258                 if (ecbdata->diff_words
1259                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1260                         fputs("~\n", ecbdata->opt->file);
1261                 return;
1262         }
1263
1264         if (ecbdata->diff_words) {
1265                 if (line[0] == '-') {
1266                         diff_words_append(line, len,
1267                                           &ecbdata->diff_words->minus);
1268                         return;
1269                 } else if (line[0] == '+') {
1270                         diff_words_append(line, len,
1271                                           &ecbdata->diff_words->plus);
1272                         return;
1273                 } else if (starts_with(line, "\\ ")) {
1274                         /*
1275                          * Eat the "no newline at eof" marker as if we
1276                          * saw a "+" or "-" line with nothing on it,
1277                          * and return without diff_words_flush() to
1278                          * defer processing. If this is the end of
1279                          * preimage, more "+" lines may come after it.
1280                          */
1281                         return;
1282                 }
1283                 diff_words_flush(ecbdata);
1284                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1285                         emit_line(ecbdata->opt, context, reset, line, len);
1286                         fputs("~\n", ecbdata->opt->file);
1287                 } else {
1288                         /*
1289                          * Skip the prefix character, if any.  With
1290                          * diff_suppress_blank_empty, there may be
1291                          * none.
1292                          */
1293                         if (line[0] != '\n') {
1294                               line++;
1295                               len--;
1296                         }
1297                         emit_line(ecbdata->opt, context, reset, line, len);
1298                 }
1299                 return;
1300         }
1301
1302         switch (line[0]) {
1303         case '+':
1304                 ecbdata->lno_in_postimage++;
1305                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1306                 break;
1307         case '-':
1308                 ecbdata->lno_in_preimage++;
1309                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1310                 break;
1311         case ' ':
1312                 ecbdata->lno_in_postimage++;
1313                 ecbdata->lno_in_preimage++;
1314                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1315                 break;
1316         default:
1317                 /* incomplete line at the end */
1318                 ecbdata->lno_in_preimage++;
1319                 emit_line(ecbdata->opt,
1320                           diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1321                           reset, line, len);
1322                 break;
1323         }
1324 }
1325
1326 static char *pprint_rename(const char *a, const char *b)
1327 {
1328         const char *old = a;
1329         const char *new = b;
1330         struct strbuf name = STRBUF_INIT;
1331         int pfx_length, sfx_length;
1332         int pfx_adjust_for_slash;
1333         int len_a = strlen(a);
1334         int len_b = strlen(b);
1335         int a_midlen, b_midlen;
1336         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1337         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1338
1339         if (qlen_a || qlen_b) {
1340                 quote_c_style(a, &name, NULL, 0);
1341                 strbuf_addstr(&name, " => ");
1342                 quote_c_style(b, &name, NULL, 0);
1343                 return strbuf_detach(&name, NULL);
1344         }
1345
1346         /* Find common prefix */
1347         pfx_length = 0;
1348         while (*old && *new && *old == *new) {
1349                 if (*old == '/')
1350                         pfx_length = old - a + 1;
1351                 old++;
1352                 new++;
1353         }
1354
1355         /* Find common suffix */
1356         old = a + len_a;
1357         new = b + len_b;
1358         sfx_length = 0;
1359         /*
1360          * If there is a common prefix, it must end in a slash.  In
1361          * that case we let this loop run 1 into the prefix to see the
1362          * same slash.
1363          *
1364          * If there is no common prefix, we cannot do this as it would
1365          * underrun the input strings.
1366          */
1367         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1368         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1369                b + pfx_length - pfx_adjust_for_slash <= new &&
1370                *old == *new) {
1371                 if (*old == '/')
1372                         sfx_length = len_a - (old - a);
1373                 old--;
1374                 new--;
1375         }
1376
1377         /*
1378          * pfx{mid-a => mid-b}sfx
1379          * {pfx-a => pfx-b}sfx
1380          * pfx{sfx-a => sfx-b}
1381          * name-a => name-b
1382          */
1383         a_midlen = len_a - pfx_length - sfx_length;
1384         b_midlen = len_b - pfx_length - sfx_length;
1385         if (a_midlen < 0)
1386                 a_midlen = 0;
1387         if (b_midlen < 0)
1388                 b_midlen = 0;
1389
1390         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1391         if (pfx_length + sfx_length) {
1392                 strbuf_add(&name, a, pfx_length);
1393                 strbuf_addch(&name, '{');
1394         }
1395         strbuf_add(&name, a + pfx_length, a_midlen);
1396         strbuf_addstr(&name, " => ");
1397         strbuf_add(&name, b + pfx_length, b_midlen);
1398         if (pfx_length + sfx_length) {
1399                 strbuf_addch(&name, '}');
1400                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1401         }
1402         return strbuf_detach(&name, NULL);
1403 }
1404
1405 struct diffstat_t {
1406         int nr;
1407         int alloc;
1408         struct diffstat_file {
1409                 char *from_name;
1410                 char *name;
1411                 char *print_name;
1412                 unsigned is_unmerged:1;
1413                 unsigned is_binary:1;
1414                 unsigned is_renamed:1;
1415                 unsigned is_interesting:1;
1416                 uintmax_t added, deleted;
1417         } **files;
1418 };
1419
1420 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1421                                           const char *name_a,
1422                                           const char *name_b)
1423 {
1424         struct diffstat_file *x;
1425         x = xcalloc(1, sizeof(*x));
1426         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1427         diffstat->files[diffstat->nr++] = x;
1428         if (name_b) {
1429                 x->from_name = xstrdup(name_a);
1430                 x->name = xstrdup(name_b);
1431                 x->is_renamed = 1;
1432         }
1433         else {
1434                 x->from_name = NULL;
1435                 x->name = xstrdup(name_a);
1436         }
1437         return x;
1438 }
1439
1440 static void diffstat_consume(void *priv, char *line, unsigned long len)
1441 {
1442         struct diffstat_t *diffstat = priv;
1443         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1444
1445         if (line[0] == '+')
1446                 x->added++;
1447         else if (line[0] == '-')
1448                 x->deleted++;
1449 }
1450
1451 const char mime_boundary_leader[] = "------------";
1452
1453 static int scale_linear(int it, int width, int max_change)
1454 {
1455         if (!it)
1456                 return 0;
1457         /*
1458          * make sure that at least one '-' or '+' is printed if
1459          * there is any change to this path. The easiest way is to
1460          * scale linearly as if the alloted width is one column shorter
1461          * than it is, and then add 1 to the result.
1462          */
1463         return 1 + (it * (width - 1) / max_change);
1464 }
1465
1466 static void show_name(FILE *file,
1467                       const char *prefix, const char *name, int len)
1468 {
1469         fprintf(file, " %s%-*s |", prefix, len, name);
1470 }
1471
1472 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1473 {
1474         if (cnt <= 0)
1475                 return;
1476         fprintf(file, "%s", set);
1477         while (cnt--)
1478                 putc(ch, file);
1479         fprintf(file, "%s", reset);
1480 }
1481
1482 static void fill_print_name(struct diffstat_file *file)
1483 {
1484         char *pname;
1485
1486         if (file->print_name)
1487                 return;
1488
1489         if (!file->is_renamed) {
1490                 struct strbuf buf = STRBUF_INIT;
1491                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1492                         pname = strbuf_detach(&buf, NULL);
1493                 } else {
1494                         pname = file->name;
1495                         strbuf_release(&buf);
1496                 }
1497         } else {
1498                 pname = pprint_rename(file->from_name, file->name);
1499         }
1500         file->print_name = pname;
1501 }
1502
1503 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1504 {
1505         struct strbuf sb = STRBUF_INIT;
1506         int ret;
1507
1508         if (!files) {
1509                 assert(insertions == 0 && deletions == 0);
1510                 return fprintf(fp, "%s\n", " 0 files changed");
1511         }
1512
1513         strbuf_addf(&sb,
1514                     (files == 1) ? " %d file changed" : " %d files changed",
1515                     files);
1516
1517         /*
1518          * For binary diff, the caller may want to print "x files
1519          * changed" with insertions == 0 && deletions == 0.
1520          *
1521          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1522          * is probably less confusing (i.e skip over "2 files changed
1523          * but nothing about added/removed lines? Is this a bug in Git?").
1524          */
1525         if (insertions || deletions == 0) {
1526                 strbuf_addf(&sb,
1527                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1528                             insertions);
1529         }
1530
1531         if (deletions || insertions == 0) {
1532                 strbuf_addf(&sb,
1533                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1534                             deletions);
1535         }
1536         strbuf_addch(&sb, '\n');
1537         ret = fputs(sb.buf, fp);
1538         strbuf_release(&sb);
1539         return ret;
1540 }
1541
1542 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1543 {
1544         int i, len, add, del, adds = 0, dels = 0;
1545         uintmax_t max_change = 0, max_len = 0;
1546         int total_files = data->nr, count;
1547         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1548         const char *reset, *add_c, *del_c;
1549         const char *line_prefix = "";
1550         int extra_shown = 0;
1551
1552         if (data->nr == 0)
1553                 return;
1554
1555         line_prefix = diff_line_prefix(options);
1556         count = options->stat_count ? options->stat_count : data->nr;
1557
1558         reset = diff_get_color_opt(options, DIFF_RESET);
1559         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1560         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1561
1562         /*
1563          * Find the longest filename and max number of changes
1564          */
1565         for (i = 0; (i < count) && (i < data->nr); i++) {
1566                 struct diffstat_file *file = data->files[i];
1567                 uintmax_t change = file->added + file->deleted;
1568
1569                 if (!file->is_interesting && (change == 0)) {
1570                         count++; /* not shown == room for one more */
1571                         continue;
1572                 }
1573                 fill_print_name(file);
1574                 len = strlen(file->print_name);
1575                 if (max_len < len)
1576                         max_len = len;
1577
1578                 if (file->is_unmerged) {
1579                         /* "Unmerged" is 8 characters */
1580                         bin_width = bin_width < 8 ? 8 : bin_width;
1581                         continue;
1582                 }
1583                 if (file->is_binary) {
1584                         /* "Bin XXX -> YYY bytes" */
1585                         int w = 14 + decimal_width(file->added)
1586                                 + decimal_width(file->deleted);
1587                         bin_width = bin_width < w ? w : bin_width;
1588                         /* Display change counts aligned with "Bin" */
1589                         number_width = 3;
1590                         continue;
1591                 }
1592
1593                 if (max_change < change)
1594                         max_change = change;
1595         }
1596         count = i; /* where we can stop scanning in data->files[] */
1597
1598         /*
1599          * We have width = stat_width or term_columns() columns total.
1600          * We want a maximum of min(max_len, stat_name_width) for the name part.
1601          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1602          * We also need 1 for " " and 4 + decimal_width(max_change)
1603          * for " | NNNN " and one the empty column at the end, altogether
1604          * 6 + decimal_width(max_change).
1605          *
1606          * If there's not enough space, we will use the smaller of
1607          * stat_name_width (if set) and 5/8*width for the filename,
1608          * and the rest for constant elements + graph part, but no more
1609          * than stat_graph_width for the graph part.
1610          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1611          * for the standard terminal size).
1612          *
1613          * In other words: stat_width limits the maximum width, and
1614          * stat_name_width fixes the maximum width of the filename,
1615          * and is also used to divide available columns if there
1616          * aren't enough.
1617          *
1618          * Binary files are displayed with "Bin XXX -> YYY bytes"
1619          * instead of the change count and graph. This part is treated
1620          * similarly to the graph part, except that it is not
1621          * "scaled". If total width is too small to accommodate the
1622          * guaranteed minimum width of the filename part and the
1623          * separators and this message, this message will "overflow"
1624          * making the line longer than the maximum width.
1625          */
1626
1627         if (options->stat_width == -1)
1628                 width = term_columns() - options->output_prefix_length;
1629         else
1630                 width = options->stat_width ? options->stat_width : 80;
1631         number_width = decimal_width(max_change) > number_width ?
1632                 decimal_width(max_change) : number_width;
1633
1634         if (options->stat_graph_width == -1)
1635                 options->stat_graph_width = diff_stat_graph_width;
1636
1637         /*
1638          * Guarantee 3/8*16==6 for the graph part
1639          * and 5/8*16==10 for the filename part
1640          */
1641         if (width < 16 + 6 + number_width)
1642                 width = 16 + 6 + number_width;
1643
1644         /*
1645          * First assign sizes that are wanted, ignoring available width.
1646          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1647          * starting from "XXX" should fit in graph_width.
1648          */
1649         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1650         if (options->stat_graph_width &&
1651             options->stat_graph_width < graph_width)
1652                 graph_width = options->stat_graph_width;
1653
1654         name_width = (options->stat_name_width > 0 &&
1655                       options->stat_name_width < max_len) ?
1656                 options->stat_name_width : max_len;
1657
1658         /*
1659          * Adjust adjustable widths not to exceed maximum width
1660          */
1661         if (name_width + number_width + 6 + graph_width > width) {
1662                 if (graph_width > width * 3/8 - number_width - 6) {
1663                         graph_width = width * 3/8 - number_width - 6;
1664                         if (graph_width < 6)
1665                                 graph_width = 6;
1666                 }
1667
1668                 if (options->stat_graph_width &&
1669                     graph_width > options->stat_graph_width)
1670                         graph_width = options->stat_graph_width;
1671                 if (name_width > width - number_width - 6 - graph_width)
1672                         name_width = width - number_width - 6 - graph_width;
1673                 else
1674                         graph_width = width - number_width - 6 - name_width;
1675         }
1676
1677         /*
1678          * From here name_width is the width of the name area,
1679          * and graph_width is the width of the graph area.
1680          * max_change is used to scale graph properly.
1681          */
1682         for (i = 0; i < count; i++) {
1683                 const char *prefix = "";
1684                 struct diffstat_file *file = data->files[i];
1685                 char *name = file->print_name;
1686                 uintmax_t added = file->added;
1687                 uintmax_t deleted = file->deleted;
1688                 int name_len;
1689
1690                 if (!file->is_interesting && (added + deleted == 0))
1691                         continue;
1692
1693                 /*
1694                  * "scale" the filename
1695                  */
1696                 len = name_width;
1697                 name_len = strlen(name);
1698                 if (name_width < name_len) {
1699                         char *slash;
1700                         prefix = "...";
1701                         len -= 3;
1702                         name += name_len - len;
1703                         slash = strchr(name, '/');
1704                         if (slash)
1705                                 name = slash;
1706                 }
1707
1708                 if (file->is_binary) {
1709                         fprintf(options->file, "%s", line_prefix);
1710                         show_name(options->file, prefix, name, len);
1711                         fprintf(options->file, " %*s", number_width, "Bin");
1712                         if (!added && !deleted) {
1713                                 putc('\n', options->file);
1714                                 continue;
1715                         }
1716                         fprintf(options->file, " %s%"PRIuMAX"%s",
1717                                 del_c, deleted, reset);
1718                         fprintf(options->file, " -> ");
1719                         fprintf(options->file, "%s%"PRIuMAX"%s",
1720                                 add_c, added, reset);
1721                         fprintf(options->file, " bytes");
1722                         fprintf(options->file, "\n");
1723                         continue;
1724                 }
1725                 else if (file->is_unmerged) {
1726                         fprintf(options->file, "%s", line_prefix);
1727                         show_name(options->file, prefix, name, len);
1728                         fprintf(options->file, " Unmerged\n");
1729                         continue;
1730                 }
1731
1732                 /*
1733                  * scale the add/delete
1734                  */
1735                 add = added;
1736                 del = deleted;
1737
1738                 if (graph_width <= max_change) {
1739                         int total = scale_linear(add + del, graph_width, max_change);
1740                         if (total < 2 && add && del)
1741                                 /* width >= 2 due to the sanity check */
1742                                 total = 2;
1743                         if (add < del) {
1744                                 add = scale_linear(add, graph_width, max_change);
1745                                 del = total - add;
1746                         } else {
1747                                 del = scale_linear(del, graph_width, max_change);
1748                                 add = total - del;
1749                         }
1750                 }
1751                 fprintf(options->file, "%s", line_prefix);
1752                 show_name(options->file, prefix, name, len);
1753                 fprintf(options->file, " %*"PRIuMAX"%s",
1754                         number_width, added + deleted,
1755                         added + deleted ? " " : "");
1756                 show_graph(options->file, '+', add, add_c, reset);
1757                 show_graph(options->file, '-', del, del_c, reset);
1758                 fprintf(options->file, "\n");
1759         }
1760
1761         for (i = 0; i < data->nr; i++) {
1762                 struct diffstat_file *file = data->files[i];
1763                 uintmax_t added = file->added;
1764                 uintmax_t deleted = file->deleted;
1765
1766                 if (file->is_unmerged ||
1767                     (!file->is_interesting && (added + deleted == 0))) {
1768                         total_files--;
1769                         continue;
1770                 }
1771
1772                 if (!file->is_binary) {
1773                         adds += added;
1774                         dels += deleted;
1775                 }
1776                 if (i < count)
1777                         continue;
1778                 if (!extra_shown)
1779                         fprintf(options->file, "%s ...\n", line_prefix);
1780                 extra_shown = 1;
1781         }
1782         fprintf(options->file, "%s", line_prefix);
1783         print_stat_summary(options->file, total_files, adds, dels);
1784 }
1785
1786 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1787 {
1788         int i, adds = 0, dels = 0, total_files = data->nr;
1789
1790         if (data->nr == 0)
1791                 return;
1792
1793         for (i = 0; i < data->nr; i++) {
1794                 int added = data->files[i]->added;
1795                 int deleted= data->files[i]->deleted;
1796
1797                 if (data->files[i]->is_unmerged ||
1798                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1799                         total_files--;
1800                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1801                         adds += added;
1802                         dels += deleted;
1803                 }
1804         }
1805         fprintf(options->file, "%s", diff_line_prefix(options));
1806         print_stat_summary(options->file, total_files, adds, dels);
1807 }
1808
1809 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1810 {
1811         int i;
1812
1813         if (data->nr == 0)
1814                 return;
1815
1816         for (i = 0; i < data->nr; i++) {
1817                 struct diffstat_file *file = data->files[i];
1818
1819                 fprintf(options->file, "%s", diff_line_prefix(options));
1820
1821                 if (file->is_binary)
1822                         fprintf(options->file, "-\t-\t");
1823                 else
1824                         fprintf(options->file,
1825                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1826                                 file->added, file->deleted);
1827                 if (options->line_termination) {
1828                         fill_print_name(file);
1829                         if (!file->is_renamed)
1830                                 write_name_quoted(file->name, options->file,
1831                                                   options->line_termination);
1832                         else {
1833                                 fputs(file->print_name, options->file);
1834                                 putc(options->line_termination, options->file);
1835                         }
1836                 } else {
1837                         if (file->is_renamed) {
1838                                 putc('\0', options->file);
1839                                 write_name_quoted(file->from_name, options->file, '\0');
1840                         }
1841                         write_name_quoted(file->name, options->file, '\0');
1842                 }
1843         }
1844 }
1845
1846 struct dirstat_file {
1847         const char *name;
1848         unsigned long changed;
1849 };
1850
1851 struct dirstat_dir {
1852         struct dirstat_file *files;
1853         int alloc, nr, permille, cumulative;
1854 };
1855
1856 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1857                 unsigned long changed, const char *base, int baselen)
1858 {
1859         unsigned long this_dir = 0;
1860         unsigned int sources = 0;
1861         const char *line_prefix = diff_line_prefix(opt);
1862
1863         while (dir->nr) {
1864                 struct dirstat_file *f = dir->files;
1865                 int namelen = strlen(f->name);
1866                 unsigned long this;
1867                 char *slash;
1868
1869                 if (namelen < baselen)
1870                         break;
1871                 if (memcmp(f->name, base, baselen))
1872                         break;
1873                 slash = strchr(f->name + baselen, '/');
1874                 if (slash) {
1875                         int newbaselen = slash + 1 - f->name;
1876                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1877                         sources++;
1878                 } else {
1879                         this = f->changed;
1880                         dir->files++;
1881                         dir->nr--;
1882                         sources += 2;
1883                 }
1884                 this_dir += this;
1885         }
1886
1887         /*
1888          * We don't report dirstat's for
1889          *  - the top level
1890          *  - or cases where everything came from a single directory
1891          *    under this directory (sources == 1).
1892          */
1893         if (baselen && sources != 1) {
1894                 if (this_dir) {
1895                         int permille = this_dir * 1000 / changed;
1896                         if (permille >= dir->permille) {
1897                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1898                                         permille / 10, permille % 10, baselen, base);
1899                                 if (!dir->cumulative)
1900                                         return 0;
1901                         }
1902                 }
1903         }
1904         return this_dir;
1905 }
1906
1907 static int dirstat_compare(const void *_a, const void *_b)
1908 {
1909         const struct dirstat_file *a = _a;
1910         const struct dirstat_file *b = _b;
1911         return strcmp(a->name, b->name);
1912 }
1913
1914 static void show_dirstat(struct diff_options *options)
1915 {
1916         int i;
1917         unsigned long changed;
1918         struct dirstat_dir dir;
1919         struct diff_queue_struct *q = &diff_queued_diff;
1920
1921         dir.files = NULL;
1922         dir.alloc = 0;
1923         dir.nr = 0;
1924         dir.permille = options->dirstat_permille;
1925         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1926
1927         changed = 0;
1928         for (i = 0; i < q->nr; i++) {
1929                 struct diff_filepair *p = q->queue[i];
1930                 const char *name;
1931                 unsigned long copied, added, damage;
1932                 int content_changed;
1933
1934                 name = p->two->path ? p->two->path : p->one->path;
1935
1936                 if (p->one->sha1_valid && p->two->sha1_valid)
1937                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1938                 else
1939                         content_changed = 1;
1940
1941                 if (!content_changed) {
1942                         /*
1943                          * The SHA1 has not changed, so pre-/post-content is
1944                          * identical. We can therefore skip looking at the
1945                          * file contents altogether.
1946                          */
1947                         damage = 0;
1948                         goto found_damage;
1949                 }
1950
1951                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1952                         /*
1953                          * In --dirstat-by-file mode, we don't really need to
1954                          * look at the actual file contents at all.
1955                          * The fact that the SHA1 changed is enough for us to
1956                          * add this file to the list of results
1957                          * (with each file contributing equal damage).
1958                          */
1959                         damage = 1;
1960                         goto found_damage;
1961                 }
1962
1963                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1964                         diff_populate_filespec(p->one, 0);
1965                         diff_populate_filespec(p->two, 0);
1966                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1967                                                &copied, &added);
1968                         diff_free_filespec_data(p->one);
1969                         diff_free_filespec_data(p->two);
1970                 } else if (DIFF_FILE_VALID(p->one)) {
1971                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1972                         copied = added = 0;
1973                         diff_free_filespec_data(p->one);
1974                 } else if (DIFF_FILE_VALID(p->two)) {
1975                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1976                         copied = 0;
1977                         added = p->two->size;
1978                         diff_free_filespec_data(p->two);
1979                 } else
1980                         continue;
1981
1982                 /*
1983                  * Original minus copied is the removed material,
1984                  * added is the new material.  They are both damages
1985                  * made to the preimage.
1986                  * If the resulting damage is zero, we know that
1987                  * diffcore_count_changes() considers the two entries to
1988                  * be identical, but since content_changed is true, we
1989                  * know that there must have been _some_ kind of change,
1990                  * so we force all entries to have damage > 0.
1991                  */
1992                 damage = (p->one->size - copied) + added;
1993                 if (!damage)
1994                         damage = 1;
1995
1996 found_damage:
1997                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1998                 dir.files[dir.nr].name = name;
1999                 dir.files[dir.nr].changed = damage;
2000                 changed += damage;
2001                 dir.nr++;
2002         }
2003
2004         /* This can happen even with many files, if everything was renames */
2005         if (!changed)
2006                 return;
2007
2008         /* Show all directories with more than x% of the changes */
2009         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2010         gather_dirstat(options, &dir, changed, "", 0);
2011 }
2012
2013 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2014 {
2015         int i;
2016         unsigned long changed;
2017         struct dirstat_dir dir;
2018
2019         if (data->nr == 0)
2020                 return;
2021
2022         dir.files = NULL;
2023         dir.alloc = 0;
2024         dir.nr = 0;
2025         dir.permille = options->dirstat_permille;
2026         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2027
2028         changed = 0;
2029         for (i = 0; i < data->nr; i++) {
2030                 struct diffstat_file *file = data->files[i];
2031                 unsigned long damage = file->added + file->deleted;
2032                 if (file->is_binary)
2033                         /*
2034                          * binary files counts bytes, not lines. Must find some
2035                          * way to normalize binary bytes vs. textual lines.
2036                          * The following heuristic assumes that there are 64
2037                          * bytes per "line".
2038                          * This is stupid and ugly, but very cheap...
2039                          */
2040                         damage = (damage + 63) / 64;
2041                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2042                 dir.files[dir.nr].name = file->name;
2043                 dir.files[dir.nr].changed = damage;
2044                 changed += damage;
2045                 dir.nr++;
2046         }
2047
2048         /* This can happen even with many files, if everything was renames */
2049         if (!changed)
2050                 return;
2051
2052         /* Show all directories with more than x% of the changes */
2053         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2054         gather_dirstat(options, &dir, changed, "", 0);
2055 }
2056
2057 static void free_diffstat_info(struct diffstat_t *diffstat)
2058 {
2059         int i;
2060         for (i = 0; i < diffstat->nr; i++) {
2061                 struct diffstat_file *f = diffstat->files[i];
2062                 if (f->name != f->print_name)
2063                         free(f->print_name);
2064                 free(f->name);
2065                 free(f->from_name);
2066                 free(f);
2067         }
2068         free(diffstat->files);
2069 }
2070
2071 struct checkdiff_t {
2072         const char *filename;
2073         int lineno;
2074         int conflict_marker_size;
2075         struct diff_options *o;
2076         unsigned ws_rule;
2077         unsigned status;
2078 };
2079
2080 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2081 {
2082         char firstchar;
2083         int cnt;
2084
2085         if (len < marker_size + 1)
2086                 return 0;
2087         firstchar = line[0];
2088         switch (firstchar) {
2089         case '=': case '>': case '<': case '|':
2090                 break;
2091         default:
2092                 return 0;
2093         }
2094         for (cnt = 1; cnt < marker_size; cnt++)
2095                 if (line[cnt] != firstchar)
2096                         return 0;
2097         /* line[1] thru line[marker_size-1] are same as firstchar */
2098         if (len < marker_size + 1 || !isspace(line[marker_size]))
2099                 return 0;
2100         return 1;
2101 }
2102
2103 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2104 {
2105         struct checkdiff_t *data = priv;
2106         int marker_size = data->conflict_marker_size;
2107         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2108         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2109         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2110         char *err;
2111         const char *line_prefix;
2112
2113         assert(data->o);
2114         line_prefix = diff_line_prefix(data->o);
2115
2116         if (line[0] == '+') {
2117                 unsigned bad;
2118                 data->lineno++;
2119                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2120                         data->status |= 1;
2121                         fprintf(data->o->file,
2122                                 "%s%s:%d: leftover conflict marker\n",
2123                                 line_prefix, data->filename, data->lineno);
2124                 }
2125                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2126                 if (!bad)
2127                         return;
2128                 data->status |= bad;
2129                 err = whitespace_error_string(bad);
2130                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2131                         line_prefix, data->filename, data->lineno, err);
2132                 free(err);
2133                 emit_line(data->o, set, reset, line, 1);
2134                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2135                               data->o->file, set, reset, ws);
2136         } else if (line[0] == ' ') {
2137                 data->lineno++;
2138         } else if (line[0] == '@') {
2139                 char *plus = strchr(line, '+');
2140                 if (plus)
2141                         data->lineno = strtol(plus, NULL, 10) - 1;
2142                 else
2143                         die("invalid diff");
2144         }
2145 }
2146
2147 static unsigned char *deflate_it(char *data,
2148                                  unsigned long size,
2149                                  unsigned long *result_size)
2150 {
2151         int bound;
2152         unsigned char *deflated;
2153         git_zstream stream;
2154
2155         git_deflate_init(&stream, zlib_compression_level);
2156         bound = git_deflate_bound(&stream, size);
2157         deflated = xmalloc(bound);
2158         stream.next_out = deflated;
2159         stream.avail_out = bound;
2160
2161         stream.next_in = (unsigned char *)data;
2162         stream.avail_in = size;
2163         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2164                 ; /* nothing */
2165         git_deflate_end(&stream);
2166         *result_size = stream.total_out;
2167         return deflated;
2168 }
2169
2170 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2171                                   const char *prefix)
2172 {
2173         void *cp;
2174         void *delta;
2175         void *deflated;
2176         void *data;
2177         unsigned long orig_size;
2178         unsigned long delta_size;
2179         unsigned long deflate_size;
2180         unsigned long data_size;
2181
2182         /* We could do deflated delta, or we could do just deflated two,
2183          * whichever is smaller.
2184          */
2185         delta = NULL;
2186         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2187         if (one->size && two->size) {
2188                 delta = diff_delta(one->ptr, one->size,
2189                                    two->ptr, two->size,
2190                                    &delta_size, deflate_size);
2191                 if (delta) {
2192                         void *to_free = delta;
2193                         orig_size = delta_size;
2194                         delta = deflate_it(delta, delta_size, &delta_size);
2195                         free(to_free);
2196                 }
2197         }
2198
2199         if (delta && delta_size < deflate_size) {
2200                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2201                 free(deflated);
2202                 data = delta;
2203                 data_size = delta_size;
2204         }
2205         else {
2206                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2207                 free(delta);
2208                 data = deflated;
2209                 data_size = deflate_size;
2210         }
2211
2212         /* emit data encoded in base85 */
2213         cp = data;
2214         while (data_size) {
2215                 int bytes = (52 < data_size) ? 52 : data_size;
2216                 char line[70];
2217                 data_size -= bytes;
2218                 if (bytes <= 26)
2219                         line[0] = bytes + 'A' - 1;
2220                 else
2221                         line[0] = bytes - 26 + 'a' - 1;
2222                 encode_85(line + 1, cp, bytes);
2223                 cp = (char *) cp + bytes;
2224                 fprintf(file, "%s", prefix);
2225                 fputs(line, file);
2226                 fputc('\n', file);
2227         }
2228         fprintf(file, "%s\n", prefix);
2229         free(data);
2230 }
2231
2232 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2233                              const char *prefix)
2234 {
2235         fprintf(file, "%sGIT binary patch\n", prefix);
2236         emit_binary_diff_body(file, one, two, prefix);
2237         emit_binary_diff_body(file, two, one, prefix);
2238 }
2239
2240 int diff_filespec_is_binary(struct diff_filespec *one)
2241 {
2242         if (one->is_binary == -1) {
2243                 diff_filespec_load_driver(one);
2244                 if (one->driver->binary != -1)
2245                         one->is_binary = one->driver->binary;
2246                 else {
2247                         if (!one->data && DIFF_FILE_VALID(one))
2248                                 diff_populate_filespec(one, CHECK_BINARY);
2249                         if (one->is_binary == -1 && one->data)
2250                                 one->is_binary = buffer_is_binary(one->data,
2251                                                 one->size);
2252                         if (one->is_binary == -1)
2253                                 one->is_binary = 0;
2254                 }
2255         }
2256         return one->is_binary;
2257 }
2258
2259 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2260 {
2261         diff_filespec_load_driver(one);
2262         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2263 }
2264
2265 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2266 {
2267         if (!options->a_prefix)
2268                 options->a_prefix = a;
2269         if (!options->b_prefix)
2270                 options->b_prefix = b;
2271 }
2272
2273 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2274 {
2275         if (!DIFF_FILE_VALID(one))
2276                 return NULL;
2277
2278         diff_filespec_load_driver(one);
2279         return userdiff_get_textconv(one->driver);
2280 }
2281
2282 static void builtin_diff(const char *name_a,
2283                          const char *name_b,
2284                          struct diff_filespec *one,
2285                          struct diff_filespec *two,
2286                          const char *xfrm_msg,
2287                          int must_show_header,
2288                          struct diff_options *o,
2289                          int complete_rewrite)
2290 {
2291         mmfile_t mf1, mf2;
2292         const char *lbl[2];
2293         char *a_one, *b_two;
2294         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2295         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2296         const char *a_prefix, *b_prefix;
2297         struct userdiff_driver *textconv_one = NULL;
2298         struct userdiff_driver *textconv_two = NULL;
2299         struct strbuf header = STRBUF_INIT;
2300         const char *line_prefix = diff_line_prefix(o);
2301
2302         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2303                         (!one->mode || S_ISGITLINK(one->mode)) &&
2304                         (!two->mode || S_ISGITLINK(two->mode))) {
2305                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2306                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2307                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2308                                 line_prefix,
2309                                 one->sha1, two->sha1, two->dirty_submodule,
2310                                 meta, del, add, reset);
2311                 return;
2312         }
2313
2314         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2315                 textconv_one = get_textconv(one);
2316                 textconv_two = get_textconv(two);
2317         }
2318
2319         diff_set_mnemonic_prefix(o, "a/", "b/");
2320         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2321                 a_prefix = o->b_prefix;
2322                 b_prefix = o->a_prefix;
2323         } else {
2324                 a_prefix = o->a_prefix;
2325                 b_prefix = o->b_prefix;
2326         }
2327
2328         /* Never use a non-valid filename anywhere if at all possible */
2329         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2330         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2331
2332         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2333         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2334         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2335         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2336         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2337         if (lbl[0][0] == '/') {
2338                 /* /dev/null */
2339                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2340                 if (xfrm_msg)
2341                         strbuf_addstr(&header, xfrm_msg);
2342                 must_show_header = 1;
2343         }
2344         else if (lbl[1][0] == '/') {
2345                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2346                 if (xfrm_msg)
2347                         strbuf_addstr(&header, xfrm_msg);
2348                 must_show_header = 1;
2349         }
2350         else {
2351                 if (one->mode != two->mode) {
2352                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2353                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2354                         must_show_header = 1;
2355                 }
2356                 if (xfrm_msg)
2357                         strbuf_addstr(&header, xfrm_msg);
2358
2359                 /*
2360                  * we do not run diff between different kind
2361                  * of objects.
2362                  */
2363                 if ((one->mode ^ two->mode) & S_IFMT)
2364                         goto free_ab_and_return;
2365                 if (complete_rewrite &&
2366                     (textconv_one || !diff_filespec_is_binary(one)) &&
2367                     (textconv_two || !diff_filespec_is_binary(two))) {
2368                         fprintf(o->file, "%s", header.buf);
2369                         strbuf_reset(&header);
2370                         emit_rewrite_diff(name_a, name_b, one, two,
2371                                                 textconv_one, textconv_two, o);
2372                         o->found_changes = 1;
2373                         goto free_ab_and_return;
2374                 }
2375         }
2376
2377         if (o->irreversible_delete && lbl[1][0] == '/') {
2378                 fprintf(o->file, "%s", header.buf);
2379                 strbuf_reset(&header);
2380                 goto free_ab_and_return;
2381         } else if (!DIFF_OPT_TST(o, TEXT) &&
2382             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2383               (!textconv_two && diff_filespec_is_binary(two)) )) {
2384                 if (!one->data && !two->data &&
2385                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2386                     !DIFF_OPT_TST(o, BINARY)) {
2387                         if (!hashcmp(one->sha1, two->sha1)) {
2388                                 if (must_show_header)
2389                                         fprintf(o->file, "%s", header.buf);
2390                                 goto free_ab_and_return;
2391                         }
2392                         fprintf(o->file, "%s", header.buf);
2393                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2394                                 line_prefix, lbl[0], lbl[1]);
2395                         goto free_ab_and_return;
2396                 }
2397                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2398                         die("unable to read files to diff");
2399                 /* Quite common confusing case */
2400                 if (mf1.size == mf2.size &&
2401                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2402                         if (must_show_header)
2403                                 fprintf(o->file, "%s", header.buf);
2404                         goto free_ab_and_return;
2405                 }
2406                 fprintf(o->file, "%s", header.buf);
2407                 strbuf_reset(&header);
2408                 if (DIFF_OPT_TST(o, BINARY))
2409                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2410                 else
2411                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2412                                 line_prefix, lbl[0], lbl[1]);
2413                 o->found_changes = 1;
2414         } else {
2415                 /* Crazy xdl interfaces.. */
2416                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2417                 const char *v;
2418                 xpparam_t xpp;
2419                 xdemitconf_t xecfg;
2420                 struct emit_callback ecbdata;
2421                 const struct userdiff_funcname *pe;
2422
2423                 if (must_show_header) {
2424                         fprintf(o->file, "%s", header.buf);
2425                         strbuf_reset(&header);
2426                 }
2427
2428                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2429                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2430
2431                 pe = diff_funcname_pattern(one);
2432                 if (!pe)
2433                         pe = diff_funcname_pattern(two);
2434
2435                 memset(&xpp, 0, sizeof(xpp));
2436                 memset(&xecfg, 0, sizeof(xecfg));
2437                 memset(&ecbdata, 0, sizeof(ecbdata));
2438                 ecbdata.label_path = lbl;
2439                 ecbdata.color_diff = want_color(o->use_color);
2440                 ecbdata.found_changesp = &o->found_changes;
2441                 ecbdata.ws_rule = whitespace_rule(name_b);
2442                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2443                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2444                 ecbdata.opt = o;
2445                 ecbdata.header = header.len ? &header : NULL;
2446                 xpp.flags = o->xdl_opts;
2447                 xecfg.ctxlen = o->context;
2448                 xecfg.interhunkctxlen = o->interhunkcontext;
2449                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2450                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2451                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2452                 if (pe)
2453                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2454                 if (!diffopts)
2455                         ;
2456                 else if (skip_prefix(diffopts, "--unified=", &v))
2457                         xecfg.ctxlen = strtoul(v, NULL, 10);
2458                 else if (skip_prefix(diffopts, "-u", &v))
2459                         xecfg.ctxlen = strtoul(v, NULL, 10);
2460                 if (o->word_diff)
2461                         init_diff_words_data(&ecbdata, o, one, two);
2462                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2463                                   &xpp, &xecfg))
2464                         die("unable to generate diff for %s", one->path);
2465                 if (o->word_diff)
2466                         free_diff_words_data(&ecbdata);
2467                 if (textconv_one)
2468                         free(mf1.ptr);
2469                 if (textconv_two)
2470                         free(mf2.ptr);
2471                 xdiff_clear_find_func(&xecfg);
2472         }
2473
2474  free_ab_and_return:
2475         strbuf_release(&header);
2476         diff_free_filespec_data(one);
2477         diff_free_filespec_data(two);
2478         free(a_one);
2479         free(b_two);
2480         return;
2481 }
2482
2483 static void builtin_diffstat(const char *name_a, const char *name_b,
2484                              struct diff_filespec *one,
2485                              struct diff_filespec *two,
2486                              struct diffstat_t *diffstat,
2487                              struct diff_options *o,
2488                              struct diff_filepair *p)
2489 {
2490         mmfile_t mf1, mf2;
2491         struct diffstat_file *data;
2492         int same_contents;
2493         int complete_rewrite = 0;
2494
2495         if (!DIFF_PAIR_UNMERGED(p)) {
2496                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2497                         complete_rewrite = 1;
2498         }
2499
2500         data = diffstat_add(diffstat, name_a, name_b);
2501         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2502
2503         if (!one || !two) {
2504                 data->is_unmerged = 1;
2505                 return;
2506         }
2507
2508         same_contents = !hashcmp(one->sha1, two->sha1);
2509
2510         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2511                 data->is_binary = 1;
2512                 if (same_contents) {
2513                         data->added = 0;
2514                         data->deleted = 0;
2515                 } else {
2516                         data->added = diff_filespec_size(two);
2517                         data->deleted = diff_filespec_size(one);
2518                 }
2519         }
2520
2521         else if (complete_rewrite) {
2522                 diff_populate_filespec(one, 0);
2523                 diff_populate_filespec(two, 0);
2524                 data->deleted = count_lines(one->data, one->size);
2525                 data->added = count_lines(two->data, two->size);
2526         }
2527
2528         else if (!same_contents) {
2529                 /* Crazy xdl interfaces.. */
2530                 xpparam_t xpp;
2531                 xdemitconf_t xecfg;
2532
2533                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2534                         die("unable to read files to diff");
2535
2536                 memset(&xpp, 0, sizeof(xpp));
2537                 memset(&xecfg, 0, sizeof(xecfg));
2538                 xpp.flags = o->xdl_opts;
2539                 xecfg.ctxlen = o->context;
2540                 xecfg.interhunkctxlen = o->interhunkcontext;
2541                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2542                                   &xpp, &xecfg))
2543                         die("unable to generate diffstat for %s", one->path);
2544         }
2545
2546         diff_free_filespec_data(one);
2547         diff_free_filespec_data(two);
2548 }
2549
2550 static void builtin_checkdiff(const char *name_a, const char *name_b,
2551                               const char *attr_path,
2552                               struct diff_filespec *one,
2553                               struct diff_filespec *two,
2554                               struct diff_options *o)
2555 {
2556         mmfile_t mf1, mf2;
2557         struct checkdiff_t data;
2558
2559         if (!two)
2560                 return;
2561
2562         memset(&data, 0, sizeof(data));
2563         data.filename = name_b ? name_b : name_a;
2564         data.lineno = 0;
2565         data.o = o;
2566         data.ws_rule = whitespace_rule(attr_path);
2567         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2568
2569         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2570                 die("unable to read files to diff");
2571
2572         /*
2573          * All the other codepaths check both sides, but not checking
2574          * the "old" side here is deliberate.  We are checking the newly
2575          * introduced changes, and as long as the "new" side is text, we
2576          * can and should check what it introduces.
2577          */
2578         if (diff_filespec_is_binary(two))
2579                 goto free_and_return;
2580         else {
2581                 /* Crazy xdl interfaces.. */
2582                 xpparam_t xpp;
2583                 xdemitconf_t xecfg;
2584
2585                 memset(&xpp, 0, sizeof(xpp));
2586                 memset(&xecfg, 0, sizeof(xecfg));
2587                 xecfg.ctxlen = 1; /* at least one context line */
2588                 xpp.flags = 0;
2589                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2590                                   &xpp, &xecfg))
2591                         die("unable to generate checkdiff for %s", one->path);
2592
2593                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2594                         struct emit_callback ecbdata;
2595                         int blank_at_eof;
2596
2597                         ecbdata.ws_rule = data.ws_rule;
2598                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2599                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2600
2601                         if (blank_at_eof) {
2602                                 static char *err;
2603                                 if (!err)
2604                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2605                                 fprintf(o->file, "%s:%d: %s.\n",
2606                                         data.filename, blank_at_eof, err);
2607                                 data.status = 1; /* report errors */
2608                         }
2609                 }
2610         }
2611  free_and_return:
2612         diff_free_filespec_data(one);
2613         diff_free_filespec_data(two);
2614         if (data.status)
2615                 DIFF_OPT_SET(o, CHECK_FAILED);
2616 }
2617
2618 struct diff_filespec *alloc_filespec(const char *path)
2619 {
2620         struct diff_filespec *spec;
2621
2622         FLEXPTR_ALLOC_STR(spec, path, path);
2623         spec->count = 1;
2624         spec->is_binary = -1;
2625         return spec;
2626 }
2627
2628 void free_filespec(struct diff_filespec *spec)
2629 {
2630         if (!--spec->count) {
2631                 diff_free_filespec_data(spec);
2632                 free(spec);
2633         }
2634 }
2635
2636 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2637                    int sha1_valid, unsigned short mode)
2638 {
2639         if (mode) {
2640                 spec->mode = canon_mode(mode);
2641                 hashcpy(spec->sha1, sha1);
2642                 spec->sha1_valid = sha1_valid;
2643         }
2644 }
2645
2646 /*
2647  * Given a name and sha1 pair, if the index tells us the file in
2648  * the work tree has that object contents, return true, so that
2649  * prepare_temp_file() does not have to inflate and extract.
2650  */
2651 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2652 {
2653         const struct cache_entry *ce;
2654         struct stat st;
2655         int pos, len;
2656
2657         /*
2658          * We do not read the cache ourselves here, because the
2659          * benchmark with my previous version that always reads cache
2660          * shows that it makes things worse for diff-tree comparing
2661          * two linux-2.6 kernel trees in an already checked out work
2662          * tree.  This is because most diff-tree comparisons deal with
2663          * only a small number of files, while reading the cache is
2664          * expensive for a large project, and its cost outweighs the
2665          * savings we get by not inflating the object to a temporary
2666          * file.  Practically, this code only helps when we are used
2667          * by diff-cache --cached, which does read the cache before
2668          * calling us.
2669          */
2670         if (!active_cache)
2671                 return 0;
2672
2673         /* We want to avoid the working directory if our caller
2674          * doesn't need the data in a normal file, this system
2675          * is rather slow with its stat/open/mmap/close syscalls,
2676          * and the object is contained in a pack file.  The pack
2677          * is probably already open and will be faster to obtain
2678          * the data through than the working directory.  Loose
2679          * objects however would tend to be slower as they need
2680          * to be individually opened and inflated.
2681          */
2682         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2683                 return 0;
2684
2685         len = strlen(name);
2686         pos = cache_name_pos(name, len);
2687         if (pos < 0)
2688                 return 0;
2689         ce = active_cache[pos];
2690
2691         /*
2692          * This is not the sha1 we are looking for, or
2693          * unreusable because it is not a regular file.
2694          */
2695         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2696                 return 0;
2697
2698         /*
2699          * If ce is marked as "assume unchanged", there is no
2700          * guarantee that work tree matches what we are looking for.
2701          */
2702         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2703                 return 0;
2704
2705         /*
2706          * If ce matches the file in the work tree, we can reuse it.
2707          */
2708         if (ce_uptodate(ce) ||
2709             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2710                 return 1;
2711
2712         return 0;
2713 }
2714
2715 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2716 {
2717         struct strbuf buf = STRBUF_INIT;
2718         char *dirty = "";
2719
2720         /* Are we looking at the work tree? */
2721         if (s->dirty_submodule)
2722                 dirty = "-dirty";
2723
2724         strbuf_addf(&buf, "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2725         s->size = buf.len;
2726         if (size_only) {
2727                 s->data = NULL;
2728                 strbuf_release(&buf);
2729         } else {
2730                 s->data = strbuf_detach(&buf, NULL);
2731                 s->should_free = 1;
2732         }
2733         return 0;
2734 }
2735
2736 /*
2737  * While doing rename detection and pickaxe operation, we may need to
2738  * grab the data for the blob (or file) for our own in-core comparison.
2739  * diff_filespec has data and size fields for this purpose.
2740  */
2741 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2742 {
2743         int size_only = flags & CHECK_SIZE_ONLY;
2744         int err = 0;
2745         /*
2746          * demote FAIL to WARN to allow inspecting the situation
2747          * instead of refusing.
2748          */
2749         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2750                                     ? SAFE_CRLF_WARN
2751                                     : safe_crlf);
2752
2753         if (!DIFF_FILE_VALID(s))
2754                 die("internal error: asking to populate invalid file.");
2755         if (S_ISDIR(s->mode))
2756                 return -1;
2757
2758         if (s->data)
2759                 return 0;
2760
2761         if (size_only && 0 < s->size)
2762                 return 0;
2763
2764         if (S_ISGITLINK(s->mode))
2765                 return diff_populate_gitlink(s, size_only);
2766
2767         if (!s->sha1_valid ||
2768             reuse_worktree_file(s->path, s->sha1, 0)) {
2769                 struct strbuf buf = STRBUF_INIT;
2770                 struct stat st;
2771                 int fd;
2772
2773                 if (lstat(s->path, &st) < 0) {
2774                         if (errno == ENOENT) {
2775                         err_empty:
2776                                 err = -1;
2777                         empty:
2778                                 s->data = (char *)"";
2779                                 s->size = 0;
2780                                 return err;
2781                         }
2782                 }
2783                 s->size = xsize_t(st.st_size);
2784                 if (!s->size)
2785                         goto empty;
2786                 if (S_ISLNK(st.st_mode)) {
2787                         struct strbuf sb = STRBUF_INIT;
2788
2789                         if (strbuf_readlink(&sb, s->path, s->size))
2790                                 goto err_empty;
2791                         s->size = sb.len;
2792                         s->data = strbuf_detach(&sb, NULL);
2793                         s->should_free = 1;
2794                         return 0;
2795                 }
2796                 if (size_only)
2797                         return 0;
2798                 if ((flags & CHECK_BINARY) &&
2799                     s->size > big_file_threshold && s->is_binary == -1) {
2800                         s->is_binary = 1;
2801                         return 0;
2802                 }
2803                 fd = open(s->path, O_RDONLY);
2804                 if (fd < 0)
2805                         goto err_empty;
2806                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2807                 close(fd);
2808                 s->should_munmap = 1;
2809
2810                 /*
2811                  * Convert from working tree format to canonical git format
2812                  */
2813                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2814                         size_t size = 0;
2815                         munmap(s->data, s->size);
2816                         s->should_munmap = 0;
2817                         s->data = strbuf_detach(&buf, &size);
2818                         s->size = size;
2819                         s->should_free = 1;
2820                 }
2821         }
2822         else {
2823                 enum object_type type;
2824                 if (size_only || (flags & CHECK_BINARY)) {
2825                         type = sha1_object_info(s->sha1, &s->size);
2826                         if (type < 0)
2827                                 die("unable to read %s", sha1_to_hex(s->sha1));
2828                         if (size_only)
2829                                 return 0;
2830                         if (s->size > big_file_threshold && s->is_binary == -1) {
2831                                 s->is_binary = 1;
2832                                 return 0;
2833                         }
2834                 }
2835                 s->data = read_sha1_file(s->sha1, &type, &s->size);
2836                 if (!s->data)
2837                         die("unable to read %s", sha1_to_hex(s->sha1));
2838                 s->should_free = 1;
2839         }
2840         return 0;
2841 }
2842
2843 void diff_free_filespec_blob(struct diff_filespec *s)
2844 {
2845         if (s->should_free)
2846                 free(s->data);
2847         else if (s->should_munmap)
2848                 munmap(s->data, s->size);
2849
2850         if (s->should_free || s->should_munmap) {
2851                 s->should_free = s->should_munmap = 0;
2852                 s->data = NULL;
2853         }
2854 }
2855
2856 void diff_free_filespec_data(struct diff_filespec *s)
2857 {
2858         diff_free_filespec_blob(s);
2859         free(s->cnt_data);
2860         s->cnt_data = NULL;
2861 }
2862
2863 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2864                            void *blob,
2865                            unsigned long size,
2866                            const unsigned char *sha1,
2867                            int mode)
2868 {
2869         int fd;
2870         struct strbuf buf = STRBUF_INIT;
2871         struct strbuf template = STRBUF_INIT;
2872         char *path_dup = xstrdup(path);
2873         const char *base = basename(path_dup);
2874
2875         /* Generate "XXXXXX_basename.ext" */
2876         strbuf_addstr(&template, "XXXXXX_");
2877         strbuf_addstr(&template, base);
2878
2879         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2880         if (fd < 0)
2881                 die_errno("unable to create temp-file");
2882         if (convert_to_working_tree(path,
2883                         (const char *)blob, (size_t)size, &buf)) {
2884                 blob = buf.buf;
2885                 size = buf.len;
2886         }
2887         if (write_in_full(fd, blob, size) != size)
2888                 die_errno("unable to write temp-file");
2889         close_tempfile(&temp->tempfile);
2890         temp->name = get_tempfile_path(&temp->tempfile);
2891         sha1_to_hex_r(temp->hex, sha1);
2892         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
2893         strbuf_release(&buf);
2894         strbuf_release(&template);
2895         free(path_dup);
2896 }
2897
2898 static struct diff_tempfile *prepare_temp_file(const char *name,
2899                 struct diff_filespec *one)
2900 {
2901         struct diff_tempfile *temp = claim_diff_tempfile();
2902
2903         if (!DIFF_FILE_VALID(one)) {
2904         not_a_valid_file:
2905                 /* A '-' entry produces this for file-2, and
2906                  * a '+' entry produces this for file-1.
2907                  */
2908                 temp->name = "/dev/null";
2909                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
2910                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
2911                 return temp;
2912         }
2913
2914         if (!S_ISGITLINK(one->mode) &&
2915             (!one->sha1_valid ||
2916              reuse_worktree_file(name, one->sha1, 1))) {
2917                 struct stat st;
2918                 if (lstat(name, &st) < 0) {
2919                         if (errno == ENOENT)
2920                                 goto not_a_valid_file;
2921                         die_errno("stat(%s)", name);
2922                 }
2923                 if (S_ISLNK(st.st_mode)) {
2924                         struct strbuf sb = STRBUF_INIT;
2925                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2926                                 die_errno("readlink(%s)", name);
2927                         prep_temp_blob(name, temp, sb.buf, sb.len,
2928                                        (one->sha1_valid ?
2929                                         one->sha1 : null_sha1),
2930                                        (one->sha1_valid ?
2931                                         one->mode : S_IFLNK));
2932                         strbuf_release(&sb);
2933                 }
2934                 else {
2935                         /* we can borrow from the file in the work tree */
2936                         temp->name = name;
2937                         if (!one->sha1_valid)
2938                                 sha1_to_hex_r(temp->hex, null_sha1);
2939                         else
2940                                 sha1_to_hex_r(temp->hex, one->sha1);
2941                         /* Even though we may sometimes borrow the
2942                          * contents from the work tree, we always want
2943                          * one->mode.  mode is trustworthy even when
2944                          * !(one->sha1_valid), as long as
2945                          * DIFF_FILE_VALID(one).
2946                          */
2947                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
2948                 }
2949                 return temp;
2950         }
2951         else {
2952                 if (diff_populate_filespec(one, 0))
2953                         die("cannot read data blob for %s", one->path);
2954                 prep_temp_blob(name, temp, one->data, one->size,
2955                                one->sha1, one->mode);
2956         }
2957         return temp;
2958 }
2959
2960 static void add_external_diff_name(struct argv_array *argv,
2961                                    const char *name,
2962                                    struct diff_filespec *df)
2963 {
2964         struct diff_tempfile *temp = prepare_temp_file(name, df);
2965         argv_array_push(argv, temp->name);
2966         argv_array_push(argv, temp->hex);
2967         argv_array_push(argv, temp->mode);
2968 }
2969
2970 /* An external diff command takes:
2971  *
2972  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2973  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2974  *
2975  */
2976 static void run_external_diff(const char *pgm,
2977                               const char *name,
2978                               const char *other,
2979                               struct diff_filespec *one,
2980                               struct diff_filespec *two,
2981                               const char *xfrm_msg,
2982                               int complete_rewrite,
2983                               struct diff_options *o)
2984 {
2985         struct argv_array argv = ARGV_ARRAY_INIT;
2986         struct argv_array env = ARGV_ARRAY_INIT;
2987         struct diff_queue_struct *q = &diff_queued_diff;
2988
2989         argv_array_push(&argv, pgm);
2990         argv_array_push(&argv, name);
2991
2992         if (one && two) {
2993                 add_external_diff_name(&argv, name, one);
2994                 if (!other)
2995                         add_external_diff_name(&argv, name, two);
2996                 else {
2997                         add_external_diff_name(&argv, other, two);
2998                         argv_array_push(&argv, other);
2999                         argv_array_push(&argv, xfrm_msg);
3000                 }
3001         }
3002
3003         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3004         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3005
3006         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3007                 die(_("external diff died, stopping at %s"), name);
3008
3009         remove_tempfile();
3010         argv_array_clear(&argv);
3011         argv_array_clear(&env);
3012 }
3013
3014 static int similarity_index(struct diff_filepair *p)
3015 {
3016         return p->score * 100 / MAX_SCORE;
3017 }
3018
3019 static void fill_metainfo(struct strbuf *msg,
3020                           const char *name,
3021                           const char *other,
3022                           struct diff_filespec *one,
3023                           struct diff_filespec *two,
3024                           struct diff_options *o,
3025                           struct diff_filepair *p,
3026                           int *must_show_header,
3027                           int use_color)
3028 {
3029         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3030         const char *reset = diff_get_color(use_color, DIFF_RESET);
3031         const char *line_prefix = diff_line_prefix(o);
3032
3033         *must_show_header = 1;
3034         strbuf_init(msg, PATH_MAX * 2 + 300);
3035         switch (p->status) {
3036         case DIFF_STATUS_COPIED:
3037                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3038                             line_prefix, set, similarity_index(p));
3039                 strbuf_addf(msg, "%s\n%s%scopy from ",
3040                             reset,  line_prefix, set);
3041                 quote_c_style(name, msg, NULL, 0);
3042                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3043                 quote_c_style(other, msg, NULL, 0);
3044                 strbuf_addf(msg, "%s\n", reset);
3045                 break;
3046         case DIFF_STATUS_RENAMED:
3047                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3048                             line_prefix, set, similarity_index(p));
3049                 strbuf_addf(msg, "%s\n%s%srename from ",
3050                             reset, line_prefix, set);
3051                 quote_c_style(name, msg, NULL, 0);
3052                 strbuf_addf(msg, "%s\n%s%srename to ",
3053                             reset, line_prefix, set);
3054                 quote_c_style(other, msg, NULL, 0);
3055                 strbuf_addf(msg, "%s\n", reset);
3056                 break;
3057         case DIFF_STATUS_MODIFIED:
3058                 if (p->score) {
3059                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3060                                     line_prefix,
3061                                     set, similarity_index(p), reset);
3062                         break;
3063                 }
3064                 /* fallthru */
3065         default:
3066                 *must_show_header = 0;
3067         }
3068         if (one && two && hashcmp(one->sha1, two->sha1)) {
3069                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3070
3071                 if (DIFF_OPT_TST(o, BINARY)) {
3072                         mmfile_t mf;
3073                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3074                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3075                                 abbrev = 40;
3076                 }
3077                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3078                             find_unique_abbrev(one->sha1, abbrev));
3079                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3080                 if (one->mode == two->mode)
3081                         strbuf_addf(msg, " %06o", one->mode);
3082                 strbuf_addf(msg, "%s\n", reset);
3083         }
3084 }
3085
3086 static void run_diff_cmd(const char *pgm,
3087                          const char *name,
3088                          const char *other,
3089                          const char *attr_path,
3090                          struct diff_filespec *one,
3091                          struct diff_filespec *two,
3092                          struct strbuf *msg,
3093                          struct diff_options *o,
3094                          struct diff_filepair *p)
3095 {
3096         const char *xfrm_msg = NULL;
3097         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3098         int must_show_header = 0;
3099
3100
3101         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3102                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3103                 if (drv && drv->external)
3104                         pgm = drv->external;
3105         }
3106
3107         if (msg) {
3108                 /*
3109                  * don't use colors when the header is intended for an
3110                  * external diff driver
3111                  */
3112                 fill_metainfo(msg, name, other, one, two, o, p,
3113                               &must_show_header,
3114                               want_color(o->use_color) && !pgm);
3115                 xfrm_msg = msg->len ? msg->buf : NULL;
3116         }
3117
3118         if (pgm) {
3119                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3120                                   complete_rewrite, o);
3121                 return;
3122         }
3123         if (one && two)
3124                 builtin_diff(name, other ? other : name,
3125                              one, two, xfrm_msg, must_show_header,
3126                              o, complete_rewrite);
3127         else
3128                 fprintf(o->file, "* Unmerged path %s\n", name);
3129 }
3130
3131 static void diff_fill_sha1_info(struct diff_filespec *one)
3132 {
3133         if (DIFF_FILE_VALID(one)) {
3134                 if (!one->sha1_valid) {
3135                         struct stat st;
3136                         if (one->is_stdin) {
3137                                 hashcpy(one->sha1, null_sha1);
3138                                 return;
3139                         }
3140                         if (lstat(one->path, &st) < 0)
3141                                 die_errno("stat '%s'", one->path);
3142                         if (index_path(one->sha1, one->path, &st, 0))
3143                                 die("cannot hash %s", one->path);
3144                 }
3145         }
3146         else
3147                 hashclr(one->sha1);
3148 }
3149
3150 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3151 {
3152         /* Strip the prefix but do not molest /dev/null and absolute paths */
3153         if (*namep && **namep != '/') {
3154                 *namep += prefix_length;
3155                 if (**namep == '/')
3156                         ++*namep;
3157         }
3158         if (*otherp && **otherp != '/') {
3159                 *otherp += prefix_length;
3160                 if (**otherp == '/')
3161                         ++*otherp;
3162         }
3163 }
3164
3165 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3166 {
3167         const char *pgm = external_diff();
3168         struct strbuf msg;
3169         struct diff_filespec *one = p->one;
3170         struct diff_filespec *two = p->two;
3171         const char *name;
3172         const char *other;
3173         const char *attr_path;
3174
3175         name  = p->one->path;
3176         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3177         attr_path = name;
3178         if (o->prefix_length)
3179                 strip_prefix(o->prefix_length, &name, &other);
3180
3181         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3182                 pgm = NULL;
3183
3184         if (DIFF_PAIR_UNMERGED(p)) {
3185                 run_diff_cmd(pgm, name, NULL, attr_path,
3186                              NULL, NULL, NULL, o, p);
3187                 return;
3188         }
3189
3190         diff_fill_sha1_info(one);
3191         diff_fill_sha1_info(two);
3192
3193         if (!pgm &&
3194             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3195             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3196                 /*
3197                  * a filepair that changes between file and symlink
3198                  * needs to be split into deletion and creation.
3199                  */
3200                 struct diff_filespec *null = alloc_filespec(two->path);
3201                 run_diff_cmd(NULL, name, other, attr_path,
3202                              one, null, &msg, o, p);
3203                 free(null);
3204                 strbuf_release(&msg);
3205
3206                 null = alloc_filespec(one->path);
3207                 run_diff_cmd(NULL, name, other, attr_path,
3208                              null, two, &msg, o, p);
3209                 free(null);
3210         }
3211         else
3212                 run_diff_cmd(pgm, name, other, attr_path,
3213                              one, two, &msg, o, p);
3214
3215         strbuf_release(&msg);
3216 }
3217
3218 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3219                          struct diffstat_t *diffstat)
3220 {
3221         const char *name;
3222         const char *other;
3223
3224         if (DIFF_PAIR_UNMERGED(p)) {
3225                 /* unmerged */
3226                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3227                 return;
3228         }
3229
3230         name = p->one->path;
3231         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3232
3233         if (o->prefix_length)
3234                 strip_prefix(o->prefix_length, &name, &other);
3235
3236         diff_fill_sha1_info(p->one);
3237         diff_fill_sha1_info(p->two);
3238
3239         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3240 }
3241
3242 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3243 {
3244         const char *name;
3245         const char *other;
3246         const char *attr_path;
3247
3248         if (DIFF_PAIR_UNMERGED(p)) {
3249                 /* unmerged */
3250                 return;
3251         }
3252
3253         name = p->one->path;
3254         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3255         attr_path = other ? other : name;
3256
3257         if (o->prefix_length)
3258                 strip_prefix(o->prefix_length, &name, &other);
3259
3260         diff_fill_sha1_info(p->one);
3261         diff_fill_sha1_info(p->two);
3262
3263         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3264 }
3265
3266 void diff_setup(struct diff_options *options)
3267 {
3268         memcpy(options, &default_diff_options, sizeof(*options));
3269
3270         options->file = stdout;
3271
3272         options->line_termination = '\n';
3273         options->break_opt = -1;
3274         options->rename_limit = -1;
3275         options->dirstat_permille = diff_dirstat_permille_default;
3276         options->context = diff_context_default;
3277         options->ws_error_highlight = WSEH_NEW;
3278         DIFF_OPT_SET(options, RENAME_EMPTY);
3279
3280         /* pathchange left =NULL by default */
3281         options->change = diff_change;
3282         options->add_remove = diff_addremove;
3283         options->use_color = diff_use_color_default;
3284         options->detect_rename = diff_detect_rename_default;
3285         options->xdl_opts |= diff_algorithm;
3286         if (diff_compaction_heuristic)
3287                 DIFF_XDL_SET(options, COMPACTION_HEURISTIC);
3288
3289         options->orderfile = diff_order_file_cfg;
3290
3291         if (diff_no_prefix) {
3292                 options->a_prefix = options->b_prefix = "";
3293         } else if (!diff_mnemonic_prefix) {
3294                 options->a_prefix = "a/";
3295                 options->b_prefix = "b/";
3296         }
3297 }
3298
3299 void diff_setup_done(struct diff_options *options)
3300 {
3301         int count = 0;
3302
3303         if (options->set_default)
3304                 options->set_default(options);
3305
3306         if (options->output_format & DIFF_FORMAT_NAME)
3307                 count++;
3308         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3309                 count++;
3310         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3311                 count++;
3312         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3313                 count++;
3314         if (count > 1)
3315                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3316
3317         /*
3318          * Most of the time we can say "there are changes"
3319          * only by checking if there are changed paths, but
3320          * --ignore-whitespace* options force us to look
3321          * inside contents.
3322          */
3323
3324         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3325             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3326             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3327                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3328         else
3329                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3330
3331         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3332                 options->detect_rename = DIFF_DETECT_COPY;
3333
3334         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3335                 options->prefix = NULL;
3336         if (options->prefix)
3337                 options->prefix_length = strlen(options->prefix);
3338         else
3339                 options->prefix_length = 0;
3340
3341         if (options->output_format & (DIFF_FORMAT_NAME |
3342                                       DIFF_FORMAT_NAME_STATUS |
3343                                       DIFF_FORMAT_CHECKDIFF |
3344                                       DIFF_FORMAT_NO_OUTPUT))
3345                 options->output_format &= ~(DIFF_FORMAT_RAW |
3346                                             DIFF_FORMAT_NUMSTAT |
3347                                             DIFF_FORMAT_DIFFSTAT |
3348                                             DIFF_FORMAT_SHORTSTAT |
3349                                             DIFF_FORMAT_DIRSTAT |
3350                                             DIFF_FORMAT_SUMMARY |
3351                                             DIFF_FORMAT_PATCH);
3352
3353         /*
3354          * These cases always need recursive; we do not drop caller-supplied
3355          * recursive bits for other formats here.
3356          */
3357         if (options->output_format & (DIFF_FORMAT_PATCH |
3358                                       DIFF_FORMAT_NUMSTAT |
3359                                       DIFF_FORMAT_DIFFSTAT |
3360                                       DIFF_FORMAT_SHORTSTAT |
3361                                       DIFF_FORMAT_DIRSTAT |
3362                                       DIFF_FORMAT_SUMMARY |
3363                                       DIFF_FORMAT_CHECKDIFF))
3364                 DIFF_OPT_SET(options, RECURSIVE);
3365         /*
3366          * Also pickaxe would not work very well if you do not say recursive
3367          */
3368         if (options->pickaxe)
3369                 DIFF_OPT_SET(options, RECURSIVE);
3370         /*
3371          * When patches are generated, submodules diffed against the work tree
3372          * must be checked for dirtiness too so it can be shown in the output
3373          */
3374         if (options->output_format & DIFF_FORMAT_PATCH)
3375                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3376
3377         if (options->detect_rename && options->rename_limit < 0)
3378                 options->rename_limit = diff_rename_limit_default;
3379         if (options->setup & DIFF_SETUP_USE_CACHE) {
3380                 if (!active_cache)
3381                         /* read-cache does not die even when it fails
3382                          * so it is safe for us to do this here.  Also
3383                          * it does not smudge active_cache or active_nr
3384                          * when it fails, so we do not have to worry about
3385                          * cleaning it up ourselves either.
3386                          */
3387                         read_cache();
3388         }
3389         if (options->abbrev <= 0 || 40 < options->abbrev)
3390                 options->abbrev = 40; /* full */
3391
3392         /*
3393          * It does not make sense to show the first hit we happened
3394          * to have found.  It does not make sense not to return with
3395          * exit code in such a case either.
3396          */
3397         if (DIFF_OPT_TST(options, QUICK)) {
3398                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3399                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3400         }
3401
3402         options->diff_path_counter = 0;
3403
3404         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3405                 die(_("--follow requires exactly one pathspec"));
3406 }
3407
3408 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3409 {
3410         char c, *eq;
3411         int len;
3412
3413         if (*arg != '-')
3414                 return 0;
3415         c = *++arg;
3416         if (!c)
3417                 return 0;
3418         if (c == arg_short) {
3419                 c = *++arg;
3420                 if (!c)
3421                         return 1;
3422                 if (val && isdigit(c)) {
3423                         char *end;
3424                         int n = strtoul(arg, &end, 10);
3425                         if (*end)
3426                                 return 0;
3427                         *val = n;
3428                         return 1;
3429                 }
3430                 return 0;
3431         }
3432         if (c != '-')
3433                 return 0;
3434         arg++;
3435         eq = strchrnul(arg, '=');
3436         len = eq - arg;
3437         if (!len || strncmp(arg, arg_long, len))
3438                 return 0;
3439         if (*eq) {
3440                 int n;
3441                 char *end;
3442                 if (!isdigit(*++eq))
3443                         return 0;
3444                 n = strtoul(eq, &end, 10);
3445                 if (*end)
3446                         return 0;
3447                 *val = n;
3448         }
3449         return 1;
3450 }
3451
3452 static int diff_scoreopt_parse(const char *opt);
3453
3454 static inline int short_opt(char opt, const char **argv,
3455                             const char **optarg)
3456 {
3457         const char *arg = argv[0];
3458         if (arg[0] != '-' || arg[1] != opt)
3459                 return 0;
3460         if (arg[2] != '\0') {
3461                 *optarg = arg + 2;
3462                 return 1;
3463         }
3464         if (!argv[1])
3465                 die("Option '%c' requires a value", opt);
3466         *optarg = argv[1];
3467         return 2;
3468 }
3469
3470 int parse_long_opt(const char *opt, const char **argv,
3471                    const char **optarg)
3472 {
3473         const char *arg = argv[0];
3474         if (!skip_prefix(arg, "--", &arg))
3475                 return 0;
3476         if (!skip_prefix(arg, opt, &arg))
3477                 return 0;
3478         if (*arg == '=') { /* stuck form: --option=value */
3479                 *optarg = arg + 1;
3480                 return 1;
3481         }
3482         if (*arg != '\0')
3483                 return 0;
3484         /* separate form: --option value */
3485         if (!argv[1])
3486                 die("Option '--%s' requires a value", opt);
3487         *optarg = argv[1];
3488         return 2;
3489 }
3490
3491 static int stat_opt(struct diff_options *options, const char **av)
3492 {
3493         const char *arg = av[0];
3494         char *end;
3495         int width = options->stat_width;
3496         int name_width = options->stat_name_width;
3497         int graph_width = options->stat_graph_width;
3498         int count = options->stat_count;
3499         int argcount = 1;
3500
3501         if (!skip_prefix(arg, "--stat", &arg))
3502                 die("BUG: stat option does not begin with --stat: %s", arg);
3503         end = (char *)arg;
3504
3505         switch (*arg) {
3506         case '-':
3507                 if (skip_prefix(arg, "-width", &arg)) {
3508                         if (*arg == '=')
3509                                 width = strtoul(arg + 1, &end, 10);
3510                         else if (!*arg && !av[1])
3511                                 die("Option '--stat-width' requires a value");
3512                         else if (!*arg) {
3513                                 width = strtoul(av[1], &end, 10);
3514                                 argcount = 2;
3515                         }
3516                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3517                         if (*arg == '=')
3518                                 name_width = strtoul(arg + 1, &end, 10);
3519                         else if (!*arg && !av[1])
3520                                 die("Option '--stat-name-width' requires a value");
3521                         else if (!*arg) {
3522                                 name_width = strtoul(av[1], &end, 10);
3523                                 argcount = 2;
3524                         }
3525                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3526                         if (*arg == '=')
3527                                 graph_width = strtoul(arg + 1, &end, 10);
3528                         else if (!*arg && !av[1])
3529                                 die("Option '--stat-graph-width' requires a value");
3530                         else if (!*arg) {
3531                                 graph_width = strtoul(av[1], &end, 10);
3532                                 argcount = 2;
3533                         }
3534                 } else if (skip_prefix(arg, "-count", &arg)) {
3535                         if (*arg == '=')
3536                                 count = strtoul(arg + 1, &end, 10);
3537                         else if (!*arg && !av[1])
3538                                 die("Option '--stat-count' requires a value");
3539                         else if (!*arg) {
3540                                 count = strtoul(av[1], &end, 10);
3541                                 argcount = 2;
3542                         }
3543                 }
3544                 break;
3545         case '=':
3546                 width = strtoul(arg+1, &end, 10);
3547                 if (*end == ',')
3548                         name_width = strtoul(end+1, &end, 10);
3549                 if (*end == ',')
3550                         count = strtoul(end+1, &end, 10);
3551         }
3552
3553         /* Important! This checks all the error cases! */
3554         if (*end)
3555                 return 0;
3556         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3557         options->stat_name_width = name_width;
3558         options->stat_graph_width = graph_width;
3559         options->stat_width = width;
3560         options->stat_count = count;
3561         return argcount;
3562 }
3563
3564 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3565 {
3566         struct strbuf errmsg = STRBUF_INIT;
3567         if (parse_dirstat_params(options, params, &errmsg))
3568                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3569                     errmsg.buf);
3570         strbuf_release(&errmsg);
3571         /*
3572          * The caller knows a dirstat-related option is given from the command
3573          * line; allow it to say "return this_function();"
3574          */
3575         options->output_format |= DIFF_FORMAT_DIRSTAT;
3576         return 1;
3577 }
3578
3579 static int parse_submodule_opt(struct diff_options *options, const char *value)
3580 {
3581         if (parse_submodule_params(options, value))
3582                 die(_("Failed to parse --submodule option parameter: '%s'"),
3583                         value);
3584         return 1;
3585 }
3586
3587 static const char diff_status_letters[] = {
3588         DIFF_STATUS_ADDED,
3589         DIFF_STATUS_COPIED,
3590         DIFF_STATUS_DELETED,
3591         DIFF_STATUS_MODIFIED,
3592         DIFF_STATUS_RENAMED,
3593         DIFF_STATUS_TYPE_CHANGED,
3594         DIFF_STATUS_UNKNOWN,
3595         DIFF_STATUS_UNMERGED,
3596         DIFF_STATUS_FILTER_AON,
3597         DIFF_STATUS_FILTER_BROKEN,
3598         '\0',
3599 };
3600
3601 static unsigned int filter_bit['Z' + 1];
3602
3603 static void prepare_filter_bits(void)
3604 {
3605         int i;
3606
3607         if (!filter_bit[DIFF_STATUS_ADDED]) {
3608                 for (i = 0; diff_status_letters[i]; i++)
3609                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3610         }
3611 }
3612
3613 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3614 {
3615         return opt->filter & filter_bit[(int) status];
3616 }
3617
3618 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3619 {
3620         int i, optch;
3621
3622         prepare_filter_bits();
3623
3624         /*
3625          * If there is a negation e.g. 'd' in the input, and we haven't
3626          * initialized the filter field with another --diff-filter, start
3627          * from full set of bits, except for AON.
3628          */
3629         if (!opt->filter) {
3630                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3631                         if (optch < 'a' || 'z' < optch)
3632                                 continue;
3633                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3634                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3635                         break;
3636                 }
3637         }
3638
3639         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3640                 unsigned int bit;
3641                 int negate;
3642
3643                 if ('a' <= optch && optch <= 'z') {
3644                         negate = 1;
3645                         optch = toupper(optch);
3646                 } else {
3647                         negate = 0;
3648                 }
3649
3650                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3651                 if (!bit)
3652                         return optarg[i];
3653                 if (negate)
3654                         opt->filter &= ~bit;
3655                 else
3656                         opt->filter |= bit;
3657         }
3658         return 0;
3659 }
3660
3661 static void enable_patch_output(int *fmt) {
3662         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3663         *fmt |= DIFF_FORMAT_PATCH;
3664 }
3665
3666 static int parse_one_token(const char **arg, const char *token)
3667 {
3668         const char *rest;
3669         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
3670                 *arg = rest;
3671                 return 1;
3672         }
3673         return 0;
3674 }
3675
3676 static int parse_ws_error_highlight(struct diff_options *opt, const char *arg)
3677 {
3678         const char *orig_arg = arg;
3679         unsigned val = 0;
3680         while (*arg) {
3681                 if (parse_one_token(&arg, "none"))
3682                         val = 0;
3683                 else if (parse_one_token(&arg, "default"))
3684                         val = WSEH_NEW;
3685                 else if (parse_one_token(&arg, "all"))
3686                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
3687                 else if (parse_one_token(&arg, "new"))
3688                         val |= WSEH_NEW;
3689                 else if (parse_one_token(&arg, "old"))
3690                         val |= WSEH_OLD;
3691                 else if (parse_one_token(&arg, "context"))
3692                         val |= WSEH_CONTEXT;
3693                 else {
3694                         error("unknown value after ws-error-highlight=%.*s",
3695                               (int)(arg - orig_arg), orig_arg);
3696                         return 0;
3697                 }
3698                 if (*arg)
3699                         arg++;
3700         }
3701         opt->ws_error_highlight = val;
3702         return 1;
3703 }
3704
3705 int diff_opt_parse(struct diff_options *options,
3706                    const char **av, int ac, const char *prefix)
3707 {
3708         const char *arg = av[0];
3709         const char *optarg;
3710         int argcount;
3711
3712         if (!prefix)
3713                 prefix = "";
3714
3715         /* Output format options */
3716         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3717             || opt_arg(arg, 'U', "unified", &options->context))
3718                 enable_patch_output(&options->output_format);
3719         else if (!strcmp(arg, "--raw"))
3720                 options->output_format |= DIFF_FORMAT_RAW;
3721         else if (!strcmp(arg, "--patch-with-raw")) {
3722                 enable_patch_output(&options->output_format);
3723                 options->output_format |= DIFF_FORMAT_RAW;
3724         } else if (!strcmp(arg, "--numstat"))
3725                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3726         else if (!strcmp(arg, "--shortstat"))
3727                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3728         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3729                 return parse_dirstat_opt(options, "");
3730         else if (skip_prefix(arg, "-X", &arg))
3731                 return parse_dirstat_opt(options, arg);
3732         else if (skip_prefix(arg, "--dirstat=", &arg))
3733                 return parse_dirstat_opt(options, arg);
3734         else if (!strcmp(arg, "--cumulative"))
3735                 return parse_dirstat_opt(options, "cumulative");
3736         else if (!strcmp(arg, "--dirstat-by-file"))
3737                 return parse_dirstat_opt(options, "files");
3738         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3739                 parse_dirstat_opt(options, "files");
3740                 return parse_dirstat_opt(options, arg);
3741         }
3742         else if (!strcmp(arg, "--check"))
3743                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3744         else if (!strcmp(arg, "--summary"))
3745                 options->output_format |= DIFF_FORMAT_SUMMARY;
3746         else if (!strcmp(arg, "--patch-with-stat")) {
3747                 enable_patch_output(&options->output_format);
3748                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3749         } else if (!strcmp(arg, "--name-only"))
3750                 options->output_format |= DIFF_FORMAT_NAME;
3751         else if (!strcmp(arg, "--name-status"))
3752                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3753         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3754                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3755         else if (starts_with(arg, "--stat"))
3756                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3757                 return stat_opt(options, av);
3758
3759         /* renames options */
3760         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3761                  !strcmp(arg, "--break-rewrites")) {
3762                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3763                         return error("invalid argument to -B: %s", arg+2);
3764         }
3765         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3766                  !strcmp(arg, "--find-renames")) {
3767                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3768                         return error("invalid argument to -M: %s", arg+2);
3769                 options->detect_rename = DIFF_DETECT_RENAME;
3770         }
3771         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3772                 options->irreversible_delete = 1;
3773         }
3774         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3775                  !strcmp(arg, "--find-copies")) {
3776                 if (options->detect_rename == DIFF_DETECT_COPY)
3777                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3778                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3779                         return error("invalid argument to -C: %s", arg+2);
3780                 options->detect_rename = DIFF_DETECT_COPY;
3781         }
3782         else if (!strcmp(arg, "--no-renames"))
3783                 options->detect_rename = 0;
3784         else if (!strcmp(arg, "--rename-empty"))
3785                 DIFF_OPT_SET(options, RENAME_EMPTY);
3786         else if (!strcmp(arg, "--no-rename-empty"))
3787                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3788         else if (!strcmp(arg, "--relative"))
3789                 DIFF_OPT_SET(options, RELATIVE_NAME);
3790         else if (skip_prefix(arg, "--relative=", &arg)) {
3791                 DIFF_OPT_SET(options, RELATIVE_NAME);
3792                 options->prefix = arg;
3793         }
3794
3795         /* xdiff options */
3796         else if (!strcmp(arg, "--minimal"))
3797                 DIFF_XDL_SET(options, NEED_MINIMAL);
3798         else if (!strcmp(arg, "--no-minimal"))
3799                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3800         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3801                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3802         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3803                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3804         else if (!strcmp(arg, "--ignore-space-at-eol"))
3805                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3806         else if (!strcmp(arg, "--ignore-blank-lines"))
3807                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3808         else if (!strcmp(arg, "--compaction-heuristic"))
3809                 DIFF_XDL_SET(options, COMPACTION_HEURISTIC);
3810         else if (!strcmp(arg, "--no-compaction-heuristic"))
3811                 DIFF_XDL_CLR(options, COMPACTION_HEURISTIC);
3812         else if (!strcmp(arg, "--patience"))
3813                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3814         else if (!strcmp(arg, "--histogram"))
3815                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3816         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3817                 long value = parse_algorithm_value(optarg);
3818                 if (value < 0)
3819                         return error("option diff-algorithm accepts \"myers\", "
3820                                      "\"minimal\", \"patience\" and \"histogram\"");
3821                 /* clear out previous settings */
3822                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3823                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3824                 options->xdl_opts |= value;
3825                 return argcount;
3826         }
3827
3828         /* flags options */
3829         else if (!strcmp(arg, "--binary")) {
3830                 enable_patch_output(&options->output_format);
3831                 DIFF_OPT_SET(options, BINARY);
3832         }
3833         else if (!strcmp(arg, "--full-index"))
3834                 DIFF_OPT_SET(options, FULL_INDEX);
3835         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3836                 DIFF_OPT_SET(options, TEXT);
3837         else if (!strcmp(arg, "-R"))
3838                 DIFF_OPT_SET(options, REVERSE_DIFF);
3839         else if (!strcmp(arg, "--find-copies-harder"))
3840                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3841         else if (!strcmp(arg, "--follow"))
3842                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3843         else if (!strcmp(arg, "--no-follow")) {
3844                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3845                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3846         } else if (!strcmp(arg, "--color"))
3847                 options->use_color = 1;
3848         else if (skip_prefix(arg, "--color=", &arg)) {
3849                 int value = git_config_colorbool(NULL, arg);
3850                 if (value < 0)
3851                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3852                 options->use_color = value;
3853         }
3854         else if (!strcmp(arg, "--no-color"))
3855                 options->use_color = 0;
3856         else if (!strcmp(arg, "--color-words")) {
3857                 options->use_color = 1;
3858                 options->word_diff = DIFF_WORDS_COLOR;
3859         }
3860         else if (skip_prefix(arg, "--color-words=", &arg)) {
3861                 options->use_color = 1;
3862                 options->word_diff = DIFF_WORDS_COLOR;
3863                 options->word_regex = arg;
3864         }
3865         else if (!strcmp(arg, "--word-diff")) {
3866                 if (options->word_diff == DIFF_WORDS_NONE)
3867                         options->word_diff = DIFF_WORDS_PLAIN;
3868         }
3869         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3870                 if (!strcmp(arg, "plain"))
3871                         options->word_diff = DIFF_WORDS_PLAIN;
3872                 else if (!strcmp(arg, "color")) {
3873                         options->use_color = 1;
3874                         options->word_diff = DIFF_WORDS_COLOR;
3875                 }
3876                 else if (!strcmp(arg, "porcelain"))
3877                         options->word_diff = DIFF_WORDS_PORCELAIN;
3878                 else if (!strcmp(arg, "none"))
3879                         options->word_diff = DIFF_WORDS_NONE;
3880                 else
3881                         die("bad --word-diff argument: %s", arg);
3882         }
3883         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3884                 if (options->word_diff == DIFF_WORDS_NONE)
3885                         options->word_diff = DIFF_WORDS_PLAIN;
3886                 options->word_regex = optarg;
3887                 return argcount;
3888         }
3889         else if (!strcmp(arg, "--exit-code"))
3890                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3891         else if (!strcmp(arg, "--quiet"))
3892                 DIFF_OPT_SET(options, QUICK);
3893         else if (!strcmp(arg, "--ext-diff"))
3894                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3895         else if (!strcmp(arg, "--no-ext-diff"))
3896                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3897         else if (!strcmp(arg, "--textconv"))
3898                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3899         else if (!strcmp(arg, "--no-textconv"))
3900                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3901         else if (!strcmp(arg, "--ignore-submodules")) {
3902                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3903                 handle_ignore_submodules_arg(options, "all");
3904         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3905                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3906                 handle_ignore_submodules_arg(options, arg);
3907         } else if (!strcmp(arg, "--submodule"))
3908                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3909         else if (skip_prefix(arg, "--submodule=", &arg))
3910                 return parse_submodule_opt(options, arg);
3911         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3912                 return parse_ws_error_highlight(options, arg);
3913
3914         /* misc options */
3915         else if (!strcmp(arg, "-z"))
3916                 options->line_termination = 0;
3917         else if ((argcount = short_opt('l', av, &optarg))) {
3918                 options->rename_limit = strtoul(optarg, NULL, 10);
3919                 return argcount;
3920         }
3921         else if ((argcount = short_opt('S', av, &optarg))) {
3922                 options->pickaxe = optarg;
3923                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3924                 return argcount;
3925         } else if ((argcount = short_opt('G', av, &optarg))) {
3926                 options->pickaxe = optarg;
3927                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3928                 return argcount;
3929         }
3930         else if (!strcmp(arg, "--pickaxe-all"))
3931                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3932         else if (!strcmp(arg, "--pickaxe-regex"))
3933                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3934         else if ((argcount = short_opt('O', av, &optarg))) {
3935                 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
3936                 options->orderfile = xstrdup(path);
3937                 return argcount;
3938         }
3939         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3940                 int offending = parse_diff_filter_opt(optarg, options);
3941                 if (offending)
3942                         die("unknown change class '%c' in --diff-filter=%s",
3943                             offending, optarg);
3944                 return argcount;
3945         }
3946         else if (!strcmp(arg, "--abbrev"))
3947                 options->abbrev = DEFAULT_ABBREV;
3948         else if (skip_prefix(arg, "--abbrev=", &arg)) {
3949                 options->abbrev = strtoul(arg, NULL, 10);
3950                 if (options->abbrev < MINIMUM_ABBREV)
3951                         options->abbrev = MINIMUM_ABBREV;
3952                 else if (40 < options->abbrev)
3953                         options->abbrev = 40;
3954         }
3955         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3956                 options->a_prefix = optarg;
3957                 return argcount;
3958         }
3959         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3960                 options->b_prefix = optarg;
3961                 return argcount;
3962         }
3963         else if (!strcmp(arg, "--no-prefix"))
3964                 options->a_prefix = options->b_prefix = "";
3965         else if (opt_arg(arg, '\0', "inter-hunk-context",
3966                          &options->interhunkcontext))
3967                 ;
3968         else if (!strcmp(arg, "-W"))
3969                 DIFF_OPT_SET(options, FUNCCONTEXT);
3970         else if (!strcmp(arg, "--function-context"))
3971                 DIFF_OPT_SET(options, FUNCCONTEXT);
3972         else if (!strcmp(arg, "--no-function-context"))
3973                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3974         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3975                 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
3976                 options->file = fopen(path, "w");
3977                 if (!options->file)
3978                         die_errno("Could not open '%s'", path);
3979                 options->close_file = 1;
3980                 return argcount;
3981         } else
3982                 return 0;
3983         return 1;
3984 }
3985
3986 int parse_rename_score(const char **cp_p)
3987 {
3988         unsigned long num, scale;
3989         int ch, dot;
3990         const char *cp = *cp_p;
3991
3992         num = 0;
3993         scale = 1;
3994         dot = 0;
3995         for (;;) {
3996                 ch = *cp;
3997                 if ( !dot && ch == '.' ) {
3998                         scale = 1;
3999                         dot = 1;
4000                 } else if ( ch == '%' ) {
4001                         scale = dot ? scale*100 : 100;
4002                         cp++;   /* % is always at the end */
4003                         break;
4004                 } else if ( ch >= '0' && ch <= '9' ) {
4005                         if ( scale < 100000 ) {
4006                                 scale *= 10;
4007                                 num = (num*10) + (ch-'0');
4008                         }
4009                 } else {
4010                         break;
4011                 }
4012                 cp++;
4013         }
4014         *cp_p = cp;
4015
4016         /* user says num divided by scale and we say internally that
4017          * is MAX_SCORE * num / scale.
4018          */
4019         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4020 }
4021
4022 static int diff_scoreopt_parse(const char *opt)
4023 {
4024         int opt1, opt2, cmd;
4025
4026         if (*opt++ != '-')
4027                 return -1;
4028         cmd = *opt++;
4029         if (cmd == '-') {
4030                 /* convert the long-form arguments into short-form versions */
4031                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4032                         if (*opt == 0 || *opt++ == '=')
4033                                 cmd = 'B';
4034                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4035                         if (*opt == 0 || *opt++ == '=')
4036                                 cmd = 'C';
4037                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4038                         if (*opt == 0 || *opt++ == '=')
4039                                 cmd = 'M';
4040                 }
4041         }
4042         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4043                 return -1; /* that is not a -M, -C, or -B option */
4044
4045         opt1 = parse_rename_score(&opt);
4046         if (cmd != 'B')
4047                 opt2 = 0;
4048         else {
4049                 if (*opt == 0)
4050                         opt2 = 0;
4051                 else if (*opt != '/')
4052                         return -1; /* we expect -B80/99 or -B80 */
4053                 else {
4054                         opt++;
4055                         opt2 = parse_rename_score(&opt);
4056                 }
4057         }
4058         if (*opt != 0)
4059                 return -1;
4060         return opt1 | (opt2 << 16);
4061 }
4062
4063 struct diff_queue_struct diff_queued_diff;
4064
4065 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4066 {
4067         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4068         queue->queue[queue->nr++] = dp;
4069 }
4070
4071 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4072                                  struct diff_filespec *one,
4073                                  struct diff_filespec *two)
4074 {
4075         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4076         dp->one = one;
4077         dp->two = two;
4078         if (queue)
4079                 diff_q(queue, dp);
4080         return dp;
4081 }
4082
4083 void diff_free_filepair(struct diff_filepair *p)
4084 {
4085         free_filespec(p->one);
4086         free_filespec(p->two);
4087         free(p);
4088 }
4089
4090 /* This is different from find_unique_abbrev() in that
4091  * it stuffs the result with dots for alignment.
4092  */
4093 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
4094 {
4095         int abblen;
4096         const char *abbrev;
4097         if (len == 40)
4098                 return sha1_to_hex(sha1);
4099
4100         abbrev = find_unique_abbrev(sha1, len);
4101         abblen = strlen(abbrev);
4102         if (abblen < 37) {
4103                 static char hex[41];
4104                 if (len < abblen && abblen <= len + 2)
4105                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4106                 else
4107                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4108                 return hex;
4109         }
4110         return sha1_to_hex(sha1);
4111 }
4112
4113 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4114 {
4115         int line_termination = opt->line_termination;
4116         int inter_name_termination = line_termination ? '\t' : '\0';
4117
4118         fprintf(opt->file, "%s", diff_line_prefix(opt));
4119         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4120                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4121                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
4122                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4123         }
4124         if (p->score) {
4125                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4126                         inter_name_termination);
4127         } else {
4128                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4129         }
4130
4131         if (p->status == DIFF_STATUS_COPIED ||
4132             p->status == DIFF_STATUS_RENAMED) {
4133                 const char *name_a, *name_b;
4134                 name_a = p->one->path;
4135                 name_b = p->two->path;
4136                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4137                 write_name_quoted(name_a, opt->file, inter_name_termination);
4138                 write_name_quoted(name_b, opt->file, line_termination);
4139         } else {
4140                 const char *name_a, *name_b;
4141                 name_a = p->one->mode ? p->one->path : p->two->path;
4142                 name_b = NULL;
4143                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4144                 write_name_quoted(name_a, opt->file, line_termination);
4145         }
4146 }
4147
4148 int diff_unmodified_pair(struct diff_filepair *p)
4149 {
4150         /* This function is written stricter than necessary to support
4151          * the currently implemented transformers, but the idea is to
4152          * let transformers to produce diff_filepairs any way they want,
4153          * and filter and clean them up here before producing the output.
4154          */
4155         struct diff_filespec *one = p->one, *two = p->two;
4156
4157         if (DIFF_PAIR_UNMERGED(p))
4158                 return 0; /* unmerged is interesting */
4159
4160         /* deletion, addition, mode or type change
4161          * and rename are all interesting.
4162          */
4163         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4164             DIFF_PAIR_MODE_CHANGED(p) ||
4165             strcmp(one->path, two->path))
4166                 return 0;
4167
4168         /* both are valid and point at the same path.  that is, we are
4169          * dealing with a change.
4170          */
4171         if (one->sha1_valid && two->sha1_valid &&
4172             !hashcmp(one->sha1, two->sha1) &&
4173             !one->dirty_submodule && !two->dirty_submodule)
4174                 return 1; /* no change */
4175         if (!one->sha1_valid && !two->sha1_valid)
4176                 return 1; /* both look at the same file on the filesystem. */
4177         return 0;
4178 }
4179
4180 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4181 {
4182         if (diff_unmodified_pair(p))
4183                 return;
4184
4185         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4186             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4187                 return; /* no tree diffs in patch format */
4188
4189         run_diff(p, o);
4190 }
4191
4192 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4193                             struct diffstat_t *diffstat)
4194 {
4195         if (diff_unmodified_pair(p))
4196                 return;
4197
4198         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4199             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4200                 return; /* no useful stat for tree diffs */
4201
4202         run_diffstat(p, o, diffstat);
4203 }
4204
4205 static void diff_flush_checkdiff(struct diff_filepair *p,
4206                 struct diff_options *o)
4207 {
4208         if (diff_unmodified_pair(p))
4209                 return;
4210
4211         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4212             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4213                 return; /* nothing to check in tree diffs */
4214
4215         run_checkdiff(p, o);
4216 }
4217
4218 int diff_queue_is_empty(void)
4219 {
4220         struct diff_queue_struct *q = &diff_queued_diff;
4221         int i;
4222         for (i = 0; i < q->nr; i++)
4223                 if (!diff_unmodified_pair(q->queue[i]))
4224                         return 0;
4225         return 1;
4226 }
4227
4228 #if DIFF_DEBUG
4229 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4230 {
4231         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4232                 x, one ? one : "",
4233                 s->path,
4234                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4235                 s->mode,
4236                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4237         fprintf(stderr, "queue[%d] %s size %lu\n",
4238                 x, one ? one : "",
4239                 s->size);
4240 }
4241
4242 void diff_debug_filepair(const struct diff_filepair *p, int i)
4243 {
4244         diff_debug_filespec(p->one, i, "one");
4245         diff_debug_filespec(p->two, i, "two");
4246         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4247                 p->score, p->status ? p->status : '?',
4248                 p->one->rename_used, p->broken_pair);
4249 }
4250
4251 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4252 {
4253         int i;
4254         if (msg)
4255                 fprintf(stderr, "%s\n", msg);
4256         fprintf(stderr, "q->nr = %d\n", q->nr);
4257         for (i = 0; i < q->nr; i++) {
4258                 struct diff_filepair *p = q->queue[i];
4259                 diff_debug_filepair(p, i);
4260         }
4261 }
4262 #endif
4263
4264 static void diff_resolve_rename_copy(void)
4265 {
4266         int i;
4267         struct diff_filepair *p;
4268         struct diff_queue_struct *q = &diff_queued_diff;
4269
4270         diff_debug_queue("resolve-rename-copy", q);
4271
4272         for (i = 0; i < q->nr; i++) {
4273                 p = q->queue[i];
4274                 p->status = 0; /* undecided */
4275                 if (DIFF_PAIR_UNMERGED(p))
4276                         p->status = DIFF_STATUS_UNMERGED;
4277                 else if (!DIFF_FILE_VALID(p->one))
4278                         p->status = DIFF_STATUS_ADDED;
4279                 else if (!DIFF_FILE_VALID(p->two))
4280                         p->status = DIFF_STATUS_DELETED;
4281                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4282                         p->status = DIFF_STATUS_TYPE_CHANGED;
4283
4284                 /* from this point on, we are dealing with a pair
4285                  * whose both sides are valid and of the same type, i.e.
4286                  * either in-place edit or rename/copy edit.
4287                  */
4288                 else if (DIFF_PAIR_RENAME(p)) {
4289                         /*
4290                          * A rename might have re-connected a broken
4291                          * pair up, causing the pathnames to be the
4292                          * same again. If so, that's not a rename at
4293                          * all, just a modification..
4294                          *
4295                          * Otherwise, see if this source was used for
4296                          * multiple renames, in which case we decrement
4297                          * the count, and call it a copy.
4298                          */
4299                         if (!strcmp(p->one->path, p->two->path))
4300                                 p->status = DIFF_STATUS_MODIFIED;
4301                         else if (--p->one->rename_used > 0)
4302                                 p->status = DIFF_STATUS_COPIED;
4303                         else
4304                                 p->status = DIFF_STATUS_RENAMED;
4305                 }
4306                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4307                          p->one->mode != p->two->mode ||
4308                          p->one->dirty_submodule ||
4309                          p->two->dirty_submodule ||
4310                          is_null_sha1(p->one->sha1))
4311                         p->status = DIFF_STATUS_MODIFIED;
4312                 else {
4313                         /* This is a "no-change" entry and should not
4314                          * happen anymore, but prepare for broken callers.
4315                          */
4316                         error("feeding unmodified %s to diffcore",
4317                               p->one->path);
4318                         p->status = DIFF_STATUS_UNKNOWN;
4319                 }
4320         }
4321         diff_debug_queue("resolve-rename-copy done", q);
4322 }
4323
4324 static int check_pair_status(struct diff_filepair *p)
4325 {
4326         switch (p->status) {
4327         case DIFF_STATUS_UNKNOWN:
4328                 return 0;
4329         case 0:
4330                 die("internal error in diff-resolve-rename-copy");
4331         default:
4332                 return 1;
4333         }
4334 }
4335
4336 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4337 {
4338         int fmt = opt->output_format;
4339
4340         if (fmt & DIFF_FORMAT_CHECKDIFF)
4341                 diff_flush_checkdiff(p, opt);
4342         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4343                 diff_flush_raw(p, opt);
4344         else if (fmt & DIFF_FORMAT_NAME) {
4345                 const char *name_a, *name_b;
4346                 name_a = p->two->path;
4347                 name_b = NULL;
4348                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4349                 write_name_quoted(name_a, opt->file, opt->line_termination);
4350         }
4351 }
4352
4353 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4354 {
4355         if (fs->mode)
4356                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4357         else
4358                 fprintf(file, " %s ", newdelete);
4359         write_name_quoted(fs->path, file, '\n');
4360 }
4361
4362
4363 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4364                 const char *line_prefix)
4365 {
4366         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4367                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4368                         p->two->mode, show_name ? ' ' : '\n');
4369                 if (show_name) {
4370                         write_name_quoted(p->two->path, file, '\n');
4371                 }
4372         }
4373 }
4374
4375 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4376                         const char *line_prefix)
4377 {
4378         char *names = pprint_rename(p->one->path, p->two->path);
4379
4380         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4381         free(names);
4382         show_mode_change(file, p, 0, line_prefix);
4383 }
4384
4385 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4386 {
4387         FILE *file = opt->file;
4388         const char *line_prefix = diff_line_prefix(opt);
4389
4390         switch(p->status) {
4391         case DIFF_STATUS_DELETED:
4392                 fputs(line_prefix, file);
4393                 show_file_mode_name(file, "delete", p->one);
4394                 break;
4395         case DIFF_STATUS_ADDED:
4396                 fputs(line_prefix, file);
4397                 show_file_mode_name(file, "create", p->two);
4398                 break;
4399         case DIFF_STATUS_COPIED:
4400                 fputs(line_prefix, file);
4401                 show_rename_copy(file, "copy", p, line_prefix);
4402                 break;
4403         case DIFF_STATUS_RENAMED:
4404                 fputs(line_prefix, file);
4405                 show_rename_copy(file, "rename", p, line_prefix);
4406                 break;
4407         default:
4408                 if (p->score) {
4409                         fprintf(file, "%s rewrite ", line_prefix);
4410                         write_name_quoted(p->two->path, file, ' ');
4411                         fprintf(file, "(%d%%)\n", similarity_index(p));
4412                 }
4413                 show_mode_change(file, p, !p->score, line_prefix);
4414                 break;
4415         }
4416 }
4417
4418 struct patch_id_t {
4419         git_SHA_CTX *ctx;
4420         int patchlen;
4421 };
4422
4423 static int remove_space(char *line, int len)
4424 {
4425         int i;
4426         char *dst = line;
4427         unsigned char c;
4428
4429         for (i = 0; i < len; i++)
4430                 if (!isspace((c = line[i])))
4431                         *dst++ = c;
4432
4433         return dst - line;
4434 }
4435
4436 static void patch_id_consume(void *priv, char *line, unsigned long len)
4437 {
4438         struct patch_id_t *data = priv;
4439         int new_len;
4440
4441         /* Ignore line numbers when computing the SHA1 of the patch */
4442         if (starts_with(line, "@@ -"))
4443                 return;
4444
4445         new_len = remove_space(line, len);
4446
4447         git_SHA1_Update(data->ctx, line, new_len);
4448         data->patchlen += new_len;
4449 }
4450
4451 /* returns 0 upon success, and writes result into sha1 */
4452 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4453 {
4454         struct diff_queue_struct *q = &diff_queued_diff;
4455         int i;
4456         git_SHA_CTX ctx;
4457         struct patch_id_t data;
4458         char buffer[PATH_MAX * 4 + 20];
4459
4460         git_SHA1_Init(&ctx);
4461         memset(&data, 0, sizeof(struct patch_id_t));
4462         data.ctx = &ctx;
4463
4464         for (i = 0; i < q->nr; i++) {
4465                 xpparam_t xpp;
4466                 xdemitconf_t xecfg;
4467                 mmfile_t mf1, mf2;
4468                 struct diff_filepair *p = q->queue[i];
4469                 int len1, len2;
4470
4471                 memset(&xpp, 0, sizeof(xpp));
4472                 memset(&xecfg, 0, sizeof(xecfg));
4473                 if (p->status == 0)
4474                         return error("internal diff status error");
4475                 if (p->status == DIFF_STATUS_UNKNOWN)
4476                         continue;
4477                 if (diff_unmodified_pair(p))
4478                         continue;
4479                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4480                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4481                         continue;
4482                 if (DIFF_PAIR_UNMERGED(p))
4483                         continue;
4484
4485                 diff_fill_sha1_info(p->one);
4486                 diff_fill_sha1_info(p->two);
4487                 if (fill_mmfile(&mf1, p->one) < 0 ||
4488                                 fill_mmfile(&mf2, p->two) < 0)
4489                         return error("unable to read files to diff");
4490
4491                 len1 = remove_space(p->one->path, strlen(p->one->path));
4492                 len2 = remove_space(p->two->path, strlen(p->two->path));
4493                 if (p->one->mode == 0)
4494                         len1 = snprintf(buffer, sizeof(buffer),
4495                                         "diff--gita/%.*sb/%.*s"
4496                                         "newfilemode%06o"
4497                                         "---/dev/null"
4498                                         "+++b/%.*s",
4499                                         len1, p->one->path,
4500                                         len2, p->two->path,
4501                                         p->two->mode,
4502                                         len2, p->two->path);
4503                 else if (p->two->mode == 0)
4504                         len1 = snprintf(buffer, sizeof(buffer),
4505                                         "diff--gita/%.*sb/%.*s"
4506                                         "deletedfilemode%06o"
4507                                         "---a/%.*s"
4508                                         "+++/dev/null",
4509                                         len1, p->one->path,
4510                                         len2, p->two->path,
4511                                         p->one->mode,
4512                                         len1, p->one->path);
4513                 else
4514                         len1 = snprintf(buffer, sizeof(buffer),
4515                                         "diff--gita/%.*sb/%.*s"
4516                                         "---a/%.*s"
4517                                         "+++b/%.*s",
4518                                         len1, p->one->path,
4519                                         len2, p->two->path,
4520                                         len1, p->one->path,
4521                                         len2, p->two->path);
4522                 git_SHA1_Update(&ctx, buffer, len1);
4523
4524                 if (diff_filespec_is_binary(p->one) ||
4525                     diff_filespec_is_binary(p->two)) {
4526                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4527                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4528                         continue;
4529                 }
4530
4531                 xpp.flags = 0;
4532                 xecfg.ctxlen = 3;
4533                 xecfg.flags = 0;
4534                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4535                                   &xpp, &xecfg))
4536                         return error("unable to generate patch-id diff for %s",
4537                                      p->one->path);
4538         }
4539
4540         git_SHA1_Final(sha1, &ctx);
4541         return 0;
4542 }
4543
4544 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4545 {
4546         struct diff_queue_struct *q = &diff_queued_diff;
4547         int i;
4548         int result = diff_get_patch_id(options, sha1);
4549
4550         for (i = 0; i < q->nr; i++)
4551                 diff_free_filepair(q->queue[i]);
4552
4553         free(q->queue);
4554         DIFF_QUEUE_CLEAR(q);
4555
4556         return result;
4557 }
4558
4559 static int is_summary_empty(const struct diff_queue_struct *q)
4560 {
4561         int i;
4562
4563         for (i = 0; i < q->nr; i++) {
4564                 const struct diff_filepair *p = q->queue[i];
4565
4566                 switch (p->status) {
4567                 case DIFF_STATUS_DELETED:
4568                 case DIFF_STATUS_ADDED:
4569                 case DIFF_STATUS_COPIED:
4570                 case DIFF_STATUS_RENAMED:
4571                         return 0;
4572                 default:
4573                         if (p->score)
4574                                 return 0;
4575                         if (p->one->mode && p->two->mode &&
4576                             p->one->mode != p->two->mode)
4577                                 return 0;
4578                         break;
4579                 }
4580         }
4581         return 1;
4582 }
4583
4584 static const char rename_limit_warning[] =
4585 "inexact rename detection was skipped due to too many files.";
4586
4587 static const char degrade_cc_to_c_warning[] =
4588 "only found copies from modified paths due to too many files.";
4589
4590 static const char rename_limit_advice[] =
4591 "you may want to set your %s variable to at least "
4592 "%d and retry the command.";
4593
4594 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4595 {
4596         if (degraded_cc)
4597                 warning(degrade_cc_to_c_warning);
4598         else if (needed)
4599                 warning(rename_limit_warning);
4600         else
4601                 return;
4602         if (0 < needed && needed < 32767)
4603                 warning(rename_limit_advice, varname, needed);
4604 }
4605
4606 void diff_flush(struct diff_options *options)
4607 {
4608         struct diff_queue_struct *q = &diff_queued_diff;
4609         int i, output_format = options->output_format;
4610         int separator = 0;
4611         int dirstat_by_line = 0;
4612
4613         /*
4614          * Order: raw, stat, summary, patch
4615          * or:    name/name-status/checkdiff (other bits clear)
4616          */
4617         if (!q->nr)
4618                 goto free_queue;
4619
4620         if (output_format & (DIFF_FORMAT_RAW |
4621                              DIFF_FORMAT_NAME |
4622                              DIFF_FORMAT_NAME_STATUS |
4623                              DIFF_FORMAT_CHECKDIFF)) {
4624                 for (i = 0; i < q->nr; i++) {
4625                         struct diff_filepair *p = q->queue[i];
4626                         if (check_pair_status(p))
4627                                 flush_one_pair(p, options);
4628                 }
4629                 separator++;
4630         }
4631
4632         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4633                 dirstat_by_line = 1;
4634
4635         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4636             dirstat_by_line) {
4637                 struct diffstat_t diffstat;
4638
4639                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4640                 for (i = 0; i < q->nr; i++) {
4641                         struct diff_filepair *p = q->queue[i];
4642                         if (check_pair_status(p))
4643                                 diff_flush_stat(p, options, &diffstat);
4644                 }
4645                 if (output_format & DIFF_FORMAT_NUMSTAT)
4646                         show_numstat(&diffstat, options);
4647                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4648                         show_stats(&diffstat, options);
4649                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4650                         show_shortstats(&diffstat, options);
4651                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4652                         show_dirstat_by_line(&diffstat, options);
4653                 free_diffstat_info(&diffstat);
4654                 separator++;
4655         }
4656         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4657                 show_dirstat(options);
4658
4659         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4660                 for (i = 0; i < q->nr; i++) {
4661                         diff_summary(options, q->queue[i]);
4662                 }
4663                 separator++;
4664         }
4665
4666         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4667             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4668             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4669                 /*
4670                  * run diff_flush_patch for the exit status. setting
4671                  * options->file to /dev/null should be safe, because we
4672                  * aren't supposed to produce any output anyway.
4673                  */
4674                 if (options->close_file)
4675                         fclose(options->file);
4676                 options->file = fopen("/dev/null", "w");
4677                 if (!options->file)
4678                         die_errno("Could not open /dev/null");
4679                 options->close_file = 1;
4680                 for (i = 0; i < q->nr; i++) {
4681                         struct diff_filepair *p = q->queue[i];
4682                         if (check_pair_status(p))
4683                                 diff_flush_patch(p, options);
4684                         if (options->found_changes)
4685                                 break;
4686                 }
4687         }
4688
4689         if (output_format & DIFF_FORMAT_PATCH) {
4690                 if (separator) {
4691                         fprintf(options->file, "%s%c",
4692                                 diff_line_prefix(options),
4693                                 options->line_termination);
4694                         if (options->stat_sep) {
4695                                 /* attach patch instead of inline */
4696                                 fputs(options->stat_sep, options->file);
4697                         }
4698                 }
4699
4700                 for (i = 0; i < q->nr; i++) {
4701                         struct diff_filepair *p = q->queue[i];
4702                         if (check_pair_status(p))
4703                                 diff_flush_patch(p, options);
4704                 }
4705         }
4706
4707         if (output_format & DIFF_FORMAT_CALLBACK)
4708                 options->format_callback(q, options, options->format_callback_data);
4709
4710         for (i = 0; i < q->nr; i++)
4711                 diff_free_filepair(q->queue[i]);
4712 free_queue:
4713         free(q->queue);
4714         DIFF_QUEUE_CLEAR(q);
4715         if (options->close_file)
4716                 fclose(options->file);
4717
4718         /*
4719          * Report the content-level differences with HAS_CHANGES;
4720          * diff_addremove/diff_change does not set the bit when
4721          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4722          */
4723         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4724                 if (options->found_changes)
4725                         DIFF_OPT_SET(options, HAS_CHANGES);
4726                 else
4727                         DIFF_OPT_CLR(options, HAS_CHANGES);
4728         }
4729 }
4730
4731 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4732 {
4733         return (((p->status == DIFF_STATUS_MODIFIED) &&
4734                  ((p->score &&
4735                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4736                   (!p->score &&
4737                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4738                 ((p->status != DIFF_STATUS_MODIFIED) &&
4739                  filter_bit_tst(p->status, options)));
4740 }
4741
4742 static void diffcore_apply_filter(struct diff_options *options)
4743 {
4744         int i;
4745         struct diff_queue_struct *q = &diff_queued_diff;
4746         struct diff_queue_struct outq;
4747
4748         DIFF_QUEUE_CLEAR(&outq);
4749
4750         if (!options->filter)
4751                 return;
4752
4753         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4754                 int found;
4755                 for (i = found = 0; !found && i < q->nr; i++) {
4756                         if (match_filter(options, q->queue[i]))
4757                                 found++;
4758                 }
4759                 if (found)
4760                         return;
4761
4762                 /* otherwise we will clear the whole queue
4763                  * by copying the empty outq at the end of this
4764                  * function, but first clear the current entries
4765                  * in the queue.
4766                  */
4767                 for (i = 0; i < q->nr; i++)
4768                         diff_free_filepair(q->queue[i]);
4769         }
4770         else {
4771                 /* Only the matching ones */
4772                 for (i = 0; i < q->nr; i++) {
4773                         struct diff_filepair *p = q->queue[i];
4774                         if (match_filter(options, p))
4775                                 diff_q(&outq, p);
4776                         else
4777                                 diff_free_filepair(p);
4778                 }
4779         }
4780         free(q->queue);
4781         *q = outq;
4782 }
4783
4784 /* Check whether two filespecs with the same mode and size are identical */
4785 static int diff_filespec_is_identical(struct diff_filespec *one,
4786                                       struct diff_filespec *two)
4787 {
4788         if (S_ISGITLINK(one->mode))
4789                 return 0;
4790         if (diff_populate_filespec(one, 0))
4791                 return 0;
4792         if (diff_populate_filespec(two, 0))
4793                 return 0;
4794         return !memcmp(one->data, two->data, one->size);
4795 }
4796
4797 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4798 {
4799         if (p->done_skip_stat_unmatch)
4800                 return p->skip_stat_unmatch_result;
4801
4802         p->done_skip_stat_unmatch = 1;
4803         p->skip_stat_unmatch_result = 0;
4804         /*
4805          * 1. Entries that come from stat info dirtiness
4806          *    always have both sides (iow, not create/delete),
4807          *    one side of the object name is unknown, with
4808          *    the same mode and size.  Keep the ones that
4809          *    do not match these criteria.  They have real
4810          *    differences.
4811          *
4812          * 2. At this point, the file is known to be modified,
4813          *    with the same mode and size, and the object
4814          *    name of one side is unknown.  Need to inspect
4815          *    the identical contents.
4816          */
4817         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4818             !DIFF_FILE_VALID(p->two) ||
4819             (p->one->sha1_valid && p->two->sha1_valid) ||
4820             (p->one->mode != p->two->mode) ||
4821             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4822             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4823             (p->one->size != p->two->size) ||
4824             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4825                 p->skip_stat_unmatch_result = 1;
4826         return p->skip_stat_unmatch_result;
4827 }
4828
4829 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4830 {
4831         int i;
4832         struct diff_queue_struct *q = &diff_queued_diff;
4833         struct diff_queue_struct outq;
4834         DIFF_QUEUE_CLEAR(&outq);
4835
4836         for (i = 0; i < q->nr; i++) {
4837                 struct diff_filepair *p = q->queue[i];
4838
4839                 if (diff_filespec_check_stat_unmatch(p))
4840                         diff_q(&outq, p);
4841                 else {
4842                         /*
4843                          * The caller can subtract 1 from skip_stat_unmatch
4844                          * to determine how many paths were dirty only
4845                          * due to stat info mismatch.
4846                          */
4847                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4848                                 diffopt->skip_stat_unmatch++;
4849                         diff_free_filepair(p);
4850                 }
4851         }
4852         free(q->queue);
4853         *q = outq;
4854 }
4855
4856 static int diffnamecmp(const void *a_, const void *b_)
4857 {
4858         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4859         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4860         const char *name_a, *name_b;
4861
4862         name_a = a->one ? a->one->path : a->two->path;
4863         name_b = b->one ? b->one->path : b->two->path;
4864         return strcmp(name_a, name_b);
4865 }
4866
4867 void diffcore_fix_diff_index(struct diff_options *options)
4868 {
4869         struct diff_queue_struct *q = &diff_queued_diff;
4870         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4871 }
4872
4873 void diffcore_std(struct diff_options *options)
4874 {
4875         /* NOTE please keep the following in sync with diff_tree_combined() */
4876         if (options->skip_stat_unmatch)
4877                 diffcore_skip_stat_unmatch(options);
4878         if (!options->found_follow) {
4879                 /* See try_to_follow_renames() in tree-diff.c */
4880                 if (options->break_opt != -1)
4881                         diffcore_break(options->break_opt);
4882                 if (options->detect_rename)
4883                         diffcore_rename(options);
4884                 if (options->break_opt != -1)
4885                         diffcore_merge_broken();
4886         }
4887         if (options->pickaxe)
4888                 diffcore_pickaxe(options);
4889         if (options->orderfile)
4890                 diffcore_order(options->orderfile);
4891         if (!options->found_follow)
4892                 /* See try_to_follow_renames() in tree-diff.c */
4893                 diff_resolve_rename_copy();
4894         diffcore_apply_filter(options);
4895
4896         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4897                 DIFF_OPT_SET(options, HAS_CHANGES);
4898         else
4899                 DIFF_OPT_CLR(options, HAS_CHANGES);
4900
4901         options->found_follow = 0;
4902 }
4903
4904 int diff_result_code(struct diff_options *opt, int status)
4905 {
4906         int result = 0;
4907
4908         diff_warn_rename_limit("diff.renameLimit",
4909                                opt->needed_rename_limit,
4910                                opt->degraded_cc_to_c);
4911         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4912             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4913                 return status;
4914         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4915             DIFF_OPT_TST(opt, HAS_CHANGES))
4916                 result |= 01;
4917         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4918             DIFF_OPT_TST(opt, CHECK_FAILED))
4919                 result |= 02;
4920         return result;
4921 }
4922
4923 int diff_can_quit_early(struct diff_options *opt)
4924 {
4925         return (DIFF_OPT_TST(opt, QUICK) &&
4926                 !opt->filter &&
4927                 DIFF_OPT_TST(opt, HAS_CHANGES));
4928 }
4929
4930 /*
4931  * Shall changes to this submodule be ignored?
4932  *
4933  * Submodule changes can be configured to be ignored separately for each path,
4934  * but that configuration can be overridden from the command line.
4935  */
4936 static int is_submodule_ignored(const char *path, struct diff_options *options)
4937 {
4938         int ignored = 0;
4939         unsigned orig_flags = options->flags;
4940         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4941                 set_diffopt_flags_from_submodule_config(options, path);
4942         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4943                 ignored = 1;
4944         options->flags = orig_flags;
4945         return ignored;
4946 }
4947
4948 void diff_addremove(struct diff_options *options,
4949                     int addremove, unsigned mode,
4950                     const unsigned char *sha1,
4951                     int sha1_valid,
4952                     const char *concatpath, unsigned dirty_submodule)
4953 {
4954         struct diff_filespec *one, *two;
4955
4956         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4957                 return;
4958
4959         /* This may look odd, but it is a preparation for
4960          * feeding "there are unchanged files which should
4961          * not produce diffs, but when you are doing copy
4962          * detection you would need them, so here they are"
4963          * entries to the diff-core.  They will be prefixed
4964          * with something like '=' or '*' (I haven't decided
4965          * which but should not make any difference).
4966          * Feeding the same new and old to diff_change()
4967          * also has the same effect.
4968          * Before the final output happens, they are pruned after
4969          * merged into rename/copy pairs as appropriate.
4970          */
4971         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4972                 addremove = (addremove == '+' ? '-' :
4973                              addremove == '-' ? '+' : addremove);
4974
4975         if (options->prefix &&
4976             strncmp(concatpath, options->prefix, options->prefix_length))
4977                 return;
4978
4979         one = alloc_filespec(concatpath);
4980         two = alloc_filespec(concatpath);
4981
4982         if (addremove != '+')
4983                 fill_filespec(one, sha1, sha1_valid, mode);
4984         if (addremove != '-') {
4985                 fill_filespec(two, sha1, sha1_valid, mode);
4986                 two->dirty_submodule = dirty_submodule;
4987         }
4988
4989         diff_queue(&diff_queued_diff, one, two);
4990         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4991                 DIFF_OPT_SET(options, HAS_CHANGES);
4992 }
4993
4994 void diff_change(struct diff_options *options,
4995                  unsigned old_mode, unsigned new_mode,
4996                  const unsigned char *old_sha1,
4997                  const unsigned char *new_sha1,
4998                  int old_sha1_valid, int new_sha1_valid,
4999                  const char *concatpath,
5000                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5001 {
5002         struct diff_filespec *one, *two;
5003         struct diff_filepair *p;
5004
5005         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5006             is_submodule_ignored(concatpath, options))
5007                 return;
5008
5009         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5010                 unsigned tmp;
5011                 const unsigned char *tmp_c;
5012                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
5013                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
5014                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
5015                         new_sha1_valid = tmp;
5016                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
5017                         new_dirty_submodule = tmp;
5018         }
5019
5020         if (options->prefix &&
5021             strncmp(concatpath, options->prefix, options->prefix_length))
5022                 return;
5023
5024         one = alloc_filespec(concatpath);
5025         two = alloc_filespec(concatpath);
5026         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5027         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5028         one->dirty_submodule = old_dirty_submodule;
5029         two->dirty_submodule = new_dirty_submodule;
5030         p = diff_queue(&diff_queued_diff, one, two);
5031
5032         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5033                 return;
5034
5035         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5036             !diff_filespec_check_stat_unmatch(p))
5037                 return;
5038
5039         DIFF_OPT_SET(options, HAS_CHANGES);
5040 }
5041
5042 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5043 {
5044         struct diff_filepair *pair;
5045         struct diff_filespec *one, *two;
5046
5047         if (options->prefix &&
5048             strncmp(path, options->prefix, options->prefix_length))
5049                 return NULL;
5050
5051         one = alloc_filespec(path);
5052         two = alloc_filespec(path);
5053         pair = diff_queue(&diff_queued_diff, one, two);
5054         pair->is_unmerged = 1;
5055         return pair;
5056 }
5057
5058 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5059                 size_t *outsize)
5060 {
5061         struct diff_tempfile *temp;
5062         const char *argv[3];
5063         const char **arg = argv;
5064         struct child_process child = CHILD_PROCESS_INIT;
5065         struct strbuf buf = STRBUF_INIT;
5066         int err = 0;
5067
5068         temp = prepare_temp_file(spec->path, spec);
5069         *arg++ = pgm;
5070         *arg++ = temp->name;
5071         *arg = NULL;
5072
5073         child.use_shell = 1;
5074         child.argv = argv;
5075         child.out = -1;
5076         if (start_command(&child)) {
5077                 remove_tempfile();
5078                 return NULL;
5079         }
5080
5081         if (strbuf_read(&buf, child.out, 0) < 0)
5082                 err = error("error reading from textconv command '%s'", pgm);
5083         close(child.out);
5084
5085         if (finish_command(&child) || err) {
5086                 strbuf_release(&buf);
5087                 remove_tempfile();
5088                 return NULL;
5089         }
5090         remove_tempfile();
5091
5092         return strbuf_detach(&buf, outsize);
5093 }
5094
5095 size_t fill_textconv(struct userdiff_driver *driver,
5096                      struct diff_filespec *df,
5097                      char **outbuf)
5098 {
5099         size_t size;
5100
5101         if (!driver) {
5102                 if (!DIFF_FILE_VALID(df)) {
5103                         *outbuf = "";
5104                         return 0;
5105                 }
5106                 if (diff_populate_filespec(df, 0))
5107                         die("unable to read files to diff");
5108                 *outbuf = df->data;
5109                 return df->size;
5110         }
5111
5112         if (!driver->textconv)
5113                 die("BUG: fill_textconv called with non-textconv driver");
5114
5115         if (driver->textconv_cache && df->sha1_valid) {
5116                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5117                                           &size);
5118                 if (*outbuf)
5119                         return size;
5120         }
5121
5122         *outbuf = run_textconv(driver->textconv, df, &size);
5123         if (!*outbuf)
5124                 die("unable to read files to diff");
5125
5126         if (driver->textconv_cache && df->sha1_valid) {
5127                 /* ignore errors, as we might be in a readonly repository */
5128                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5129                                 size);
5130                 /*
5131                  * we could save up changes and flush them all at the end,
5132                  * but we would need an extra call after all diffing is done.
5133                  * Since generating a cache entry is the slow path anyway,
5134                  * this extra overhead probably isn't a big deal.
5135                  */
5136                 notes_cache_write(driver->textconv_cache);
5137         }
5138
5139         return size;
5140 }
5141
5142 void setup_diff_pager(struct diff_options *opt)
5143 {
5144         /*
5145          * If the user asked for our exit code, then either they want --quiet
5146          * or --exit-code. We should definitely not bother with a pager in the
5147          * former case, as we will generate no output. Since we still properly
5148          * report our exit code even when a pager is run, we _could_ run a
5149          * pager with --exit-code. But since we have not done so historically,
5150          * and because it is easy to find people oneline advising "git diff
5151          * --exit-code" in hooks and other scripts, we do not do so.
5152          */
5153         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5154             check_pager_config("diff") != 0)
5155                 setup_pager();
5156 }