Merge branch 'ad/sha1-update-chunked' into maint
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "tempfile.h"
6 #include "quote.h"
7 #include "diff.h"
8 #include "diffcore.h"
9 #include "delta.h"
10 #include "xdiff-interface.h"
11 #include "color.h"
12 #include "attr.h"
13 #include "run-command.h"
14 #include "utf8.h"
15 #include "userdiff.h"
16 #include "sigchain.h"
17 #include "submodule-config.h"
18 #include "submodule.h"
19 #include "ll-merge.h"
20 #include "string-list.h"
21 #include "argv-array.h"
22
23 #ifdef NO_FAST_WORKING_DIRECTORY
24 #define FAST_WORKING_DIRECTORY 0
25 #else
26 #define FAST_WORKING_DIRECTORY 1
27 #endif
28
29 static int diff_detect_rename_default;
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 int git_diff_ui_config(const char *var, const char *value, void *cb)
173 {
174         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
175                 diff_use_color_default = git_config_colorbool(var, value);
176                 return 0;
177         }
178         if (!strcmp(var, "diff.context")) {
179                 diff_context_default = git_config_int(var, value);
180                 if (diff_context_default < 0)
181                         return -1;
182                 return 0;
183         }
184         if (!strcmp(var, "diff.renames")) {
185                 diff_detect_rename_default = git_config_rename(var, value);
186                 return 0;
187         }
188         if (!strcmp(var, "diff.autorefreshindex")) {
189                 diff_auto_refresh_index = git_config_bool(var, value);
190                 return 0;
191         }
192         if (!strcmp(var, "diff.mnemonicprefix")) {
193                 diff_mnemonic_prefix = git_config_bool(var, value);
194                 return 0;
195         }
196         if (!strcmp(var, "diff.noprefix")) {
197                 diff_no_prefix = git_config_bool(var, value);
198                 return 0;
199         }
200         if (!strcmp(var, "diff.statgraphwidth")) {
201                 diff_stat_graph_width = git_config_int(var, value);
202                 return 0;
203         }
204         if (!strcmp(var, "diff.external"))
205                 return git_config_string(&external_diff_cmd_cfg, var, value);
206         if (!strcmp(var, "diff.wordregex"))
207                 return git_config_string(&diff_word_regex_cfg, var, value);
208         if (!strcmp(var, "diff.orderfile"))
209                 return git_config_pathname(&diff_order_file_cfg, var, value);
210
211         if (!strcmp(var, "diff.ignoresubmodules"))
212                 handle_ignore_submodules_arg(&default_diff_options, value);
213
214         if (!strcmp(var, "diff.submodule")) {
215                 if (parse_submodule_params(&default_diff_options, value))
216                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
217                                 value);
218                 return 0;
219         }
220
221         if (!strcmp(var, "diff.algorithm")) {
222                 diff_algorithm = parse_algorithm_value(value);
223                 if (diff_algorithm < 0)
224                         return -1;
225                 return 0;
226         }
227
228         if (git_color_config(var, value, cb) < 0)
229                 return -1;
230
231         return git_diff_basic_config(var, value, cb);
232 }
233
234 int git_diff_basic_config(const char *var, const char *value, void *cb)
235 {
236         const char *name;
237
238         if (!strcmp(var, "diff.renamelimit")) {
239                 diff_rename_limit_default = git_config_int(var, value);
240                 return 0;
241         }
242
243         if (userdiff_config(var, value) < 0)
244                 return -1;
245
246         if (skip_prefix(var, "diff.color.", &name) ||
247             skip_prefix(var, "color.diff.", &name)) {
248                 int slot = parse_diff_color_slot(name);
249                 if (slot < 0)
250                         return 0;
251                 if (!value)
252                         return config_error_nonbool(var);
253                 return color_parse(value, diff_colors[slot]);
254         }
255
256         /* like GNU diff's --suppress-blank-empty option  */
257         if (!strcmp(var, "diff.suppressblankempty") ||
258                         /* for backwards compatibility */
259                         !strcmp(var, "diff.suppress-blank-empty")) {
260                 diff_suppress_blank_empty = git_config_bool(var, value);
261                 return 0;
262         }
263
264         if (!strcmp(var, "diff.dirstat")) {
265                 struct strbuf errmsg = STRBUF_INIT;
266                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
267                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
268                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
269                                 errmsg.buf);
270                 strbuf_release(&errmsg);
271                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
272                 return 0;
273         }
274
275         if (starts_with(var, "submodule."))
276                 return parse_submodule_config_option(var, value);
277
278         return git_default_config(var, value, cb);
279 }
280
281 static char *quote_two(const char *one, const char *two)
282 {
283         int need_one = quote_c_style(one, NULL, NULL, 1);
284         int need_two = quote_c_style(two, NULL, NULL, 1);
285         struct strbuf res = STRBUF_INIT;
286
287         if (need_one + need_two) {
288                 strbuf_addch(&res, '"');
289                 quote_c_style(one, &res, NULL, 1);
290                 quote_c_style(two, &res, NULL, 1);
291                 strbuf_addch(&res, '"');
292         } else {
293                 strbuf_addstr(&res, one);
294                 strbuf_addstr(&res, two);
295         }
296         return strbuf_detach(&res, NULL);
297 }
298
299 static const char *external_diff(void)
300 {
301         static const char *external_diff_cmd = NULL;
302         static int done_preparing = 0;
303
304         if (done_preparing)
305                 return external_diff_cmd;
306         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
307         if (!external_diff_cmd)
308                 external_diff_cmd = external_diff_cmd_cfg;
309         done_preparing = 1;
310         return external_diff_cmd;
311 }
312
313 /*
314  * Keep track of files used for diffing. Sometimes such an entry
315  * refers to a temporary file, sometimes to an existing file, and
316  * sometimes to "/dev/null".
317  */
318 static struct diff_tempfile {
319         /*
320          * filename external diff should read from, or NULL if this
321          * entry is currently not in use:
322          */
323         const char *name;
324
325         char hex[41];
326         char mode[10];
327
328         /*
329          * If this diff_tempfile instance refers to a temporary file,
330          * this tempfile object is used to manage its lifetime.
331          */
332         struct tempfile tempfile;
333 } diff_temp[2];
334
335 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
336
337 struct emit_callback {
338         int color_diff;
339         unsigned ws_rule;
340         int blank_at_eof_in_preimage;
341         int blank_at_eof_in_postimage;
342         int lno_in_preimage;
343         int lno_in_postimage;
344         sane_truncate_fn truncate;
345         const char **label_path;
346         struct diff_words_data *diff_words;
347         struct diff_options *opt;
348         int *found_changesp;
349         struct strbuf *header;
350 };
351
352 static int count_lines(const char *data, int size)
353 {
354         int count, ch, completely_empty = 1, nl_just_seen = 0;
355         count = 0;
356         while (0 < size--) {
357                 ch = *data++;
358                 if (ch == '\n') {
359                         count++;
360                         nl_just_seen = 1;
361                         completely_empty = 0;
362                 }
363                 else {
364                         nl_just_seen = 0;
365                         completely_empty = 0;
366                 }
367         }
368         if (completely_empty)
369                 return 0;
370         if (!nl_just_seen)
371                 count++; /* no trailing newline */
372         return count;
373 }
374
375 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
376 {
377         if (!DIFF_FILE_VALID(one)) {
378                 mf->ptr = (char *)""; /* does not matter */
379                 mf->size = 0;
380                 return 0;
381         }
382         else if (diff_populate_filespec(one, 0))
383                 return -1;
384
385         mf->ptr = one->data;
386         mf->size = one->size;
387         return 0;
388 }
389
390 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
391 static unsigned long diff_filespec_size(struct diff_filespec *one)
392 {
393         if (!DIFF_FILE_VALID(one))
394                 return 0;
395         diff_populate_filespec(one, CHECK_SIZE_ONLY);
396         return one->size;
397 }
398
399 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
400 {
401         char *ptr = mf->ptr;
402         long size = mf->size;
403         int cnt = 0;
404
405         if (!size)
406                 return cnt;
407         ptr += size - 1; /* pointing at the very end */
408         if (*ptr != '\n')
409                 ; /* incomplete line */
410         else
411                 ptr--; /* skip the last LF */
412         while (mf->ptr < ptr) {
413                 char *prev_eol;
414                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
415                         if (*prev_eol == '\n')
416                                 break;
417                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
418                         break;
419                 cnt++;
420                 ptr = prev_eol - 1;
421         }
422         return cnt;
423 }
424
425 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
426                                struct emit_callback *ecbdata)
427 {
428         int l1, l2, at;
429         unsigned ws_rule = ecbdata->ws_rule;
430         l1 = count_trailing_blank(mf1, ws_rule);
431         l2 = count_trailing_blank(mf2, ws_rule);
432         if (l2 <= l1) {
433                 ecbdata->blank_at_eof_in_preimage = 0;
434                 ecbdata->blank_at_eof_in_postimage = 0;
435                 return;
436         }
437         at = count_lines(mf1->ptr, mf1->size);
438         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
439
440         at = count_lines(mf2->ptr, mf2->size);
441         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
442 }
443
444 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
445                         int first, const char *line, int len)
446 {
447         int has_trailing_newline, has_trailing_carriage_return;
448         int nofirst;
449         FILE *file = o->file;
450
451         fputs(diff_line_prefix(o), file);
452
453         if (len == 0) {
454                 has_trailing_newline = (first == '\n');
455                 has_trailing_carriage_return = (!has_trailing_newline &&
456                                                 (first == '\r'));
457                 nofirst = has_trailing_newline || has_trailing_carriage_return;
458         } else {
459                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
460                 if (has_trailing_newline)
461                         len--;
462                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
463                 if (has_trailing_carriage_return)
464                         len--;
465                 nofirst = 0;
466         }
467
468         if (len || !nofirst) {
469                 fputs(set, file);
470                 if (!nofirst)
471                         fputc(first, file);
472                 fwrite(line, len, 1, file);
473                 fputs(reset, file);
474         }
475         if (has_trailing_carriage_return)
476                 fputc('\r', file);
477         if (has_trailing_newline)
478                 fputc('\n', file);
479 }
480
481 static void emit_line(struct diff_options *o, const char *set, const char *reset,
482                       const char *line, int len)
483 {
484         emit_line_0(o, set, reset, line[0], line+1, len-1);
485 }
486
487 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
488 {
489         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
490               ecbdata->blank_at_eof_in_preimage &&
491               ecbdata->blank_at_eof_in_postimage &&
492               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
493               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
494                 return 0;
495         return ws_blank_line(line, len, ecbdata->ws_rule);
496 }
497
498 static void emit_line_checked(const char *reset,
499                               struct emit_callback *ecbdata,
500                               const char *line, int len,
501                               enum color_diff color,
502                               unsigned ws_error_highlight,
503                               char sign)
504 {
505         const char *set = diff_get_color(ecbdata->color_diff, color);
506         const char *ws = NULL;
507
508         if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
509                 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
510                 if (!*ws)
511                         ws = NULL;
512         }
513
514         if (!ws)
515                 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
516         else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
517                 /* Blank line at EOF - paint '+' as well */
518                 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
519         else {
520                 /* Emit just the prefix, then the rest. */
521                 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
522                 ws_check_emit(line, len, ecbdata->ws_rule,
523                               ecbdata->opt->file, set, reset, ws);
524         }
525 }
526
527 static void emit_add_line(const char *reset,
528                           struct emit_callback *ecbdata,
529                           const char *line, int len)
530 {
531         emit_line_checked(reset, ecbdata, line, len,
532                           DIFF_FILE_NEW, WSEH_NEW, '+');
533 }
534
535 static void emit_del_line(const char *reset,
536                           struct emit_callback *ecbdata,
537                           const char *line, int len)
538 {
539         emit_line_checked(reset, ecbdata, line, len,
540                           DIFF_FILE_OLD, WSEH_OLD, '-');
541 }
542
543 static void emit_context_line(const char *reset,
544                               struct emit_callback *ecbdata,
545                               const char *line, int len)
546 {
547         emit_line_checked(reset, ecbdata, line, len,
548                           DIFF_CONTEXT, WSEH_CONTEXT, ' ');
549 }
550
551 static void emit_hunk_header(struct emit_callback *ecbdata,
552                              const char *line, int len)
553 {
554         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
555         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
556         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
557         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
558         static const char atat[2] = { '@', '@' };
559         const char *cp, *ep;
560         struct strbuf msgbuf = STRBUF_INIT;
561         int org_len = len;
562         int i = 1;
563
564         /*
565          * As a hunk header must begin with "@@ -<old>, +<new> @@",
566          * it always is at least 10 bytes long.
567          */
568         if (len < 10 ||
569             memcmp(line, atat, 2) ||
570             !(ep = memmem(line + 2, len - 2, atat, 2))) {
571                 emit_line(ecbdata->opt, context, reset, line, len);
572                 return;
573         }
574         ep += 2; /* skip over @@ */
575
576         /* The hunk header in fraginfo color */
577         strbuf_addstr(&msgbuf, frag);
578         strbuf_add(&msgbuf, line, ep - line);
579         strbuf_addstr(&msgbuf, reset);
580
581         /*
582          * trailing "\r\n"
583          */
584         for ( ; i < 3; i++)
585                 if (line[len - i] == '\r' || line[len - i] == '\n')
586                         len--;
587
588         /* blank before the func header */
589         for (cp = ep; ep - line < len; ep++)
590                 if (*ep != ' ' && *ep != '\t')
591                         break;
592         if (ep != cp) {
593                 strbuf_addstr(&msgbuf, context);
594                 strbuf_add(&msgbuf, cp, ep - cp);
595                 strbuf_addstr(&msgbuf, reset);
596         }
597
598         if (ep < line + len) {
599                 strbuf_addstr(&msgbuf, func);
600                 strbuf_add(&msgbuf, ep, line + len - ep);
601                 strbuf_addstr(&msgbuf, reset);
602         }
603
604         strbuf_add(&msgbuf, line + len, org_len - len);
605         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
606         strbuf_release(&msgbuf);
607 }
608
609 static struct diff_tempfile *claim_diff_tempfile(void) {
610         int i;
611         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
612                 if (!diff_temp[i].name)
613                         return diff_temp + i;
614         die("BUG: diff is failing to clean up its tempfiles");
615 }
616
617 static void remove_tempfile(void)
618 {
619         int i;
620         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
621                 if (is_tempfile_active(&diff_temp[i].tempfile))
622                         delete_tempfile(&diff_temp[i].tempfile);
623                 diff_temp[i].name = NULL;
624         }
625 }
626
627 static void print_line_count(FILE *file, int count)
628 {
629         switch (count) {
630         case 0:
631                 fprintf(file, "0,0");
632                 break;
633         case 1:
634                 fprintf(file, "1");
635                 break;
636         default:
637                 fprintf(file, "1,%d", count);
638                 break;
639         }
640 }
641
642 static void emit_rewrite_lines(struct emit_callback *ecb,
643                                int prefix, const char *data, int size)
644 {
645         const char *endp = NULL;
646         static const char *nneof = " No newline at end of file\n";
647         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
648
649         while (0 < size) {
650                 int len;
651
652                 endp = memchr(data, '\n', size);
653                 len = endp ? (endp - data + 1) : size;
654                 if (prefix != '+') {
655                         ecb->lno_in_preimage++;
656                         emit_del_line(reset, ecb, data, len);
657                 } else {
658                         ecb->lno_in_postimage++;
659                         emit_add_line(reset, ecb, data, len);
660                 }
661                 size -= len;
662                 data += len;
663         }
664         if (!endp) {
665                 const char *context = diff_get_color(ecb->color_diff,
666                                                      DIFF_CONTEXT);
667                 putc('\n', ecb->opt->file);
668                 emit_line_0(ecb->opt, context, reset, '\\',
669                             nneof, strlen(nneof));
670         }
671 }
672
673 static void emit_rewrite_diff(const char *name_a,
674                               const char *name_b,
675                               struct diff_filespec *one,
676                               struct diff_filespec *two,
677                               struct userdiff_driver *textconv_one,
678                               struct userdiff_driver *textconv_two,
679                               struct diff_options *o)
680 {
681         int lc_a, lc_b;
682         const char *name_a_tab, *name_b_tab;
683         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
684         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
685         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
686         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
687         const char *a_prefix, *b_prefix;
688         char *data_one, *data_two;
689         size_t size_one, size_two;
690         struct emit_callback ecbdata;
691         const char *line_prefix = diff_line_prefix(o);
692
693         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
694                 a_prefix = o->b_prefix;
695                 b_prefix = o->a_prefix;
696         } else {
697                 a_prefix = o->a_prefix;
698                 b_prefix = o->b_prefix;
699         }
700
701         name_a += (*name_a == '/');
702         name_b += (*name_b == '/');
703         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
704         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
705
706         strbuf_reset(&a_name);
707         strbuf_reset(&b_name);
708         quote_two_c_style(&a_name, a_prefix, name_a, 0);
709         quote_two_c_style(&b_name, b_prefix, name_b, 0);
710
711         size_one = fill_textconv(textconv_one, one, &data_one);
712         size_two = fill_textconv(textconv_two, two, &data_two);
713
714         memset(&ecbdata, 0, sizeof(ecbdata));
715         ecbdata.color_diff = want_color(o->use_color);
716         ecbdata.found_changesp = &o->found_changes;
717         ecbdata.ws_rule = whitespace_rule(name_b);
718         ecbdata.opt = o;
719         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
720                 mmfile_t mf1, mf2;
721                 mf1.ptr = (char *)data_one;
722                 mf2.ptr = (char *)data_two;
723                 mf1.size = size_one;
724                 mf2.size = size_two;
725                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
726         }
727         ecbdata.lno_in_preimage = 1;
728         ecbdata.lno_in_postimage = 1;
729
730         lc_a = count_lines(data_one, size_one);
731         lc_b = count_lines(data_two, size_two);
732         fprintf(o->file,
733                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
734                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
735                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
736                 line_prefix, fraginfo);
737         if (!o->irreversible_delete)
738                 print_line_count(o->file, lc_a);
739         else
740                 fprintf(o->file, "?,?");
741         fprintf(o->file, " +");
742         print_line_count(o->file, lc_b);
743         fprintf(o->file, " @@%s\n", reset);
744         if (lc_a && !o->irreversible_delete)
745                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
746         if (lc_b)
747                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
748         if (textconv_one)
749                 free((char *)data_one);
750         if (textconv_two)
751                 free((char *)data_two);
752 }
753
754 struct diff_words_buffer {
755         mmfile_t text;
756         long alloc;
757         struct diff_words_orig {
758                 const char *begin, *end;
759         } *orig;
760         int orig_nr, orig_alloc;
761 };
762
763 static void diff_words_append(char *line, unsigned long len,
764                 struct diff_words_buffer *buffer)
765 {
766         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
767         line++;
768         len--;
769         memcpy(buffer->text.ptr + buffer->text.size, line, len);
770         buffer->text.size += len;
771         buffer->text.ptr[buffer->text.size] = '\0';
772 }
773
774 struct diff_words_style_elem {
775         const char *prefix;
776         const char *suffix;
777         const char *color; /* NULL; filled in by the setup code if
778                             * color is enabled */
779 };
780
781 struct diff_words_style {
782         enum diff_words_type type;
783         struct diff_words_style_elem new, old, ctx;
784         const char *newline;
785 };
786
787 static struct diff_words_style diff_words_styles[] = {
788         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
789         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
790         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
791 };
792
793 struct diff_words_data {
794         struct diff_words_buffer minus, plus;
795         const char *current_plus;
796         int last_minus;
797         struct diff_options *opt;
798         regex_t *word_regex;
799         enum diff_words_type type;
800         struct diff_words_style *style;
801 };
802
803 static int fn_out_diff_words_write_helper(FILE *fp,
804                                           struct diff_words_style_elem *st_el,
805                                           const char *newline,
806                                           size_t count, const char *buf,
807                                           const char *line_prefix)
808 {
809         int print = 0;
810
811         while (count) {
812                 char *p = memchr(buf, '\n', count);
813                 if (print)
814                         fputs(line_prefix, fp);
815                 if (p != buf) {
816                         if (st_el->color && fputs(st_el->color, fp) < 0)
817                                 return -1;
818                         if (fputs(st_el->prefix, fp) < 0 ||
819                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
820                             fputs(st_el->suffix, fp) < 0)
821                                 return -1;
822                         if (st_el->color && *st_el->color
823                             && fputs(GIT_COLOR_RESET, fp) < 0)
824                                 return -1;
825                 }
826                 if (!p)
827                         return 0;
828                 if (fputs(newline, fp) < 0)
829                         return -1;
830                 count -= p + 1 - buf;
831                 buf = p + 1;
832                 print = 1;
833         }
834         return 0;
835 }
836
837 /*
838  * '--color-words' algorithm can be described as:
839  *
840  *   1. collect a the minus/plus lines of a diff hunk, divided into
841  *      minus-lines and plus-lines;
842  *
843  *   2. break both minus-lines and plus-lines into words and
844  *      place them into two mmfile_t with one word for each line;
845  *
846  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
847  *
848  * And for the common parts of the both file, we output the plus side text.
849  * diff_words->current_plus is used to trace the current position of the plus file
850  * which printed. diff_words->last_minus is used to trace the last minus word
851  * printed.
852  *
853  * For '--graph' to work with '--color-words', we need to output the graph prefix
854  * on each line of color words output. Generally, there are two conditions on
855  * which we should output the prefix.
856  *
857  *   1. diff_words->last_minus == 0 &&
858  *      diff_words->current_plus == diff_words->plus.text.ptr
859  *
860  *      that is: the plus text must start as a new line, and if there is no minus
861  *      word printed, a graph prefix must be printed.
862  *
863  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
864  *      *(diff_words->current_plus - 1) == '\n'
865  *
866  *      that is: a graph prefix must be printed following a '\n'
867  */
868 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
869 {
870         if ((diff_words->last_minus == 0 &&
871                 diff_words->current_plus == diff_words->plus.text.ptr) ||
872                 (diff_words->current_plus > diff_words->plus.text.ptr &&
873                 *(diff_words->current_plus - 1) == '\n')) {
874                 return 1;
875         } else {
876                 return 0;
877         }
878 }
879
880 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
881 {
882         struct diff_words_data *diff_words = priv;
883         struct diff_words_style *style = diff_words->style;
884         int minus_first, minus_len, plus_first, plus_len;
885         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
886         struct diff_options *opt = diff_words->opt;
887         const char *line_prefix;
888
889         if (line[0] != '@' || parse_hunk_header(line, len,
890                         &minus_first, &minus_len, &plus_first, &plus_len))
891                 return;
892
893         assert(opt);
894         line_prefix = diff_line_prefix(opt);
895
896         /* POSIX requires that first be decremented by one if len == 0... */
897         if (minus_len) {
898                 minus_begin = diff_words->minus.orig[minus_first].begin;
899                 minus_end =
900                         diff_words->minus.orig[minus_first + minus_len - 1].end;
901         } else
902                 minus_begin = minus_end =
903                         diff_words->minus.orig[minus_first].end;
904
905         if (plus_len) {
906                 plus_begin = diff_words->plus.orig[plus_first].begin;
907                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
908         } else
909                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
910
911         if (color_words_output_graph_prefix(diff_words)) {
912                 fputs(line_prefix, diff_words->opt->file);
913         }
914         if (diff_words->current_plus != plus_begin) {
915                 fn_out_diff_words_write_helper(diff_words->opt->file,
916                                 &style->ctx, style->newline,
917                                 plus_begin - diff_words->current_plus,
918                                 diff_words->current_plus, line_prefix);
919                 if (*(plus_begin - 1) == '\n')
920                         fputs(line_prefix, diff_words->opt->file);
921         }
922         if (minus_begin != minus_end) {
923                 fn_out_diff_words_write_helper(diff_words->opt->file,
924                                 &style->old, style->newline,
925                                 minus_end - minus_begin, minus_begin,
926                                 line_prefix);
927         }
928         if (plus_begin != plus_end) {
929                 fn_out_diff_words_write_helper(diff_words->opt->file,
930                                 &style->new, style->newline,
931                                 plus_end - plus_begin, plus_begin,
932                                 line_prefix);
933         }
934
935         diff_words->current_plus = plus_end;
936         diff_words->last_minus = minus_first;
937 }
938
939 /* This function starts looking at *begin, and returns 0 iff a word was found. */
940 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
941                 int *begin, int *end)
942 {
943         if (word_regex && *begin < buffer->size) {
944                 regmatch_t match[1];
945                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
946                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
947                                         '\n', match[0].rm_eo - match[0].rm_so);
948                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
949                         *begin += match[0].rm_so;
950                         return *begin >= *end;
951                 }
952                 return -1;
953         }
954
955         /* find the next word */
956         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
957                 (*begin)++;
958         if (*begin >= buffer->size)
959                 return -1;
960
961         /* find the end of the word */
962         *end = *begin + 1;
963         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
964                 (*end)++;
965
966         return 0;
967 }
968
969 /*
970  * This function splits the words in buffer->text, stores the list with
971  * newline separator into out, and saves the offsets of the original words
972  * in buffer->orig.
973  */
974 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
975                 regex_t *word_regex)
976 {
977         int i, j;
978         long alloc = 0;
979
980         out->size = 0;
981         out->ptr = NULL;
982
983         /* fake an empty "0th" word */
984         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
985         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
986         buffer->orig_nr = 1;
987
988         for (i = 0; i < buffer->text.size; i++) {
989                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
990                         return;
991
992                 /* store original boundaries */
993                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
994                                 buffer->orig_alloc);
995                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
996                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
997                 buffer->orig_nr++;
998
999                 /* store one word */
1000                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1001                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1002                 out->ptr[out->size + j - i] = '\n';
1003                 out->size += j - i + 1;
1004
1005                 i = j - 1;
1006         }
1007 }
1008
1009 /* this executes the word diff on the accumulated buffers */
1010 static void diff_words_show(struct diff_words_data *diff_words)
1011 {
1012         xpparam_t xpp;
1013         xdemitconf_t xecfg;
1014         mmfile_t minus, plus;
1015         struct diff_words_style *style = diff_words->style;
1016
1017         struct diff_options *opt = diff_words->opt;
1018         const char *line_prefix;
1019
1020         assert(opt);
1021         line_prefix = diff_line_prefix(opt);
1022
1023         /* special case: only removal */
1024         if (!diff_words->plus.text.size) {
1025                 fputs(line_prefix, diff_words->opt->file);
1026                 fn_out_diff_words_write_helper(diff_words->opt->file,
1027                         &style->old, style->newline,
1028                         diff_words->minus.text.size,
1029                         diff_words->minus.text.ptr, line_prefix);
1030                 diff_words->minus.text.size = 0;
1031                 return;
1032         }
1033
1034         diff_words->current_plus = diff_words->plus.text.ptr;
1035         diff_words->last_minus = 0;
1036
1037         memset(&xpp, 0, sizeof(xpp));
1038         memset(&xecfg, 0, sizeof(xecfg));
1039         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1040         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1041         xpp.flags = 0;
1042         /* as only the hunk header will be parsed, we need a 0-context */
1043         xecfg.ctxlen = 0;
1044         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1045                           &xpp, &xecfg))
1046                 die("unable to generate word diff");
1047         free(minus.ptr);
1048         free(plus.ptr);
1049         if (diff_words->current_plus != diff_words->plus.text.ptr +
1050                         diff_words->plus.text.size) {
1051                 if (color_words_output_graph_prefix(diff_words))
1052                         fputs(line_prefix, diff_words->opt->file);
1053                 fn_out_diff_words_write_helper(diff_words->opt->file,
1054                         &style->ctx, style->newline,
1055                         diff_words->plus.text.ptr + diff_words->plus.text.size
1056                         - diff_words->current_plus, diff_words->current_plus,
1057                         line_prefix);
1058         }
1059         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1060 }
1061
1062 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1063 static void diff_words_flush(struct emit_callback *ecbdata)
1064 {
1065         if (ecbdata->diff_words->minus.text.size ||
1066             ecbdata->diff_words->plus.text.size)
1067                 diff_words_show(ecbdata->diff_words);
1068 }
1069
1070 static void diff_filespec_load_driver(struct diff_filespec *one)
1071 {
1072         /* Use already-loaded driver */
1073         if (one->driver)
1074                 return;
1075
1076         if (S_ISREG(one->mode))
1077                 one->driver = userdiff_find_by_path(one->path);
1078
1079         /* Fallback to default settings */
1080         if (!one->driver)
1081                 one->driver = userdiff_find_by_name("default");
1082 }
1083
1084 static const char *userdiff_word_regex(struct diff_filespec *one)
1085 {
1086         diff_filespec_load_driver(one);
1087         return one->driver->word_regex;
1088 }
1089
1090 static void init_diff_words_data(struct emit_callback *ecbdata,
1091                                  struct diff_options *orig_opts,
1092                                  struct diff_filespec *one,
1093                                  struct diff_filespec *two)
1094 {
1095         int i;
1096         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1097         memcpy(o, orig_opts, sizeof(struct diff_options));
1098
1099         ecbdata->diff_words =
1100                 xcalloc(1, sizeof(struct diff_words_data));
1101         ecbdata->diff_words->type = o->word_diff;
1102         ecbdata->diff_words->opt = o;
1103         if (!o->word_regex)
1104                 o->word_regex = userdiff_word_regex(one);
1105         if (!o->word_regex)
1106                 o->word_regex = userdiff_word_regex(two);
1107         if (!o->word_regex)
1108                 o->word_regex = diff_word_regex_cfg;
1109         if (o->word_regex) {
1110                 ecbdata->diff_words->word_regex = (regex_t *)
1111                         xmalloc(sizeof(regex_t));
1112                 if (regcomp(ecbdata->diff_words->word_regex,
1113                             o->word_regex,
1114                             REG_EXTENDED | REG_NEWLINE))
1115                         die ("Invalid regular expression: %s",
1116                              o->word_regex);
1117         }
1118         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1119                 if (o->word_diff == diff_words_styles[i].type) {
1120                         ecbdata->diff_words->style =
1121                                 &diff_words_styles[i];
1122                         break;
1123                 }
1124         }
1125         if (want_color(o->use_color)) {
1126                 struct diff_words_style *st = ecbdata->diff_words->style;
1127                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1128                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1129                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1130         }
1131 }
1132
1133 static void free_diff_words_data(struct emit_callback *ecbdata)
1134 {
1135         if (ecbdata->diff_words) {
1136                 diff_words_flush(ecbdata);
1137                 free (ecbdata->diff_words->opt);
1138                 free (ecbdata->diff_words->minus.text.ptr);
1139                 free (ecbdata->diff_words->minus.orig);
1140                 free (ecbdata->diff_words->plus.text.ptr);
1141                 free (ecbdata->diff_words->plus.orig);
1142                 if (ecbdata->diff_words->word_regex) {
1143                         regfree(ecbdata->diff_words->word_regex);
1144                         free(ecbdata->diff_words->word_regex);
1145                 }
1146                 free(ecbdata->diff_words);
1147                 ecbdata->diff_words = NULL;
1148         }
1149 }
1150
1151 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1152 {
1153         if (want_color(diff_use_color))
1154                 return diff_colors[ix];
1155         return "";
1156 }
1157
1158 const char *diff_line_prefix(struct diff_options *opt)
1159 {
1160         struct strbuf *msgbuf;
1161         if (!opt->output_prefix)
1162                 return "";
1163
1164         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1165         return msgbuf->buf;
1166 }
1167
1168 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1169 {
1170         const char *cp;
1171         unsigned long allot;
1172         size_t l = len;
1173
1174         if (ecb->truncate)
1175                 return ecb->truncate(line, len);
1176         cp = line;
1177         allot = l;
1178         while (0 < l) {
1179                 (void) utf8_width(&cp, &l);
1180                 if (!cp)
1181                         break; /* truncated in the middle? */
1182         }
1183         return allot - l;
1184 }
1185
1186 static void find_lno(const char *line, struct emit_callback *ecbdata)
1187 {
1188         const char *p;
1189         ecbdata->lno_in_preimage = 0;
1190         ecbdata->lno_in_postimage = 0;
1191         p = strchr(line, '-');
1192         if (!p)
1193                 return; /* cannot happen */
1194         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1195         p = strchr(p, '+');
1196         if (!p)
1197                 return; /* cannot happen */
1198         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1199 }
1200
1201 static void fn_out_consume(void *priv, char *line, unsigned long len)
1202 {
1203         struct emit_callback *ecbdata = priv;
1204         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1205         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1206         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1207         struct diff_options *o = ecbdata->opt;
1208         const char *line_prefix = diff_line_prefix(o);
1209
1210         if (ecbdata->header) {
1211                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1212                 strbuf_reset(ecbdata->header);
1213                 ecbdata->header = NULL;
1214         }
1215         *(ecbdata->found_changesp) = 1;
1216
1217         if (ecbdata->label_path[0]) {
1218                 const char *name_a_tab, *name_b_tab;
1219
1220                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1221                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1222
1223                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1224                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1225                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1226                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1227                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1228         }
1229
1230         if (diff_suppress_blank_empty
1231             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1232                 line[0] = '\n';
1233                 len = 1;
1234         }
1235
1236         if (line[0] == '@') {
1237                 if (ecbdata->diff_words)
1238                         diff_words_flush(ecbdata);
1239                 len = sane_truncate_line(ecbdata, line, len);
1240                 find_lno(line, ecbdata);
1241                 emit_hunk_header(ecbdata, line, len);
1242                 if (line[len-1] != '\n')
1243                         putc('\n', ecbdata->opt->file);
1244                 return;
1245         }
1246
1247         if (len < 1) {
1248                 emit_line(ecbdata->opt, reset, reset, line, len);
1249                 if (ecbdata->diff_words
1250                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1251                         fputs("~\n", ecbdata->opt->file);
1252                 return;
1253         }
1254
1255         if (ecbdata->diff_words) {
1256                 if (line[0] == '-') {
1257                         diff_words_append(line, len,
1258                                           &ecbdata->diff_words->minus);
1259                         return;
1260                 } else if (line[0] == '+') {
1261                         diff_words_append(line, len,
1262                                           &ecbdata->diff_words->plus);
1263                         return;
1264                 } else if (starts_with(line, "\\ ")) {
1265                         /*
1266                          * Eat the "no newline at eof" marker as if we
1267                          * saw a "+" or "-" line with nothing on it,
1268                          * and return without diff_words_flush() to
1269                          * defer processing. If this is the end of
1270                          * preimage, more "+" lines may come after it.
1271                          */
1272                         return;
1273                 }
1274                 diff_words_flush(ecbdata);
1275                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1276                         emit_line(ecbdata->opt, context, reset, line, len);
1277                         fputs("~\n", ecbdata->opt->file);
1278                 } else {
1279                         /*
1280                          * Skip the prefix character, if any.  With
1281                          * diff_suppress_blank_empty, there may be
1282                          * none.
1283                          */
1284                         if (line[0] != '\n') {
1285                               line++;
1286                               len--;
1287                         }
1288                         emit_line(ecbdata->opt, context, reset, line, len);
1289                 }
1290                 return;
1291         }
1292
1293         switch (line[0]) {
1294         case '+':
1295                 ecbdata->lno_in_postimage++;
1296                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1297                 break;
1298         case '-':
1299                 ecbdata->lno_in_preimage++;
1300                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1301                 break;
1302         case ' ':
1303                 ecbdata->lno_in_postimage++;
1304                 ecbdata->lno_in_preimage++;
1305                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1306                 break;
1307         default:
1308                 /* incomplete line at the end */
1309                 ecbdata->lno_in_preimage++;
1310                 emit_line(ecbdata->opt,
1311                           diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1312                           reset, line, len);
1313                 break;
1314         }
1315 }
1316
1317 static char *pprint_rename(const char *a, const char *b)
1318 {
1319         const char *old = a;
1320         const char *new = b;
1321         struct strbuf name = STRBUF_INIT;
1322         int pfx_length, sfx_length;
1323         int pfx_adjust_for_slash;
1324         int len_a = strlen(a);
1325         int len_b = strlen(b);
1326         int a_midlen, b_midlen;
1327         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1328         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1329
1330         if (qlen_a || qlen_b) {
1331                 quote_c_style(a, &name, NULL, 0);
1332                 strbuf_addstr(&name, " => ");
1333                 quote_c_style(b, &name, NULL, 0);
1334                 return strbuf_detach(&name, NULL);
1335         }
1336
1337         /* Find common prefix */
1338         pfx_length = 0;
1339         while (*old && *new && *old == *new) {
1340                 if (*old == '/')
1341                         pfx_length = old - a + 1;
1342                 old++;
1343                 new++;
1344         }
1345
1346         /* Find common suffix */
1347         old = a + len_a;
1348         new = b + len_b;
1349         sfx_length = 0;
1350         /*
1351          * If there is a common prefix, it must end in a slash.  In
1352          * that case we let this loop run 1 into the prefix to see the
1353          * same slash.
1354          *
1355          * If there is no common prefix, we cannot do this as it would
1356          * underrun the input strings.
1357          */
1358         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1359         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1360                b + pfx_length - pfx_adjust_for_slash <= new &&
1361                *old == *new) {
1362                 if (*old == '/')
1363                         sfx_length = len_a - (old - a);
1364                 old--;
1365                 new--;
1366         }
1367
1368         /*
1369          * pfx{mid-a => mid-b}sfx
1370          * {pfx-a => pfx-b}sfx
1371          * pfx{sfx-a => sfx-b}
1372          * name-a => name-b
1373          */
1374         a_midlen = len_a - pfx_length - sfx_length;
1375         b_midlen = len_b - pfx_length - sfx_length;
1376         if (a_midlen < 0)
1377                 a_midlen = 0;
1378         if (b_midlen < 0)
1379                 b_midlen = 0;
1380
1381         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1382         if (pfx_length + sfx_length) {
1383                 strbuf_add(&name, a, pfx_length);
1384                 strbuf_addch(&name, '{');
1385         }
1386         strbuf_add(&name, a + pfx_length, a_midlen);
1387         strbuf_addstr(&name, " => ");
1388         strbuf_add(&name, b + pfx_length, b_midlen);
1389         if (pfx_length + sfx_length) {
1390                 strbuf_addch(&name, '}');
1391                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1392         }
1393         return strbuf_detach(&name, NULL);
1394 }
1395
1396 struct diffstat_t {
1397         int nr;
1398         int alloc;
1399         struct diffstat_file {
1400                 char *from_name;
1401                 char *name;
1402                 char *print_name;
1403                 unsigned is_unmerged:1;
1404                 unsigned is_binary:1;
1405                 unsigned is_renamed:1;
1406                 unsigned is_interesting:1;
1407                 uintmax_t added, deleted;
1408         } **files;
1409 };
1410
1411 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1412                                           const char *name_a,
1413                                           const char *name_b)
1414 {
1415         struct diffstat_file *x;
1416         x = xcalloc(1, sizeof(*x));
1417         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1418         diffstat->files[diffstat->nr++] = x;
1419         if (name_b) {
1420                 x->from_name = xstrdup(name_a);
1421                 x->name = xstrdup(name_b);
1422                 x->is_renamed = 1;
1423         }
1424         else {
1425                 x->from_name = NULL;
1426                 x->name = xstrdup(name_a);
1427         }
1428         return x;
1429 }
1430
1431 static void diffstat_consume(void *priv, char *line, unsigned long len)
1432 {
1433         struct diffstat_t *diffstat = priv;
1434         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1435
1436         if (line[0] == '+')
1437                 x->added++;
1438         else if (line[0] == '-')
1439                 x->deleted++;
1440 }
1441
1442 const char mime_boundary_leader[] = "------------";
1443
1444 static int scale_linear(int it, int width, int max_change)
1445 {
1446         if (!it)
1447                 return 0;
1448         /*
1449          * make sure that at least one '-' or '+' is printed if
1450          * there is any change to this path. The easiest way is to
1451          * scale linearly as if the alloted width is one column shorter
1452          * than it is, and then add 1 to the result.
1453          */
1454         return 1 + (it * (width - 1) / max_change);
1455 }
1456
1457 static void show_name(FILE *file,
1458                       const char *prefix, const char *name, int len)
1459 {
1460         fprintf(file, " %s%-*s |", prefix, len, name);
1461 }
1462
1463 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1464 {
1465         if (cnt <= 0)
1466                 return;
1467         fprintf(file, "%s", set);
1468         while (cnt--)
1469                 putc(ch, file);
1470         fprintf(file, "%s", reset);
1471 }
1472
1473 static void fill_print_name(struct diffstat_file *file)
1474 {
1475         char *pname;
1476
1477         if (file->print_name)
1478                 return;
1479
1480         if (!file->is_renamed) {
1481                 struct strbuf buf = STRBUF_INIT;
1482                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1483                         pname = strbuf_detach(&buf, NULL);
1484                 } else {
1485                         pname = file->name;
1486                         strbuf_release(&buf);
1487                 }
1488         } else {
1489                 pname = pprint_rename(file->from_name, file->name);
1490         }
1491         file->print_name = pname;
1492 }
1493
1494 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1495 {
1496         struct strbuf sb = STRBUF_INIT;
1497         int ret;
1498
1499         if (!files) {
1500                 assert(insertions == 0 && deletions == 0);
1501                 return fprintf(fp, "%s\n", " 0 files changed");
1502         }
1503
1504         strbuf_addf(&sb,
1505                     (files == 1) ? " %d file changed" : " %d files changed",
1506                     files);
1507
1508         /*
1509          * For binary diff, the caller may want to print "x files
1510          * changed" with insertions == 0 && deletions == 0.
1511          *
1512          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1513          * is probably less confusing (i.e skip over "2 files changed
1514          * but nothing about added/removed lines? Is this a bug in Git?").
1515          */
1516         if (insertions || deletions == 0) {
1517                 strbuf_addf(&sb,
1518                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1519                             insertions);
1520         }
1521
1522         if (deletions || insertions == 0) {
1523                 strbuf_addf(&sb,
1524                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1525                             deletions);
1526         }
1527         strbuf_addch(&sb, '\n');
1528         ret = fputs(sb.buf, fp);
1529         strbuf_release(&sb);
1530         return ret;
1531 }
1532
1533 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1534 {
1535         int i, len, add, del, adds = 0, dels = 0;
1536         uintmax_t max_change = 0, max_len = 0;
1537         int total_files = data->nr, count;
1538         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1539         const char *reset, *add_c, *del_c;
1540         const char *line_prefix = "";
1541         int extra_shown = 0;
1542
1543         if (data->nr == 0)
1544                 return;
1545
1546         line_prefix = diff_line_prefix(options);
1547         count = options->stat_count ? options->stat_count : data->nr;
1548
1549         reset = diff_get_color_opt(options, DIFF_RESET);
1550         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1551         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1552
1553         /*
1554          * Find the longest filename and max number of changes
1555          */
1556         for (i = 0; (i < count) && (i < data->nr); i++) {
1557                 struct diffstat_file *file = data->files[i];
1558                 uintmax_t change = file->added + file->deleted;
1559
1560                 if (!file->is_interesting && (change == 0)) {
1561                         count++; /* not shown == room for one more */
1562                         continue;
1563                 }
1564                 fill_print_name(file);
1565                 len = strlen(file->print_name);
1566                 if (max_len < len)
1567                         max_len = len;
1568
1569                 if (file->is_unmerged) {
1570                         /* "Unmerged" is 8 characters */
1571                         bin_width = bin_width < 8 ? 8 : bin_width;
1572                         continue;
1573                 }
1574                 if (file->is_binary) {
1575                         /* "Bin XXX -> YYY bytes" */
1576                         int w = 14 + decimal_width(file->added)
1577                                 + decimal_width(file->deleted);
1578                         bin_width = bin_width < w ? w : bin_width;
1579                         /* Display change counts aligned with "Bin" */
1580                         number_width = 3;
1581                         continue;
1582                 }
1583
1584                 if (max_change < change)
1585                         max_change = change;
1586         }
1587         count = i; /* where we can stop scanning in data->files[] */
1588
1589         /*
1590          * We have width = stat_width or term_columns() columns total.
1591          * We want a maximum of min(max_len, stat_name_width) for the name part.
1592          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1593          * We also need 1 for " " and 4 + decimal_width(max_change)
1594          * for " | NNNN " and one the empty column at the end, altogether
1595          * 6 + decimal_width(max_change).
1596          *
1597          * If there's not enough space, we will use the smaller of
1598          * stat_name_width (if set) and 5/8*width for the filename,
1599          * and the rest for constant elements + graph part, but no more
1600          * than stat_graph_width for the graph part.
1601          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1602          * for the standard terminal size).
1603          *
1604          * In other words: stat_width limits the maximum width, and
1605          * stat_name_width fixes the maximum width of the filename,
1606          * and is also used to divide available columns if there
1607          * aren't enough.
1608          *
1609          * Binary files are displayed with "Bin XXX -> YYY bytes"
1610          * instead of the change count and graph. This part is treated
1611          * similarly to the graph part, except that it is not
1612          * "scaled". If total width is too small to accommodate the
1613          * guaranteed minimum width of the filename part and the
1614          * separators and this message, this message will "overflow"
1615          * making the line longer than the maximum width.
1616          */
1617
1618         if (options->stat_width == -1)
1619                 width = term_columns() - options->output_prefix_length;
1620         else
1621                 width = options->stat_width ? options->stat_width : 80;
1622         number_width = decimal_width(max_change) > number_width ?
1623                 decimal_width(max_change) : number_width;
1624
1625         if (options->stat_graph_width == -1)
1626                 options->stat_graph_width = diff_stat_graph_width;
1627
1628         /*
1629          * Guarantee 3/8*16==6 for the graph part
1630          * and 5/8*16==10 for the filename part
1631          */
1632         if (width < 16 + 6 + number_width)
1633                 width = 16 + 6 + number_width;
1634
1635         /*
1636          * First assign sizes that are wanted, ignoring available width.
1637          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1638          * starting from "XXX" should fit in graph_width.
1639          */
1640         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1641         if (options->stat_graph_width &&
1642             options->stat_graph_width < graph_width)
1643                 graph_width = options->stat_graph_width;
1644
1645         name_width = (options->stat_name_width > 0 &&
1646                       options->stat_name_width < max_len) ?
1647                 options->stat_name_width : max_len;
1648
1649         /*
1650          * Adjust adjustable widths not to exceed maximum width
1651          */
1652         if (name_width + number_width + 6 + graph_width > width) {
1653                 if (graph_width > width * 3/8 - number_width - 6) {
1654                         graph_width = width * 3/8 - number_width - 6;
1655                         if (graph_width < 6)
1656                                 graph_width = 6;
1657                 }
1658
1659                 if (options->stat_graph_width &&
1660                     graph_width > options->stat_graph_width)
1661                         graph_width = options->stat_graph_width;
1662                 if (name_width > width - number_width - 6 - graph_width)
1663                         name_width = width - number_width - 6 - graph_width;
1664                 else
1665                         graph_width = width - number_width - 6 - name_width;
1666         }
1667
1668         /*
1669          * From here name_width is the width of the name area,
1670          * and graph_width is the width of the graph area.
1671          * max_change is used to scale graph properly.
1672          */
1673         for (i = 0; i < count; i++) {
1674                 const char *prefix = "";
1675                 struct diffstat_file *file = data->files[i];
1676                 char *name = file->print_name;
1677                 uintmax_t added = file->added;
1678                 uintmax_t deleted = file->deleted;
1679                 int name_len;
1680
1681                 if (!file->is_interesting && (added + deleted == 0))
1682                         continue;
1683
1684                 /*
1685                  * "scale" the filename
1686                  */
1687                 len = name_width;
1688                 name_len = strlen(name);
1689                 if (name_width < name_len) {
1690                         char *slash;
1691                         prefix = "...";
1692                         len -= 3;
1693                         name += name_len - len;
1694                         slash = strchr(name, '/');
1695                         if (slash)
1696                                 name = slash;
1697                 }
1698
1699                 if (file->is_binary) {
1700                         fprintf(options->file, "%s", line_prefix);
1701                         show_name(options->file, prefix, name, len);
1702                         fprintf(options->file, " %*s", number_width, "Bin");
1703                         if (!added && !deleted) {
1704                                 putc('\n', options->file);
1705                                 continue;
1706                         }
1707                         fprintf(options->file, " %s%"PRIuMAX"%s",
1708                                 del_c, deleted, reset);
1709                         fprintf(options->file, " -> ");
1710                         fprintf(options->file, "%s%"PRIuMAX"%s",
1711                                 add_c, added, reset);
1712                         fprintf(options->file, " bytes");
1713                         fprintf(options->file, "\n");
1714                         continue;
1715                 }
1716                 else if (file->is_unmerged) {
1717                         fprintf(options->file, "%s", line_prefix);
1718                         show_name(options->file, prefix, name, len);
1719                         fprintf(options->file, " Unmerged\n");
1720                         continue;
1721                 }
1722
1723                 /*
1724                  * scale the add/delete
1725                  */
1726                 add = added;
1727                 del = deleted;
1728
1729                 if (graph_width <= max_change) {
1730                         int total = scale_linear(add + del, graph_width, max_change);
1731                         if (total < 2 && add && del)
1732                                 /* width >= 2 due to the sanity check */
1733                                 total = 2;
1734                         if (add < del) {
1735                                 add = scale_linear(add, graph_width, max_change);
1736                                 del = total - add;
1737                         } else {
1738                                 del = scale_linear(del, graph_width, max_change);
1739                                 add = total - del;
1740                         }
1741                 }
1742                 fprintf(options->file, "%s", line_prefix);
1743                 show_name(options->file, prefix, name, len);
1744                 fprintf(options->file, " %*"PRIuMAX"%s",
1745                         number_width, added + deleted,
1746                         added + deleted ? " " : "");
1747                 show_graph(options->file, '+', add, add_c, reset);
1748                 show_graph(options->file, '-', del, del_c, reset);
1749                 fprintf(options->file, "\n");
1750         }
1751
1752         for (i = 0; i < data->nr; i++) {
1753                 struct diffstat_file *file = data->files[i];
1754                 uintmax_t added = file->added;
1755                 uintmax_t deleted = file->deleted;
1756
1757                 if (file->is_unmerged ||
1758                     (!file->is_interesting && (added + deleted == 0))) {
1759                         total_files--;
1760                         continue;
1761                 }
1762
1763                 if (!file->is_binary) {
1764                         adds += added;
1765                         dels += deleted;
1766                 }
1767                 if (i < count)
1768                         continue;
1769                 if (!extra_shown)
1770                         fprintf(options->file, "%s ...\n", line_prefix);
1771                 extra_shown = 1;
1772         }
1773         fprintf(options->file, "%s", line_prefix);
1774         print_stat_summary(options->file, total_files, adds, dels);
1775 }
1776
1777 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1778 {
1779         int i, adds = 0, dels = 0, total_files = data->nr;
1780
1781         if (data->nr == 0)
1782                 return;
1783
1784         for (i = 0; i < data->nr; i++) {
1785                 int added = data->files[i]->added;
1786                 int deleted= data->files[i]->deleted;
1787
1788                 if (data->files[i]->is_unmerged ||
1789                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1790                         total_files--;
1791                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1792                         adds += added;
1793                         dels += deleted;
1794                 }
1795         }
1796         fprintf(options->file, "%s", diff_line_prefix(options));
1797         print_stat_summary(options->file, total_files, adds, dels);
1798 }
1799
1800 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1801 {
1802         int i;
1803
1804         if (data->nr == 0)
1805                 return;
1806
1807         for (i = 0; i < data->nr; i++) {
1808                 struct diffstat_file *file = data->files[i];
1809
1810                 fprintf(options->file, "%s", diff_line_prefix(options));
1811
1812                 if (file->is_binary)
1813                         fprintf(options->file, "-\t-\t");
1814                 else
1815                         fprintf(options->file,
1816                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1817                                 file->added, file->deleted);
1818                 if (options->line_termination) {
1819                         fill_print_name(file);
1820                         if (!file->is_renamed)
1821                                 write_name_quoted(file->name, options->file,
1822                                                   options->line_termination);
1823                         else {
1824                                 fputs(file->print_name, options->file);
1825                                 putc(options->line_termination, options->file);
1826                         }
1827                 } else {
1828                         if (file->is_renamed) {
1829                                 putc('\0', options->file);
1830                                 write_name_quoted(file->from_name, options->file, '\0');
1831                         }
1832                         write_name_quoted(file->name, options->file, '\0');
1833                 }
1834         }
1835 }
1836
1837 struct dirstat_file {
1838         const char *name;
1839         unsigned long changed;
1840 };
1841
1842 struct dirstat_dir {
1843         struct dirstat_file *files;
1844         int alloc, nr, permille, cumulative;
1845 };
1846
1847 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1848                 unsigned long changed, const char *base, int baselen)
1849 {
1850         unsigned long this_dir = 0;
1851         unsigned int sources = 0;
1852         const char *line_prefix = diff_line_prefix(opt);
1853
1854         while (dir->nr) {
1855                 struct dirstat_file *f = dir->files;
1856                 int namelen = strlen(f->name);
1857                 unsigned long this;
1858                 char *slash;
1859
1860                 if (namelen < baselen)
1861                         break;
1862                 if (memcmp(f->name, base, baselen))
1863                         break;
1864                 slash = strchr(f->name + baselen, '/');
1865                 if (slash) {
1866                         int newbaselen = slash + 1 - f->name;
1867                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1868                         sources++;
1869                 } else {
1870                         this = f->changed;
1871                         dir->files++;
1872                         dir->nr--;
1873                         sources += 2;
1874                 }
1875                 this_dir += this;
1876         }
1877
1878         /*
1879          * We don't report dirstat's for
1880          *  - the top level
1881          *  - or cases where everything came from a single directory
1882          *    under this directory (sources == 1).
1883          */
1884         if (baselen && sources != 1) {
1885                 if (this_dir) {
1886                         int permille = this_dir * 1000 / changed;
1887                         if (permille >= dir->permille) {
1888                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1889                                         permille / 10, permille % 10, baselen, base);
1890                                 if (!dir->cumulative)
1891                                         return 0;
1892                         }
1893                 }
1894         }
1895         return this_dir;
1896 }
1897
1898 static int dirstat_compare(const void *_a, const void *_b)
1899 {
1900         const struct dirstat_file *a = _a;
1901         const struct dirstat_file *b = _b;
1902         return strcmp(a->name, b->name);
1903 }
1904
1905 static void show_dirstat(struct diff_options *options)
1906 {
1907         int i;
1908         unsigned long changed;
1909         struct dirstat_dir dir;
1910         struct diff_queue_struct *q = &diff_queued_diff;
1911
1912         dir.files = NULL;
1913         dir.alloc = 0;
1914         dir.nr = 0;
1915         dir.permille = options->dirstat_permille;
1916         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1917
1918         changed = 0;
1919         for (i = 0; i < q->nr; i++) {
1920                 struct diff_filepair *p = q->queue[i];
1921                 const char *name;
1922                 unsigned long copied, added, damage;
1923                 int content_changed;
1924
1925                 name = p->two->path ? p->two->path : p->one->path;
1926
1927                 if (p->one->sha1_valid && p->two->sha1_valid)
1928                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1929                 else
1930                         content_changed = 1;
1931
1932                 if (!content_changed) {
1933                         /*
1934                          * The SHA1 has not changed, so pre-/post-content is
1935                          * identical. We can therefore skip looking at the
1936                          * file contents altogether.
1937                          */
1938                         damage = 0;
1939                         goto found_damage;
1940                 }
1941
1942                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1943                         /*
1944                          * In --dirstat-by-file mode, we don't really need to
1945                          * look at the actual file contents at all.
1946                          * The fact that the SHA1 changed is enough for us to
1947                          * add this file to the list of results
1948                          * (with each file contributing equal damage).
1949                          */
1950                         damage = 1;
1951                         goto found_damage;
1952                 }
1953
1954                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1955                         diff_populate_filespec(p->one, 0);
1956                         diff_populate_filespec(p->two, 0);
1957                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1958                                                &copied, &added);
1959                         diff_free_filespec_data(p->one);
1960                         diff_free_filespec_data(p->two);
1961                 } else if (DIFF_FILE_VALID(p->one)) {
1962                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1963                         copied = added = 0;
1964                         diff_free_filespec_data(p->one);
1965                 } else if (DIFF_FILE_VALID(p->two)) {
1966                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1967                         copied = 0;
1968                         added = p->two->size;
1969                         diff_free_filespec_data(p->two);
1970                 } else
1971                         continue;
1972
1973                 /*
1974                  * Original minus copied is the removed material,
1975                  * added is the new material.  They are both damages
1976                  * made to the preimage.
1977                  * If the resulting damage is zero, we know that
1978                  * diffcore_count_changes() considers the two entries to
1979                  * be identical, but since content_changed is true, we
1980                  * know that there must have been _some_ kind of change,
1981                  * so we force all entries to have damage > 0.
1982                  */
1983                 damage = (p->one->size - copied) + added;
1984                 if (!damage)
1985                         damage = 1;
1986
1987 found_damage:
1988                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1989                 dir.files[dir.nr].name = name;
1990                 dir.files[dir.nr].changed = damage;
1991                 changed += damage;
1992                 dir.nr++;
1993         }
1994
1995         /* This can happen even with many files, if everything was renames */
1996         if (!changed)
1997                 return;
1998
1999         /* Show all directories with more than x% of the changes */
2000         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2001         gather_dirstat(options, &dir, changed, "", 0);
2002 }
2003
2004 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2005 {
2006         int i;
2007         unsigned long changed;
2008         struct dirstat_dir dir;
2009
2010         if (data->nr == 0)
2011                 return;
2012
2013         dir.files = NULL;
2014         dir.alloc = 0;
2015         dir.nr = 0;
2016         dir.permille = options->dirstat_permille;
2017         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2018
2019         changed = 0;
2020         for (i = 0; i < data->nr; i++) {
2021                 struct diffstat_file *file = data->files[i];
2022                 unsigned long damage = file->added + file->deleted;
2023                 if (file->is_binary)
2024                         /*
2025                          * binary files counts bytes, not lines. Must find some
2026                          * way to normalize binary bytes vs. textual lines.
2027                          * The following heuristic assumes that there are 64
2028                          * bytes per "line".
2029                          * This is stupid and ugly, but very cheap...
2030                          */
2031                         damage = (damage + 63) / 64;
2032                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2033                 dir.files[dir.nr].name = file->name;
2034                 dir.files[dir.nr].changed = damage;
2035                 changed += damage;
2036                 dir.nr++;
2037         }
2038
2039         /* This can happen even with many files, if everything was renames */
2040         if (!changed)
2041                 return;
2042
2043         /* Show all directories with more than x% of the changes */
2044         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2045         gather_dirstat(options, &dir, changed, "", 0);
2046 }
2047
2048 static void free_diffstat_info(struct diffstat_t *diffstat)
2049 {
2050         int i;
2051         for (i = 0; i < diffstat->nr; i++) {
2052                 struct diffstat_file *f = diffstat->files[i];
2053                 if (f->name != f->print_name)
2054                         free(f->print_name);
2055                 free(f->name);
2056                 free(f->from_name);
2057                 free(f);
2058         }
2059         free(diffstat->files);
2060 }
2061
2062 struct checkdiff_t {
2063         const char *filename;
2064         int lineno;
2065         int conflict_marker_size;
2066         struct diff_options *o;
2067         unsigned ws_rule;
2068         unsigned status;
2069 };
2070
2071 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2072 {
2073         char firstchar;
2074         int cnt;
2075
2076         if (len < marker_size + 1)
2077                 return 0;
2078         firstchar = line[0];
2079         switch (firstchar) {
2080         case '=': case '>': case '<': case '|':
2081                 break;
2082         default:
2083                 return 0;
2084         }
2085         for (cnt = 1; cnt < marker_size; cnt++)
2086                 if (line[cnt] != firstchar)
2087                         return 0;
2088         /* line[1] thru line[marker_size-1] are same as firstchar */
2089         if (len < marker_size + 1 || !isspace(line[marker_size]))
2090                 return 0;
2091         return 1;
2092 }
2093
2094 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2095 {
2096         struct checkdiff_t *data = priv;
2097         int marker_size = data->conflict_marker_size;
2098         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2099         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2100         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2101         char *err;
2102         const char *line_prefix;
2103
2104         assert(data->o);
2105         line_prefix = diff_line_prefix(data->o);
2106
2107         if (line[0] == '+') {
2108                 unsigned bad;
2109                 data->lineno++;
2110                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2111                         data->status |= 1;
2112                         fprintf(data->o->file,
2113                                 "%s%s:%d: leftover conflict marker\n",
2114                                 line_prefix, data->filename, data->lineno);
2115                 }
2116                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2117                 if (!bad)
2118                         return;
2119                 data->status |= bad;
2120                 err = whitespace_error_string(bad);
2121                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2122                         line_prefix, data->filename, data->lineno, err);
2123                 free(err);
2124                 emit_line(data->o, set, reset, line, 1);
2125                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2126                               data->o->file, set, reset, ws);
2127         } else if (line[0] == ' ') {
2128                 data->lineno++;
2129         } else if (line[0] == '@') {
2130                 char *plus = strchr(line, '+');
2131                 if (plus)
2132                         data->lineno = strtol(plus, NULL, 10) - 1;
2133                 else
2134                         die("invalid diff");
2135         }
2136 }
2137
2138 static unsigned char *deflate_it(char *data,
2139                                  unsigned long size,
2140                                  unsigned long *result_size)
2141 {
2142         int bound;
2143         unsigned char *deflated;
2144         git_zstream stream;
2145
2146         git_deflate_init(&stream, zlib_compression_level);
2147         bound = git_deflate_bound(&stream, size);
2148         deflated = xmalloc(bound);
2149         stream.next_out = deflated;
2150         stream.avail_out = bound;
2151
2152         stream.next_in = (unsigned char *)data;
2153         stream.avail_in = size;
2154         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2155                 ; /* nothing */
2156         git_deflate_end(&stream);
2157         *result_size = stream.total_out;
2158         return deflated;
2159 }
2160
2161 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2162                                   const char *prefix)
2163 {
2164         void *cp;
2165         void *delta;
2166         void *deflated;
2167         void *data;
2168         unsigned long orig_size;
2169         unsigned long delta_size;
2170         unsigned long deflate_size;
2171         unsigned long data_size;
2172
2173         /* We could do deflated delta, or we could do just deflated two,
2174          * whichever is smaller.
2175          */
2176         delta = NULL;
2177         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2178         if (one->size && two->size) {
2179                 delta = diff_delta(one->ptr, one->size,
2180                                    two->ptr, two->size,
2181                                    &delta_size, deflate_size);
2182                 if (delta) {
2183                         void *to_free = delta;
2184                         orig_size = delta_size;
2185                         delta = deflate_it(delta, delta_size, &delta_size);
2186                         free(to_free);
2187                 }
2188         }
2189
2190         if (delta && delta_size < deflate_size) {
2191                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2192                 free(deflated);
2193                 data = delta;
2194                 data_size = delta_size;
2195         }
2196         else {
2197                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2198                 free(delta);
2199                 data = deflated;
2200                 data_size = deflate_size;
2201         }
2202
2203         /* emit data encoded in base85 */
2204         cp = data;
2205         while (data_size) {
2206                 int bytes = (52 < data_size) ? 52 : data_size;
2207                 char line[70];
2208                 data_size -= bytes;
2209                 if (bytes <= 26)
2210                         line[0] = bytes + 'A' - 1;
2211                 else
2212                         line[0] = bytes - 26 + 'a' - 1;
2213                 encode_85(line + 1, cp, bytes);
2214                 cp = (char *) cp + bytes;
2215                 fprintf(file, "%s", prefix);
2216                 fputs(line, file);
2217                 fputc('\n', file);
2218         }
2219         fprintf(file, "%s\n", prefix);
2220         free(data);
2221 }
2222
2223 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2224                              const char *prefix)
2225 {
2226         fprintf(file, "%sGIT binary patch\n", prefix);
2227         emit_binary_diff_body(file, one, two, prefix);
2228         emit_binary_diff_body(file, two, one, prefix);
2229 }
2230
2231 int diff_filespec_is_binary(struct diff_filespec *one)
2232 {
2233         if (one->is_binary == -1) {
2234                 diff_filespec_load_driver(one);
2235                 if (one->driver->binary != -1)
2236                         one->is_binary = one->driver->binary;
2237                 else {
2238                         if (!one->data && DIFF_FILE_VALID(one))
2239                                 diff_populate_filespec(one, CHECK_BINARY);
2240                         if (one->is_binary == -1 && one->data)
2241                                 one->is_binary = buffer_is_binary(one->data,
2242                                                 one->size);
2243                         if (one->is_binary == -1)
2244                                 one->is_binary = 0;
2245                 }
2246         }
2247         return one->is_binary;
2248 }
2249
2250 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2251 {
2252         diff_filespec_load_driver(one);
2253         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2254 }
2255
2256 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2257 {
2258         if (!options->a_prefix)
2259                 options->a_prefix = a;
2260         if (!options->b_prefix)
2261                 options->b_prefix = b;
2262 }
2263
2264 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2265 {
2266         if (!DIFF_FILE_VALID(one))
2267                 return NULL;
2268
2269         diff_filespec_load_driver(one);
2270         return userdiff_get_textconv(one->driver);
2271 }
2272
2273 static void builtin_diff(const char *name_a,
2274                          const char *name_b,
2275                          struct diff_filespec *one,
2276                          struct diff_filespec *two,
2277                          const char *xfrm_msg,
2278                          int must_show_header,
2279                          struct diff_options *o,
2280                          int complete_rewrite)
2281 {
2282         mmfile_t mf1, mf2;
2283         const char *lbl[2];
2284         char *a_one, *b_two;
2285         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2286         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2287         const char *a_prefix, *b_prefix;
2288         struct userdiff_driver *textconv_one = NULL;
2289         struct userdiff_driver *textconv_two = NULL;
2290         struct strbuf header = STRBUF_INIT;
2291         const char *line_prefix = diff_line_prefix(o);
2292
2293         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2294                         (!one->mode || S_ISGITLINK(one->mode)) &&
2295                         (!two->mode || S_ISGITLINK(two->mode))) {
2296                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2297                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2298                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2299                                 line_prefix,
2300                                 one->sha1, two->sha1, two->dirty_submodule,
2301                                 meta, del, add, reset);
2302                 return;
2303         }
2304
2305         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2306                 textconv_one = get_textconv(one);
2307                 textconv_two = get_textconv(two);
2308         }
2309
2310         diff_set_mnemonic_prefix(o, "a/", "b/");
2311         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2312                 a_prefix = o->b_prefix;
2313                 b_prefix = o->a_prefix;
2314         } else {
2315                 a_prefix = o->a_prefix;
2316                 b_prefix = o->b_prefix;
2317         }
2318
2319         /* Never use a non-valid filename anywhere if at all possible */
2320         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2321         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2322
2323         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2324         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2325         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2326         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2327         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2328         if (lbl[0][0] == '/') {
2329                 /* /dev/null */
2330                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2331                 if (xfrm_msg)
2332                         strbuf_addstr(&header, xfrm_msg);
2333                 must_show_header = 1;
2334         }
2335         else if (lbl[1][0] == '/') {
2336                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2337                 if (xfrm_msg)
2338                         strbuf_addstr(&header, xfrm_msg);
2339                 must_show_header = 1;
2340         }
2341         else {
2342                 if (one->mode != two->mode) {
2343                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2344                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2345                         must_show_header = 1;
2346                 }
2347                 if (xfrm_msg)
2348                         strbuf_addstr(&header, xfrm_msg);
2349
2350                 /*
2351                  * we do not run diff between different kind
2352                  * of objects.
2353                  */
2354                 if ((one->mode ^ two->mode) & S_IFMT)
2355                         goto free_ab_and_return;
2356                 if (complete_rewrite &&
2357                     (textconv_one || !diff_filespec_is_binary(one)) &&
2358                     (textconv_two || !diff_filespec_is_binary(two))) {
2359                         fprintf(o->file, "%s", header.buf);
2360                         strbuf_reset(&header);
2361                         emit_rewrite_diff(name_a, name_b, one, two,
2362                                                 textconv_one, textconv_two, o);
2363                         o->found_changes = 1;
2364                         goto free_ab_and_return;
2365                 }
2366         }
2367
2368         if (o->irreversible_delete && lbl[1][0] == '/') {
2369                 fprintf(o->file, "%s", header.buf);
2370                 strbuf_reset(&header);
2371                 goto free_ab_and_return;
2372         } else if (!DIFF_OPT_TST(o, TEXT) &&
2373             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2374               (!textconv_two && diff_filespec_is_binary(two)) )) {
2375                 if (!one->data && !two->data &&
2376                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2377                     !DIFF_OPT_TST(o, BINARY)) {
2378                         if (!hashcmp(one->sha1, two->sha1)) {
2379                                 if (must_show_header)
2380                                         fprintf(o->file, "%s", header.buf);
2381                                 goto free_ab_and_return;
2382                         }
2383                         fprintf(o->file, "%s", header.buf);
2384                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2385                                 line_prefix, lbl[0], lbl[1]);
2386                         goto free_ab_and_return;
2387                 }
2388                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2389                         die("unable to read files to diff");
2390                 /* Quite common confusing case */
2391                 if (mf1.size == mf2.size &&
2392                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2393                         if (must_show_header)
2394                                 fprintf(o->file, "%s", header.buf);
2395                         goto free_ab_and_return;
2396                 }
2397                 fprintf(o->file, "%s", header.buf);
2398                 strbuf_reset(&header);
2399                 if (DIFF_OPT_TST(o, BINARY))
2400                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2401                 else
2402                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2403                                 line_prefix, lbl[0], lbl[1]);
2404                 o->found_changes = 1;
2405         } else {
2406                 /* Crazy xdl interfaces.. */
2407                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2408                 const char *v;
2409                 xpparam_t xpp;
2410                 xdemitconf_t xecfg;
2411                 struct emit_callback ecbdata;
2412                 const struct userdiff_funcname *pe;
2413
2414                 if (must_show_header) {
2415                         fprintf(o->file, "%s", header.buf);
2416                         strbuf_reset(&header);
2417                 }
2418
2419                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2420                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2421
2422                 pe = diff_funcname_pattern(one);
2423                 if (!pe)
2424                         pe = diff_funcname_pattern(two);
2425
2426                 memset(&xpp, 0, sizeof(xpp));
2427                 memset(&xecfg, 0, sizeof(xecfg));
2428                 memset(&ecbdata, 0, sizeof(ecbdata));
2429                 ecbdata.label_path = lbl;
2430                 ecbdata.color_diff = want_color(o->use_color);
2431                 ecbdata.found_changesp = &o->found_changes;
2432                 ecbdata.ws_rule = whitespace_rule(name_b);
2433                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2434                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2435                 ecbdata.opt = o;
2436                 ecbdata.header = header.len ? &header : NULL;
2437                 xpp.flags = o->xdl_opts;
2438                 xecfg.ctxlen = o->context;
2439                 xecfg.interhunkctxlen = o->interhunkcontext;
2440                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2441                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2442                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2443                 if (pe)
2444                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2445                 if (!diffopts)
2446                         ;
2447                 else if (skip_prefix(diffopts, "--unified=", &v))
2448                         xecfg.ctxlen = strtoul(v, NULL, 10);
2449                 else if (skip_prefix(diffopts, "-u", &v))
2450                         xecfg.ctxlen = strtoul(v, NULL, 10);
2451                 if (o->word_diff)
2452                         init_diff_words_data(&ecbdata, o, one, two);
2453                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2454                                   &xpp, &xecfg))
2455                         die("unable to generate diff for %s", one->path);
2456                 if (o->word_diff)
2457                         free_diff_words_data(&ecbdata);
2458                 if (textconv_one)
2459                         free(mf1.ptr);
2460                 if (textconv_two)
2461                         free(mf2.ptr);
2462                 xdiff_clear_find_func(&xecfg);
2463         }
2464
2465  free_ab_and_return:
2466         strbuf_release(&header);
2467         diff_free_filespec_data(one);
2468         diff_free_filespec_data(two);
2469         free(a_one);
2470         free(b_two);
2471         return;
2472 }
2473
2474 static void builtin_diffstat(const char *name_a, const char *name_b,
2475                              struct diff_filespec *one,
2476                              struct diff_filespec *two,
2477                              struct diffstat_t *diffstat,
2478                              struct diff_options *o,
2479                              struct diff_filepair *p)
2480 {
2481         mmfile_t mf1, mf2;
2482         struct diffstat_file *data;
2483         int same_contents;
2484         int complete_rewrite = 0;
2485
2486         if (!DIFF_PAIR_UNMERGED(p)) {
2487                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2488                         complete_rewrite = 1;
2489         }
2490
2491         data = diffstat_add(diffstat, name_a, name_b);
2492         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2493
2494         if (!one || !two) {
2495                 data->is_unmerged = 1;
2496                 return;
2497         }
2498
2499         same_contents = !hashcmp(one->sha1, two->sha1);
2500
2501         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2502                 data->is_binary = 1;
2503                 if (same_contents) {
2504                         data->added = 0;
2505                         data->deleted = 0;
2506                 } else {
2507                         data->added = diff_filespec_size(two);
2508                         data->deleted = diff_filespec_size(one);
2509                 }
2510         }
2511
2512         else if (complete_rewrite) {
2513                 diff_populate_filespec(one, 0);
2514                 diff_populate_filespec(two, 0);
2515                 data->deleted = count_lines(one->data, one->size);
2516                 data->added = count_lines(two->data, two->size);
2517         }
2518
2519         else if (!same_contents) {
2520                 /* Crazy xdl interfaces.. */
2521                 xpparam_t xpp;
2522                 xdemitconf_t xecfg;
2523
2524                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2525                         die("unable to read files to diff");
2526
2527                 memset(&xpp, 0, sizeof(xpp));
2528                 memset(&xecfg, 0, sizeof(xecfg));
2529                 xpp.flags = o->xdl_opts;
2530                 xecfg.ctxlen = o->context;
2531                 xecfg.interhunkctxlen = o->interhunkcontext;
2532                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2533                                   &xpp, &xecfg))
2534                         die("unable to generate diffstat for %s", one->path);
2535         }
2536
2537         diff_free_filespec_data(one);
2538         diff_free_filespec_data(two);
2539 }
2540
2541 static void builtin_checkdiff(const char *name_a, const char *name_b,
2542                               const char *attr_path,
2543                               struct diff_filespec *one,
2544                               struct diff_filespec *two,
2545                               struct diff_options *o)
2546 {
2547         mmfile_t mf1, mf2;
2548         struct checkdiff_t data;
2549
2550         if (!two)
2551                 return;
2552
2553         memset(&data, 0, sizeof(data));
2554         data.filename = name_b ? name_b : name_a;
2555         data.lineno = 0;
2556         data.o = o;
2557         data.ws_rule = whitespace_rule(attr_path);
2558         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2559
2560         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2561                 die("unable to read files to diff");
2562
2563         /*
2564          * All the other codepaths check both sides, but not checking
2565          * the "old" side here is deliberate.  We are checking the newly
2566          * introduced changes, and as long as the "new" side is text, we
2567          * can and should check what it introduces.
2568          */
2569         if (diff_filespec_is_binary(two))
2570                 goto free_and_return;
2571         else {
2572                 /* Crazy xdl interfaces.. */
2573                 xpparam_t xpp;
2574                 xdemitconf_t xecfg;
2575
2576                 memset(&xpp, 0, sizeof(xpp));
2577                 memset(&xecfg, 0, sizeof(xecfg));
2578                 xecfg.ctxlen = 1; /* at least one context line */
2579                 xpp.flags = 0;
2580                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2581                                   &xpp, &xecfg))
2582                         die("unable to generate checkdiff for %s", one->path);
2583
2584                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2585                         struct emit_callback ecbdata;
2586                         int blank_at_eof;
2587
2588                         ecbdata.ws_rule = data.ws_rule;
2589                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2590                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2591
2592                         if (blank_at_eof) {
2593                                 static char *err;
2594                                 if (!err)
2595                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2596                                 fprintf(o->file, "%s:%d: %s.\n",
2597                                         data.filename, blank_at_eof, err);
2598                                 data.status = 1; /* report errors */
2599                         }
2600                 }
2601         }
2602  free_and_return:
2603         diff_free_filespec_data(one);
2604         diff_free_filespec_data(two);
2605         if (data.status)
2606                 DIFF_OPT_SET(o, CHECK_FAILED);
2607 }
2608
2609 struct diff_filespec *alloc_filespec(const char *path)
2610 {
2611         int namelen = strlen(path);
2612         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2613
2614         memset(spec, 0, sizeof(*spec));
2615         spec->path = (char *)(spec + 1);
2616         memcpy(spec->path, path, namelen+1);
2617         spec->count = 1;
2618         spec->is_binary = -1;
2619         return spec;
2620 }
2621
2622 void free_filespec(struct diff_filespec *spec)
2623 {
2624         if (!--spec->count) {
2625                 diff_free_filespec_data(spec);
2626                 free(spec);
2627         }
2628 }
2629
2630 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2631                    int sha1_valid, unsigned short mode)
2632 {
2633         if (mode) {
2634                 spec->mode = canon_mode(mode);
2635                 hashcpy(spec->sha1, sha1);
2636                 spec->sha1_valid = sha1_valid;
2637         }
2638 }
2639
2640 /*
2641  * Given a name and sha1 pair, if the index tells us the file in
2642  * the work tree has that object contents, return true, so that
2643  * prepare_temp_file() does not have to inflate and extract.
2644  */
2645 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2646 {
2647         const struct cache_entry *ce;
2648         struct stat st;
2649         int pos, len;
2650
2651         /*
2652          * We do not read the cache ourselves here, because the
2653          * benchmark with my previous version that always reads cache
2654          * shows that it makes things worse for diff-tree comparing
2655          * two linux-2.6 kernel trees in an already checked out work
2656          * tree.  This is because most diff-tree comparisons deal with
2657          * only a small number of files, while reading the cache is
2658          * expensive for a large project, and its cost outweighs the
2659          * savings we get by not inflating the object to a temporary
2660          * file.  Practically, this code only helps when we are used
2661          * by diff-cache --cached, which does read the cache before
2662          * calling us.
2663          */
2664         if (!active_cache)
2665                 return 0;
2666
2667         /* We want to avoid the working directory if our caller
2668          * doesn't need the data in a normal file, this system
2669          * is rather slow with its stat/open/mmap/close syscalls,
2670          * and the object is contained in a pack file.  The pack
2671          * is probably already open and will be faster to obtain
2672          * the data through than the working directory.  Loose
2673          * objects however would tend to be slower as they need
2674          * to be individually opened and inflated.
2675          */
2676         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2677                 return 0;
2678
2679         len = strlen(name);
2680         pos = cache_name_pos(name, len);
2681         if (pos < 0)
2682                 return 0;
2683         ce = active_cache[pos];
2684
2685         /*
2686          * This is not the sha1 we are looking for, or
2687          * unreusable because it is not a regular file.
2688          */
2689         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2690                 return 0;
2691
2692         /*
2693          * If ce is marked as "assume unchanged", there is no
2694          * guarantee that work tree matches what we are looking for.
2695          */
2696         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2697                 return 0;
2698
2699         /*
2700          * If ce matches the file in the work tree, we can reuse it.
2701          */
2702         if (ce_uptodate(ce) ||
2703             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2704                 return 1;
2705
2706         return 0;
2707 }
2708
2709 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2710 {
2711         int len;
2712         char *data = xmalloc(100), *dirty = "";
2713
2714         /* Are we looking at the work tree? */
2715         if (s->dirty_submodule)
2716                 dirty = "-dirty";
2717
2718         len = snprintf(data, 100,
2719                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2720         s->data = data;
2721         s->size = len;
2722         s->should_free = 1;
2723         if (size_only) {
2724                 s->data = NULL;
2725                 free(data);
2726         }
2727         return 0;
2728 }
2729
2730 /*
2731  * While doing rename detection and pickaxe operation, we may need to
2732  * grab the data for the blob (or file) for our own in-core comparison.
2733  * diff_filespec has data and size fields for this purpose.
2734  */
2735 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2736 {
2737         int size_only = flags & CHECK_SIZE_ONLY;
2738         int err = 0;
2739         /*
2740          * demote FAIL to WARN to allow inspecting the situation
2741          * instead of refusing.
2742          */
2743         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2744                                     ? SAFE_CRLF_WARN
2745                                     : safe_crlf);
2746
2747         if (!DIFF_FILE_VALID(s))
2748                 die("internal error: asking to populate invalid file.");
2749         if (S_ISDIR(s->mode))
2750                 return -1;
2751
2752         if (s->data)
2753                 return 0;
2754
2755         if (size_only && 0 < s->size)
2756                 return 0;
2757
2758         if (S_ISGITLINK(s->mode))
2759                 return diff_populate_gitlink(s, size_only);
2760
2761         if (!s->sha1_valid ||
2762             reuse_worktree_file(s->path, s->sha1, 0)) {
2763                 struct strbuf buf = STRBUF_INIT;
2764                 struct stat st;
2765                 int fd;
2766
2767                 if (lstat(s->path, &st) < 0) {
2768                         if (errno == ENOENT) {
2769                         err_empty:
2770                                 err = -1;
2771                         empty:
2772                                 s->data = (char *)"";
2773                                 s->size = 0;
2774                                 return err;
2775                         }
2776                 }
2777                 s->size = xsize_t(st.st_size);
2778                 if (!s->size)
2779                         goto empty;
2780                 if (S_ISLNK(st.st_mode)) {
2781                         struct strbuf sb = STRBUF_INIT;
2782
2783                         if (strbuf_readlink(&sb, s->path, s->size))
2784                                 goto err_empty;
2785                         s->size = sb.len;
2786                         s->data = strbuf_detach(&sb, NULL);
2787                         s->should_free = 1;
2788                         return 0;
2789                 }
2790                 if (size_only)
2791                         return 0;
2792                 if ((flags & CHECK_BINARY) &&
2793                     s->size > big_file_threshold && s->is_binary == -1) {
2794                         s->is_binary = 1;
2795                         return 0;
2796                 }
2797                 fd = open(s->path, O_RDONLY);
2798                 if (fd < 0)
2799                         goto err_empty;
2800                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2801                 close(fd);
2802                 s->should_munmap = 1;
2803
2804                 /*
2805                  * Convert from working tree format to canonical git format
2806                  */
2807                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2808                         size_t size = 0;
2809                         munmap(s->data, s->size);
2810                         s->should_munmap = 0;
2811                         s->data = strbuf_detach(&buf, &size);
2812                         s->size = size;
2813                         s->should_free = 1;
2814                 }
2815         }
2816         else {
2817                 enum object_type type;
2818                 if (size_only || (flags & CHECK_BINARY)) {
2819                         type = sha1_object_info(s->sha1, &s->size);
2820                         if (type < 0)
2821                                 die("unable to read %s", sha1_to_hex(s->sha1));
2822                         if (size_only)
2823                                 return 0;
2824                         if (s->size > big_file_threshold && s->is_binary == -1) {
2825                                 s->is_binary = 1;
2826                                 return 0;
2827                         }
2828                 }
2829                 s->data = read_sha1_file(s->sha1, &type, &s->size);
2830                 if (!s->data)
2831                         die("unable to read %s", sha1_to_hex(s->sha1));
2832                 s->should_free = 1;
2833         }
2834         return 0;
2835 }
2836
2837 void diff_free_filespec_blob(struct diff_filespec *s)
2838 {
2839         if (s->should_free)
2840                 free(s->data);
2841         else if (s->should_munmap)
2842                 munmap(s->data, s->size);
2843
2844         if (s->should_free || s->should_munmap) {
2845                 s->should_free = s->should_munmap = 0;
2846                 s->data = NULL;
2847         }
2848 }
2849
2850 void diff_free_filespec_data(struct diff_filespec *s)
2851 {
2852         diff_free_filespec_blob(s);
2853         free(s->cnt_data);
2854         s->cnt_data = NULL;
2855 }
2856
2857 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2858                            void *blob,
2859                            unsigned long size,
2860                            const unsigned char *sha1,
2861                            int mode)
2862 {
2863         int fd;
2864         struct strbuf buf = STRBUF_INIT;
2865         struct strbuf template = STRBUF_INIT;
2866         char *path_dup = xstrdup(path);
2867         const char *base = basename(path_dup);
2868
2869         /* Generate "XXXXXX_basename.ext" */
2870         strbuf_addstr(&template, "XXXXXX_");
2871         strbuf_addstr(&template, base);
2872
2873         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2874         if (fd < 0)
2875                 die_errno("unable to create temp-file");
2876         if (convert_to_working_tree(path,
2877                         (const char *)blob, (size_t)size, &buf)) {
2878                 blob = buf.buf;
2879                 size = buf.len;
2880         }
2881         if (write_in_full(fd, blob, size) != size)
2882                 die_errno("unable to write temp-file");
2883         close_tempfile(&temp->tempfile);
2884         temp->name = get_tempfile_path(&temp->tempfile);
2885         strcpy(temp->hex, sha1_to_hex(sha1));
2886         temp->hex[40] = 0;
2887         sprintf(temp->mode, "%06o", mode);
2888         strbuf_release(&buf);
2889         strbuf_release(&template);
2890         free(path_dup);
2891 }
2892
2893 static struct diff_tempfile *prepare_temp_file(const char *name,
2894                 struct diff_filespec *one)
2895 {
2896         struct diff_tempfile *temp = claim_diff_tempfile();
2897
2898         if (!DIFF_FILE_VALID(one)) {
2899         not_a_valid_file:
2900                 /* A '-' entry produces this for file-2, and
2901                  * a '+' entry produces this for file-1.
2902                  */
2903                 temp->name = "/dev/null";
2904                 strcpy(temp->hex, ".");
2905                 strcpy(temp->mode, ".");
2906                 return temp;
2907         }
2908
2909         if (!S_ISGITLINK(one->mode) &&
2910             (!one->sha1_valid ||
2911              reuse_worktree_file(name, one->sha1, 1))) {
2912                 struct stat st;
2913                 if (lstat(name, &st) < 0) {
2914                         if (errno == ENOENT)
2915                                 goto not_a_valid_file;
2916                         die_errno("stat(%s)", name);
2917                 }
2918                 if (S_ISLNK(st.st_mode)) {
2919                         struct strbuf sb = STRBUF_INIT;
2920                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2921                                 die_errno("readlink(%s)", name);
2922                         prep_temp_blob(name, temp, sb.buf, sb.len,
2923                                        (one->sha1_valid ?
2924                                         one->sha1 : null_sha1),
2925                                        (one->sha1_valid ?
2926                                         one->mode : S_IFLNK));
2927                         strbuf_release(&sb);
2928                 }
2929                 else {
2930                         /* we can borrow from the file in the work tree */
2931                         temp->name = name;
2932                         if (!one->sha1_valid)
2933                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2934                         else
2935                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2936                         /* Even though we may sometimes borrow the
2937                          * contents from the work tree, we always want
2938                          * one->mode.  mode is trustworthy even when
2939                          * !(one->sha1_valid), as long as
2940                          * DIFF_FILE_VALID(one).
2941                          */
2942                         sprintf(temp->mode, "%06o", one->mode);
2943                 }
2944                 return temp;
2945         }
2946         else {
2947                 if (diff_populate_filespec(one, 0))
2948                         die("cannot read data blob for %s", one->path);
2949                 prep_temp_blob(name, temp, one->data, one->size,
2950                                one->sha1, one->mode);
2951         }
2952         return temp;
2953 }
2954
2955 static void add_external_diff_name(struct argv_array *argv,
2956                                    const char *name,
2957                                    struct diff_filespec *df)
2958 {
2959         struct diff_tempfile *temp = prepare_temp_file(name, df);
2960         argv_array_push(argv, temp->name);
2961         argv_array_push(argv, temp->hex);
2962         argv_array_push(argv, temp->mode);
2963 }
2964
2965 /* An external diff command takes:
2966  *
2967  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2968  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2969  *
2970  */
2971 static void run_external_diff(const char *pgm,
2972                               const char *name,
2973                               const char *other,
2974                               struct diff_filespec *one,
2975                               struct diff_filespec *two,
2976                               const char *xfrm_msg,
2977                               int complete_rewrite,
2978                               struct diff_options *o)
2979 {
2980         struct argv_array argv = ARGV_ARRAY_INIT;
2981         struct argv_array env = ARGV_ARRAY_INIT;
2982         struct diff_queue_struct *q = &diff_queued_diff;
2983
2984         argv_array_push(&argv, pgm);
2985         argv_array_push(&argv, name);
2986
2987         if (one && two) {
2988                 add_external_diff_name(&argv, name, one);
2989                 if (!other)
2990                         add_external_diff_name(&argv, name, two);
2991                 else {
2992                         add_external_diff_name(&argv, other, two);
2993                         argv_array_push(&argv, other);
2994                         argv_array_push(&argv, xfrm_msg);
2995                 }
2996         }
2997
2998         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
2999         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3000
3001         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3002                 die(_("external diff died, stopping at %s"), name);
3003
3004         remove_tempfile();
3005         argv_array_clear(&argv);
3006         argv_array_clear(&env);
3007 }
3008
3009 static int similarity_index(struct diff_filepair *p)
3010 {
3011         return p->score * 100 / MAX_SCORE;
3012 }
3013
3014 static void fill_metainfo(struct strbuf *msg,
3015                           const char *name,
3016                           const char *other,
3017                           struct diff_filespec *one,
3018                           struct diff_filespec *two,
3019                           struct diff_options *o,
3020                           struct diff_filepair *p,
3021                           int *must_show_header,
3022                           int use_color)
3023 {
3024         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3025         const char *reset = diff_get_color(use_color, DIFF_RESET);
3026         const char *line_prefix = diff_line_prefix(o);
3027
3028         *must_show_header = 1;
3029         strbuf_init(msg, PATH_MAX * 2 + 300);
3030         switch (p->status) {
3031         case DIFF_STATUS_COPIED:
3032                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3033                             line_prefix, set, similarity_index(p));
3034                 strbuf_addf(msg, "%s\n%s%scopy from ",
3035                             reset,  line_prefix, set);
3036                 quote_c_style(name, msg, NULL, 0);
3037                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3038                 quote_c_style(other, msg, NULL, 0);
3039                 strbuf_addf(msg, "%s\n", reset);
3040                 break;
3041         case DIFF_STATUS_RENAMED:
3042                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3043                             line_prefix, set, similarity_index(p));
3044                 strbuf_addf(msg, "%s\n%s%srename from ",
3045                             reset, line_prefix, set);
3046                 quote_c_style(name, msg, NULL, 0);
3047                 strbuf_addf(msg, "%s\n%s%srename to ",
3048                             reset, line_prefix, set);
3049                 quote_c_style(other, msg, NULL, 0);
3050                 strbuf_addf(msg, "%s\n", reset);
3051                 break;
3052         case DIFF_STATUS_MODIFIED:
3053                 if (p->score) {
3054                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3055                                     line_prefix,
3056                                     set, similarity_index(p), reset);
3057                         break;
3058                 }
3059                 /* fallthru */
3060         default:
3061                 *must_show_header = 0;
3062         }
3063         if (one && two && hashcmp(one->sha1, two->sha1)) {
3064                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3065
3066                 if (DIFF_OPT_TST(o, BINARY)) {
3067                         mmfile_t mf;
3068                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3069                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3070                                 abbrev = 40;
3071                 }
3072                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3073                             find_unique_abbrev(one->sha1, abbrev));
3074                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3075                 if (one->mode == two->mode)
3076                         strbuf_addf(msg, " %06o", one->mode);
3077                 strbuf_addf(msg, "%s\n", reset);
3078         }
3079 }
3080
3081 static void run_diff_cmd(const char *pgm,
3082                          const char *name,
3083                          const char *other,
3084                          const char *attr_path,
3085                          struct diff_filespec *one,
3086                          struct diff_filespec *two,
3087                          struct strbuf *msg,
3088                          struct diff_options *o,
3089                          struct diff_filepair *p)
3090 {
3091         const char *xfrm_msg = NULL;
3092         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3093         int must_show_header = 0;
3094
3095
3096         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3097                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3098                 if (drv && drv->external)
3099                         pgm = drv->external;
3100         }
3101
3102         if (msg) {
3103                 /*
3104                  * don't use colors when the header is intended for an
3105                  * external diff driver
3106                  */
3107                 fill_metainfo(msg, name, other, one, two, o, p,
3108                               &must_show_header,
3109                               want_color(o->use_color) && !pgm);
3110                 xfrm_msg = msg->len ? msg->buf : NULL;
3111         }
3112
3113         if (pgm) {
3114                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3115                                   complete_rewrite, o);
3116                 return;
3117         }
3118         if (one && two)
3119                 builtin_diff(name, other ? other : name,
3120                              one, two, xfrm_msg, must_show_header,
3121                              o, complete_rewrite);
3122         else
3123                 fprintf(o->file, "* Unmerged path %s\n", name);
3124 }
3125
3126 static void diff_fill_sha1_info(struct diff_filespec *one)
3127 {
3128         if (DIFF_FILE_VALID(one)) {
3129                 if (!one->sha1_valid) {
3130                         struct stat st;
3131                         if (one->is_stdin) {
3132                                 hashcpy(one->sha1, null_sha1);
3133                                 return;
3134                         }
3135                         if (lstat(one->path, &st) < 0)
3136                                 die_errno("stat '%s'", one->path);
3137                         if (index_path(one->sha1, one->path, &st, 0))
3138                                 die("cannot hash %s", one->path);
3139                 }
3140         }
3141         else
3142                 hashclr(one->sha1);
3143 }
3144
3145 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3146 {
3147         /* Strip the prefix but do not molest /dev/null and absolute paths */
3148         if (*namep && **namep != '/') {
3149                 *namep += prefix_length;
3150                 if (**namep == '/')
3151                         ++*namep;
3152         }
3153         if (*otherp && **otherp != '/') {
3154                 *otherp += prefix_length;
3155                 if (**otherp == '/')
3156                         ++*otherp;
3157         }
3158 }
3159
3160 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3161 {
3162         const char *pgm = external_diff();
3163         struct strbuf msg;
3164         struct diff_filespec *one = p->one;
3165         struct diff_filespec *two = p->two;
3166         const char *name;
3167         const char *other;
3168         const char *attr_path;
3169
3170         name  = p->one->path;
3171         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3172         attr_path = name;
3173         if (o->prefix_length)
3174                 strip_prefix(o->prefix_length, &name, &other);
3175
3176         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3177                 pgm = NULL;
3178
3179         if (DIFF_PAIR_UNMERGED(p)) {
3180                 run_diff_cmd(pgm, name, NULL, attr_path,
3181                              NULL, NULL, NULL, o, p);
3182                 return;
3183         }
3184
3185         diff_fill_sha1_info(one);
3186         diff_fill_sha1_info(two);
3187
3188         if (!pgm &&
3189             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3190             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3191                 /*
3192                  * a filepair that changes between file and symlink
3193                  * needs to be split into deletion and creation.
3194                  */
3195                 struct diff_filespec *null = alloc_filespec(two->path);
3196                 run_diff_cmd(NULL, name, other, attr_path,
3197                              one, null, &msg, o, p);
3198                 free(null);
3199                 strbuf_release(&msg);
3200
3201                 null = alloc_filespec(one->path);
3202                 run_diff_cmd(NULL, name, other, attr_path,
3203                              null, two, &msg, o, p);
3204                 free(null);
3205         }
3206         else
3207                 run_diff_cmd(pgm, name, other, attr_path,
3208                              one, two, &msg, o, p);
3209
3210         strbuf_release(&msg);
3211 }
3212
3213 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3214                          struct diffstat_t *diffstat)
3215 {
3216         const char *name;
3217         const char *other;
3218
3219         if (DIFF_PAIR_UNMERGED(p)) {
3220                 /* unmerged */
3221                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3222                 return;
3223         }
3224
3225         name = p->one->path;
3226         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3227
3228         if (o->prefix_length)
3229                 strip_prefix(o->prefix_length, &name, &other);
3230
3231         diff_fill_sha1_info(p->one);
3232         diff_fill_sha1_info(p->two);
3233
3234         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3235 }
3236
3237 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3238 {
3239         const char *name;
3240         const char *other;
3241         const char *attr_path;
3242
3243         if (DIFF_PAIR_UNMERGED(p)) {
3244                 /* unmerged */
3245                 return;
3246         }
3247
3248         name = p->one->path;
3249         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3250         attr_path = other ? other : name;
3251
3252         if (o->prefix_length)
3253                 strip_prefix(o->prefix_length, &name, &other);
3254
3255         diff_fill_sha1_info(p->one);
3256         diff_fill_sha1_info(p->two);
3257
3258         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3259 }
3260
3261 void diff_setup(struct diff_options *options)
3262 {
3263         memcpy(options, &default_diff_options, sizeof(*options));
3264
3265         options->file = stdout;
3266
3267         options->line_termination = '\n';
3268         options->break_opt = -1;
3269         options->rename_limit = -1;
3270         options->dirstat_permille = diff_dirstat_permille_default;
3271         options->context = diff_context_default;
3272         options->ws_error_highlight = WSEH_NEW;
3273         DIFF_OPT_SET(options, RENAME_EMPTY);
3274
3275         /* pathchange left =NULL by default */
3276         options->change = diff_change;
3277         options->add_remove = diff_addremove;
3278         options->use_color = diff_use_color_default;
3279         options->detect_rename = diff_detect_rename_default;
3280         options->xdl_opts |= diff_algorithm;
3281
3282         options->orderfile = diff_order_file_cfg;
3283
3284         if (diff_no_prefix) {
3285                 options->a_prefix = options->b_prefix = "";
3286         } else if (!diff_mnemonic_prefix) {
3287                 options->a_prefix = "a/";
3288                 options->b_prefix = "b/";
3289         }
3290 }
3291
3292 void diff_setup_done(struct diff_options *options)
3293 {
3294         int count = 0;
3295
3296         if (options->set_default)
3297                 options->set_default(options);
3298
3299         if (options->output_format & DIFF_FORMAT_NAME)
3300                 count++;
3301         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3302                 count++;
3303         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3304                 count++;
3305         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3306                 count++;
3307         if (count > 1)
3308                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3309
3310         /*
3311          * Most of the time we can say "there are changes"
3312          * only by checking if there are changed paths, but
3313          * --ignore-whitespace* options force us to look
3314          * inside contents.
3315          */
3316
3317         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3318             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3319             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3320                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3321         else
3322                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3323
3324         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3325                 options->detect_rename = DIFF_DETECT_COPY;
3326
3327         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3328                 options->prefix = NULL;
3329         if (options->prefix)
3330                 options->prefix_length = strlen(options->prefix);
3331         else
3332                 options->prefix_length = 0;
3333
3334         if (options->output_format & (DIFF_FORMAT_NAME |
3335                                       DIFF_FORMAT_NAME_STATUS |
3336                                       DIFF_FORMAT_CHECKDIFF |
3337                                       DIFF_FORMAT_NO_OUTPUT))
3338                 options->output_format &= ~(DIFF_FORMAT_RAW |
3339                                             DIFF_FORMAT_NUMSTAT |
3340                                             DIFF_FORMAT_DIFFSTAT |
3341                                             DIFF_FORMAT_SHORTSTAT |
3342                                             DIFF_FORMAT_DIRSTAT |
3343                                             DIFF_FORMAT_SUMMARY |
3344                                             DIFF_FORMAT_PATCH);
3345
3346         /*
3347          * These cases always need recursive; we do not drop caller-supplied
3348          * recursive bits for other formats here.
3349          */
3350         if (options->output_format & (DIFF_FORMAT_PATCH |
3351                                       DIFF_FORMAT_NUMSTAT |
3352                                       DIFF_FORMAT_DIFFSTAT |
3353                                       DIFF_FORMAT_SHORTSTAT |
3354                                       DIFF_FORMAT_DIRSTAT |
3355                                       DIFF_FORMAT_SUMMARY |
3356                                       DIFF_FORMAT_CHECKDIFF))
3357                 DIFF_OPT_SET(options, RECURSIVE);
3358         /*
3359          * Also pickaxe would not work very well if you do not say recursive
3360          */
3361         if (options->pickaxe)
3362                 DIFF_OPT_SET(options, RECURSIVE);
3363         /*
3364          * When patches are generated, submodules diffed against the work tree
3365          * must be checked for dirtiness too so it can be shown in the output
3366          */
3367         if (options->output_format & DIFF_FORMAT_PATCH)
3368                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3369
3370         if (options->detect_rename && options->rename_limit < 0)
3371                 options->rename_limit = diff_rename_limit_default;
3372         if (options->setup & DIFF_SETUP_USE_CACHE) {
3373                 if (!active_cache)
3374                         /* read-cache does not die even when it fails
3375                          * so it is safe for us to do this here.  Also
3376                          * it does not smudge active_cache or active_nr
3377                          * when it fails, so we do not have to worry about
3378                          * cleaning it up ourselves either.
3379                          */
3380                         read_cache();
3381         }
3382         if (options->abbrev <= 0 || 40 < options->abbrev)
3383                 options->abbrev = 40; /* full */
3384
3385         /*
3386          * It does not make sense to show the first hit we happened
3387          * to have found.  It does not make sense not to return with
3388          * exit code in such a case either.
3389          */
3390         if (DIFF_OPT_TST(options, QUICK)) {
3391                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3392                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3393         }
3394
3395         options->diff_path_counter = 0;
3396
3397         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3398                 die(_("--follow requires exactly one pathspec"));
3399 }
3400
3401 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3402 {
3403         char c, *eq;
3404         int len;
3405
3406         if (*arg != '-')
3407                 return 0;
3408         c = *++arg;
3409         if (!c)
3410                 return 0;
3411         if (c == arg_short) {
3412                 c = *++arg;
3413                 if (!c)
3414                         return 1;
3415                 if (val && isdigit(c)) {
3416                         char *end;
3417                         int n = strtoul(arg, &end, 10);
3418                         if (*end)
3419                                 return 0;
3420                         *val = n;
3421                         return 1;
3422                 }
3423                 return 0;
3424         }
3425         if (c != '-')
3426                 return 0;
3427         arg++;
3428         eq = strchrnul(arg, '=');
3429         len = eq - arg;
3430         if (!len || strncmp(arg, arg_long, len))
3431                 return 0;
3432         if (*eq) {
3433                 int n;
3434                 char *end;
3435                 if (!isdigit(*++eq))
3436                         return 0;
3437                 n = strtoul(eq, &end, 10);
3438                 if (*end)
3439                         return 0;
3440                 *val = n;
3441         }
3442         return 1;
3443 }
3444
3445 static int diff_scoreopt_parse(const char *opt);
3446
3447 static inline int short_opt(char opt, const char **argv,
3448                             const char **optarg)
3449 {
3450         const char *arg = argv[0];
3451         if (arg[0] != '-' || arg[1] != opt)
3452                 return 0;
3453         if (arg[2] != '\0') {
3454                 *optarg = arg + 2;
3455                 return 1;
3456         }
3457         if (!argv[1])
3458                 die("Option '%c' requires a value", opt);
3459         *optarg = argv[1];
3460         return 2;
3461 }
3462
3463 int parse_long_opt(const char *opt, const char **argv,
3464                    const char **optarg)
3465 {
3466         const char *arg = argv[0];
3467         if (!skip_prefix(arg, "--", &arg))
3468                 return 0;
3469         if (!skip_prefix(arg, opt, &arg))
3470                 return 0;
3471         if (*arg == '=') { /* stuck form: --option=value */
3472                 *optarg = arg + 1;
3473                 return 1;
3474         }
3475         if (*arg != '\0')
3476                 return 0;
3477         /* separate form: --option value */
3478         if (!argv[1])
3479                 die("Option '--%s' requires a value", opt);
3480         *optarg = argv[1];
3481         return 2;
3482 }
3483
3484 static int stat_opt(struct diff_options *options, const char **av)
3485 {
3486         const char *arg = av[0];
3487         char *end;
3488         int width = options->stat_width;
3489         int name_width = options->stat_name_width;
3490         int graph_width = options->stat_graph_width;
3491         int count = options->stat_count;
3492         int argcount = 1;
3493
3494         if (!skip_prefix(arg, "--stat", &arg))
3495                 die("BUG: stat option does not begin with --stat: %s", arg);
3496         end = (char *)arg;
3497
3498         switch (*arg) {
3499         case '-':
3500                 if (skip_prefix(arg, "-width", &arg)) {
3501                         if (*arg == '=')
3502                                 width = strtoul(arg + 1, &end, 10);
3503                         else if (!*arg && !av[1])
3504                                 die("Option '--stat-width' requires a value");
3505                         else if (!*arg) {
3506                                 width = strtoul(av[1], &end, 10);
3507                                 argcount = 2;
3508                         }
3509                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3510                         if (*arg == '=')
3511                                 name_width = strtoul(arg + 1, &end, 10);
3512                         else if (!*arg && !av[1])
3513                                 die("Option '--stat-name-width' requires a value");
3514                         else if (!*arg) {
3515                                 name_width = strtoul(av[1], &end, 10);
3516                                 argcount = 2;
3517                         }
3518                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3519                         if (*arg == '=')
3520                                 graph_width = strtoul(arg + 1, &end, 10);
3521                         else if (!*arg && !av[1])
3522                                 die("Option '--stat-graph-width' requires a value");
3523                         else if (!*arg) {
3524                                 graph_width = strtoul(av[1], &end, 10);
3525                                 argcount = 2;
3526                         }
3527                 } else if (skip_prefix(arg, "-count", &arg)) {
3528                         if (*arg == '=')
3529                                 count = strtoul(arg + 1, &end, 10);
3530                         else if (!*arg && !av[1])
3531                                 die("Option '--stat-count' requires a value");
3532                         else if (!*arg) {
3533                                 count = strtoul(av[1], &end, 10);
3534                                 argcount = 2;
3535                         }
3536                 }
3537                 break;
3538         case '=':
3539                 width = strtoul(arg+1, &end, 10);
3540                 if (*end == ',')
3541                         name_width = strtoul(end+1, &end, 10);
3542                 if (*end == ',')
3543                         count = strtoul(end+1, &end, 10);
3544         }
3545
3546         /* Important! This checks all the error cases! */
3547         if (*end)
3548                 return 0;
3549         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3550         options->stat_name_width = name_width;
3551         options->stat_graph_width = graph_width;
3552         options->stat_width = width;
3553         options->stat_count = count;
3554         return argcount;
3555 }
3556
3557 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3558 {
3559         struct strbuf errmsg = STRBUF_INIT;
3560         if (parse_dirstat_params(options, params, &errmsg))
3561                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3562                     errmsg.buf);
3563         strbuf_release(&errmsg);
3564         /*
3565          * The caller knows a dirstat-related option is given from the command
3566          * line; allow it to say "return this_function();"
3567          */
3568         options->output_format |= DIFF_FORMAT_DIRSTAT;
3569         return 1;
3570 }
3571
3572 static int parse_submodule_opt(struct diff_options *options, const char *value)
3573 {
3574         if (parse_submodule_params(options, value))
3575                 die(_("Failed to parse --submodule option parameter: '%s'"),
3576                         value);
3577         return 1;
3578 }
3579
3580 static const char diff_status_letters[] = {
3581         DIFF_STATUS_ADDED,
3582         DIFF_STATUS_COPIED,
3583         DIFF_STATUS_DELETED,
3584         DIFF_STATUS_MODIFIED,
3585         DIFF_STATUS_RENAMED,
3586         DIFF_STATUS_TYPE_CHANGED,
3587         DIFF_STATUS_UNKNOWN,
3588         DIFF_STATUS_UNMERGED,
3589         DIFF_STATUS_FILTER_AON,
3590         DIFF_STATUS_FILTER_BROKEN,
3591         '\0',
3592 };
3593
3594 static unsigned int filter_bit['Z' + 1];
3595
3596 static void prepare_filter_bits(void)
3597 {
3598         int i;
3599
3600         if (!filter_bit[DIFF_STATUS_ADDED]) {
3601                 for (i = 0; diff_status_letters[i]; i++)
3602                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3603         }
3604 }
3605
3606 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3607 {
3608         return opt->filter & filter_bit[(int) status];
3609 }
3610
3611 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3612 {
3613         int i, optch;
3614
3615         prepare_filter_bits();
3616
3617         /*
3618          * If there is a negation e.g. 'd' in the input, and we haven't
3619          * initialized the filter field with another --diff-filter, start
3620          * from full set of bits, except for AON.
3621          */
3622         if (!opt->filter) {
3623                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3624                         if (optch < 'a' || 'z' < optch)
3625                                 continue;
3626                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3627                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3628                         break;
3629                 }
3630         }
3631
3632         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3633                 unsigned int bit;
3634                 int negate;
3635
3636                 if ('a' <= optch && optch <= 'z') {
3637                         negate = 1;
3638                         optch = toupper(optch);
3639                 } else {
3640                         negate = 0;
3641                 }
3642
3643                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3644                 if (!bit)
3645                         return optarg[i];
3646                 if (negate)
3647                         opt->filter &= ~bit;
3648                 else
3649                         opt->filter |= bit;
3650         }
3651         return 0;
3652 }
3653
3654 static void enable_patch_output(int *fmt) {
3655         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3656         *fmt |= DIFF_FORMAT_PATCH;
3657 }
3658
3659 static int parse_one_token(const char **arg, const char *token)
3660 {
3661         const char *rest;
3662         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
3663                 *arg = rest;
3664                 return 1;
3665         }
3666         return 0;
3667 }
3668
3669 static int parse_ws_error_highlight(struct diff_options *opt, const char *arg)
3670 {
3671         const char *orig_arg = arg;
3672         unsigned val = 0;
3673         while (*arg) {
3674                 if (parse_one_token(&arg, "none"))
3675                         val = 0;
3676                 else if (parse_one_token(&arg, "default"))
3677                         val = WSEH_NEW;
3678                 else if (parse_one_token(&arg, "all"))
3679                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
3680                 else if (parse_one_token(&arg, "new"))
3681                         val |= WSEH_NEW;
3682                 else if (parse_one_token(&arg, "old"))
3683                         val |= WSEH_OLD;
3684                 else if (parse_one_token(&arg, "context"))
3685                         val |= WSEH_CONTEXT;
3686                 else {
3687                         error("unknown value after ws-error-highlight=%.*s",
3688                               (int)(arg - orig_arg), orig_arg);
3689                         return 0;
3690                 }
3691                 if (*arg)
3692                         arg++;
3693         }
3694         opt->ws_error_highlight = val;
3695         return 1;
3696 }
3697
3698 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3699 {
3700         const char *arg = av[0];
3701         const char *optarg;
3702         int argcount;
3703
3704         /* Output format options */
3705         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3706             || opt_arg(arg, 'U', "unified", &options->context))
3707                 enable_patch_output(&options->output_format);
3708         else if (!strcmp(arg, "--raw"))
3709                 options->output_format |= DIFF_FORMAT_RAW;
3710         else if (!strcmp(arg, "--patch-with-raw")) {
3711                 enable_patch_output(&options->output_format);
3712                 options->output_format |= DIFF_FORMAT_RAW;
3713         } else if (!strcmp(arg, "--numstat"))
3714                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3715         else if (!strcmp(arg, "--shortstat"))
3716                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3717         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3718                 return parse_dirstat_opt(options, "");
3719         else if (skip_prefix(arg, "-X", &arg))
3720                 return parse_dirstat_opt(options, arg);
3721         else if (skip_prefix(arg, "--dirstat=", &arg))
3722                 return parse_dirstat_opt(options, arg);
3723         else if (!strcmp(arg, "--cumulative"))
3724                 return parse_dirstat_opt(options, "cumulative");
3725         else if (!strcmp(arg, "--dirstat-by-file"))
3726                 return parse_dirstat_opt(options, "files");
3727         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3728                 parse_dirstat_opt(options, "files");
3729                 return parse_dirstat_opt(options, arg);
3730         }
3731         else if (!strcmp(arg, "--check"))
3732                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3733         else if (!strcmp(arg, "--summary"))
3734                 options->output_format |= DIFF_FORMAT_SUMMARY;
3735         else if (!strcmp(arg, "--patch-with-stat")) {
3736                 enable_patch_output(&options->output_format);
3737                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3738         } else if (!strcmp(arg, "--name-only"))
3739                 options->output_format |= DIFF_FORMAT_NAME;
3740         else if (!strcmp(arg, "--name-status"))
3741                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3742         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3743                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3744         else if (starts_with(arg, "--stat"))
3745                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3746                 return stat_opt(options, av);
3747
3748         /* renames options */
3749         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3750                  !strcmp(arg, "--break-rewrites")) {
3751                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3752                         return error("invalid argument to -B: %s", arg+2);
3753         }
3754         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3755                  !strcmp(arg, "--find-renames")) {
3756                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3757                         return error("invalid argument to -M: %s", arg+2);
3758                 options->detect_rename = DIFF_DETECT_RENAME;
3759         }
3760         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3761                 options->irreversible_delete = 1;
3762         }
3763         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3764                  !strcmp(arg, "--find-copies")) {
3765                 if (options->detect_rename == DIFF_DETECT_COPY)
3766                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3767                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3768                         return error("invalid argument to -C: %s", arg+2);
3769                 options->detect_rename = DIFF_DETECT_COPY;
3770         }
3771         else if (!strcmp(arg, "--no-renames"))
3772                 options->detect_rename = 0;
3773         else if (!strcmp(arg, "--rename-empty"))
3774                 DIFF_OPT_SET(options, RENAME_EMPTY);
3775         else if (!strcmp(arg, "--no-rename-empty"))
3776                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3777         else if (!strcmp(arg, "--relative"))
3778                 DIFF_OPT_SET(options, RELATIVE_NAME);
3779         else if (skip_prefix(arg, "--relative=", &arg)) {
3780                 DIFF_OPT_SET(options, RELATIVE_NAME);
3781                 options->prefix = arg;
3782         }
3783
3784         /* xdiff options */
3785         else if (!strcmp(arg, "--minimal"))
3786                 DIFF_XDL_SET(options, NEED_MINIMAL);
3787         else if (!strcmp(arg, "--no-minimal"))
3788                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3789         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3790                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3791         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3792                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3793         else if (!strcmp(arg, "--ignore-space-at-eol"))
3794                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3795         else if (!strcmp(arg, "--ignore-blank-lines"))
3796                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3797         else if (!strcmp(arg, "--patience"))
3798                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3799         else if (!strcmp(arg, "--histogram"))
3800                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3801         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3802                 long value = parse_algorithm_value(optarg);
3803                 if (value < 0)
3804                         return error("option diff-algorithm accepts \"myers\", "
3805                                      "\"minimal\", \"patience\" and \"histogram\"");
3806                 /* clear out previous settings */
3807                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3808                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3809                 options->xdl_opts |= value;
3810                 return argcount;
3811         }
3812
3813         /* flags options */
3814         else if (!strcmp(arg, "--binary")) {
3815                 enable_patch_output(&options->output_format);
3816                 DIFF_OPT_SET(options, BINARY);
3817         }
3818         else if (!strcmp(arg, "--full-index"))
3819                 DIFF_OPT_SET(options, FULL_INDEX);
3820         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3821                 DIFF_OPT_SET(options, TEXT);
3822         else if (!strcmp(arg, "-R"))
3823                 DIFF_OPT_SET(options, REVERSE_DIFF);
3824         else if (!strcmp(arg, "--find-copies-harder"))
3825                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3826         else if (!strcmp(arg, "--follow"))
3827                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3828         else if (!strcmp(arg, "--no-follow")) {
3829                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3830                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3831         } else if (!strcmp(arg, "--color"))
3832                 options->use_color = 1;
3833         else if (skip_prefix(arg, "--color=", &arg)) {
3834                 int value = git_config_colorbool(NULL, arg);
3835                 if (value < 0)
3836                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3837                 options->use_color = value;
3838         }
3839         else if (!strcmp(arg, "--no-color"))
3840                 options->use_color = 0;
3841         else if (!strcmp(arg, "--color-words")) {
3842                 options->use_color = 1;
3843                 options->word_diff = DIFF_WORDS_COLOR;
3844         }
3845         else if (skip_prefix(arg, "--color-words=", &arg)) {
3846                 options->use_color = 1;
3847                 options->word_diff = DIFF_WORDS_COLOR;
3848                 options->word_regex = arg;
3849         }
3850         else if (!strcmp(arg, "--word-diff")) {
3851                 if (options->word_diff == DIFF_WORDS_NONE)
3852                         options->word_diff = DIFF_WORDS_PLAIN;
3853         }
3854         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3855                 if (!strcmp(arg, "plain"))
3856                         options->word_diff = DIFF_WORDS_PLAIN;
3857                 else if (!strcmp(arg, "color")) {
3858                         options->use_color = 1;
3859                         options->word_diff = DIFF_WORDS_COLOR;
3860                 }
3861                 else if (!strcmp(arg, "porcelain"))
3862                         options->word_diff = DIFF_WORDS_PORCELAIN;
3863                 else if (!strcmp(arg, "none"))
3864                         options->word_diff = DIFF_WORDS_NONE;
3865                 else
3866                         die("bad --word-diff argument: %s", arg);
3867         }
3868         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3869                 if (options->word_diff == DIFF_WORDS_NONE)
3870                         options->word_diff = DIFF_WORDS_PLAIN;
3871                 options->word_regex = optarg;
3872                 return argcount;
3873         }
3874         else if (!strcmp(arg, "--exit-code"))
3875                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3876         else if (!strcmp(arg, "--quiet"))
3877                 DIFF_OPT_SET(options, QUICK);
3878         else if (!strcmp(arg, "--ext-diff"))
3879                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3880         else if (!strcmp(arg, "--no-ext-diff"))
3881                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3882         else if (!strcmp(arg, "--textconv"))
3883                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3884         else if (!strcmp(arg, "--no-textconv"))
3885                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3886         else if (!strcmp(arg, "--ignore-submodules")) {
3887                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3888                 handle_ignore_submodules_arg(options, "all");
3889         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3890                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3891                 handle_ignore_submodules_arg(options, arg);
3892         } else if (!strcmp(arg, "--submodule"))
3893                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3894         else if (skip_prefix(arg, "--submodule=", &arg))
3895                 return parse_submodule_opt(options, arg);
3896         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3897                 return parse_ws_error_highlight(options, arg);
3898
3899         /* misc options */
3900         else if (!strcmp(arg, "-z"))
3901                 options->line_termination = 0;
3902         else if ((argcount = short_opt('l', av, &optarg))) {
3903                 options->rename_limit = strtoul(optarg, NULL, 10);
3904                 return argcount;
3905         }
3906         else if ((argcount = short_opt('S', av, &optarg))) {
3907                 options->pickaxe = optarg;
3908                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3909                 return argcount;
3910         } else if ((argcount = short_opt('G', av, &optarg))) {
3911                 options->pickaxe = optarg;
3912                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3913                 return argcount;
3914         }
3915         else if (!strcmp(arg, "--pickaxe-all"))
3916                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3917         else if (!strcmp(arg, "--pickaxe-regex"))
3918                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3919         else if ((argcount = short_opt('O', av, &optarg))) {
3920                 options->orderfile = optarg;
3921                 return argcount;
3922         }
3923         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3924                 int offending = parse_diff_filter_opt(optarg, options);
3925                 if (offending)
3926                         die("unknown change class '%c' in --diff-filter=%s",
3927                             offending, optarg);
3928                 return argcount;
3929         }
3930         else if (!strcmp(arg, "--abbrev"))
3931                 options->abbrev = DEFAULT_ABBREV;
3932         else if (skip_prefix(arg, "--abbrev=", &arg)) {
3933                 options->abbrev = strtoul(arg, NULL, 10);
3934                 if (options->abbrev < MINIMUM_ABBREV)
3935                         options->abbrev = MINIMUM_ABBREV;
3936                 else if (40 < options->abbrev)
3937                         options->abbrev = 40;
3938         }
3939         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3940                 options->a_prefix = optarg;
3941                 return argcount;
3942         }
3943         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3944                 options->b_prefix = optarg;
3945                 return argcount;
3946         }
3947         else if (!strcmp(arg, "--no-prefix"))
3948                 options->a_prefix = options->b_prefix = "";
3949         else if (opt_arg(arg, '\0', "inter-hunk-context",
3950                          &options->interhunkcontext))
3951                 ;
3952         else if (!strcmp(arg, "-W"))
3953                 DIFF_OPT_SET(options, FUNCCONTEXT);
3954         else if (!strcmp(arg, "--function-context"))
3955                 DIFF_OPT_SET(options, FUNCCONTEXT);
3956         else if (!strcmp(arg, "--no-function-context"))
3957                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3958         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3959                 options->file = fopen(optarg, "w");
3960                 if (!options->file)
3961                         die_errno("Could not open '%s'", optarg);
3962                 options->close_file = 1;
3963                 return argcount;
3964         } else
3965                 return 0;
3966         return 1;
3967 }
3968
3969 int parse_rename_score(const char **cp_p)
3970 {
3971         unsigned long num, scale;
3972         int ch, dot;
3973         const char *cp = *cp_p;
3974
3975         num = 0;
3976         scale = 1;
3977         dot = 0;
3978         for (;;) {
3979                 ch = *cp;
3980                 if ( !dot && ch == '.' ) {
3981                         scale = 1;
3982                         dot = 1;
3983                 } else if ( ch == '%' ) {
3984                         scale = dot ? scale*100 : 100;
3985                         cp++;   /* % is always at the end */
3986                         break;
3987                 } else if ( ch >= '0' && ch <= '9' ) {
3988                         if ( scale < 100000 ) {
3989                                 scale *= 10;
3990                                 num = (num*10) + (ch-'0');
3991                         }
3992                 } else {
3993                         break;
3994                 }
3995                 cp++;
3996         }
3997         *cp_p = cp;
3998
3999         /* user says num divided by scale and we say internally that
4000          * is MAX_SCORE * num / scale.
4001          */
4002         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4003 }
4004
4005 static int diff_scoreopt_parse(const char *opt)
4006 {
4007         int opt1, opt2, cmd;
4008
4009         if (*opt++ != '-')
4010                 return -1;
4011         cmd = *opt++;
4012         if (cmd == '-') {
4013                 /* convert the long-form arguments into short-form versions */
4014                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4015                         if (*opt == 0 || *opt++ == '=')
4016                                 cmd = 'B';
4017                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4018                         if (*opt == 0 || *opt++ == '=')
4019                                 cmd = 'C';
4020                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4021                         if (*opt == 0 || *opt++ == '=')
4022                                 cmd = 'M';
4023                 }
4024         }
4025         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4026                 return -1; /* that is not a -M, -C, or -B option */
4027
4028         opt1 = parse_rename_score(&opt);
4029         if (cmd != 'B')
4030                 opt2 = 0;
4031         else {
4032                 if (*opt == 0)
4033                         opt2 = 0;
4034                 else if (*opt != '/')
4035                         return -1; /* we expect -B80/99 or -B80 */
4036                 else {
4037                         opt++;
4038                         opt2 = parse_rename_score(&opt);
4039                 }
4040         }
4041         if (*opt != 0)
4042                 return -1;
4043         return opt1 | (opt2 << 16);
4044 }
4045
4046 struct diff_queue_struct diff_queued_diff;
4047
4048 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4049 {
4050         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4051         queue->queue[queue->nr++] = dp;
4052 }
4053
4054 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4055                                  struct diff_filespec *one,
4056                                  struct diff_filespec *two)
4057 {
4058         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4059         dp->one = one;
4060         dp->two = two;
4061         if (queue)
4062                 diff_q(queue, dp);
4063         return dp;
4064 }
4065
4066 void diff_free_filepair(struct diff_filepair *p)
4067 {
4068         free_filespec(p->one);
4069         free_filespec(p->two);
4070         free(p);
4071 }
4072
4073 /* This is different from find_unique_abbrev() in that
4074  * it stuffs the result with dots for alignment.
4075  */
4076 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
4077 {
4078         int abblen;
4079         const char *abbrev;
4080         if (len == 40)
4081                 return sha1_to_hex(sha1);
4082
4083         abbrev = find_unique_abbrev(sha1, len);
4084         abblen = strlen(abbrev);
4085         if (abblen < 37) {
4086                 static char hex[41];
4087                 if (len < abblen && abblen <= len + 2)
4088                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
4089                 else
4090                         sprintf(hex, "%s...", abbrev);
4091                 return hex;
4092         }
4093         return sha1_to_hex(sha1);
4094 }
4095
4096 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4097 {
4098         int line_termination = opt->line_termination;
4099         int inter_name_termination = line_termination ? '\t' : '\0';
4100
4101         fprintf(opt->file, "%s", diff_line_prefix(opt));
4102         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4103                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4104                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
4105                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4106         }
4107         if (p->score) {
4108                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4109                         inter_name_termination);
4110         } else {
4111                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4112         }
4113
4114         if (p->status == DIFF_STATUS_COPIED ||
4115             p->status == DIFF_STATUS_RENAMED) {
4116                 const char *name_a, *name_b;
4117                 name_a = p->one->path;
4118                 name_b = p->two->path;
4119                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4120                 write_name_quoted(name_a, opt->file, inter_name_termination);
4121                 write_name_quoted(name_b, opt->file, line_termination);
4122         } else {
4123                 const char *name_a, *name_b;
4124                 name_a = p->one->mode ? p->one->path : p->two->path;
4125                 name_b = NULL;
4126                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4127                 write_name_quoted(name_a, opt->file, line_termination);
4128         }
4129 }
4130
4131 int diff_unmodified_pair(struct diff_filepair *p)
4132 {
4133         /* This function is written stricter than necessary to support
4134          * the currently implemented transformers, but the idea is to
4135          * let transformers to produce diff_filepairs any way they want,
4136          * and filter and clean them up here before producing the output.
4137          */
4138         struct diff_filespec *one = p->one, *two = p->two;
4139
4140         if (DIFF_PAIR_UNMERGED(p))
4141                 return 0; /* unmerged is interesting */
4142
4143         /* deletion, addition, mode or type change
4144          * and rename are all interesting.
4145          */
4146         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4147             DIFF_PAIR_MODE_CHANGED(p) ||
4148             strcmp(one->path, two->path))
4149                 return 0;
4150
4151         /* both are valid and point at the same path.  that is, we are
4152          * dealing with a change.
4153          */
4154         if (one->sha1_valid && two->sha1_valid &&
4155             !hashcmp(one->sha1, two->sha1) &&
4156             !one->dirty_submodule && !two->dirty_submodule)
4157                 return 1; /* no change */
4158         if (!one->sha1_valid && !two->sha1_valid)
4159                 return 1; /* both look at the same file on the filesystem. */
4160         return 0;
4161 }
4162
4163 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4164 {
4165         if (diff_unmodified_pair(p))
4166                 return;
4167
4168         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4169             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4170                 return; /* no tree diffs in patch format */
4171
4172         run_diff(p, o);
4173 }
4174
4175 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4176                             struct diffstat_t *diffstat)
4177 {
4178         if (diff_unmodified_pair(p))
4179                 return;
4180
4181         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4182             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4183                 return; /* no useful stat for tree diffs */
4184
4185         run_diffstat(p, o, diffstat);
4186 }
4187
4188 static void diff_flush_checkdiff(struct diff_filepair *p,
4189                 struct diff_options *o)
4190 {
4191         if (diff_unmodified_pair(p))
4192                 return;
4193
4194         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4195             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4196                 return; /* nothing to check in tree diffs */
4197
4198         run_checkdiff(p, o);
4199 }
4200
4201 int diff_queue_is_empty(void)
4202 {
4203         struct diff_queue_struct *q = &diff_queued_diff;
4204         int i;
4205         for (i = 0; i < q->nr; i++)
4206                 if (!diff_unmodified_pair(q->queue[i]))
4207                         return 0;
4208         return 1;
4209 }
4210
4211 #if DIFF_DEBUG
4212 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4213 {
4214         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4215                 x, one ? one : "",
4216                 s->path,
4217                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4218                 s->mode,
4219                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4220         fprintf(stderr, "queue[%d] %s size %lu\n",
4221                 x, one ? one : "",
4222                 s->size);
4223 }
4224
4225 void diff_debug_filepair(const struct diff_filepair *p, int i)
4226 {
4227         diff_debug_filespec(p->one, i, "one");
4228         diff_debug_filespec(p->two, i, "two");
4229         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4230                 p->score, p->status ? p->status : '?',
4231                 p->one->rename_used, p->broken_pair);
4232 }
4233
4234 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4235 {
4236         int i;
4237         if (msg)
4238                 fprintf(stderr, "%s\n", msg);
4239         fprintf(stderr, "q->nr = %d\n", q->nr);
4240         for (i = 0; i < q->nr; i++) {
4241                 struct diff_filepair *p = q->queue[i];
4242                 diff_debug_filepair(p, i);
4243         }
4244 }
4245 #endif
4246
4247 static void diff_resolve_rename_copy(void)
4248 {
4249         int i;
4250         struct diff_filepair *p;
4251         struct diff_queue_struct *q = &diff_queued_diff;
4252
4253         diff_debug_queue("resolve-rename-copy", q);
4254
4255         for (i = 0; i < q->nr; i++) {
4256                 p = q->queue[i];
4257                 p->status = 0; /* undecided */
4258                 if (DIFF_PAIR_UNMERGED(p))
4259                         p->status = DIFF_STATUS_UNMERGED;
4260                 else if (!DIFF_FILE_VALID(p->one))
4261                         p->status = DIFF_STATUS_ADDED;
4262                 else if (!DIFF_FILE_VALID(p->two))
4263                         p->status = DIFF_STATUS_DELETED;
4264                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4265                         p->status = DIFF_STATUS_TYPE_CHANGED;
4266
4267                 /* from this point on, we are dealing with a pair
4268                  * whose both sides are valid and of the same type, i.e.
4269                  * either in-place edit or rename/copy edit.
4270                  */
4271                 else if (DIFF_PAIR_RENAME(p)) {
4272                         /*
4273                          * A rename might have re-connected a broken
4274                          * pair up, causing the pathnames to be the
4275                          * same again. If so, that's not a rename at
4276                          * all, just a modification..
4277                          *
4278                          * Otherwise, see if this source was used for
4279                          * multiple renames, in which case we decrement
4280                          * the count, and call it a copy.
4281                          */
4282                         if (!strcmp(p->one->path, p->two->path))
4283                                 p->status = DIFF_STATUS_MODIFIED;
4284                         else if (--p->one->rename_used > 0)
4285                                 p->status = DIFF_STATUS_COPIED;
4286                         else
4287                                 p->status = DIFF_STATUS_RENAMED;
4288                 }
4289                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4290                          p->one->mode != p->two->mode ||
4291                          p->one->dirty_submodule ||
4292                          p->two->dirty_submodule ||
4293                          is_null_sha1(p->one->sha1))
4294                         p->status = DIFF_STATUS_MODIFIED;
4295                 else {
4296                         /* This is a "no-change" entry and should not
4297                          * happen anymore, but prepare for broken callers.
4298                          */
4299                         error("feeding unmodified %s to diffcore",
4300                               p->one->path);
4301                         p->status = DIFF_STATUS_UNKNOWN;
4302                 }
4303         }
4304         diff_debug_queue("resolve-rename-copy done", q);
4305 }
4306
4307 static int check_pair_status(struct diff_filepair *p)
4308 {
4309         switch (p->status) {
4310         case DIFF_STATUS_UNKNOWN:
4311                 return 0;
4312         case 0:
4313                 die("internal error in diff-resolve-rename-copy");
4314         default:
4315                 return 1;
4316         }
4317 }
4318
4319 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4320 {
4321         int fmt = opt->output_format;
4322
4323         if (fmt & DIFF_FORMAT_CHECKDIFF)
4324                 diff_flush_checkdiff(p, opt);
4325         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4326                 diff_flush_raw(p, opt);
4327         else if (fmt & DIFF_FORMAT_NAME) {
4328                 const char *name_a, *name_b;
4329                 name_a = p->two->path;
4330                 name_b = NULL;
4331                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4332                 write_name_quoted(name_a, opt->file, opt->line_termination);
4333         }
4334 }
4335
4336 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4337 {
4338         if (fs->mode)
4339                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4340         else
4341                 fprintf(file, " %s ", newdelete);
4342         write_name_quoted(fs->path, file, '\n');
4343 }
4344
4345
4346 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4347                 const char *line_prefix)
4348 {
4349         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4350                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4351                         p->two->mode, show_name ? ' ' : '\n');
4352                 if (show_name) {
4353                         write_name_quoted(p->two->path, file, '\n');
4354                 }
4355         }
4356 }
4357
4358 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4359                         const char *line_prefix)
4360 {
4361         char *names = pprint_rename(p->one->path, p->two->path);
4362
4363         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4364         free(names);
4365         show_mode_change(file, p, 0, line_prefix);
4366 }
4367
4368 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4369 {
4370         FILE *file = opt->file;
4371         const char *line_prefix = diff_line_prefix(opt);
4372
4373         switch(p->status) {
4374         case DIFF_STATUS_DELETED:
4375                 fputs(line_prefix, file);
4376                 show_file_mode_name(file, "delete", p->one);
4377                 break;
4378         case DIFF_STATUS_ADDED:
4379                 fputs(line_prefix, file);
4380                 show_file_mode_name(file, "create", p->two);
4381                 break;
4382         case DIFF_STATUS_COPIED:
4383                 fputs(line_prefix, file);
4384                 show_rename_copy(file, "copy", p, line_prefix);
4385                 break;
4386         case DIFF_STATUS_RENAMED:
4387                 fputs(line_prefix, file);
4388                 show_rename_copy(file, "rename", p, line_prefix);
4389                 break;
4390         default:
4391                 if (p->score) {
4392                         fprintf(file, "%s rewrite ", line_prefix);
4393                         write_name_quoted(p->two->path, file, ' ');
4394                         fprintf(file, "(%d%%)\n", similarity_index(p));
4395                 }
4396                 show_mode_change(file, p, !p->score, line_prefix);
4397                 break;
4398         }
4399 }
4400
4401 struct patch_id_t {
4402         git_SHA_CTX *ctx;
4403         int patchlen;
4404 };
4405
4406 static int remove_space(char *line, int len)
4407 {
4408         int i;
4409         char *dst = line;
4410         unsigned char c;
4411
4412         for (i = 0; i < len; i++)
4413                 if (!isspace((c = line[i])))
4414                         *dst++ = c;
4415
4416         return dst - line;
4417 }
4418
4419 static void patch_id_consume(void *priv, char *line, unsigned long len)
4420 {
4421         struct patch_id_t *data = priv;
4422         int new_len;
4423
4424         /* Ignore line numbers when computing the SHA1 of the patch */
4425         if (starts_with(line, "@@ -"))
4426                 return;
4427
4428         new_len = remove_space(line, len);
4429
4430         git_SHA1_Update(data->ctx, line, new_len);
4431         data->patchlen += new_len;
4432 }
4433
4434 /* returns 0 upon success, and writes result into sha1 */
4435 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4436 {
4437         struct diff_queue_struct *q = &diff_queued_diff;
4438         int i;
4439         git_SHA_CTX ctx;
4440         struct patch_id_t data;
4441         char buffer[PATH_MAX * 4 + 20];
4442
4443         git_SHA1_Init(&ctx);
4444         memset(&data, 0, sizeof(struct patch_id_t));
4445         data.ctx = &ctx;
4446
4447         for (i = 0; i < q->nr; i++) {
4448                 xpparam_t xpp;
4449                 xdemitconf_t xecfg;
4450                 mmfile_t mf1, mf2;
4451                 struct diff_filepair *p = q->queue[i];
4452                 int len1, len2;
4453
4454                 memset(&xpp, 0, sizeof(xpp));
4455                 memset(&xecfg, 0, sizeof(xecfg));
4456                 if (p->status == 0)
4457                         return error("internal diff status error");
4458                 if (p->status == DIFF_STATUS_UNKNOWN)
4459                         continue;
4460                 if (diff_unmodified_pair(p))
4461                         continue;
4462                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4463                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4464                         continue;
4465                 if (DIFF_PAIR_UNMERGED(p))
4466                         continue;
4467
4468                 diff_fill_sha1_info(p->one);
4469                 diff_fill_sha1_info(p->two);
4470                 if (fill_mmfile(&mf1, p->one) < 0 ||
4471                                 fill_mmfile(&mf2, p->two) < 0)
4472                         return error("unable to read files to diff");
4473
4474                 len1 = remove_space(p->one->path, strlen(p->one->path));
4475                 len2 = remove_space(p->two->path, strlen(p->two->path));
4476                 if (p->one->mode == 0)
4477                         len1 = snprintf(buffer, sizeof(buffer),
4478                                         "diff--gita/%.*sb/%.*s"
4479                                         "newfilemode%06o"
4480                                         "---/dev/null"
4481                                         "+++b/%.*s",
4482                                         len1, p->one->path,
4483                                         len2, p->two->path,
4484                                         p->two->mode,
4485                                         len2, p->two->path);
4486                 else if (p->two->mode == 0)
4487                         len1 = snprintf(buffer, sizeof(buffer),
4488                                         "diff--gita/%.*sb/%.*s"
4489                                         "deletedfilemode%06o"
4490                                         "---a/%.*s"
4491                                         "+++/dev/null",
4492                                         len1, p->one->path,
4493                                         len2, p->two->path,
4494                                         p->one->mode,
4495                                         len1, p->one->path);
4496                 else
4497                         len1 = snprintf(buffer, sizeof(buffer),
4498                                         "diff--gita/%.*sb/%.*s"
4499                                         "---a/%.*s"
4500                                         "+++b/%.*s",
4501                                         len1, p->one->path,
4502                                         len2, p->two->path,
4503                                         len1, p->one->path,
4504                                         len2, p->two->path);
4505                 git_SHA1_Update(&ctx, buffer, len1);
4506
4507                 if (diff_filespec_is_binary(p->one) ||
4508                     diff_filespec_is_binary(p->two)) {
4509                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4510                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4511                         continue;
4512                 }
4513
4514                 xpp.flags = 0;
4515                 xecfg.ctxlen = 3;
4516                 xecfg.flags = 0;
4517                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4518                                   &xpp, &xecfg))
4519                         return error("unable to generate patch-id diff for %s",
4520                                      p->one->path);
4521         }
4522
4523         git_SHA1_Final(sha1, &ctx);
4524         return 0;
4525 }
4526
4527 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4528 {
4529         struct diff_queue_struct *q = &diff_queued_diff;
4530         int i;
4531         int result = diff_get_patch_id(options, sha1);
4532
4533         for (i = 0; i < q->nr; i++)
4534                 diff_free_filepair(q->queue[i]);
4535
4536         free(q->queue);
4537         DIFF_QUEUE_CLEAR(q);
4538
4539         return result;
4540 }
4541
4542 static int is_summary_empty(const struct diff_queue_struct *q)
4543 {
4544         int i;
4545
4546         for (i = 0; i < q->nr; i++) {
4547                 const struct diff_filepair *p = q->queue[i];
4548
4549                 switch (p->status) {
4550                 case DIFF_STATUS_DELETED:
4551                 case DIFF_STATUS_ADDED:
4552                 case DIFF_STATUS_COPIED:
4553                 case DIFF_STATUS_RENAMED:
4554                         return 0;
4555                 default:
4556                         if (p->score)
4557                                 return 0;
4558                         if (p->one->mode && p->two->mode &&
4559                             p->one->mode != p->two->mode)
4560                                 return 0;
4561                         break;
4562                 }
4563         }
4564         return 1;
4565 }
4566
4567 static const char rename_limit_warning[] =
4568 "inexact rename detection was skipped due to too many files.";
4569
4570 static const char degrade_cc_to_c_warning[] =
4571 "only found copies from modified paths due to too many files.";
4572
4573 static const char rename_limit_advice[] =
4574 "you may want to set your %s variable to at least "
4575 "%d and retry the command.";
4576
4577 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4578 {
4579         if (degraded_cc)
4580                 warning(degrade_cc_to_c_warning);
4581         else if (needed)
4582                 warning(rename_limit_warning);
4583         else
4584                 return;
4585         if (0 < needed && needed < 32767)
4586                 warning(rename_limit_advice, varname, needed);
4587 }
4588
4589 void diff_flush(struct diff_options *options)
4590 {
4591         struct diff_queue_struct *q = &diff_queued_diff;
4592         int i, output_format = options->output_format;
4593         int separator = 0;
4594         int dirstat_by_line = 0;
4595
4596         /*
4597          * Order: raw, stat, summary, patch
4598          * or:    name/name-status/checkdiff (other bits clear)
4599          */
4600         if (!q->nr)
4601                 goto free_queue;
4602
4603         if (output_format & (DIFF_FORMAT_RAW |
4604                              DIFF_FORMAT_NAME |
4605                              DIFF_FORMAT_NAME_STATUS |
4606                              DIFF_FORMAT_CHECKDIFF)) {
4607                 for (i = 0; i < q->nr; i++) {
4608                         struct diff_filepair *p = q->queue[i];
4609                         if (check_pair_status(p))
4610                                 flush_one_pair(p, options);
4611                 }
4612                 separator++;
4613         }
4614
4615         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4616                 dirstat_by_line = 1;
4617
4618         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4619             dirstat_by_line) {
4620                 struct diffstat_t diffstat;
4621
4622                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4623                 for (i = 0; i < q->nr; i++) {
4624                         struct diff_filepair *p = q->queue[i];
4625                         if (check_pair_status(p))
4626                                 diff_flush_stat(p, options, &diffstat);
4627                 }
4628                 if (output_format & DIFF_FORMAT_NUMSTAT)
4629                         show_numstat(&diffstat, options);
4630                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4631                         show_stats(&diffstat, options);
4632                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4633                         show_shortstats(&diffstat, options);
4634                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4635                         show_dirstat_by_line(&diffstat, options);
4636                 free_diffstat_info(&diffstat);
4637                 separator++;
4638         }
4639         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4640                 show_dirstat(options);
4641
4642         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4643                 for (i = 0; i < q->nr; i++) {
4644                         diff_summary(options, q->queue[i]);
4645                 }
4646                 separator++;
4647         }
4648
4649         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4650             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4651             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4652                 /*
4653                  * run diff_flush_patch for the exit status. setting
4654                  * options->file to /dev/null should be safe, because we
4655                  * aren't supposed to produce any output anyway.
4656                  */
4657                 if (options->close_file)
4658                         fclose(options->file);
4659                 options->file = fopen("/dev/null", "w");
4660                 if (!options->file)
4661                         die_errno("Could not open /dev/null");
4662                 options->close_file = 1;
4663                 for (i = 0; i < q->nr; i++) {
4664                         struct diff_filepair *p = q->queue[i];
4665                         if (check_pair_status(p))
4666                                 diff_flush_patch(p, options);
4667                         if (options->found_changes)
4668                                 break;
4669                 }
4670         }
4671
4672         if (output_format & DIFF_FORMAT_PATCH) {
4673                 if (separator) {
4674                         fprintf(options->file, "%s%c",
4675                                 diff_line_prefix(options),
4676                                 options->line_termination);
4677                         if (options->stat_sep) {
4678                                 /* attach patch instead of inline */
4679                                 fputs(options->stat_sep, options->file);
4680                         }
4681                 }
4682
4683                 for (i = 0; i < q->nr; i++) {
4684                         struct diff_filepair *p = q->queue[i];
4685                         if (check_pair_status(p))
4686                                 diff_flush_patch(p, options);
4687                 }
4688         }
4689
4690         if (output_format & DIFF_FORMAT_CALLBACK)
4691                 options->format_callback(q, options, options->format_callback_data);
4692
4693         for (i = 0; i < q->nr; i++)
4694                 diff_free_filepair(q->queue[i]);
4695 free_queue:
4696         free(q->queue);
4697         DIFF_QUEUE_CLEAR(q);
4698         if (options->close_file)
4699                 fclose(options->file);
4700
4701         /*
4702          * Report the content-level differences with HAS_CHANGES;
4703          * diff_addremove/diff_change does not set the bit when
4704          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4705          */
4706         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4707                 if (options->found_changes)
4708                         DIFF_OPT_SET(options, HAS_CHANGES);
4709                 else
4710                         DIFF_OPT_CLR(options, HAS_CHANGES);
4711         }
4712 }
4713
4714 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4715 {
4716         return (((p->status == DIFF_STATUS_MODIFIED) &&
4717                  ((p->score &&
4718                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4719                   (!p->score &&
4720                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4721                 ((p->status != DIFF_STATUS_MODIFIED) &&
4722                  filter_bit_tst(p->status, options)));
4723 }
4724
4725 static void diffcore_apply_filter(struct diff_options *options)
4726 {
4727         int i;
4728         struct diff_queue_struct *q = &diff_queued_diff;
4729         struct diff_queue_struct outq;
4730
4731         DIFF_QUEUE_CLEAR(&outq);
4732
4733         if (!options->filter)
4734                 return;
4735
4736         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4737                 int found;
4738                 for (i = found = 0; !found && i < q->nr; i++) {
4739                         if (match_filter(options, q->queue[i]))
4740                                 found++;
4741                 }
4742                 if (found)
4743                         return;
4744
4745                 /* otherwise we will clear the whole queue
4746                  * by copying the empty outq at the end of this
4747                  * function, but first clear the current entries
4748                  * in the queue.
4749                  */
4750                 for (i = 0; i < q->nr; i++)
4751                         diff_free_filepair(q->queue[i]);
4752         }
4753         else {
4754                 /* Only the matching ones */
4755                 for (i = 0; i < q->nr; i++) {
4756                         struct diff_filepair *p = q->queue[i];
4757                         if (match_filter(options, p))
4758                                 diff_q(&outq, p);
4759                         else
4760                                 diff_free_filepair(p);
4761                 }
4762         }
4763         free(q->queue);
4764         *q = outq;
4765 }
4766
4767 /* Check whether two filespecs with the same mode and size are identical */
4768 static int diff_filespec_is_identical(struct diff_filespec *one,
4769                                       struct diff_filespec *two)
4770 {
4771         if (S_ISGITLINK(one->mode))
4772                 return 0;
4773         if (diff_populate_filespec(one, 0))
4774                 return 0;
4775         if (diff_populate_filespec(two, 0))
4776                 return 0;
4777         return !memcmp(one->data, two->data, one->size);
4778 }
4779
4780 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4781 {
4782         if (p->done_skip_stat_unmatch)
4783                 return p->skip_stat_unmatch_result;
4784
4785         p->done_skip_stat_unmatch = 1;
4786         p->skip_stat_unmatch_result = 0;
4787         /*
4788          * 1. Entries that come from stat info dirtiness
4789          *    always have both sides (iow, not create/delete),
4790          *    one side of the object name is unknown, with
4791          *    the same mode and size.  Keep the ones that
4792          *    do not match these criteria.  They have real
4793          *    differences.
4794          *
4795          * 2. At this point, the file is known to be modified,
4796          *    with the same mode and size, and the object
4797          *    name of one side is unknown.  Need to inspect
4798          *    the identical contents.
4799          */
4800         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4801             !DIFF_FILE_VALID(p->two) ||
4802             (p->one->sha1_valid && p->two->sha1_valid) ||
4803             (p->one->mode != p->two->mode) ||
4804             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4805             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4806             (p->one->size != p->two->size) ||
4807             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4808                 p->skip_stat_unmatch_result = 1;
4809         return p->skip_stat_unmatch_result;
4810 }
4811
4812 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4813 {
4814         int i;
4815         struct diff_queue_struct *q = &diff_queued_diff;
4816         struct diff_queue_struct outq;
4817         DIFF_QUEUE_CLEAR(&outq);
4818
4819         for (i = 0; i < q->nr; i++) {
4820                 struct diff_filepair *p = q->queue[i];
4821
4822                 if (diff_filespec_check_stat_unmatch(p))
4823                         diff_q(&outq, p);
4824                 else {
4825                         /*
4826                          * The caller can subtract 1 from skip_stat_unmatch
4827                          * to determine how many paths were dirty only
4828                          * due to stat info mismatch.
4829                          */
4830                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4831                                 diffopt->skip_stat_unmatch++;
4832                         diff_free_filepair(p);
4833                 }
4834         }
4835         free(q->queue);
4836         *q = outq;
4837 }
4838
4839 static int diffnamecmp(const void *a_, const void *b_)
4840 {
4841         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4842         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4843         const char *name_a, *name_b;
4844
4845         name_a = a->one ? a->one->path : a->two->path;
4846         name_b = b->one ? b->one->path : b->two->path;
4847         return strcmp(name_a, name_b);
4848 }
4849
4850 void diffcore_fix_diff_index(struct diff_options *options)
4851 {
4852         struct diff_queue_struct *q = &diff_queued_diff;
4853         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4854 }
4855
4856 void diffcore_std(struct diff_options *options)
4857 {
4858         /* NOTE please keep the following in sync with diff_tree_combined() */
4859         if (options->skip_stat_unmatch)
4860                 diffcore_skip_stat_unmatch(options);
4861         if (!options->found_follow) {
4862                 /* See try_to_follow_renames() in tree-diff.c */
4863                 if (options->break_opt != -1)
4864                         diffcore_break(options->break_opt);
4865                 if (options->detect_rename)
4866                         diffcore_rename(options);
4867                 if (options->break_opt != -1)
4868                         diffcore_merge_broken();
4869         }
4870         if (options->pickaxe)
4871                 diffcore_pickaxe(options);
4872         if (options->orderfile)
4873                 diffcore_order(options->orderfile);
4874         if (!options->found_follow)
4875                 /* See try_to_follow_renames() in tree-diff.c */
4876                 diff_resolve_rename_copy();
4877         diffcore_apply_filter(options);
4878
4879         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4880                 DIFF_OPT_SET(options, HAS_CHANGES);
4881         else
4882                 DIFF_OPT_CLR(options, HAS_CHANGES);
4883
4884         options->found_follow = 0;
4885 }
4886
4887 int diff_result_code(struct diff_options *opt, int status)
4888 {
4889         int result = 0;
4890
4891         diff_warn_rename_limit("diff.renameLimit",
4892                                opt->needed_rename_limit,
4893                                opt->degraded_cc_to_c);
4894         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4895             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4896                 return status;
4897         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4898             DIFF_OPT_TST(opt, HAS_CHANGES))
4899                 result |= 01;
4900         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4901             DIFF_OPT_TST(opt, CHECK_FAILED))
4902                 result |= 02;
4903         return result;
4904 }
4905
4906 int diff_can_quit_early(struct diff_options *opt)
4907 {
4908         return (DIFF_OPT_TST(opt, QUICK) &&
4909                 !opt->filter &&
4910                 DIFF_OPT_TST(opt, HAS_CHANGES));
4911 }
4912
4913 /*
4914  * Shall changes to this submodule be ignored?
4915  *
4916  * Submodule changes can be configured to be ignored separately for each path,
4917  * but that configuration can be overridden from the command line.
4918  */
4919 static int is_submodule_ignored(const char *path, struct diff_options *options)
4920 {
4921         int ignored = 0;
4922         unsigned orig_flags = options->flags;
4923         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4924                 set_diffopt_flags_from_submodule_config(options, path);
4925         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4926                 ignored = 1;
4927         options->flags = orig_flags;
4928         return ignored;
4929 }
4930
4931 void diff_addremove(struct diff_options *options,
4932                     int addremove, unsigned mode,
4933                     const unsigned char *sha1,
4934                     int sha1_valid,
4935                     const char *concatpath, unsigned dirty_submodule)
4936 {
4937         struct diff_filespec *one, *two;
4938
4939         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4940                 return;
4941
4942         /* This may look odd, but it is a preparation for
4943          * feeding "there are unchanged files which should
4944          * not produce diffs, but when you are doing copy
4945          * detection you would need them, so here they are"
4946          * entries to the diff-core.  They will be prefixed
4947          * with something like '=' or '*' (I haven't decided
4948          * which but should not make any difference).
4949          * Feeding the same new and old to diff_change()
4950          * also has the same effect.
4951          * Before the final output happens, they are pruned after
4952          * merged into rename/copy pairs as appropriate.
4953          */
4954         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4955                 addremove = (addremove == '+' ? '-' :
4956                              addremove == '-' ? '+' : addremove);
4957
4958         if (options->prefix &&
4959             strncmp(concatpath, options->prefix, options->prefix_length))
4960                 return;
4961
4962         one = alloc_filespec(concatpath);
4963         two = alloc_filespec(concatpath);
4964
4965         if (addremove != '+')
4966                 fill_filespec(one, sha1, sha1_valid, mode);
4967         if (addremove != '-') {
4968                 fill_filespec(two, sha1, sha1_valid, mode);
4969                 two->dirty_submodule = dirty_submodule;
4970         }
4971
4972         diff_queue(&diff_queued_diff, one, two);
4973         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4974                 DIFF_OPT_SET(options, HAS_CHANGES);
4975 }
4976
4977 void diff_change(struct diff_options *options,
4978                  unsigned old_mode, unsigned new_mode,
4979                  const unsigned char *old_sha1,
4980                  const unsigned char *new_sha1,
4981                  int old_sha1_valid, int new_sha1_valid,
4982                  const char *concatpath,
4983                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4984 {
4985         struct diff_filespec *one, *two;
4986         struct diff_filepair *p;
4987
4988         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4989             is_submodule_ignored(concatpath, options))
4990                 return;
4991
4992         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4993                 unsigned tmp;
4994                 const unsigned char *tmp_c;
4995                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4996                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4997                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4998                         new_sha1_valid = tmp;
4999                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
5000                         new_dirty_submodule = tmp;
5001         }
5002
5003         if (options->prefix &&
5004             strncmp(concatpath, options->prefix, options->prefix_length))
5005                 return;
5006
5007         one = alloc_filespec(concatpath);
5008         two = alloc_filespec(concatpath);
5009         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5010         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5011         one->dirty_submodule = old_dirty_submodule;
5012         two->dirty_submodule = new_dirty_submodule;
5013         p = diff_queue(&diff_queued_diff, one, two);
5014
5015         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5016                 return;
5017
5018         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5019             !diff_filespec_check_stat_unmatch(p))
5020                 return;
5021
5022         DIFF_OPT_SET(options, HAS_CHANGES);
5023 }
5024
5025 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5026 {
5027         struct diff_filepair *pair;
5028         struct diff_filespec *one, *two;
5029
5030         if (options->prefix &&
5031             strncmp(path, options->prefix, options->prefix_length))
5032                 return NULL;
5033
5034         one = alloc_filespec(path);
5035         two = alloc_filespec(path);
5036         pair = diff_queue(&diff_queued_diff, one, two);
5037         pair->is_unmerged = 1;
5038         return pair;
5039 }
5040
5041 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5042                 size_t *outsize)
5043 {
5044         struct diff_tempfile *temp;
5045         const char *argv[3];
5046         const char **arg = argv;
5047         struct child_process child = CHILD_PROCESS_INIT;
5048         struct strbuf buf = STRBUF_INIT;
5049         int err = 0;
5050
5051         temp = prepare_temp_file(spec->path, spec);
5052         *arg++ = pgm;
5053         *arg++ = temp->name;
5054         *arg = NULL;
5055
5056         child.use_shell = 1;
5057         child.argv = argv;
5058         child.out = -1;
5059         if (start_command(&child)) {
5060                 remove_tempfile();
5061                 return NULL;
5062         }
5063
5064         if (strbuf_read(&buf, child.out, 0) < 0)
5065                 err = error("error reading from textconv command '%s'", pgm);
5066         close(child.out);
5067
5068         if (finish_command(&child) || err) {
5069                 strbuf_release(&buf);
5070                 remove_tempfile();
5071                 return NULL;
5072         }
5073         remove_tempfile();
5074
5075         return strbuf_detach(&buf, outsize);
5076 }
5077
5078 size_t fill_textconv(struct userdiff_driver *driver,
5079                      struct diff_filespec *df,
5080                      char **outbuf)
5081 {
5082         size_t size;
5083
5084         if (!driver || !driver->textconv) {
5085                 if (!DIFF_FILE_VALID(df)) {
5086                         *outbuf = "";
5087                         return 0;
5088                 }
5089                 if (diff_populate_filespec(df, 0))
5090                         die("unable to read files to diff");
5091                 *outbuf = df->data;
5092                 return df->size;
5093         }
5094
5095         if (driver->textconv_cache && df->sha1_valid) {
5096                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5097                                           &size);
5098                 if (*outbuf)
5099                         return size;
5100         }
5101
5102         *outbuf = run_textconv(driver->textconv, df, &size);
5103         if (!*outbuf)
5104                 die("unable to read files to diff");
5105
5106         if (driver->textconv_cache && df->sha1_valid) {
5107                 /* ignore errors, as we might be in a readonly repository */
5108                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5109                                 size);
5110                 /*
5111                  * we could save up changes and flush them all at the end,
5112                  * but we would need an extra call after all diffing is done.
5113                  * Since generating a cache entry is the slow path anyway,
5114                  * this extra overhead probably isn't a big deal.
5115                  */
5116                 notes_cache_write(driver->textconv_cache);
5117         }
5118
5119         return size;
5120 }
5121
5122 void setup_diff_pager(struct diff_options *opt)
5123 {
5124         /*
5125          * If the user asked for our exit code, then either they want --quiet
5126          * or --exit-code. We should definitely not bother with a pager in the
5127          * former case, as we will generate no output. Since we still properly
5128          * report our exit code even when a pager is run, we _could_ run a
5129          * pager with --exit-code. But since we have not done so historically,
5130          * and because it is easy to find people oneline advising "git diff
5131          * --exit-code" in hooks and other scripts, we do not do so.
5132          */
5133         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5134             check_pager_config("diff") != 0)
5135                 setup_pager();
5136 }