Merge branch 'for-junio' of git://bogomips.org/git-svn
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
14 #include "userdiff.h"
15 #include "sigchain.h"
16 #include "submodule.h"
17 #include "ll-merge.h"
18 #include "string-list.h"
19 #include "argv-array.h"
20
21 #ifdef NO_FAST_WORKING_DIRECTORY
22 #define FAST_WORKING_DIRECTORY 0
23 #else
24 #define FAST_WORKING_DIRECTORY 1
25 #endif
26
27 static int diff_detect_rename_default;
28 static int diff_rename_limit_default = 400;
29 static int diff_suppress_blank_empty;
30 static int diff_use_color_default = -1;
31 static int diff_context_default = 3;
32 static const char *diff_word_regex_cfg;
33 static const char *external_diff_cmd_cfg;
34 static const char *diff_order_file_cfg;
35 int diff_auto_refresh_index = 1;
36 static int diff_mnemonic_prefix;
37 static int diff_no_prefix;
38 static int diff_stat_graph_width;
39 static int diff_dirstat_permille_default = 30;
40 static struct diff_options default_diff_options;
41 static long diff_algorithm;
42
43 static char diff_colors[][COLOR_MAXLEN] = {
44         GIT_COLOR_RESET,
45         GIT_COLOR_NORMAL,       /* PLAIN */
46         GIT_COLOR_BOLD,         /* METAINFO */
47         GIT_COLOR_CYAN,         /* FRAGINFO */
48         GIT_COLOR_RED,          /* OLD */
49         GIT_COLOR_GREEN,        /* NEW */
50         GIT_COLOR_YELLOW,       /* COMMIT */
51         GIT_COLOR_BG_RED,       /* WHITESPACE */
52         GIT_COLOR_NORMAL,       /* FUNCINFO */
53 };
54
55 static int parse_diff_color_slot(const char *var)
56 {
57         if (!strcasecmp(var, "plain"))
58                 return DIFF_PLAIN;
59         if (!strcasecmp(var, "meta"))
60                 return DIFF_METAINFO;
61         if (!strcasecmp(var, "frag"))
62                 return DIFF_FRAGINFO;
63         if (!strcasecmp(var, "old"))
64                 return DIFF_FILE_OLD;
65         if (!strcasecmp(var, "new"))
66                 return DIFF_FILE_NEW;
67         if (!strcasecmp(var, "commit"))
68                 return DIFF_COMMIT;
69         if (!strcasecmp(var, "whitespace"))
70                 return DIFF_WHITESPACE;
71         if (!strcasecmp(var, "func"))
72                 return DIFF_FUNCINFO;
73         return -1;
74 }
75
76 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
77                                 struct strbuf *errmsg)
78 {
79         char *params_copy = xstrdup(params_string);
80         struct string_list params = STRING_LIST_INIT_NODUP;
81         int ret = 0;
82         int i;
83
84         if (*params_copy)
85                 string_list_split_in_place(&params, params_copy, ',', -1);
86         for (i = 0; i < params.nr; i++) {
87                 const char *p = params.items[i].string;
88                 if (!strcmp(p, "changes")) {
89                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
90                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
91                 } else if (!strcmp(p, "lines")) {
92                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
93                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
94                 } else if (!strcmp(p, "files")) {
95                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
96                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
97                 } else if (!strcmp(p, "noncumulative")) {
98                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
99                 } else if (!strcmp(p, "cumulative")) {
100                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
101                 } else if (isdigit(*p)) {
102                         char *end;
103                         int permille = strtoul(p, &end, 10) * 10;
104                         if (*end == '.' && isdigit(*++end)) {
105                                 /* only use first digit */
106                                 permille += *end - '0';
107                                 /* .. and ignore any further digits */
108                                 while (isdigit(*++end))
109                                         ; /* nothing */
110                         }
111                         if (!*end)
112                                 options->dirstat_permille = permille;
113                         else {
114                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
115                                             p);
116                                 ret++;
117                         }
118                 } else {
119                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
120                         ret++;
121                 }
122
123         }
124         string_list_clear(&params, 0);
125         free(params_copy);
126         return ret;
127 }
128
129 static int parse_submodule_params(struct diff_options *options, const char *value)
130 {
131         if (!strcmp(value, "log"))
132                 DIFF_OPT_SET(options, SUBMODULE_LOG);
133         else if (!strcmp(value, "short"))
134                 DIFF_OPT_CLR(options, SUBMODULE_LOG);
135         else
136                 return -1;
137         return 0;
138 }
139
140 static int git_config_rename(const char *var, const char *value)
141 {
142         if (!value)
143                 return DIFF_DETECT_RENAME;
144         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
145                 return  DIFF_DETECT_COPY;
146         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
147 }
148
149 long parse_algorithm_value(const char *value)
150 {
151         if (!value)
152                 return -1;
153         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
154                 return 0;
155         else if (!strcasecmp(value, "minimal"))
156                 return XDF_NEED_MINIMAL;
157         else if (!strcasecmp(value, "patience"))
158                 return XDF_PATIENCE_DIFF;
159         else if (!strcasecmp(value, "histogram"))
160                 return XDF_HISTOGRAM_DIFF;
161         return -1;
162 }
163
164 /*
165  * These are to give UI layer defaults.
166  * The core-level commands such as git-diff-files should
167  * never be affected by the setting of diff.renames
168  * the user happens to have in the configuration file.
169  */
170 int git_diff_ui_config(const char *var, const char *value, void *cb)
171 {
172         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
173                 diff_use_color_default = git_config_colorbool(var, value);
174                 return 0;
175         }
176         if (!strcmp(var, "diff.context")) {
177                 diff_context_default = git_config_int(var, value);
178                 if (diff_context_default < 0)
179                         return -1;
180                 return 0;
181         }
182         if (!strcmp(var, "diff.renames")) {
183                 diff_detect_rename_default = git_config_rename(var, value);
184                 return 0;
185         }
186         if (!strcmp(var, "diff.autorefreshindex")) {
187                 diff_auto_refresh_index = git_config_bool(var, value);
188                 return 0;
189         }
190         if (!strcmp(var, "diff.mnemonicprefix")) {
191                 diff_mnemonic_prefix = git_config_bool(var, value);
192                 return 0;
193         }
194         if (!strcmp(var, "diff.noprefix")) {
195                 diff_no_prefix = git_config_bool(var, value);
196                 return 0;
197         }
198         if (!strcmp(var, "diff.statgraphwidth")) {
199                 diff_stat_graph_width = git_config_int(var, value);
200                 return 0;
201         }
202         if (!strcmp(var, "diff.external"))
203                 return git_config_string(&external_diff_cmd_cfg, var, value);
204         if (!strcmp(var, "diff.wordregex"))
205                 return git_config_string(&diff_word_regex_cfg, var, value);
206         if (!strcmp(var, "diff.orderfile"))
207                 return git_config_pathname(&diff_order_file_cfg, var, value);
208
209         if (!strcmp(var, "diff.ignoresubmodules"))
210                 handle_ignore_submodules_arg(&default_diff_options, value);
211
212         if (!strcmp(var, "diff.submodule")) {
213                 if (parse_submodule_params(&default_diff_options, value))
214                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
215                                 value);
216                 return 0;
217         }
218
219         if (!strcmp(var, "diff.algorithm")) {
220                 diff_algorithm = parse_algorithm_value(value);
221                 if (diff_algorithm < 0)
222                         return -1;
223                 return 0;
224         }
225
226         if (git_color_config(var, value, cb) < 0)
227                 return -1;
228
229         return git_diff_basic_config(var, value, cb);
230 }
231
232 int git_diff_basic_config(const char *var, const char *value, void *cb)
233 {
234         const char *name;
235
236         if (!strcmp(var, "diff.renamelimit")) {
237                 diff_rename_limit_default = git_config_int(var, value);
238                 return 0;
239         }
240
241         if (userdiff_config(var, value) < 0)
242                 return -1;
243
244         if (skip_prefix(var, "diff.color.", &name) ||
245             skip_prefix(var, "color.diff.", &name)) {
246                 int slot = parse_diff_color_slot(name);
247                 if (slot < 0)
248                         return 0;
249                 if (!value)
250                         return config_error_nonbool(var);
251                 return color_parse(value, diff_colors[slot]);
252         }
253
254         /* like GNU diff's --suppress-blank-empty option  */
255         if (!strcmp(var, "diff.suppressblankempty") ||
256                         /* for backwards compatibility */
257                         !strcmp(var, "diff.suppress-blank-empty")) {
258                 diff_suppress_blank_empty = git_config_bool(var, value);
259                 return 0;
260         }
261
262         if (!strcmp(var, "diff.dirstat")) {
263                 struct strbuf errmsg = STRBUF_INIT;
264                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
265                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
266                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
267                                 errmsg.buf);
268                 strbuf_release(&errmsg);
269                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
270                 return 0;
271         }
272
273         if (starts_with(var, "submodule."))
274                 return parse_submodule_config_option(var, value);
275
276         return git_default_config(var, value, cb);
277 }
278
279 static char *quote_two(const char *one, const char *two)
280 {
281         int need_one = quote_c_style(one, NULL, NULL, 1);
282         int need_two = quote_c_style(two, NULL, NULL, 1);
283         struct strbuf res = STRBUF_INIT;
284
285         if (need_one + need_two) {
286                 strbuf_addch(&res, '"');
287                 quote_c_style(one, &res, NULL, 1);
288                 quote_c_style(two, &res, NULL, 1);
289                 strbuf_addch(&res, '"');
290         } else {
291                 strbuf_addstr(&res, one);
292                 strbuf_addstr(&res, two);
293         }
294         return strbuf_detach(&res, NULL);
295 }
296
297 static const char *external_diff(void)
298 {
299         static const char *external_diff_cmd = NULL;
300         static int done_preparing = 0;
301
302         if (done_preparing)
303                 return external_diff_cmd;
304         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
305         if (!external_diff_cmd)
306                 external_diff_cmd = external_diff_cmd_cfg;
307         done_preparing = 1;
308         return external_diff_cmd;
309 }
310
311 static struct diff_tempfile {
312         const char *name; /* filename external diff should read from */
313         char hex[41];
314         char mode[10];
315         char tmp_path[PATH_MAX];
316 } diff_temp[2];
317
318 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
319
320 struct emit_callback {
321         int color_diff;
322         unsigned ws_rule;
323         int blank_at_eof_in_preimage;
324         int blank_at_eof_in_postimage;
325         int lno_in_preimage;
326         int lno_in_postimage;
327         sane_truncate_fn truncate;
328         const char **label_path;
329         struct diff_words_data *diff_words;
330         struct diff_options *opt;
331         int *found_changesp;
332         struct strbuf *header;
333 };
334
335 static int count_lines(const char *data, int size)
336 {
337         int count, ch, completely_empty = 1, nl_just_seen = 0;
338         count = 0;
339         while (0 < size--) {
340                 ch = *data++;
341                 if (ch == '\n') {
342                         count++;
343                         nl_just_seen = 1;
344                         completely_empty = 0;
345                 }
346                 else {
347                         nl_just_seen = 0;
348                         completely_empty = 0;
349                 }
350         }
351         if (completely_empty)
352                 return 0;
353         if (!nl_just_seen)
354                 count++; /* no trailing newline */
355         return count;
356 }
357
358 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
359 {
360         if (!DIFF_FILE_VALID(one)) {
361                 mf->ptr = (char *)""; /* does not matter */
362                 mf->size = 0;
363                 return 0;
364         }
365         else if (diff_populate_filespec(one, 0))
366                 return -1;
367
368         mf->ptr = one->data;
369         mf->size = one->size;
370         return 0;
371 }
372
373 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
374 static unsigned long diff_filespec_size(struct diff_filespec *one)
375 {
376         if (!DIFF_FILE_VALID(one))
377                 return 0;
378         diff_populate_filespec(one, CHECK_SIZE_ONLY);
379         return one->size;
380 }
381
382 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
383 {
384         char *ptr = mf->ptr;
385         long size = mf->size;
386         int cnt = 0;
387
388         if (!size)
389                 return cnt;
390         ptr += size - 1; /* pointing at the very end */
391         if (*ptr != '\n')
392                 ; /* incomplete line */
393         else
394                 ptr--; /* skip the last LF */
395         while (mf->ptr < ptr) {
396                 char *prev_eol;
397                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
398                         if (*prev_eol == '\n')
399                                 break;
400                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
401                         break;
402                 cnt++;
403                 ptr = prev_eol - 1;
404         }
405         return cnt;
406 }
407
408 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
409                                struct emit_callback *ecbdata)
410 {
411         int l1, l2, at;
412         unsigned ws_rule = ecbdata->ws_rule;
413         l1 = count_trailing_blank(mf1, ws_rule);
414         l2 = count_trailing_blank(mf2, ws_rule);
415         if (l2 <= l1) {
416                 ecbdata->blank_at_eof_in_preimage = 0;
417                 ecbdata->blank_at_eof_in_postimage = 0;
418                 return;
419         }
420         at = count_lines(mf1->ptr, mf1->size);
421         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
422
423         at = count_lines(mf2->ptr, mf2->size);
424         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
425 }
426
427 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
428                         int first, const char *line, int len)
429 {
430         int has_trailing_newline, has_trailing_carriage_return;
431         int nofirst;
432         FILE *file = o->file;
433
434         fputs(diff_line_prefix(o), file);
435
436         if (len == 0) {
437                 has_trailing_newline = (first == '\n');
438                 has_trailing_carriage_return = (!has_trailing_newline &&
439                                                 (first == '\r'));
440                 nofirst = has_trailing_newline || has_trailing_carriage_return;
441         } else {
442                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
443                 if (has_trailing_newline)
444                         len--;
445                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
446                 if (has_trailing_carriage_return)
447                         len--;
448                 nofirst = 0;
449         }
450
451         if (len || !nofirst) {
452                 fputs(set, file);
453                 if (!nofirst)
454                         fputc(first, file);
455                 fwrite(line, len, 1, file);
456                 fputs(reset, file);
457         }
458         if (has_trailing_carriage_return)
459                 fputc('\r', file);
460         if (has_trailing_newline)
461                 fputc('\n', file);
462 }
463
464 static void emit_line(struct diff_options *o, const char *set, const char *reset,
465                       const char *line, int len)
466 {
467         emit_line_0(o, set, reset, line[0], line+1, len-1);
468 }
469
470 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
471 {
472         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
473               ecbdata->blank_at_eof_in_preimage &&
474               ecbdata->blank_at_eof_in_postimage &&
475               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
476               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
477                 return 0;
478         return ws_blank_line(line, len, ecbdata->ws_rule);
479 }
480
481 static void emit_add_line(const char *reset,
482                           struct emit_callback *ecbdata,
483                           const char *line, int len)
484 {
485         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
486         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
487
488         if (!*ws)
489                 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
490         else if (new_blank_line_at_eof(ecbdata, line, len))
491                 /* Blank line at EOF - paint '+' as well */
492                 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
493         else {
494                 /* Emit just the prefix, then the rest. */
495                 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
496                 ws_check_emit(line, len, ecbdata->ws_rule,
497                               ecbdata->opt->file, set, reset, ws);
498         }
499 }
500
501 static void emit_hunk_header(struct emit_callback *ecbdata,
502                              const char *line, int len)
503 {
504         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
505         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
506         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
507         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
508         static const char atat[2] = { '@', '@' };
509         const char *cp, *ep;
510         struct strbuf msgbuf = STRBUF_INIT;
511         int org_len = len;
512         int i = 1;
513
514         /*
515          * As a hunk header must begin with "@@ -<old>, +<new> @@",
516          * it always is at least 10 bytes long.
517          */
518         if (len < 10 ||
519             memcmp(line, atat, 2) ||
520             !(ep = memmem(line + 2, len - 2, atat, 2))) {
521                 emit_line(ecbdata->opt, plain, reset, line, len);
522                 return;
523         }
524         ep += 2; /* skip over @@ */
525
526         /* The hunk header in fraginfo color */
527         strbuf_addstr(&msgbuf, frag);
528         strbuf_add(&msgbuf, line, ep - line);
529         strbuf_addstr(&msgbuf, reset);
530
531         /*
532          * trailing "\r\n"
533          */
534         for ( ; i < 3; i++)
535                 if (line[len - i] == '\r' || line[len - i] == '\n')
536                         len--;
537
538         /* blank before the func header */
539         for (cp = ep; ep - line < len; ep++)
540                 if (*ep != ' ' && *ep != '\t')
541                         break;
542         if (ep != cp) {
543                 strbuf_addstr(&msgbuf, plain);
544                 strbuf_add(&msgbuf, cp, ep - cp);
545                 strbuf_addstr(&msgbuf, reset);
546         }
547
548         if (ep < line + len) {
549                 strbuf_addstr(&msgbuf, func);
550                 strbuf_add(&msgbuf, ep, line + len - ep);
551                 strbuf_addstr(&msgbuf, reset);
552         }
553
554         strbuf_add(&msgbuf, line + len, org_len - len);
555         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
556         strbuf_release(&msgbuf);
557 }
558
559 static struct diff_tempfile *claim_diff_tempfile(void) {
560         int i;
561         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
562                 if (!diff_temp[i].name)
563                         return diff_temp + i;
564         die("BUG: diff is failing to clean up its tempfiles");
565 }
566
567 static int remove_tempfile_installed;
568
569 static void remove_tempfile(void)
570 {
571         int i;
572         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
573                 if (diff_temp[i].name == diff_temp[i].tmp_path)
574                         unlink_or_warn(diff_temp[i].name);
575                 diff_temp[i].name = NULL;
576         }
577 }
578
579 static void remove_tempfile_on_signal(int signo)
580 {
581         remove_tempfile();
582         sigchain_pop(signo);
583         raise(signo);
584 }
585
586 static void print_line_count(FILE *file, int count)
587 {
588         switch (count) {
589         case 0:
590                 fprintf(file, "0,0");
591                 break;
592         case 1:
593                 fprintf(file, "1");
594                 break;
595         default:
596                 fprintf(file, "1,%d", count);
597                 break;
598         }
599 }
600
601 static void emit_rewrite_lines(struct emit_callback *ecb,
602                                int prefix, const char *data, int size)
603 {
604         const char *endp = NULL;
605         static const char *nneof = " No newline at end of file\n";
606         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
607         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
608
609         while (0 < size) {
610                 int len;
611
612                 endp = memchr(data, '\n', size);
613                 len = endp ? (endp - data + 1) : size;
614                 if (prefix != '+') {
615                         ecb->lno_in_preimage++;
616                         emit_line_0(ecb->opt, old, reset, '-',
617                                     data, len);
618                 } else {
619                         ecb->lno_in_postimage++;
620                         emit_add_line(reset, ecb, data, len);
621                 }
622                 size -= len;
623                 data += len;
624         }
625         if (!endp) {
626                 const char *plain = diff_get_color(ecb->color_diff,
627                                                    DIFF_PLAIN);
628                 putc('\n', ecb->opt->file);
629                 emit_line_0(ecb->opt, plain, reset, '\\',
630                             nneof, strlen(nneof));
631         }
632 }
633
634 static void emit_rewrite_diff(const char *name_a,
635                               const char *name_b,
636                               struct diff_filespec *one,
637                               struct diff_filespec *two,
638                               struct userdiff_driver *textconv_one,
639                               struct userdiff_driver *textconv_two,
640                               struct diff_options *o)
641 {
642         int lc_a, lc_b;
643         const char *name_a_tab, *name_b_tab;
644         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
645         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
646         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
647         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
648         const char *a_prefix, *b_prefix;
649         char *data_one, *data_two;
650         size_t size_one, size_two;
651         struct emit_callback ecbdata;
652         const char *line_prefix = diff_line_prefix(o);
653
654         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
655                 a_prefix = o->b_prefix;
656                 b_prefix = o->a_prefix;
657         } else {
658                 a_prefix = o->a_prefix;
659                 b_prefix = o->b_prefix;
660         }
661
662         name_a += (*name_a == '/');
663         name_b += (*name_b == '/');
664         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
665         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
666
667         strbuf_reset(&a_name);
668         strbuf_reset(&b_name);
669         quote_two_c_style(&a_name, a_prefix, name_a, 0);
670         quote_two_c_style(&b_name, b_prefix, name_b, 0);
671
672         size_one = fill_textconv(textconv_one, one, &data_one);
673         size_two = fill_textconv(textconv_two, two, &data_two);
674
675         memset(&ecbdata, 0, sizeof(ecbdata));
676         ecbdata.color_diff = want_color(o->use_color);
677         ecbdata.found_changesp = &o->found_changes;
678         ecbdata.ws_rule = whitespace_rule(name_b);
679         ecbdata.opt = o;
680         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
681                 mmfile_t mf1, mf2;
682                 mf1.ptr = (char *)data_one;
683                 mf2.ptr = (char *)data_two;
684                 mf1.size = size_one;
685                 mf2.size = size_two;
686                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
687         }
688         ecbdata.lno_in_preimage = 1;
689         ecbdata.lno_in_postimage = 1;
690
691         lc_a = count_lines(data_one, size_one);
692         lc_b = count_lines(data_two, size_two);
693         fprintf(o->file,
694                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
695                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
696                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
697                 line_prefix, fraginfo);
698         if (!o->irreversible_delete)
699                 print_line_count(o->file, lc_a);
700         else
701                 fprintf(o->file, "?,?");
702         fprintf(o->file, " +");
703         print_line_count(o->file, lc_b);
704         fprintf(o->file, " @@%s\n", reset);
705         if (lc_a && !o->irreversible_delete)
706                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
707         if (lc_b)
708                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
709         if (textconv_one)
710                 free((char *)data_one);
711         if (textconv_two)
712                 free((char *)data_two);
713 }
714
715 struct diff_words_buffer {
716         mmfile_t text;
717         long alloc;
718         struct diff_words_orig {
719                 const char *begin, *end;
720         } *orig;
721         int orig_nr, orig_alloc;
722 };
723
724 static void diff_words_append(char *line, unsigned long len,
725                 struct diff_words_buffer *buffer)
726 {
727         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
728         line++;
729         len--;
730         memcpy(buffer->text.ptr + buffer->text.size, line, len);
731         buffer->text.size += len;
732         buffer->text.ptr[buffer->text.size] = '\0';
733 }
734
735 struct diff_words_style_elem {
736         const char *prefix;
737         const char *suffix;
738         const char *color; /* NULL; filled in by the setup code if
739                             * color is enabled */
740 };
741
742 struct diff_words_style {
743         enum diff_words_type type;
744         struct diff_words_style_elem new, old, ctx;
745         const char *newline;
746 };
747
748 static struct diff_words_style diff_words_styles[] = {
749         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
750         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
751         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
752 };
753
754 struct diff_words_data {
755         struct diff_words_buffer minus, plus;
756         const char *current_plus;
757         int last_minus;
758         struct diff_options *opt;
759         regex_t *word_regex;
760         enum diff_words_type type;
761         struct diff_words_style *style;
762 };
763
764 static int fn_out_diff_words_write_helper(FILE *fp,
765                                           struct diff_words_style_elem *st_el,
766                                           const char *newline,
767                                           size_t count, const char *buf,
768                                           const char *line_prefix)
769 {
770         int print = 0;
771
772         while (count) {
773                 char *p = memchr(buf, '\n', count);
774                 if (print)
775                         fputs(line_prefix, fp);
776                 if (p != buf) {
777                         if (st_el->color && fputs(st_el->color, fp) < 0)
778                                 return -1;
779                         if (fputs(st_el->prefix, fp) < 0 ||
780                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
781                             fputs(st_el->suffix, fp) < 0)
782                                 return -1;
783                         if (st_el->color && *st_el->color
784                             && fputs(GIT_COLOR_RESET, fp) < 0)
785                                 return -1;
786                 }
787                 if (!p)
788                         return 0;
789                 if (fputs(newline, fp) < 0)
790                         return -1;
791                 count -= p + 1 - buf;
792                 buf = p + 1;
793                 print = 1;
794         }
795         return 0;
796 }
797
798 /*
799  * '--color-words' algorithm can be described as:
800  *
801  *   1. collect a the minus/plus lines of a diff hunk, divided into
802  *      minus-lines and plus-lines;
803  *
804  *   2. break both minus-lines and plus-lines into words and
805  *      place them into two mmfile_t with one word for each line;
806  *
807  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
808  *
809  * And for the common parts of the both file, we output the plus side text.
810  * diff_words->current_plus is used to trace the current position of the plus file
811  * which printed. diff_words->last_minus is used to trace the last minus word
812  * printed.
813  *
814  * For '--graph' to work with '--color-words', we need to output the graph prefix
815  * on each line of color words output. Generally, there are two conditions on
816  * which we should output the prefix.
817  *
818  *   1. diff_words->last_minus == 0 &&
819  *      diff_words->current_plus == diff_words->plus.text.ptr
820  *
821  *      that is: the plus text must start as a new line, and if there is no minus
822  *      word printed, a graph prefix must be printed.
823  *
824  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
825  *      *(diff_words->current_plus - 1) == '\n'
826  *
827  *      that is: a graph prefix must be printed following a '\n'
828  */
829 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
830 {
831         if ((diff_words->last_minus == 0 &&
832                 diff_words->current_plus == diff_words->plus.text.ptr) ||
833                 (diff_words->current_plus > diff_words->plus.text.ptr &&
834                 *(diff_words->current_plus - 1) == '\n')) {
835                 return 1;
836         } else {
837                 return 0;
838         }
839 }
840
841 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
842 {
843         struct diff_words_data *diff_words = priv;
844         struct diff_words_style *style = diff_words->style;
845         int minus_first, minus_len, plus_first, plus_len;
846         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
847         struct diff_options *opt = diff_words->opt;
848         const char *line_prefix;
849
850         if (line[0] != '@' || parse_hunk_header(line, len,
851                         &minus_first, &minus_len, &plus_first, &plus_len))
852                 return;
853
854         assert(opt);
855         line_prefix = diff_line_prefix(opt);
856
857         /* POSIX requires that first be decremented by one if len == 0... */
858         if (minus_len) {
859                 minus_begin = diff_words->minus.orig[minus_first].begin;
860                 minus_end =
861                         diff_words->minus.orig[minus_first + minus_len - 1].end;
862         } else
863                 minus_begin = minus_end =
864                         diff_words->minus.orig[minus_first].end;
865
866         if (plus_len) {
867                 plus_begin = diff_words->plus.orig[plus_first].begin;
868                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
869         } else
870                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
871
872         if (color_words_output_graph_prefix(diff_words)) {
873                 fputs(line_prefix, diff_words->opt->file);
874         }
875         if (diff_words->current_plus != plus_begin) {
876                 fn_out_diff_words_write_helper(diff_words->opt->file,
877                                 &style->ctx, style->newline,
878                                 plus_begin - diff_words->current_plus,
879                                 diff_words->current_plus, line_prefix);
880                 if (*(plus_begin - 1) == '\n')
881                         fputs(line_prefix, diff_words->opt->file);
882         }
883         if (minus_begin != minus_end) {
884                 fn_out_diff_words_write_helper(diff_words->opt->file,
885                                 &style->old, style->newline,
886                                 minus_end - minus_begin, minus_begin,
887                                 line_prefix);
888         }
889         if (plus_begin != plus_end) {
890                 fn_out_diff_words_write_helper(diff_words->opt->file,
891                                 &style->new, style->newline,
892                                 plus_end - plus_begin, plus_begin,
893                                 line_prefix);
894         }
895
896         diff_words->current_plus = plus_end;
897         diff_words->last_minus = minus_first;
898 }
899
900 /* This function starts looking at *begin, and returns 0 iff a word was found. */
901 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
902                 int *begin, int *end)
903 {
904         if (word_regex && *begin < buffer->size) {
905                 regmatch_t match[1];
906                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
907                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
908                                         '\n', match[0].rm_eo - match[0].rm_so);
909                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
910                         *begin += match[0].rm_so;
911                         return *begin >= *end;
912                 }
913                 return -1;
914         }
915
916         /* find the next word */
917         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
918                 (*begin)++;
919         if (*begin >= buffer->size)
920                 return -1;
921
922         /* find the end of the word */
923         *end = *begin + 1;
924         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
925                 (*end)++;
926
927         return 0;
928 }
929
930 /*
931  * This function splits the words in buffer->text, stores the list with
932  * newline separator into out, and saves the offsets of the original words
933  * in buffer->orig.
934  */
935 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
936                 regex_t *word_regex)
937 {
938         int i, j;
939         long alloc = 0;
940
941         out->size = 0;
942         out->ptr = NULL;
943
944         /* fake an empty "0th" word */
945         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
946         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
947         buffer->orig_nr = 1;
948
949         for (i = 0; i < buffer->text.size; i++) {
950                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
951                         return;
952
953                 /* store original boundaries */
954                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
955                                 buffer->orig_alloc);
956                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
957                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
958                 buffer->orig_nr++;
959
960                 /* store one word */
961                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
962                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
963                 out->ptr[out->size + j - i] = '\n';
964                 out->size += j - i + 1;
965
966                 i = j - 1;
967         }
968 }
969
970 /* this executes the word diff on the accumulated buffers */
971 static void diff_words_show(struct diff_words_data *diff_words)
972 {
973         xpparam_t xpp;
974         xdemitconf_t xecfg;
975         mmfile_t minus, plus;
976         struct diff_words_style *style = diff_words->style;
977
978         struct diff_options *opt = diff_words->opt;
979         const char *line_prefix;
980
981         assert(opt);
982         line_prefix = diff_line_prefix(opt);
983
984         /* special case: only removal */
985         if (!diff_words->plus.text.size) {
986                 fputs(line_prefix, diff_words->opt->file);
987                 fn_out_diff_words_write_helper(diff_words->opt->file,
988                         &style->old, style->newline,
989                         diff_words->minus.text.size,
990                         diff_words->minus.text.ptr, line_prefix);
991                 diff_words->minus.text.size = 0;
992                 return;
993         }
994
995         diff_words->current_plus = diff_words->plus.text.ptr;
996         diff_words->last_minus = 0;
997
998         memset(&xpp, 0, sizeof(xpp));
999         memset(&xecfg, 0, sizeof(xecfg));
1000         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1001         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1002         xpp.flags = 0;
1003         /* as only the hunk header will be parsed, we need a 0-context */
1004         xecfg.ctxlen = 0;
1005         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1006                       &xpp, &xecfg);
1007         free(minus.ptr);
1008         free(plus.ptr);
1009         if (diff_words->current_plus != diff_words->plus.text.ptr +
1010                         diff_words->plus.text.size) {
1011                 if (color_words_output_graph_prefix(diff_words))
1012                         fputs(line_prefix, diff_words->opt->file);
1013                 fn_out_diff_words_write_helper(diff_words->opt->file,
1014                         &style->ctx, style->newline,
1015                         diff_words->plus.text.ptr + diff_words->plus.text.size
1016                         - diff_words->current_plus, diff_words->current_plus,
1017                         line_prefix);
1018         }
1019         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1020 }
1021
1022 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1023 static void diff_words_flush(struct emit_callback *ecbdata)
1024 {
1025         if (ecbdata->diff_words->minus.text.size ||
1026             ecbdata->diff_words->plus.text.size)
1027                 diff_words_show(ecbdata->diff_words);
1028 }
1029
1030 static void diff_filespec_load_driver(struct diff_filespec *one)
1031 {
1032         /* Use already-loaded driver */
1033         if (one->driver)
1034                 return;
1035
1036         if (S_ISREG(one->mode))
1037                 one->driver = userdiff_find_by_path(one->path);
1038
1039         /* Fallback to default settings */
1040         if (!one->driver)
1041                 one->driver = userdiff_find_by_name("default");
1042 }
1043
1044 static const char *userdiff_word_regex(struct diff_filespec *one)
1045 {
1046         diff_filespec_load_driver(one);
1047         return one->driver->word_regex;
1048 }
1049
1050 static void init_diff_words_data(struct emit_callback *ecbdata,
1051                                  struct diff_options *orig_opts,
1052                                  struct diff_filespec *one,
1053                                  struct diff_filespec *two)
1054 {
1055         int i;
1056         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1057         memcpy(o, orig_opts, sizeof(struct diff_options));
1058
1059         ecbdata->diff_words =
1060                 xcalloc(1, sizeof(struct diff_words_data));
1061         ecbdata->diff_words->type = o->word_diff;
1062         ecbdata->diff_words->opt = o;
1063         if (!o->word_regex)
1064                 o->word_regex = userdiff_word_regex(one);
1065         if (!o->word_regex)
1066                 o->word_regex = userdiff_word_regex(two);
1067         if (!o->word_regex)
1068                 o->word_regex = diff_word_regex_cfg;
1069         if (o->word_regex) {
1070                 ecbdata->diff_words->word_regex = (regex_t *)
1071                         xmalloc(sizeof(regex_t));
1072                 if (regcomp(ecbdata->diff_words->word_regex,
1073                             o->word_regex,
1074                             REG_EXTENDED | REG_NEWLINE))
1075                         die ("Invalid regular expression: %s",
1076                              o->word_regex);
1077         }
1078         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1079                 if (o->word_diff == diff_words_styles[i].type) {
1080                         ecbdata->diff_words->style =
1081                                 &diff_words_styles[i];
1082                         break;
1083                 }
1084         }
1085         if (want_color(o->use_color)) {
1086                 struct diff_words_style *st = ecbdata->diff_words->style;
1087                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1088                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1089                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1090         }
1091 }
1092
1093 static void free_diff_words_data(struct emit_callback *ecbdata)
1094 {
1095         if (ecbdata->diff_words) {
1096                 diff_words_flush(ecbdata);
1097                 free (ecbdata->diff_words->opt);
1098                 free (ecbdata->diff_words->minus.text.ptr);
1099                 free (ecbdata->diff_words->minus.orig);
1100                 free (ecbdata->diff_words->plus.text.ptr);
1101                 free (ecbdata->diff_words->plus.orig);
1102                 if (ecbdata->diff_words->word_regex) {
1103                         regfree(ecbdata->diff_words->word_regex);
1104                         free(ecbdata->diff_words->word_regex);
1105                 }
1106                 free(ecbdata->diff_words);
1107                 ecbdata->diff_words = NULL;
1108         }
1109 }
1110
1111 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1112 {
1113         if (want_color(diff_use_color))
1114                 return diff_colors[ix];
1115         return "";
1116 }
1117
1118 const char *diff_line_prefix(struct diff_options *opt)
1119 {
1120         struct strbuf *msgbuf;
1121         if (!opt->output_prefix)
1122                 return "";
1123
1124         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1125         return msgbuf->buf;
1126 }
1127
1128 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1129 {
1130         const char *cp;
1131         unsigned long allot;
1132         size_t l = len;
1133
1134         if (ecb->truncate)
1135                 return ecb->truncate(line, len);
1136         cp = line;
1137         allot = l;
1138         while (0 < l) {
1139                 (void) utf8_width(&cp, &l);
1140                 if (!cp)
1141                         break; /* truncated in the middle? */
1142         }
1143         return allot - l;
1144 }
1145
1146 static void find_lno(const char *line, struct emit_callback *ecbdata)
1147 {
1148         const char *p;
1149         ecbdata->lno_in_preimage = 0;
1150         ecbdata->lno_in_postimage = 0;
1151         p = strchr(line, '-');
1152         if (!p)
1153                 return; /* cannot happen */
1154         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1155         p = strchr(p, '+');
1156         if (!p)
1157                 return; /* cannot happen */
1158         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1159 }
1160
1161 static void fn_out_consume(void *priv, char *line, unsigned long len)
1162 {
1163         struct emit_callback *ecbdata = priv;
1164         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1165         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1166         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1167         struct diff_options *o = ecbdata->opt;
1168         const char *line_prefix = diff_line_prefix(o);
1169
1170         if (ecbdata->header) {
1171                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1172                 strbuf_reset(ecbdata->header);
1173                 ecbdata->header = NULL;
1174         }
1175         *(ecbdata->found_changesp) = 1;
1176
1177         if (ecbdata->label_path[0]) {
1178                 const char *name_a_tab, *name_b_tab;
1179
1180                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1181                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1182
1183                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1184                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1185                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1186                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1187                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1188         }
1189
1190         if (diff_suppress_blank_empty
1191             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1192                 line[0] = '\n';
1193                 len = 1;
1194         }
1195
1196         if (line[0] == '@') {
1197                 if (ecbdata->diff_words)
1198                         diff_words_flush(ecbdata);
1199                 len = sane_truncate_line(ecbdata, line, len);
1200                 find_lno(line, ecbdata);
1201                 emit_hunk_header(ecbdata, line, len);
1202                 if (line[len-1] != '\n')
1203                         putc('\n', ecbdata->opt->file);
1204                 return;
1205         }
1206
1207         if (len < 1) {
1208                 emit_line(ecbdata->opt, reset, reset, line, len);
1209                 if (ecbdata->diff_words
1210                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1211                         fputs("~\n", ecbdata->opt->file);
1212                 return;
1213         }
1214
1215         if (ecbdata->diff_words) {
1216                 if (line[0] == '-') {
1217                         diff_words_append(line, len,
1218                                           &ecbdata->diff_words->minus);
1219                         return;
1220                 } else if (line[0] == '+') {
1221                         diff_words_append(line, len,
1222                                           &ecbdata->diff_words->plus);
1223                         return;
1224                 } else if (starts_with(line, "\\ ")) {
1225                         /*
1226                          * Eat the "no newline at eof" marker as if we
1227                          * saw a "+" or "-" line with nothing on it,
1228                          * and return without diff_words_flush() to
1229                          * defer processing. If this is the end of
1230                          * preimage, more "+" lines may come after it.
1231                          */
1232                         return;
1233                 }
1234                 diff_words_flush(ecbdata);
1235                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1236                         emit_line(ecbdata->opt, plain, reset, line, len);
1237                         fputs("~\n", ecbdata->opt->file);
1238                 } else {
1239                         /*
1240                          * Skip the prefix character, if any.  With
1241                          * diff_suppress_blank_empty, there may be
1242                          * none.
1243                          */
1244                         if (line[0] != '\n') {
1245                               line++;
1246                               len--;
1247                         }
1248                         emit_line(ecbdata->opt, plain, reset, line, len);
1249                 }
1250                 return;
1251         }
1252
1253         if (line[0] != '+') {
1254                 const char *color =
1255                         diff_get_color(ecbdata->color_diff,
1256                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1257                 ecbdata->lno_in_preimage++;
1258                 if (line[0] == ' ')
1259                         ecbdata->lno_in_postimage++;
1260                 emit_line(ecbdata->opt, color, reset, line, len);
1261         } else {
1262                 ecbdata->lno_in_postimage++;
1263                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1264         }
1265 }
1266
1267 static char *pprint_rename(const char *a, const char *b)
1268 {
1269         const char *old = a;
1270         const char *new = b;
1271         struct strbuf name = STRBUF_INIT;
1272         int pfx_length, sfx_length;
1273         int pfx_adjust_for_slash;
1274         int len_a = strlen(a);
1275         int len_b = strlen(b);
1276         int a_midlen, b_midlen;
1277         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1278         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1279
1280         if (qlen_a || qlen_b) {
1281                 quote_c_style(a, &name, NULL, 0);
1282                 strbuf_addstr(&name, " => ");
1283                 quote_c_style(b, &name, NULL, 0);
1284                 return strbuf_detach(&name, NULL);
1285         }
1286
1287         /* Find common prefix */
1288         pfx_length = 0;
1289         while (*old && *new && *old == *new) {
1290                 if (*old == '/')
1291                         pfx_length = old - a + 1;
1292                 old++;
1293                 new++;
1294         }
1295
1296         /* Find common suffix */
1297         old = a + len_a;
1298         new = b + len_b;
1299         sfx_length = 0;
1300         /*
1301          * If there is a common prefix, it must end in a slash.  In
1302          * that case we let this loop run 1 into the prefix to see the
1303          * same slash.
1304          *
1305          * If there is no common prefix, we cannot do this as it would
1306          * underrun the input strings.
1307          */
1308         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1309         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1310                b + pfx_length - pfx_adjust_for_slash <= new &&
1311                *old == *new) {
1312                 if (*old == '/')
1313                         sfx_length = len_a - (old - a);
1314                 old--;
1315                 new--;
1316         }
1317
1318         /*
1319          * pfx{mid-a => mid-b}sfx
1320          * {pfx-a => pfx-b}sfx
1321          * pfx{sfx-a => sfx-b}
1322          * name-a => name-b
1323          */
1324         a_midlen = len_a - pfx_length - sfx_length;
1325         b_midlen = len_b - pfx_length - sfx_length;
1326         if (a_midlen < 0)
1327                 a_midlen = 0;
1328         if (b_midlen < 0)
1329                 b_midlen = 0;
1330
1331         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1332         if (pfx_length + sfx_length) {
1333                 strbuf_add(&name, a, pfx_length);
1334                 strbuf_addch(&name, '{');
1335         }
1336         strbuf_add(&name, a + pfx_length, a_midlen);
1337         strbuf_addstr(&name, " => ");
1338         strbuf_add(&name, b + pfx_length, b_midlen);
1339         if (pfx_length + sfx_length) {
1340                 strbuf_addch(&name, '}');
1341                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1342         }
1343         return strbuf_detach(&name, NULL);
1344 }
1345
1346 struct diffstat_t {
1347         int nr;
1348         int alloc;
1349         struct diffstat_file {
1350                 char *from_name;
1351                 char *name;
1352                 char *print_name;
1353                 unsigned is_unmerged:1;
1354                 unsigned is_binary:1;
1355                 unsigned is_renamed:1;
1356                 unsigned is_interesting:1;
1357                 uintmax_t added, deleted;
1358         } **files;
1359 };
1360
1361 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1362                                           const char *name_a,
1363                                           const char *name_b)
1364 {
1365         struct diffstat_file *x;
1366         x = xcalloc(1, sizeof(*x));
1367         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1368         diffstat->files[diffstat->nr++] = x;
1369         if (name_b) {
1370                 x->from_name = xstrdup(name_a);
1371                 x->name = xstrdup(name_b);
1372                 x->is_renamed = 1;
1373         }
1374         else {
1375                 x->from_name = NULL;
1376                 x->name = xstrdup(name_a);
1377         }
1378         return x;
1379 }
1380
1381 static void diffstat_consume(void *priv, char *line, unsigned long len)
1382 {
1383         struct diffstat_t *diffstat = priv;
1384         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1385
1386         if (line[0] == '+')
1387                 x->added++;
1388         else if (line[0] == '-')
1389                 x->deleted++;
1390 }
1391
1392 const char mime_boundary_leader[] = "------------";
1393
1394 static int scale_linear(int it, int width, int max_change)
1395 {
1396         if (!it)
1397                 return 0;
1398         /*
1399          * make sure that at least one '-' or '+' is printed if
1400          * there is any change to this path. The easiest way is to
1401          * scale linearly as if the alloted width is one column shorter
1402          * than it is, and then add 1 to the result.
1403          */
1404         return 1 + (it * (width - 1) / max_change);
1405 }
1406
1407 static void show_name(FILE *file,
1408                       const char *prefix, const char *name, int len)
1409 {
1410         fprintf(file, " %s%-*s |", prefix, len, name);
1411 }
1412
1413 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1414 {
1415         if (cnt <= 0)
1416                 return;
1417         fprintf(file, "%s", set);
1418         while (cnt--)
1419                 putc(ch, file);
1420         fprintf(file, "%s", reset);
1421 }
1422
1423 static void fill_print_name(struct diffstat_file *file)
1424 {
1425         char *pname;
1426
1427         if (file->print_name)
1428                 return;
1429
1430         if (!file->is_renamed) {
1431                 struct strbuf buf = STRBUF_INIT;
1432                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1433                         pname = strbuf_detach(&buf, NULL);
1434                 } else {
1435                         pname = file->name;
1436                         strbuf_release(&buf);
1437                 }
1438         } else {
1439                 pname = pprint_rename(file->from_name, file->name);
1440         }
1441         file->print_name = pname;
1442 }
1443
1444 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1445 {
1446         struct strbuf sb = STRBUF_INIT;
1447         int ret;
1448
1449         if (!files) {
1450                 assert(insertions == 0 && deletions == 0);
1451                 return fprintf(fp, "%s\n", " 0 files changed");
1452         }
1453
1454         strbuf_addf(&sb,
1455                     (files == 1) ? " %d file changed" : " %d files changed",
1456                     files);
1457
1458         /*
1459          * For binary diff, the caller may want to print "x files
1460          * changed" with insertions == 0 && deletions == 0.
1461          *
1462          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1463          * is probably less confusing (i.e skip over "2 files changed
1464          * but nothing about added/removed lines? Is this a bug in Git?").
1465          */
1466         if (insertions || deletions == 0) {
1467                 strbuf_addf(&sb,
1468                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1469                             insertions);
1470         }
1471
1472         if (deletions || insertions == 0) {
1473                 strbuf_addf(&sb,
1474                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1475                             deletions);
1476         }
1477         strbuf_addch(&sb, '\n');
1478         ret = fputs(sb.buf, fp);
1479         strbuf_release(&sb);
1480         return ret;
1481 }
1482
1483 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1484 {
1485         int i, len, add, del, adds = 0, dels = 0;
1486         uintmax_t max_change = 0, max_len = 0;
1487         int total_files = data->nr, count;
1488         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1489         const char *reset, *add_c, *del_c;
1490         const char *line_prefix = "";
1491         int extra_shown = 0;
1492
1493         if (data->nr == 0)
1494                 return;
1495
1496         line_prefix = diff_line_prefix(options);
1497         count = options->stat_count ? options->stat_count : data->nr;
1498
1499         reset = diff_get_color_opt(options, DIFF_RESET);
1500         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1501         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1502
1503         /*
1504          * Find the longest filename and max number of changes
1505          */
1506         for (i = 0; (i < count) && (i < data->nr); i++) {
1507                 struct diffstat_file *file = data->files[i];
1508                 uintmax_t change = file->added + file->deleted;
1509
1510                 if (!file->is_interesting && (change == 0)) {
1511                         count++; /* not shown == room for one more */
1512                         continue;
1513                 }
1514                 fill_print_name(file);
1515                 len = strlen(file->print_name);
1516                 if (max_len < len)
1517                         max_len = len;
1518
1519                 if (file->is_unmerged) {
1520                         /* "Unmerged" is 8 characters */
1521                         bin_width = bin_width < 8 ? 8 : bin_width;
1522                         continue;
1523                 }
1524                 if (file->is_binary) {
1525                         /* "Bin XXX -> YYY bytes" */
1526                         int w = 14 + decimal_width(file->added)
1527                                 + decimal_width(file->deleted);
1528                         bin_width = bin_width < w ? w : bin_width;
1529                         /* Display change counts aligned with "Bin" */
1530                         number_width = 3;
1531                         continue;
1532                 }
1533
1534                 if (max_change < change)
1535                         max_change = change;
1536         }
1537         count = i; /* where we can stop scanning in data->files[] */
1538
1539         /*
1540          * We have width = stat_width or term_columns() columns total.
1541          * We want a maximum of min(max_len, stat_name_width) for the name part.
1542          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1543          * We also need 1 for " " and 4 + decimal_width(max_change)
1544          * for " | NNNN " and one the empty column at the end, altogether
1545          * 6 + decimal_width(max_change).
1546          *
1547          * If there's not enough space, we will use the smaller of
1548          * stat_name_width (if set) and 5/8*width for the filename,
1549          * and the rest for constant elements + graph part, but no more
1550          * than stat_graph_width for the graph part.
1551          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1552          * for the standard terminal size).
1553          *
1554          * In other words: stat_width limits the maximum width, and
1555          * stat_name_width fixes the maximum width of the filename,
1556          * and is also used to divide available columns if there
1557          * aren't enough.
1558          *
1559          * Binary files are displayed with "Bin XXX -> YYY bytes"
1560          * instead of the change count and graph. This part is treated
1561          * similarly to the graph part, except that it is not
1562          * "scaled". If total width is too small to accommodate the
1563          * guaranteed minimum width of the filename part and the
1564          * separators and this message, this message will "overflow"
1565          * making the line longer than the maximum width.
1566          */
1567
1568         if (options->stat_width == -1)
1569                 width = term_columns() - options->output_prefix_length;
1570         else
1571                 width = options->stat_width ? options->stat_width : 80;
1572         number_width = decimal_width(max_change) > number_width ?
1573                 decimal_width(max_change) : number_width;
1574
1575         if (options->stat_graph_width == -1)
1576                 options->stat_graph_width = diff_stat_graph_width;
1577
1578         /*
1579          * Guarantee 3/8*16==6 for the graph part
1580          * and 5/8*16==10 for the filename part
1581          */
1582         if (width < 16 + 6 + number_width)
1583                 width = 16 + 6 + number_width;
1584
1585         /*
1586          * First assign sizes that are wanted, ignoring available width.
1587          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1588          * starting from "XXX" should fit in graph_width.
1589          */
1590         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1591         if (options->stat_graph_width &&
1592             options->stat_graph_width < graph_width)
1593                 graph_width = options->stat_graph_width;
1594
1595         name_width = (options->stat_name_width > 0 &&
1596                       options->stat_name_width < max_len) ?
1597                 options->stat_name_width : max_len;
1598
1599         /*
1600          * Adjust adjustable widths not to exceed maximum width
1601          */
1602         if (name_width + number_width + 6 + graph_width > width) {
1603                 if (graph_width > width * 3/8 - number_width - 6) {
1604                         graph_width = width * 3/8 - number_width - 6;
1605                         if (graph_width < 6)
1606                                 graph_width = 6;
1607                 }
1608
1609                 if (options->stat_graph_width &&
1610                     graph_width > options->stat_graph_width)
1611                         graph_width = options->stat_graph_width;
1612                 if (name_width > width - number_width - 6 - graph_width)
1613                         name_width = width - number_width - 6 - graph_width;
1614                 else
1615                         graph_width = width - number_width - 6 - name_width;
1616         }
1617
1618         /*
1619          * From here name_width is the width of the name area,
1620          * and graph_width is the width of the graph area.
1621          * max_change is used to scale graph properly.
1622          */
1623         for (i = 0; i < count; i++) {
1624                 const char *prefix = "";
1625                 struct diffstat_file *file = data->files[i];
1626                 char *name = file->print_name;
1627                 uintmax_t added = file->added;
1628                 uintmax_t deleted = file->deleted;
1629                 int name_len;
1630
1631                 if (!file->is_interesting && (added + deleted == 0))
1632                         continue;
1633
1634                 /*
1635                  * "scale" the filename
1636                  */
1637                 len = name_width;
1638                 name_len = strlen(name);
1639                 if (name_width < name_len) {
1640                         char *slash;
1641                         prefix = "...";
1642                         len -= 3;
1643                         name += name_len - len;
1644                         slash = strchr(name, '/');
1645                         if (slash)
1646                                 name = slash;
1647                 }
1648
1649                 if (file->is_binary) {
1650                         fprintf(options->file, "%s", line_prefix);
1651                         show_name(options->file, prefix, name, len);
1652                         fprintf(options->file, " %*s", number_width, "Bin");
1653                         if (!added && !deleted) {
1654                                 putc('\n', options->file);
1655                                 continue;
1656                         }
1657                         fprintf(options->file, " %s%"PRIuMAX"%s",
1658                                 del_c, deleted, reset);
1659                         fprintf(options->file, " -> ");
1660                         fprintf(options->file, "%s%"PRIuMAX"%s",
1661                                 add_c, added, reset);
1662                         fprintf(options->file, " bytes");
1663                         fprintf(options->file, "\n");
1664                         continue;
1665                 }
1666                 else if (file->is_unmerged) {
1667                         fprintf(options->file, "%s", line_prefix);
1668                         show_name(options->file, prefix, name, len);
1669                         fprintf(options->file, " Unmerged\n");
1670                         continue;
1671                 }
1672
1673                 /*
1674                  * scale the add/delete
1675                  */
1676                 add = added;
1677                 del = deleted;
1678
1679                 if (graph_width <= max_change) {
1680                         int total = scale_linear(add + del, graph_width, max_change);
1681                         if (total < 2 && add && del)
1682                                 /* width >= 2 due to the sanity check */
1683                                 total = 2;
1684                         if (add < del) {
1685                                 add = scale_linear(add, graph_width, max_change);
1686                                 del = total - add;
1687                         } else {
1688                                 del = scale_linear(del, graph_width, max_change);
1689                                 add = total - del;
1690                         }
1691                 }
1692                 fprintf(options->file, "%s", line_prefix);
1693                 show_name(options->file, prefix, name, len);
1694                 fprintf(options->file, " %*"PRIuMAX"%s",
1695                         number_width, added + deleted,
1696                         added + deleted ? " " : "");
1697                 show_graph(options->file, '+', add, add_c, reset);
1698                 show_graph(options->file, '-', del, del_c, reset);
1699                 fprintf(options->file, "\n");
1700         }
1701
1702         for (i = 0; i < data->nr; i++) {
1703                 struct diffstat_file *file = data->files[i];
1704                 uintmax_t added = file->added;
1705                 uintmax_t deleted = file->deleted;
1706
1707                 if (file->is_unmerged ||
1708                     (!file->is_interesting && (added + deleted == 0))) {
1709                         total_files--;
1710                         continue;
1711                 }
1712
1713                 if (!file->is_binary) {
1714                         adds += added;
1715                         dels += deleted;
1716                 }
1717                 if (i < count)
1718                         continue;
1719                 if (!extra_shown)
1720                         fprintf(options->file, "%s ...\n", line_prefix);
1721                 extra_shown = 1;
1722         }
1723         fprintf(options->file, "%s", line_prefix);
1724         print_stat_summary(options->file, total_files, adds, dels);
1725 }
1726
1727 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1728 {
1729         int i, adds = 0, dels = 0, total_files = data->nr;
1730
1731         if (data->nr == 0)
1732                 return;
1733
1734         for (i = 0; i < data->nr; i++) {
1735                 int added = data->files[i]->added;
1736                 int deleted= data->files[i]->deleted;
1737
1738                 if (data->files[i]->is_unmerged ||
1739                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1740                         total_files--;
1741                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1742                         adds += added;
1743                         dels += deleted;
1744                 }
1745         }
1746         fprintf(options->file, "%s", diff_line_prefix(options));
1747         print_stat_summary(options->file, total_files, adds, dels);
1748 }
1749
1750 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1751 {
1752         int i;
1753
1754         if (data->nr == 0)
1755                 return;
1756
1757         for (i = 0; i < data->nr; i++) {
1758                 struct diffstat_file *file = data->files[i];
1759
1760                 fprintf(options->file, "%s", diff_line_prefix(options));
1761
1762                 if (file->is_binary)
1763                         fprintf(options->file, "-\t-\t");
1764                 else
1765                         fprintf(options->file,
1766                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1767                                 file->added, file->deleted);
1768                 if (options->line_termination) {
1769                         fill_print_name(file);
1770                         if (!file->is_renamed)
1771                                 write_name_quoted(file->name, options->file,
1772                                                   options->line_termination);
1773                         else {
1774                                 fputs(file->print_name, options->file);
1775                                 putc(options->line_termination, options->file);
1776                         }
1777                 } else {
1778                         if (file->is_renamed) {
1779                                 putc('\0', options->file);
1780                                 write_name_quoted(file->from_name, options->file, '\0');
1781                         }
1782                         write_name_quoted(file->name, options->file, '\0');
1783                 }
1784         }
1785 }
1786
1787 struct dirstat_file {
1788         const char *name;
1789         unsigned long changed;
1790 };
1791
1792 struct dirstat_dir {
1793         struct dirstat_file *files;
1794         int alloc, nr, permille, cumulative;
1795 };
1796
1797 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1798                 unsigned long changed, const char *base, int baselen)
1799 {
1800         unsigned long this_dir = 0;
1801         unsigned int sources = 0;
1802         const char *line_prefix = diff_line_prefix(opt);
1803
1804         while (dir->nr) {
1805                 struct dirstat_file *f = dir->files;
1806                 int namelen = strlen(f->name);
1807                 unsigned long this;
1808                 char *slash;
1809
1810                 if (namelen < baselen)
1811                         break;
1812                 if (memcmp(f->name, base, baselen))
1813                         break;
1814                 slash = strchr(f->name + baselen, '/');
1815                 if (slash) {
1816                         int newbaselen = slash + 1 - f->name;
1817                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1818                         sources++;
1819                 } else {
1820                         this = f->changed;
1821                         dir->files++;
1822                         dir->nr--;
1823                         sources += 2;
1824                 }
1825                 this_dir += this;
1826         }
1827
1828         /*
1829          * We don't report dirstat's for
1830          *  - the top level
1831          *  - or cases where everything came from a single directory
1832          *    under this directory (sources == 1).
1833          */
1834         if (baselen && sources != 1) {
1835                 if (this_dir) {
1836                         int permille = this_dir * 1000 / changed;
1837                         if (permille >= dir->permille) {
1838                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1839                                         permille / 10, permille % 10, baselen, base);
1840                                 if (!dir->cumulative)
1841                                         return 0;
1842                         }
1843                 }
1844         }
1845         return this_dir;
1846 }
1847
1848 static int dirstat_compare(const void *_a, const void *_b)
1849 {
1850         const struct dirstat_file *a = _a;
1851         const struct dirstat_file *b = _b;
1852         return strcmp(a->name, b->name);
1853 }
1854
1855 static void show_dirstat(struct diff_options *options)
1856 {
1857         int i;
1858         unsigned long changed;
1859         struct dirstat_dir dir;
1860         struct diff_queue_struct *q = &diff_queued_diff;
1861
1862         dir.files = NULL;
1863         dir.alloc = 0;
1864         dir.nr = 0;
1865         dir.permille = options->dirstat_permille;
1866         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1867
1868         changed = 0;
1869         for (i = 0; i < q->nr; i++) {
1870                 struct diff_filepair *p = q->queue[i];
1871                 const char *name;
1872                 unsigned long copied, added, damage;
1873                 int content_changed;
1874
1875                 name = p->two->path ? p->two->path : p->one->path;
1876
1877                 if (p->one->sha1_valid && p->two->sha1_valid)
1878                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1879                 else
1880                         content_changed = 1;
1881
1882                 if (!content_changed) {
1883                         /*
1884                          * The SHA1 has not changed, so pre-/post-content is
1885                          * identical. We can therefore skip looking at the
1886                          * file contents altogether.
1887                          */
1888                         damage = 0;
1889                         goto found_damage;
1890                 }
1891
1892                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1893                         /*
1894                          * In --dirstat-by-file mode, we don't really need to
1895                          * look at the actual file contents at all.
1896                          * The fact that the SHA1 changed is enough for us to
1897                          * add this file to the list of results
1898                          * (with each file contributing equal damage).
1899                          */
1900                         damage = 1;
1901                         goto found_damage;
1902                 }
1903
1904                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1905                         diff_populate_filespec(p->one, 0);
1906                         diff_populate_filespec(p->two, 0);
1907                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1908                                                &copied, &added);
1909                         diff_free_filespec_data(p->one);
1910                         diff_free_filespec_data(p->two);
1911                 } else if (DIFF_FILE_VALID(p->one)) {
1912                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1913                         copied = added = 0;
1914                         diff_free_filespec_data(p->one);
1915                 } else if (DIFF_FILE_VALID(p->two)) {
1916                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1917                         copied = 0;
1918                         added = p->two->size;
1919                         diff_free_filespec_data(p->two);
1920                 } else
1921                         continue;
1922
1923                 /*
1924                  * Original minus copied is the removed material,
1925                  * added is the new material.  They are both damages
1926                  * made to the preimage.
1927                  * If the resulting damage is zero, we know that
1928                  * diffcore_count_changes() considers the two entries to
1929                  * be identical, but since content_changed is true, we
1930                  * know that there must have been _some_ kind of change,
1931                  * so we force all entries to have damage > 0.
1932                  */
1933                 damage = (p->one->size - copied) + added;
1934                 if (!damage)
1935                         damage = 1;
1936
1937 found_damage:
1938                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1939                 dir.files[dir.nr].name = name;
1940                 dir.files[dir.nr].changed = damage;
1941                 changed += damage;
1942                 dir.nr++;
1943         }
1944
1945         /* This can happen even with many files, if everything was renames */
1946         if (!changed)
1947                 return;
1948
1949         /* Show all directories with more than x% of the changes */
1950         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1951         gather_dirstat(options, &dir, changed, "", 0);
1952 }
1953
1954 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1955 {
1956         int i;
1957         unsigned long changed;
1958         struct dirstat_dir dir;
1959
1960         if (data->nr == 0)
1961                 return;
1962
1963         dir.files = NULL;
1964         dir.alloc = 0;
1965         dir.nr = 0;
1966         dir.permille = options->dirstat_permille;
1967         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1968
1969         changed = 0;
1970         for (i = 0; i < data->nr; i++) {
1971                 struct diffstat_file *file = data->files[i];
1972                 unsigned long damage = file->added + file->deleted;
1973                 if (file->is_binary)
1974                         /*
1975                          * binary files counts bytes, not lines. Must find some
1976                          * way to normalize binary bytes vs. textual lines.
1977                          * The following heuristic assumes that there are 64
1978                          * bytes per "line".
1979                          * This is stupid and ugly, but very cheap...
1980                          */
1981                         damage = (damage + 63) / 64;
1982                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1983                 dir.files[dir.nr].name = file->name;
1984                 dir.files[dir.nr].changed = damage;
1985                 changed += damage;
1986                 dir.nr++;
1987         }
1988
1989         /* This can happen even with many files, if everything was renames */
1990         if (!changed)
1991                 return;
1992
1993         /* Show all directories with more than x% of the changes */
1994         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1995         gather_dirstat(options, &dir, changed, "", 0);
1996 }
1997
1998 static void free_diffstat_info(struct diffstat_t *diffstat)
1999 {
2000         int i;
2001         for (i = 0; i < diffstat->nr; i++) {
2002                 struct diffstat_file *f = diffstat->files[i];
2003                 if (f->name != f->print_name)
2004                         free(f->print_name);
2005                 free(f->name);
2006                 free(f->from_name);
2007                 free(f);
2008         }
2009         free(diffstat->files);
2010 }
2011
2012 struct checkdiff_t {
2013         const char *filename;
2014         int lineno;
2015         int conflict_marker_size;
2016         struct diff_options *o;
2017         unsigned ws_rule;
2018         unsigned status;
2019 };
2020
2021 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2022 {
2023         char firstchar;
2024         int cnt;
2025
2026         if (len < marker_size + 1)
2027                 return 0;
2028         firstchar = line[0];
2029         switch (firstchar) {
2030         case '=': case '>': case '<': case '|':
2031                 break;
2032         default:
2033                 return 0;
2034         }
2035         for (cnt = 1; cnt < marker_size; cnt++)
2036                 if (line[cnt] != firstchar)
2037                         return 0;
2038         /* line[1] thru line[marker_size-1] are same as firstchar */
2039         if (len < marker_size + 1 || !isspace(line[marker_size]))
2040                 return 0;
2041         return 1;
2042 }
2043
2044 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2045 {
2046         struct checkdiff_t *data = priv;
2047         int marker_size = data->conflict_marker_size;
2048         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2049         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2050         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2051         char *err;
2052         const char *line_prefix;
2053
2054         assert(data->o);
2055         line_prefix = diff_line_prefix(data->o);
2056
2057         if (line[0] == '+') {
2058                 unsigned bad;
2059                 data->lineno++;
2060                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2061                         data->status |= 1;
2062                         fprintf(data->o->file,
2063                                 "%s%s:%d: leftover conflict marker\n",
2064                                 line_prefix, data->filename, data->lineno);
2065                 }
2066                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2067                 if (!bad)
2068                         return;
2069                 data->status |= bad;
2070                 err = whitespace_error_string(bad);
2071                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2072                         line_prefix, data->filename, data->lineno, err);
2073                 free(err);
2074                 emit_line(data->o, set, reset, line, 1);
2075                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2076                               data->o->file, set, reset, ws);
2077         } else if (line[0] == ' ') {
2078                 data->lineno++;
2079         } else if (line[0] == '@') {
2080                 char *plus = strchr(line, '+');
2081                 if (plus)
2082                         data->lineno = strtol(plus, NULL, 10) - 1;
2083                 else
2084                         die("invalid diff");
2085         }
2086 }
2087
2088 static unsigned char *deflate_it(char *data,
2089                                  unsigned long size,
2090                                  unsigned long *result_size)
2091 {
2092         int bound;
2093         unsigned char *deflated;
2094         git_zstream stream;
2095
2096         memset(&stream, 0, sizeof(stream));
2097         git_deflate_init(&stream, zlib_compression_level);
2098         bound = git_deflate_bound(&stream, size);
2099         deflated = xmalloc(bound);
2100         stream.next_out = deflated;
2101         stream.avail_out = bound;
2102
2103         stream.next_in = (unsigned char *)data;
2104         stream.avail_in = size;
2105         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2106                 ; /* nothing */
2107         git_deflate_end(&stream);
2108         *result_size = stream.total_out;
2109         return deflated;
2110 }
2111
2112 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2113                                   const char *prefix)
2114 {
2115         void *cp;
2116         void *delta;
2117         void *deflated;
2118         void *data;
2119         unsigned long orig_size;
2120         unsigned long delta_size;
2121         unsigned long deflate_size;
2122         unsigned long data_size;
2123
2124         /* We could do deflated delta, or we could do just deflated two,
2125          * whichever is smaller.
2126          */
2127         delta = NULL;
2128         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2129         if (one->size && two->size) {
2130                 delta = diff_delta(one->ptr, one->size,
2131                                    two->ptr, two->size,
2132                                    &delta_size, deflate_size);
2133                 if (delta) {
2134                         void *to_free = delta;
2135                         orig_size = delta_size;
2136                         delta = deflate_it(delta, delta_size, &delta_size);
2137                         free(to_free);
2138                 }
2139         }
2140
2141         if (delta && delta_size < deflate_size) {
2142                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2143                 free(deflated);
2144                 data = delta;
2145                 data_size = delta_size;
2146         }
2147         else {
2148                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2149                 free(delta);
2150                 data = deflated;
2151                 data_size = deflate_size;
2152         }
2153
2154         /* emit data encoded in base85 */
2155         cp = data;
2156         while (data_size) {
2157                 int bytes = (52 < data_size) ? 52 : data_size;
2158                 char line[70];
2159                 data_size -= bytes;
2160                 if (bytes <= 26)
2161                         line[0] = bytes + 'A' - 1;
2162                 else
2163                         line[0] = bytes - 26 + 'a' - 1;
2164                 encode_85(line + 1, cp, bytes);
2165                 cp = (char *) cp + bytes;
2166                 fprintf(file, "%s", prefix);
2167                 fputs(line, file);
2168                 fputc('\n', file);
2169         }
2170         fprintf(file, "%s\n", prefix);
2171         free(data);
2172 }
2173
2174 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2175                              const char *prefix)
2176 {
2177         fprintf(file, "%sGIT binary patch\n", prefix);
2178         emit_binary_diff_body(file, one, two, prefix);
2179         emit_binary_diff_body(file, two, one, prefix);
2180 }
2181
2182 int diff_filespec_is_binary(struct diff_filespec *one)
2183 {
2184         if (one->is_binary == -1) {
2185                 diff_filespec_load_driver(one);
2186                 if (one->driver->binary != -1)
2187                         one->is_binary = one->driver->binary;
2188                 else {
2189                         if (!one->data && DIFF_FILE_VALID(one))
2190                                 diff_populate_filespec(one, CHECK_BINARY);
2191                         if (one->is_binary == -1 && one->data)
2192                                 one->is_binary = buffer_is_binary(one->data,
2193                                                 one->size);
2194                         if (one->is_binary == -1)
2195                                 one->is_binary = 0;
2196                 }
2197         }
2198         return one->is_binary;
2199 }
2200
2201 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2202 {
2203         diff_filespec_load_driver(one);
2204         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2205 }
2206
2207 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2208 {
2209         if (!options->a_prefix)
2210                 options->a_prefix = a;
2211         if (!options->b_prefix)
2212                 options->b_prefix = b;
2213 }
2214
2215 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2216 {
2217         if (!DIFF_FILE_VALID(one))
2218                 return NULL;
2219
2220         diff_filespec_load_driver(one);
2221         return userdiff_get_textconv(one->driver);
2222 }
2223
2224 static void builtin_diff(const char *name_a,
2225                          const char *name_b,
2226                          struct diff_filespec *one,
2227                          struct diff_filespec *two,
2228                          const char *xfrm_msg,
2229                          int must_show_header,
2230                          struct diff_options *o,
2231                          int complete_rewrite)
2232 {
2233         mmfile_t mf1, mf2;
2234         const char *lbl[2];
2235         char *a_one, *b_two;
2236         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2237         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2238         const char *a_prefix, *b_prefix;
2239         struct userdiff_driver *textconv_one = NULL;
2240         struct userdiff_driver *textconv_two = NULL;
2241         struct strbuf header = STRBUF_INIT;
2242         const char *line_prefix = diff_line_prefix(o);
2243
2244         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2245                         (!one->mode || S_ISGITLINK(one->mode)) &&
2246                         (!two->mode || S_ISGITLINK(two->mode))) {
2247                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2248                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2249                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2250                                 line_prefix,
2251                                 one->sha1, two->sha1, two->dirty_submodule,
2252                                 meta, del, add, reset);
2253                 return;
2254         }
2255
2256         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2257                 textconv_one = get_textconv(one);
2258                 textconv_two = get_textconv(two);
2259         }
2260
2261         diff_set_mnemonic_prefix(o, "a/", "b/");
2262         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2263                 a_prefix = o->b_prefix;
2264                 b_prefix = o->a_prefix;
2265         } else {
2266                 a_prefix = o->a_prefix;
2267                 b_prefix = o->b_prefix;
2268         }
2269
2270         /* Never use a non-valid filename anywhere if at all possible */
2271         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2272         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2273
2274         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2275         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2276         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2277         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2278         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2279         if (lbl[0][0] == '/') {
2280                 /* /dev/null */
2281                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2282                 if (xfrm_msg)
2283                         strbuf_addstr(&header, xfrm_msg);
2284                 must_show_header = 1;
2285         }
2286         else if (lbl[1][0] == '/') {
2287                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2288                 if (xfrm_msg)
2289                         strbuf_addstr(&header, xfrm_msg);
2290                 must_show_header = 1;
2291         }
2292         else {
2293                 if (one->mode != two->mode) {
2294                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2295                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2296                         must_show_header = 1;
2297                 }
2298                 if (xfrm_msg)
2299                         strbuf_addstr(&header, xfrm_msg);
2300
2301                 /*
2302                  * we do not run diff between different kind
2303                  * of objects.
2304                  */
2305                 if ((one->mode ^ two->mode) & S_IFMT)
2306                         goto free_ab_and_return;
2307                 if (complete_rewrite &&
2308                     (textconv_one || !diff_filespec_is_binary(one)) &&
2309                     (textconv_two || !diff_filespec_is_binary(two))) {
2310                         fprintf(o->file, "%s", header.buf);
2311                         strbuf_reset(&header);
2312                         emit_rewrite_diff(name_a, name_b, one, two,
2313                                                 textconv_one, textconv_two, o);
2314                         o->found_changes = 1;
2315                         goto free_ab_and_return;
2316                 }
2317         }
2318
2319         if (o->irreversible_delete && lbl[1][0] == '/') {
2320                 fprintf(o->file, "%s", header.buf);
2321                 strbuf_reset(&header);
2322                 goto free_ab_and_return;
2323         } else if (!DIFF_OPT_TST(o, TEXT) &&
2324             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2325               (!textconv_two && diff_filespec_is_binary(two)) )) {
2326                 if (!one->data && !two->data &&
2327                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2328                     !DIFF_OPT_TST(o, BINARY)) {
2329                         if (!hashcmp(one->sha1, two->sha1)) {
2330                                 if (must_show_header)
2331                                         fprintf(o->file, "%s", header.buf);
2332                                 goto free_ab_and_return;
2333                         }
2334                         fprintf(o->file, "%s", header.buf);
2335                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2336                                 line_prefix, lbl[0], lbl[1]);
2337                         goto free_ab_and_return;
2338                 }
2339                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2340                         die("unable to read files to diff");
2341                 /* Quite common confusing case */
2342                 if (mf1.size == mf2.size &&
2343                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2344                         if (must_show_header)
2345                                 fprintf(o->file, "%s", header.buf);
2346                         goto free_ab_and_return;
2347                 }
2348                 fprintf(o->file, "%s", header.buf);
2349                 strbuf_reset(&header);
2350                 if (DIFF_OPT_TST(o, BINARY))
2351                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2352                 else
2353                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2354                                 line_prefix, lbl[0], lbl[1]);
2355                 o->found_changes = 1;
2356         } else {
2357                 /* Crazy xdl interfaces.. */
2358                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2359                 const char *v;
2360                 xpparam_t xpp;
2361                 xdemitconf_t xecfg;
2362                 struct emit_callback ecbdata;
2363                 const struct userdiff_funcname *pe;
2364
2365                 if (must_show_header) {
2366                         fprintf(o->file, "%s", header.buf);
2367                         strbuf_reset(&header);
2368                 }
2369
2370                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2371                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2372
2373                 pe = diff_funcname_pattern(one);
2374                 if (!pe)
2375                         pe = diff_funcname_pattern(two);
2376
2377                 memset(&xpp, 0, sizeof(xpp));
2378                 memset(&xecfg, 0, sizeof(xecfg));
2379                 memset(&ecbdata, 0, sizeof(ecbdata));
2380                 ecbdata.label_path = lbl;
2381                 ecbdata.color_diff = want_color(o->use_color);
2382                 ecbdata.found_changesp = &o->found_changes;
2383                 ecbdata.ws_rule = whitespace_rule(name_b);
2384                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2385                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2386                 ecbdata.opt = o;
2387                 ecbdata.header = header.len ? &header : NULL;
2388                 xpp.flags = o->xdl_opts;
2389                 xecfg.ctxlen = o->context;
2390                 xecfg.interhunkctxlen = o->interhunkcontext;
2391                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2392                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2393                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2394                 if (pe)
2395                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2396                 if (!diffopts)
2397                         ;
2398                 else if (skip_prefix(diffopts, "--unified=", &v))
2399                         xecfg.ctxlen = strtoul(v, NULL, 10);
2400                 else if (skip_prefix(diffopts, "-u", &v))
2401                         xecfg.ctxlen = strtoul(v, NULL, 10);
2402                 if (o->word_diff)
2403                         init_diff_words_data(&ecbdata, o, one, two);
2404                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2405                               &xpp, &xecfg);
2406                 if (o->word_diff)
2407                         free_diff_words_data(&ecbdata);
2408                 if (textconv_one)
2409                         free(mf1.ptr);
2410                 if (textconv_two)
2411                         free(mf2.ptr);
2412                 xdiff_clear_find_func(&xecfg);
2413         }
2414
2415  free_ab_and_return:
2416         strbuf_release(&header);
2417         diff_free_filespec_data(one);
2418         diff_free_filespec_data(two);
2419         free(a_one);
2420         free(b_two);
2421         return;
2422 }
2423
2424 static void builtin_diffstat(const char *name_a, const char *name_b,
2425                              struct diff_filespec *one,
2426                              struct diff_filespec *two,
2427                              struct diffstat_t *diffstat,
2428                              struct diff_options *o,
2429                              struct diff_filepair *p)
2430 {
2431         mmfile_t mf1, mf2;
2432         struct diffstat_file *data;
2433         int same_contents;
2434         int complete_rewrite = 0;
2435
2436         if (!DIFF_PAIR_UNMERGED(p)) {
2437                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2438                         complete_rewrite = 1;
2439         }
2440
2441         data = diffstat_add(diffstat, name_a, name_b);
2442         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2443
2444         if (!one || !two) {
2445                 data->is_unmerged = 1;
2446                 return;
2447         }
2448
2449         same_contents = !hashcmp(one->sha1, two->sha1);
2450
2451         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2452                 data->is_binary = 1;
2453                 if (same_contents) {
2454                         data->added = 0;
2455                         data->deleted = 0;
2456                 } else {
2457                         data->added = diff_filespec_size(two);
2458                         data->deleted = diff_filespec_size(one);
2459                 }
2460         }
2461
2462         else if (complete_rewrite) {
2463                 diff_populate_filespec(one, 0);
2464                 diff_populate_filespec(two, 0);
2465                 data->deleted = count_lines(one->data, one->size);
2466                 data->added = count_lines(two->data, two->size);
2467         }
2468
2469         else if (!same_contents) {
2470                 /* Crazy xdl interfaces.. */
2471                 xpparam_t xpp;
2472                 xdemitconf_t xecfg;
2473
2474                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2475                         die("unable to read files to diff");
2476
2477                 memset(&xpp, 0, sizeof(xpp));
2478                 memset(&xecfg, 0, sizeof(xecfg));
2479                 xpp.flags = o->xdl_opts;
2480                 xecfg.ctxlen = o->context;
2481                 xecfg.interhunkctxlen = o->interhunkcontext;
2482                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2483                               &xpp, &xecfg);
2484         }
2485
2486         diff_free_filespec_data(one);
2487         diff_free_filespec_data(two);
2488 }
2489
2490 static void builtin_checkdiff(const char *name_a, const char *name_b,
2491                               const char *attr_path,
2492                               struct diff_filespec *one,
2493                               struct diff_filespec *two,
2494                               struct diff_options *o)
2495 {
2496         mmfile_t mf1, mf2;
2497         struct checkdiff_t data;
2498
2499         if (!two)
2500                 return;
2501
2502         memset(&data, 0, sizeof(data));
2503         data.filename = name_b ? name_b : name_a;
2504         data.lineno = 0;
2505         data.o = o;
2506         data.ws_rule = whitespace_rule(attr_path);
2507         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2508
2509         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2510                 die("unable to read files to diff");
2511
2512         /*
2513          * All the other codepaths check both sides, but not checking
2514          * the "old" side here is deliberate.  We are checking the newly
2515          * introduced changes, and as long as the "new" side is text, we
2516          * can and should check what it introduces.
2517          */
2518         if (diff_filespec_is_binary(two))
2519                 goto free_and_return;
2520         else {
2521                 /* Crazy xdl interfaces.. */
2522                 xpparam_t xpp;
2523                 xdemitconf_t xecfg;
2524
2525                 memset(&xpp, 0, sizeof(xpp));
2526                 memset(&xecfg, 0, sizeof(xecfg));
2527                 xecfg.ctxlen = 1; /* at least one context line */
2528                 xpp.flags = 0;
2529                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2530                               &xpp, &xecfg);
2531
2532                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2533                         struct emit_callback ecbdata;
2534                         int blank_at_eof;
2535
2536                         ecbdata.ws_rule = data.ws_rule;
2537                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2538                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2539
2540                         if (blank_at_eof) {
2541                                 static char *err;
2542                                 if (!err)
2543                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2544                                 fprintf(o->file, "%s:%d: %s.\n",
2545                                         data.filename, blank_at_eof, err);
2546                                 data.status = 1; /* report errors */
2547                         }
2548                 }
2549         }
2550  free_and_return:
2551         diff_free_filespec_data(one);
2552         diff_free_filespec_data(two);
2553         if (data.status)
2554                 DIFF_OPT_SET(o, CHECK_FAILED);
2555 }
2556
2557 struct diff_filespec *alloc_filespec(const char *path)
2558 {
2559         int namelen = strlen(path);
2560         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2561
2562         memset(spec, 0, sizeof(*spec));
2563         spec->path = (char *)(spec + 1);
2564         memcpy(spec->path, path, namelen+1);
2565         spec->count = 1;
2566         spec->is_binary = -1;
2567         return spec;
2568 }
2569
2570 void free_filespec(struct diff_filespec *spec)
2571 {
2572         if (!--spec->count) {
2573                 diff_free_filespec_data(spec);
2574                 free(spec);
2575         }
2576 }
2577
2578 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2579                    int sha1_valid, unsigned short mode)
2580 {
2581         if (mode) {
2582                 spec->mode = canon_mode(mode);
2583                 hashcpy(spec->sha1, sha1);
2584                 spec->sha1_valid = sha1_valid;
2585         }
2586 }
2587
2588 /*
2589  * Given a name and sha1 pair, if the index tells us the file in
2590  * the work tree has that object contents, return true, so that
2591  * prepare_temp_file() does not have to inflate and extract.
2592  */
2593 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2594 {
2595         const struct cache_entry *ce;
2596         struct stat st;
2597         int pos, len;
2598
2599         /*
2600          * We do not read the cache ourselves here, because the
2601          * benchmark with my previous version that always reads cache
2602          * shows that it makes things worse for diff-tree comparing
2603          * two linux-2.6 kernel trees in an already checked out work
2604          * tree.  This is because most diff-tree comparisons deal with
2605          * only a small number of files, while reading the cache is
2606          * expensive for a large project, and its cost outweighs the
2607          * savings we get by not inflating the object to a temporary
2608          * file.  Practically, this code only helps when we are used
2609          * by diff-cache --cached, which does read the cache before
2610          * calling us.
2611          */
2612         if (!active_cache)
2613                 return 0;
2614
2615         /* We want to avoid the working directory if our caller
2616          * doesn't need the data in a normal file, this system
2617          * is rather slow with its stat/open/mmap/close syscalls,
2618          * and the object is contained in a pack file.  The pack
2619          * is probably already open and will be faster to obtain
2620          * the data through than the working directory.  Loose
2621          * objects however would tend to be slower as they need
2622          * to be individually opened and inflated.
2623          */
2624         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2625                 return 0;
2626
2627         len = strlen(name);
2628         pos = cache_name_pos(name, len);
2629         if (pos < 0)
2630                 return 0;
2631         ce = active_cache[pos];
2632
2633         /*
2634          * This is not the sha1 we are looking for, or
2635          * unreusable because it is not a regular file.
2636          */
2637         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2638                 return 0;
2639
2640         /*
2641          * If ce is marked as "assume unchanged", there is no
2642          * guarantee that work tree matches what we are looking for.
2643          */
2644         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2645                 return 0;
2646
2647         /*
2648          * If ce matches the file in the work tree, we can reuse it.
2649          */
2650         if (ce_uptodate(ce) ||
2651             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2652                 return 1;
2653
2654         return 0;
2655 }
2656
2657 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2658 {
2659         int len;
2660         char *data = xmalloc(100), *dirty = "";
2661
2662         /* Are we looking at the work tree? */
2663         if (s->dirty_submodule)
2664                 dirty = "-dirty";
2665
2666         len = snprintf(data, 100,
2667                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2668         s->data = data;
2669         s->size = len;
2670         s->should_free = 1;
2671         if (size_only) {
2672                 s->data = NULL;
2673                 free(data);
2674         }
2675         return 0;
2676 }
2677
2678 /*
2679  * While doing rename detection and pickaxe operation, we may need to
2680  * grab the data for the blob (or file) for our own in-core comparison.
2681  * diff_filespec has data and size fields for this purpose.
2682  */
2683 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2684 {
2685         int size_only = flags & CHECK_SIZE_ONLY;
2686         int err = 0;
2687         /*
2688          * demote FAIL to WARN to allow inspecting the situation
2689          * instead of refusing.
2690          */
2691         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2692                                     ? SAFE_CRLF_WARN
2693                                     : safe_crlf);
2694
2695         if (!DIFF_FILE_VALID(s))
2696                 die("internal error: asking to populate invalid file.");
2697         if (S_ISDIR(s->mode))
2698                 return -1;
2699
2700         if (s->data)
2701                 return 0;
2702
2703         if (size_only && 0 < s->size)
2704                 return 0;
2705
2706         if (S_ISGITLINK(s->mode))
2707                 return diff_populate_gitlink(s, size_only);
2708
2709         if (!s->sha1_valid ||
2710             reuse_worktree_file(s->path, s->sha1, 0)) {
2711                 struct strbuf buf = STRBUF_INIT;
2712                 struct stat st;
2713                 int fd;
2714
2715                 if (lstat(s->path, &st) < 0) {
2716                         if (errno == ENOENT) {
2717                         err_empty:
2718                                 err = -1;
2719                         empty:
2720                                 s->data = (char *)"";
2721                                 s->size = 0;
2722                                 return err;
2723                         }
2724                 }
2725                 s->size = xsize_t(st.st_size);
2726                 if (!s->size)
2727                         goto empty;
2728                 if (S_ISLNK(st.st_mode)) {
2729                         struct strbuf sb = STRBUF_INIT;
2730
2731                         if (strbuf_readlink(&sb, s->path, s->size))
2732                                 goto err_empty;
2733                         s->size = sb.len;
2734                         s->data = strbuf_detach(&sb, NULL);
2735                         s->should_free = 1;
2736                         return 0;
2737                 }
2738                 if (size_only)
2739                         return 0;
2740                 if ((flags & CHECK_BINARY) &&
2741                     s->size > big_file_threshold && s->is_binary == -1) {
2742                         s->is_binary = 1;
2743                         return 0;
2744                 }
2745                 fd = open(s->path, O_RDONLY);
2746                 if (fd < 0)
2747                         goto err_empty;
2748                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2749                 close(fd);
2750                 s->should_munmap = 1;
2751
2752                 /*
2753                  * Convert from working tree format to canonical git format
2754                  */
2755                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2756                         size_t size = 0;
2757                         munmap(s->data, s->size);
2758                         s->should_munmap = 0;
2759                         s->data = strbuf_detach(&buf, &size);
2760                         s->size = size;
2761                         s->should_free = 1;
2762                 }
2763         }
2764         else {
2765                 enum object_type type;
2766                 if (size_only || (flags & CHECK_BINARY)) {
2767                         type = sha1_object_info(s->sha1, &s->size);
2768                         if (type < 0)
2769                                 die("unable to read %s", sha1_to_hex(s->sha1));
2770                         if (size_only)
2771                                 return 0;
2772                         if (s->size > big_file_threshold && s->is_binary == -1) {
2773                                 s->is_binary = 1;
2774                                 return 0;
2775                         }
2776                 }
2777                 s->data = read_sha1_file(s->sha1, &type, &s->size);
2778                 if (!s->data)
2779                         die("unable to read %s", sha1_to_hex(s->sha1));
2780                 s->should_free = 1;
2781         }
2782         return 0;
2783 }
2784
2785 void diff_free_filespec_blob(struct diff_filespec *s)
2786 {
2787         if (s->should_free)
2788                 free(s->data);
2789         else if (s->should_munmap)
2790                 munmap(s->data, s->size);
2791
2792         if (s->should_free || s->should_munmap) {
2793                 s->should_free = s->should_munmap = 0;
2794                 s->data = NULL;
2795         }
2796 }
2797
2798 void diff_free_filespec_data(struct diff_filespec *s)
2799 {
2800         diff_free_filespec_blob(s);
2801         free(s->cnt_data);
2802         s->cnt_data = NULL;
2803 }
2804
2805 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2806                            void *blob,
2807                            unsigned long size,
2808                            const unsigned char *sha1,
2809                            int mode)
2810 {
2811         int fd;
2812         struct strbuf buf = STRBUF_INIT;
2813         struct strbuf template = STRBUF_INIT;
2814         char *path_dup = xstrdup(path);
2815         const char *base = basename(path_dup);
2816
2817         /* Generate "XXXXXX_basename.ext" */
2818         strbuf_addstr(&template, "XXXXXX_");
2819         strbuf_addstr(&template, base);
2820
2821         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2822                         strlen(base) + 1);
2823         if (fd < 0)
2824                 die_errno("unable to create temp-file");
2825         if (convert_to_working_tree(path,
2826                         (const char *)blob, (size_t)size, &buf)) {
2827                 blob = buf.buf;
2828                 size = buf.len;
2829         }
2830         if (write_in_full(fd, blob, size) != size)
2831                 die_errno("unable to write temp-file");
2832         close(fd);
2833         temp->name = temp->tmp_path;
2834         strcpy(temp->hex, sha1_to_hex(sha1));
2835         temp->hex[40] = 0;
2836         sprintf(temp->mode, "%06o", mode);
2837         strbuf_release(&buf);
2838         strbuf_release(&template);
2839         free(path_dup);
2840 }
2841
2842 static struct diff_tempfile *prepare_temp_file(const char *name,
2843                 struct diff_filespec *one)
2844 {
2845         struct diff_tempfile *temp = claim_diff_tempfile();
2846
2847         if (!DIFF_FILE_VALID(one)) {
2848         not_a_valid_file:
2849                 /* A '-' entry produces this for file-2, and
2850                  * a '+' entry produces this for file-1.
2851                  */
2852                 temp->name = "/dev/null";
2853                 strcpy(temp->hex, ".");
2854                 strcpy(temp->mode, ".");
2855                 return temp;
2856         }
2857
2858         if (!remove_tempfile_installed) {
2859                 atexit(remove_tempfile);
2860                 sigchain_push_common(remove_tempfile_on_signal);
2861                 remove_tempfile_installed = 1;
2862         }
2863
2864         if (!S_ISGITLINK(one->mode) &&
2865             (!one->sha1_valid ||
2866              reuse_worktree_file(name, one->sha1, 1))) {
2867                 struct stat st;
2868                 if (lstat(name, &st) < 0) {
2869                         if (errno == ENOENT)
2870                                 goto not_a_valid_file;
2871                         die_errno("stat(%s)", name);
2872                 }
2873                 if (S_ISLNK(st.st_mode)) {
2874                         struct strbuf sb = STRBUF_INIT;
2875                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2876                                 die_errno("readlink(%s)", name);
2877                         prep_temp_blob(name, temp, sb.buf, sb.len,
2878                                        (one->sha1_valid ?
2879                                         one->sha1 : null_sha1),
2880                                        (one->sha1_valid ?
2881                                         one->mode : S_IFLNK));
2882                         strbuf_release(&sb);
2883                 }
2884                 else {
2885                         /* we can borrow from the file in the work tree */
2886                         temp->name = name;
2887                         if (!one->sha1_valid)
2888                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2889                         else
2890                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2891                         /* Even though we may sometimes borrow the
2892                          * contents from the work tree, we always want
2893                          * one->mode.  mode is trustworthy even when
2894                          * !(one->sha1_valid), as long as
2895                          * DIFF_FILE_VALID(one).
2896                          */
2897                         sprintf(temp->mode, "%06o", one->mode);
2898                 }
2899                 return temp;
2900         }
2901         else {
2902                 if (diff_populate_filespec(one, 0))
2903                         die("cannot read data blob for %s", one->path);
2904                 prep_temp_blob(name, temp, one->data, one->size,
2905                                one->sha1, one->mode);
2906         }
2907         return temp;
2908 }
2909
2910 static void add_external_diff_name(struct argv_array *argv,
2911                                    const char *name,
2912                                    struct diff_filespec *df)
2913 {
2914         struct diff_tempfile *temp = prepare_temp_file(name, df);
2915         argv_array_push(argv, temp->name);
2916         argv_array_push(argv, temp->hex);
2917         argv_array_push(argv, temp->mode);
2918 }
2919
2920 /* An external diff command takes:
2921  *
2922  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2923  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2924  *
2925  */
2926 static void run_external_diff(const char *pgm,
2927                               const char *name,
2928                               const char *other,
2929                               struct diff_filespec *one,
2930                               struct diff_filespec *two,
2931                               const char *xfrm_msg,
2932                               int complete_rewrite,
2933                               struct diff_options *o)
2934 {
2935         struct argv_array argv = ARGV_ARRAY_INIT;
2936         struct argv_array env = ARGV_ARRAY_INIT;
2937         struct diff_queue_struct *q = &diff_queued_diff;
2938
2939         argv_array_push(&argv, pgm);
2940         argv_array_push(&argv, name);
2941
2942         if (one && two) {
2943                 add_external_diff_name(&argv, name, one);
2944                 if (!other)
2945                         add_external_diff_name(&argv, name, two);
2946                 else {
2947                         add_external_diff_name(&argv, other, two);
2948                         argv_array_push(&argv, other);
2949                         argv_array_push(&argv, xfrm_msg);
2950                 }
2951         }
2952
2953         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
2954         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2955
2956         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
2957                 die(_("external diff died, stopping at %s"), name);
2958
2959         remove_tempfile();
2960         argv_array_clear(&argv);
2961         argv_array_clear(&env);
2962 }
2963
2964 static int similarity_index(struct diff_filepair *p)
2965 {
2966         return p->score * 100 / MAX_SCORE;
2967 }
2968
2969 static void fill_metainfo(struct strbuf *msg,
2970                           const char *name,
2971                           const char *other,
2972                           struct diff_filespec *one,
2973                           struct diff_filespec *two,
2974                           struct diff_options *o,
2975                           struct diff_filepair *p,
2976                           int *must_show_header,
2977                           int use_color)
2978 {
2979         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2980         const char *reset = diff_get_color(use_color, DIFF_RESET);
2981         const char *line_prefix = diff_line_prefix(o);
2982
2983         *must_show_header = 1;
2984         strbuf_init(msg, PATH_MAX * 2 + 300);
2985         switch (p->status) {
2986         case DIFF_STATUS_COPIED:
2987                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2988                             line_prefix, set, similarity_index(p));
2989                 strbuf_addf(msg, "%s\n%s%scopy from ",
2990                             reset,  line_prefix, set);
2991                 quote_c_style(name, msg, NULL, 0);
2992                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2993                 quote_c_style(other, msg, NULL, 0);
2994                 strbuf_addf(msg, "%s\n", reset);
2995                 break;
2996         case DIFF_STATUS_RENAMED:
2997                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2998                             line_prefix, set, similarity_index(p));
2999                 strbuf_addf(msg, "%s\n%s%srename from ",
3000                             reset, line_prefix, set);
3001                 quote_c_style(name, msg, NULL, 0);
3002                 strbuf_addf(msg, "%s\n%s%srename to ",
3003                             reset, line_prefix, set);
3004                 quote_c_style(other, msg, NULL, 0);
3005                 strbuf_addf(msg, "%s\n", reset);
3006                 break;
3007         case DIFF_STATUS_MODIFIED:
3008                 if (p->score) {
3009                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3010                                     line_prefix,
3011                                     set, similarity_index(p), reset);
3012                         break;
3013                 }
3014                 /* fallthru */
3015         default:
3016                 *must_show_header = 0;
3017         }
3018         if (one && two && hashcmp(one->sha1, two->sha1)) {
3019                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3020
3021                 if (DIFF_OPT_TST(o, BINARY)) {
3022                         mmfile_t mf;
3023                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3024                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3025                                 abbrev = 40;
3026                 }
3027                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3028                             find_unique_abbrev(one->sha1, abbrev));
3029                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3030                 if (one->mode == two->mode)
3031                         strbuf_addf(msg, " %06o", one->mode);
3032                 strbuf_addf(msg, "%s\n", reset);
3033         }
3034 }
3035
3036 static void run_diff_cmd(const char *pgm,
3037                          const char *name,
3038                          const char *other,
3039                          const char *attr_path,
3040                          struct diff_filespec *one,
3041                          struct diff_filespec *two,
3042                          struct strbuf *msg,
3043                          struct diff_options *o,
3044                          struct diff_filepair *p)
3045 {
3046         const char *xfrm_msg = NULL;
3047         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3048         int must_show_header = 0;
3049
3050
3051         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3052                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3053                 if (drv && drv->external)
3054                         pgm = drv->external;
3055         }
3056
3057         if (msg) {
3058                 /*
3059                  * don't use colors when the header is intended for an
3060                  * external diff driver
3061                  */
3062                 fill_metainfo(msg, name, other, one, two, o, p,
3063                               &must_show_header,
3064                               want_color(o->use_color) && !pgm);
3065                 xfrm_msg = msg->len ? msg->buf : NULL;
3066         }
3067
3068         if (pgm) {
3069                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3070                                   complete_rewrite, o);
3071                 return;
3072         }
3073         if (one && two)
3074                 builtin_diff(name, other ? other : name,
3075                              one, two, xfrm_msg, must_show_header,
3076                              o, complete_rewrite);
3077         else
3078                 fprintf(o->file, "* Unmerged path %s\n", name);
3079 }
3080
3081 static void diff_fill_sha1_info(struct diff_filespec *one)
3082 {
3083         if (DIFF_FILE_VALID(one)) {
3084                 if (!one->sha1_valid) {
3085                         struct stat st;
3086                         if (one->is_stdin) {
3087                                 hashcpy(one->sha1, null_sha1);
3088                                 return;
3089                         }
3090                         if (lstat(one->path, &st) < 0)
3091                                 die_errno("stat '%s'", one->path);
3092                         if (index_path(one->sha1, one->path, &st, 0))
3093                                 die("cannot hash %s", one->path);
3094                 }
3095         }
3096         else
3097                 hashclr(one->sha1);
3098 }
3099
3100 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3101 {
3102         /* Strip the prefix but do not molest /dev/null and absolute paths */
3103         if (*namep && **namep != '/') {
3104                 *namep += prefix_length;
3105                 if (**namep == '/')
3106                         ++*namep;
3107         }
3108         if (*otherp && **otherp != '/') {
3109                 *otherp += prefix_length;
3110                 if (**otherp == '/')
3111                         ++*otherp;
3112         }
3113 }
3114
3115 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3116 {
3117         const char *pgm = external_diff();
3118         struct strbuf msg;
3119         struct diff_filespec *one = p->one;
3120         struct diff_filespec *two = p->two;
3121         const char *name;
3122         const char *other;
3123         const char *attr_path;
3124
3125         name  = p->one->path;
3126         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3127         attr_path = name;
3128         if (o->prefix_length)
3129                 strip_prefix(o->prefix_length, &name, &other);
3130
3131         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3132                 pgm = NULL;
3133
3134         if (DIFF_PAIR_UNMERGED(p)) {
3135                 run_diff_cmd(pgm, name, NULL, attr_path,
3136                              NULL, NULL, NULL, o, p);
3137                 return;
3138         }
3139
3140         diff_fill_sha1_info(one);
3141         diff_fill_sha1_info(two);
3142
3143         if (!pgm &&
3144             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3145             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3146                 /*
3147                  * a filepair that changes between file and symlink
3148                  * needs to be split into deletion and creation.
3149                  */
3150                 struct diff_filespec *null = alloc_filespec(two->path);
3151                 run_diff_cmd(NULL, name, other, attr_path,
3152                              one, null, &msg, o, p);
3153                 free(null);
3154                 strbuf_release(&msg);
3155
3156                 null = alloc_filespec(one->path);
3157                 run_diff_cmd(NULL, name, other, attr_path,
3158                              null, two, &msg, o, p);
3159                 free(null);
3160         }
3161         else
3162                 run_diff_cmd(pgm, name, other, attr_path,
3163                              one, two, &msg, o, p);
3164
3165         strbuf_release(&msg);
3166 }
3167
3168 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3169                          struct diffstat_t *diffstat)
3170 {
3171         const char *name;
3172         const char *other;
3173
3174         if (DIFF_PAIR_UNMERGED(p)) {
3175                 /* unmerged */
3176                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3177                 return;
3178         }
3179
3180         name = p->one->path;
3181         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3182
3183         if (o->prefix_length)
3184                 strip_prefix(o->prefix_length, &name, &other);
3185
3186         diff_fill_sha1_info(p->one);
3187         diff_fill_sha1_info(p->two);
3188
3189         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3190 }
3191
3192 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3193 {
3194         const char *name;
3195         const char *other;
3196         const char *attr_path;
3197
3198         if (DIFF_PAIR_UNMERGED(p)) {
3199                 /* unmerged */
3200                 return;
3201         }
3202
3203         name = p->one->path;
3204         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3205         attr_path = other ? other : name;
3206
3207         if (o->prefix_length)
3208                 strip_prefix(o->prefix_length, &name, &other);
3209
3210         diff_fill_sha1_info(p->one);
3211         diff_fill_sha1_info(p->two);
3212
3213         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3214 }
3215
3216 void diff_setup(struct diff_options *options)
3217 {
3218         memcpy(options, &default_diff_options, sizeof(*options));
3219
3220         options->file = stdout;
3221
3222         options->line_termination = '\n';
3223         options->break_opt = -1;
3224         options->rename_limit = -1;
3225         options->dirstat_permille = diff_dirstat_permille_default;
3226         options->context = diff_context_default;
3227         DIFF_OPT_SET(options, RENAME_EMPTY);
3228
3229         /* pathchange left =NULL by default */
3230         options->change = diff_change;
3231         options->add_remove = diff_addremove;
3232         options->use_color = diff_use_color_default;
3233         options->detect_rename = diff_detect_rename_default;
3234         options->xdl_opts |= diff_algorithm;
3235
3236         options->orderfile = diff_order_file_cfg;
3237
3238         if (diff_no_prefix) {
3239                 options->a_prefix = options->b_prefix = "";
3240         } else if (!diff_mnemonic_prefix) {
3241                 options->a_prefix = "a/";
3242                 options->b_prefix = "b/";
3243         }
3244 }
3245
3246 void diff_setup_done(struct diff_options *options)
3247 {
3248         int count = 0;
3249
3250         if (options->set_default)
3251                 options->set_default(options);
3252
3253         if (options->output_format & DIFF_FORMAT_NAME)
3254                 count++;
3255         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3256                 count++;
3257         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3258                 count++;
3259         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3260                 count++;
3261         if (count > 1)
3262                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3263
3264         /*
3265          * Most of the time we can say "there are changes"
3266          * only by checking if there are changed paths, but
3267          * --ignore-whitespace* options force us to look
3268          * inside contents.
3269          */
3270
3271         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3272             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3273             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3274                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3275         else
3276                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3277
3278         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3279                 options->detect_rename = DIFF_DETECT_COPY;
3280
3281         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3282                 options->prefix = NULL;
3283         if (options->prefix)
3284                 options->prefix_length = strlen(options->prefix);
3285         else
3286                 options->prefix_length = 0;
3287
3288         if (options->output_format & (DIFF_FORMAT_NAME |
3289                                       DIFF_FORMAT_NAME_STATUS |
3290                                       DIFF_FORMAT_CHECKDIFF |
3291                                       DIFF_FORMAT_NO_OUTPUT))
3292                 options->output_format &= ~(DIFF_FORMAT_RAW |
3293                                             DIFF_FORMAT_NUMSTAT |
3294                                             DIFF_FORMAT_DIFFSTAT |
3295                                             DIFF_FORMAT_SHORTSTAT |
3296                                             DIFF_FORMAT_DIRSTAT |
3297                                             DIFF_FORMAT_SUMMARY |
3298                                             DIFF_FORMAT_PATCH);
3299
3300         /*
3301          * These cases always need recursive; we do not drop caller-supplied
3302          * recursive bits for other formats here.
3303          */
3304         if (options->output_format & (DIFF_FORMAT_PATCH |
3305                                       DIFF_FORMAT_NUMSTAT |
3306                                       DIFF_FORMAT_DIFFSTAT |
3307                                       DIFF_FORMAT_SHORTSTAT |
3308                                       DIFF_FORMAT_DIRSTAT |
3309                                       DIFF_FORMAT_SUMMARY |
3310                                       DIFF_FORMAT_CHECKDIFF))
3311                 DIFF_OPT_SET(options, RECURSIVE);
3312         /*
3313          * Also pickaxe would not work very well if you do not say recursive
3314          */
3315         if (options->pickaxe)
3316                 DIFF_OPT_SET(options, RECURSIVE);
3317         /*
3318          * When patches are generated, submodules diffed against the work tree
3319          * must be checked for dirtiness too so it can be shown in the output
3320          */
3321         if (options->output_format & DIFF_FORMAT_PATCH)
3322                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3323
3324         if (options->detect_rename && options->rename_limit < 0)
3325                 options->rename_limit = diff_rename_limit_default;
3326         if (options->setup & DIFF_SETUP_USE_CACHE) {
3327                 if (!active_cache)
3328                         /* read-cache does not die even when it fails
3329                          * so it is safe for us to do this here.  Also
3330                          * it does not smudge active_cache or active_nr
3331                          * when it fails, so we do not have to worry about
3332                          * cleaning it up ourselves either.
3333                          */
3334                         read_cache();
3335         }
3336         if (options->abbrev <= 0 || 40 < options->abbrev)
3337                 options->abbrev = 40; /* full */
3338
3339         /*
3340          * It does not make sense to show the first hit we happened
3341          * to have found.  It does not make sense not to return with
3342          * exit code in such a case either.
3343          */
3344         if (DIFF_OPT_TST(options, QUICK)) {
3345                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3346                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3347         }
3348
3349         options->diff_path_counter = 0;
3350
3351         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3352                 die(_("--follow requires exactly one pathspec"));
3353 }
3354
3355 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3356 {
3357         char c, *eq;
3358         int len;
3359
3360         if (*arg != '-')
3361                 return 0;
3362         c = *++arg;
3363         if (!c)
3364                 return 0;
3365         if (c == arg_short) {
3366                 c = *++arg;
3367                 if (!c)
3368                         return 1;
3369                 if (val && isdigit(c)) {
3370                         char *end;
3371                         int n = strtoul(arg, &end, 10);
3372                         if (*end)
3373                                 return 0;
3374                         *val = n;
3375                         return 1;
3376                 }
3377                 return 0;
3378         }
3379         if (c != '-')
3380                 return 0;
3381         arg++;
3382         eq = strchrnul(arg, '=');
3383         len = eq - arg;
3384         if (!len || strncmp(arg, arg_long, len))
3385                 return 0;
3386         if (*eq) {
3387                 int n;
3388                 char *end;
3389                 if (!isdigit(*++eq))
3390                         return 0;
3391                 n = strtoul(eq, &end, 10);
3392                 if (*end)
3393                         return 0;
3394                 *val = n;
3395         }
3396         return 1;
3397 }
3398
3399 static int diff_scoreopt_parse(const char *opt);
3400
3401 static inline int short_opt(char opt, const char **argv,
3402                             const char **optarg)
3403 {
3404         const char *arg = argv[0];
3405         if (arg[0] != '-' || arg[1] != opt)
3406                 return 0;
3407         if (arg[2] != '\0') {
3408                 *optarg = arg + 2;
3409                 return 1;
3410         }
3411         if (!argv[1])
3412                 die("Option '%c' requires a value", opt);
3413         *optarg = argv[1];
3414         return 2;
3415 }
3416
3417 int parse_long_opt(const char *opt, const char **argv,
3418                    const char **optarg)
3419 {
3420         const char *arg = argv[0];
3421         if (!skip_prefix(arg, "--", &arg))
3422                 return 0;
3423         if (!skip_prefix(arg, opt, &arg))
3424                 return 0;
3425         if (*arg == '=') { /* stuck form: --option=value */
3426                 *optarg = arg + 1;
3427                 return 1;
3428         }
3429         if (*arg != '\0')
3430                 return 0;
3431         /* separate form: --option value */
3432         if (!argv[1])
3433                 die("Option '--%s' requires a value", opt);
3434         *optarg = argv[1];
3435         return 2;
3436 }
3437
3438 static int stat_opt(struct diff_options *options, const char **av)
3439 {
3440         const char *arg = av[0];
3441         char *end;
3442         int width = options->stat_width;
3443         int name_width = options->stat_name_width;
3444         int graph_width = options->stat_graph_width;
3445         int count = options->stat_count;
3446         int argcount = 1;
3447
3448         if (!skip_prefix(arg, "--stat", &arg))
3449                 die("BUG: stat option does not begin with --stat: %s", arg);
3450         end = (char *)arg;
3451
3452         switch (*arg) {
3453         case '-':
3454                 if (skip_prefix(arg, "-width", &arg)) {
3455                         if (*arg == '=')
3456                                 width = strtoul(arg + 1, &end, 10);
3457                         else if (!*arg && !av[1])
3458                                 die("Option '--stat-width' requires a value");
3459                         else if (!*arg) {
3460                                 width = strtoul(av[1], &end, 10);
3461                                 argcount = 2;
3462                         }
3463                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3464                         if (*arg == '=')
3465                                 name_width = strtoul(arg + 1, &end, 10);
3466                         else if (!*arg && !av[1])
3467                                 die("Option '--stat-name-width' requires a value");
3468                         else if (!*arg) {
3469                                 name_width = strtoul(av[1], &end, 10);
3470                                 argcount = 2;
3471                         }
3472                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3473                         if (*arg == '=')
3474                                 graph_width = strtoul(arg + 1, &end, 10);
3475                         else if (!*arg && !av[1])
3476                                 die("Option '--stat-graph-width' requires a value");
3477                         else if (!*arg) {
3478                                 graph_width = strtoul(av[1], &end, 10);
3479                                 argcount = 2;
3480                         }
3481                 } else if (skip_prefix(arg, "-count", &arg)) {
3482                         if (*arg == '=')
3483                                 count = strtoul(arg + 1, &end, 10);
3484                         else if (!*arg && !av[1])
3485                                 die("Option '--stat-count' requires a value");
3486                         else if (!*arg) {
3487                                 count = strtoul(av[1], &end, 10);
3488                                 argcount = 2;
3489                         }
3490                 }
3491                 break;
3492         case '=':
3493                 width = strtoul(arg+1, &end, 10);
3494                 if (*end == ',')
3495                         name_width = strtoul(end+1, &end, 10);
3496                 if (*end == ',')
3497                         count = strtoul(end+1, &end, 10);
3498         }
3499
3500         /* Important! This checks all the error cases! */
3501         if (*end)
3502                 return 0;
3503         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3504         options->stat_name_width = name_width;
3505         options->stat_graph_width = graph_width;
3506         options->stat_width = width;
3507         options->stat_count = count;
3508         return argcount;
3509 }
3510
3511 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3512 {
3513         struct strbuf errmsg = STRBUF_INIT;
3514         if (parse_dirstat_params(options, params, &errmsg))
3515                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3516                     errmsg.buf);
3517         strbuf_release(&errmsg);
3518         /*
3519          * The caller knows a dirstat-related option is given from the command
3520          * line; allow it to say "return this_function();"
3521          */
3522         options->output_format |= DIFF_FORMAT_DIRSTAT;
3523         return 1;
3524 }
3525
3526 static int parse_submodule_opt(struct diff_options *options, const char *value)
3527 {
3528         if (parse_submodule_params(options, value))
3529                 die(_("Failed to parse --submodule option parameter: '%s'"),
3530                         value);
3531         return 1;
3532 }
3533
3534 static const char diff_status_letters[] = {
3535         DIFF_STATUS_ADDED,
3536         DIFF_STATUS_COPIED,
3537         DIFF_STATUS_DELETED,
3538         DIFF_STATUS_MODIFIED,
3539         DIFF_STATUS_RENAMED,
3540         DIFF_STATUS_TYPE_CHANGED,
3541         DIFF_STATUS_UNKNOWN,
3542         DIFF_STATUS_UNMERGED,
3543         DIFF_STATUS_FILTER_AON,
3544         DIFF_STATUS_FILTER_BROKEN,
3545         '\0',
3546 };
3547
3548 static unsigned int filter_bit['Z' + 1];
3549
3550 static void prepare_filter_bits(void)
3551 {
3552         int i;
3553
3554         if (!filter_bit[DIFF_STATUS_ADDED]) {
3555                 for (i = 0; diff_status_letters[i]; i++)
3556                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3557         }
3558 }
3559
3560 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3561 {
3562         return opt->filter & filter_bit[(int) status];
3563 }
3564
3565 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3566 {
3567         int i, optch;
3568
3569         prepare_filter_bits();
3570
3571         /*
3572          * If there is a negation e.g. 'd' in the input, and we haven't
3573          * initialized the filter field with another --diff-filter, start
3574          * from full set of bits, except for AON.
3575          */
3576         if (!opt->filter) {
3577                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3578                         if (optch < 'a' || 'z' < optch)
3579                                 continue;
3580                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3581                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3582                         break;
3583                 }
3584         }
3585
3586         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3587                 unsigned int bit;
3588                 int negate;
3589
3590                 if ('a' <= optch && optch <= 'z') {
3591                         negate = 1;
3592                         optch = toupper(optch);
3593                 } else {
3594                         negate = 0;
3595                 }
3596
3597                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3598                 if (!bit)
3599                         return optarg[i];
3600                 if (negate)
3601                         opt->filter &= ~bit;
3602                 else
3603                         opt->filter |= bit;
3604         }
3605         return 0;
3606 }
3607
3608 static void enable_patch_output(int *fmt) {
3609         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3610         *fmt |= DIFF_FORMAT_PATCH;
3611 }
3612
3613 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3614 {
3615         const char *arg = av[0];
3616         const char *optarg;
3617         int argcount;
3618
3619         /* Output format options */
3620         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3621             || opt_arg(arg, 'U', "unified", &options->context))
3622                 enable_patch_output(&options->output_format);
3623         else if (!strcmp(arg, "--raw"))
3624                 options->output_format |= DIFF_FORMAT_RAW;
3625         else if (!strcmp(arg, "--patch-with-raw")) {
3626                 enable_patch_output(&options->output_format);
3627                 options->output_format |= DIFF_FORMAT_RAW;
3628         } else if (!strcmp(arg, "--numstat"))
3629                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3630         else if (!strcmp(arg, "--shortstat"))
3631                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3632         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3633                 return parse_dirstat_opt(options, "");
3634         else if (skip_prefix(arg, "-X", &arg))
3635                 return parse_dirstat_opt(options, arg);
3636         else if (skip_prefix(arg, "--dirstat=", &arg))
3637                 return parse_dirstat_opt(options, arg);
3638         else if (!strcmp(arg, "--cumulative"))
3639                 return parse_dirstat_opt(options, "cumulative");
3640         else if (!strcmp(arg, "--dirstat-by-file"))
3641                 return parse_dirstat_opt(options, "files");
3642         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3643                 parse_dirstat_opt(options, "files");
3644                 return parse_dirstat_opt(options, arg);
3645         }
3646         else if (!strcmp(arg, "--check"))
3647                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3648         else if (!strcmp(arg, "--summary"))
3649                 options->output_format |= DIFF_FORMAT_SUMMARY;
3650         else if (!strcmp(arg, "--patch-with-stat")) {
3651                 enable_patch_output(&options->output_format);
3652                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3653         } else if (!strcmp(arg, "--name-only"))
3654                 options->output_format |= DIFF_FORMAT_NAME;
3655         else if (!strcmp(arg, "--name-status"))
3656                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3657         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3658                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3659         else if (starts_with(arg, "--stat"))
3660                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3661                 return stat_opt(options, av);
3662
3663         /* renames options */
3664         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3665                  !strcmp(arg, "--break-rewrites")) {
3666                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3667                         return error("invalid argument to -B: %s", arg+2);
3668         }
3669         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3670                  !strcmp(arg, "--find-renames")) {
3671                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3672                         return error("invalid argument to -M: %s", arg+2);
3673                 options->detect_rename = DIFF_DETECT_RENAME;
3674         }
3675         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3676                 options->irreversible_delete = 1;
3677         }
3678         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3679                  !strcmp(arg, "--find-copies")) {
3680                 if (options->detect_rename == DIFF_DETECT_COPY)
3681                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3682                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3683                         return error("invalid argument to -C: %s", arg+2);
3684                 options->detect_rename = DIFF_DETECT_COPY;
3685         }
3686         else if (!strcmp(arg, "--no-renames"))
3687                 options->detect_rename = 0;
3688         else if (!strcmp(arg, "--rename-empty"))
3689                 DIFF_OPT_SET(options, RENAME_EMPTY);
3690         else if (!strcmp(arg, "--no-rename-empty"))
3691                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3692         else if (!strcmp(arg, "--relative"))
3693                 DIFF_OPT_SET(options, RELATIVE_NAME);
3694         else if (skip_prefix(arg, "--relative=", &arg)) {
3695                 DIFF_OPT_SET(options, RELATIVE_NAME);
3696                 options->prefix = arg;
3697         }
3698
3699         /* xdiff options */
3700         else if (!strcmp(arg, "--minimal"))
3701                 DIFF_XDL_SET(options, NEED_MINIMAL);
3702         else if (!strcmp(arg, "--no-minimal"))
3703                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3704         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3705                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3706         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3707                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3708         else if (!strcmp(arg, "--ignore-space-at-eol"))
3709                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3710         else if (!strcmp(arg, "--ignore-blank-lines"))
3711                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3712         else if (!strcmp(arg, "--patience"))
3713                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3714         else if (!strcmp(arg, "--histogram"))
3715                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3716         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3717                 long value = parse_algorithm_value(optarg);
3718                 if (value < 0)
3719                         return error("option diff-algorithm accepts \"myers\", "
3720                                      "\"minimal\", \"patience\" and \"histogram\"");
3721                 /* clear out previous settings */
3722                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3723                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3724                 options->xdl_opts |= value;
3725                 return argcount;
3726         }
3727
3728         /* flags options */
3729         else if (!strcmp(arg, "--binary")) {
3730                 enable_patch_output(&options->output_format);
3731                 DIFF_OPT_SET(options, BINARY);
3732         }
3733         else if (!strcmp(arg, "--full-index"))
3734                 DIFF_OPT_SET(options, FULL_INDEX);
3735         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3736                 DIFF_OPT_SET(options, TEXT);
3737         else if (!strcmp(arg, "-R"))
3738                 DIFF_OPT_SET(options, REVERSE_DIFF);
3739         else if (!strcmp(arg, "--find-copies-harder"))
3740                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3741         else if (!strcmp(arg, "--follow"))
3742                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3743         else if (!strcmp(arg, "--no-follow"))
3744                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3745         else if (!strcmp(arg, "--color"))
3746                 options->use_color = 1;
3747         else if (skip_prefix(arg, "--color=", &arg)) {
3748                 int value = git_config_colorbool(NULL, arg);
3749                 if (value < 0)
3750                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3751                 options->use_color = value;
3752         }
3753         else if (!strcmp(arg, "--no-color"))
3754                 options->use_color = 0;
3755         else if (!strcmp(arg, "--color-words")) {
3756                 options->use_color = 1;
3757                 options->word_diff = DIFF_WORDS_COLOR;
3758         }
3759         else if (skip_prefix(arg, "--color-words=", &arg)) {
3760                 options->use_color = 1;
3761                 options->word_diff = DIFF_WORDS_COLOR;
3762                 options->word_regex = arg;
3763         }
3764         else if (!strcmp(arg, "--word-diff")) {
3765                 if (options->word_diff == DIFF_WORDS_NONE)
3766                         options->word_diff = DIFF_WORDS_PLAIN;
3767         }
3768         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3769                 if (!strcmp(arg, "plain"))
3770                         options->word_diff = DIFF_WORDS_PLAIN;
3771                 else if (!strcmp(arg, "color")) {
3772                         options->use_color = 1;
3773                         options->word_diff = DIFF_WORDS_COLOR;
3774                 }
3775                 else if (!strcmp(arg, "porcelain"))
3776                         options->word_diff = DIFF_WORDS_PORCELAIN;
3777                 else if (!strcmp(arg, "none"))
3778                         options->word_diff = DIFF_WORDS_NONE;
3779                 else
3780                         die("bad --word-diff argument: %s", arg);
3781         }
3782         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3783                 if (options->word_diff == DIFF_WORDS_NONE)
3784                         options->word_diff = DIFF_WORDS_PLAIN;
3785                 options->word_regex = optarg;
3786                 return argcount;
3787         }
3788         else if (!strcmp(arg, "--exit-code"))
3789                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3790         else if (!strcmp(arg, "--quiet"))
3791                 DIFF_OPT_SET(options, QUICK);
3792         else if (!strcmp(arg, "--ext-diff"))
3793                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3794         else if (!strcmp(arg, "--no-ext-diff"))
3795                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3796         else if (!strcmp(arg, "--textconv"))
3797                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3798         else if (!strcmp(arg, "--no-textconv"))
3799                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3800         else if (!strcmp(arg, "--ignore-submodules")) {
3801                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3802                 handle_ignore_submodules_arg(options, "all");
3803         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3804                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3805                 handle_ignore_submodules_arg(options, arg);
3806         } else if (!strcmp(arg, "--submodule"))
3807                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3808         else if (skip_prefix(arg, "--submodule=", &arg))
3809                 return parse_submodule_opt(options, arg);
3810
3811         /* misc options */
3812         else if (!strcmp(arg, "-z"))
3813                 options->line_termination = 0;
3814         else if ((argcount = short_opt('l', av, &optarg))) {
3815                 options->rename_limit = strtoul(optarg, NULL, 10);
3816                 return argcount;
3817         }
3818         else if ((argcount = short_opt('S', av, &optarg))) {
3819                 options->pickaxe = optarg;
3820                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3821                 return argcount;
3822         } else if ((argcount = short_opt('G', av, &optarg))) {
3823                 options->pickaxe = optarg;
3824                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3825                 return argcount;
3826         }
3827         else if (!strcmp(arg, "--pickaxe-all"))
3828                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3829         else if (!strcmp(arg, "--pickaxe-regex"))
3830                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3831         else if ((argcount = short_opt('O', av, &optarg))) {
3832                 options->orderfile = optarg;
3833                 return argcount;
3834         }
3835         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3836                 int offending = parse_diff_filter_opt(optarg, options);
3837                 if (offending)
3838                         die("unknown change class '%c' in --diff-filter=%s",
3839                             offending, optarg);
3840                 return argcount;
3841         }
3842         else if (!strcmp(arg, "--abbrev"))
3843                 options->abbrev = DEFAULT_ABBREV;
3844         else if (skip_prefix(arg, "--abbrev=", &arg)) {
3845                 options->abbrev = strtoul(arg, NULL, 10);
3846                 if (options->abbrev < MINIMUM_ABBREV)
3847                         options->abbrev = MINIMUM_ABBREV;
3848                 else if (40 < options->abbrev)
3849                         options->abbrev = 40;
3850         }
3851         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3852                 options->a_prefix = optarg;
3853                 return argcount;
3854         }
3855         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3856                 options->b_prefix = optarg;
3857                 return argcount;
3858         }
3859         else if (!strcmp(arg, "--no-prefix"))
3860                 options->a_prefix = options->b_prefix = "";
3861         else if (opt_arg(arg, '\0', "inter-hunk-context",
3862                          &options->interhunkcontext))
3863                 ;
3864         else if (!strcmp(arg, "-W"))
3865                 DIFF_OPT_SET(options, FUNCCONTEXT);
3866         else if (!strcmp(arg, "--function-context"))
3867                 DIFF_OPT_SET(options, FUNCCONTEXT);
3868         else if (!strcmp(arg, "--no-function-context"))
3869                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3870         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3871                 options->file = fopen(optarg, "w");
3872                 if (!options->file)
3873                         die_errno("Could not open '%s'", optarg);
3874                 options->close_file = 1;
3875                 return argcount;
3876         } else
3877                 return 0;
3878         return 1;
3879 }
3880
3881 int parse_rename_score(const char **cp_p)
3882 {
3883         unsigned long num, scale;
3884         int ch, dot;
3885         const char *cp = *cp_p;
3886
3887         num = 0;
3888         scale = 1;
3889         dot = 0;
3890         for (;;) {
3891                 ch = *cp;
3892                 if ( !dot && ch == '.' ) {
3893                         scale = 1;
3894                         dot = 1;
3895                 } else if ( ch == '%' ) {
3896                         scale = dot ? scale*100 : 100;
3897                         cp++;   /* % is always at the end */
3898                         break;
3899                 } else if ( ch >= '0' && ch <= '9' ) {
3900                         if ( scale < 100000 ) {
3901                                 scale *= 10;
3902                                 num = (num*10) + (ch-'0');
3903                         }
3904                 } else {
3905                         break;
3906                 }
3907                 cp++;
3908         }
3909         *cp_p = cp;
3910
3911         /* user says num divided by scale and we say internally that
3912          * is MAX_SCORE * num / scale.
3913          */
3914         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3915 }
3916
3917 static int diff_scoreopt_parse(const char *opt)
3918 {
3919         int opt1, opt2, cmd;
3920
3921         if (*opt++ != '-')
3922                 return -1;
3923         cmd = *opt++;
3924         if (cmd == '-') {
3925                 /* convert the long-form arguments into short-form versions */
3926                 if (skip_prefix(opt, "break-rewrites", &opt)) {
3927                         if (*opt == 0 || *opt++ == '=')
3928                                 cmd = 'B';
3929                 } else if (skip_prefix(opt, "find-copies", &opt)) {
3930                         if (*opt == 0 || *opt++ == '=')
3931                                 cmd = 'C';
3932                 } else if (skip_prefix(opt, "find-renames", &opt)) {
3933                         if (*opt == 0 || *opt++ == '=')
3934                                 cmd = 'M';
3935                 }
3936         }
3937         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3938                 return -1; /* that is not a -M, -C, or -B option */
3939
3940         opt1 = parse_rename_score(&opt);
3941         if (cmd != 'B')
3942                 opt2 = 0;
3943         else {
3944                 if (*opt == 0)
3945                         opt2 = 0;
3946                 else if (*opt != '/')
3947                         return -1; /* we expect -B80/99 or -B80 */
3948                 else {
3949                         opt++;
3950                         opt2 = parse_rename_score(&opt);
3951                 }
3952         }
3953         if (*opt != 0)
3954                 return -1;
3955         return opt1 | (opt2 << 16);
3956 }
3957
3958 struct diff_queue_struct diff_queued_diff;
3959
3960 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3961 {
3962         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
3963         queue->queue[queue->nr++] = dp;
3964 }
3965
3966 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3967                                  struct diff_filespec *one,
3968                                  struct diff_filespec *two)
3969 {
3970         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3971         dp->one = one;
3972         dp->two = two;
3973         if (queue)
3974                 diff_q(queue, dp);
3975         return dp;
3976 }
3977
3978 void diff_free_filepair(struct diff_filepair *p)
3979 {
3980         free_filespec(p->one);
3981         free_filespec(p->two);
3982         free(p);
3983 }
3984
3985 /* This is different from find_unique_abbrev() in that
3986  * it stuffs the result with dots for alignment.
3987  */
3988 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3989 {
3990         int abblen;
3991         const char *abbrev;
3992         if (len == 40)
3993                 return sha1_to_hex(sha1);
3994
3995         abbrev = find_unique_abbrev(sha1, len);
3996         abblen = strlen(abbrev);
3997         if (abblen < 37) {
3998                 static char hex[41];
3999                 if (len < abblen && abblen <= len + 2)
4000                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
4001                 else
4002                         sprintf(hex, "%s...", abbrev);
4003                 return hex;
4004         }
4005         return sha1_to_hex(sha1);
4006 }
4007
4008 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4009 {
4010         int line_termination = opt->line_termination;
4011         int inter_name_termination = line_termination ? '\t' : '\0';
4012
4013         fprintf(opt->file, "%s", diff_line_prefix(opt));
4014         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4015                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4016                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
4017                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4018         }
4019         if (p->score) {
4020                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4021                         inter_name_termination);
4022         } else {
4023                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4024         }
4025
4026         if (p->status == DIFF_STATUS_COPIED ||
4027             p->status == DIFF_STATUS_RENAMED) {
4028                 const char *name_a, *name_b;
4029                 name_a = p->one->path;
4030                 name_b = p->two->path;
4031                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4032                 write_name_quoted(name_a, opt->file, inter_name_termination);
4033                 write_name_quoted(name_b, opt->file, line_termination);
4034         } else {
4035                 const char *name_a, *name_b;
4036                 name_a = p->one->mode ? p->one->path : p->two->path;
4037                 name_b = NULL;
4038                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4039                 write_name_quoted(name_a, opt->file, line_termination);
4040         }
4041 }
4042
4043 int diff_unmodified_pair(struct diff_filepair *p)
4044 {
4045         /* This function is written stricter than necessary to support
4046          * the currently implemented transformers, but the idea is to
4047          * let transformers to produce diff_filepairs any way they want,
4048          * and filter and clean them up here before producing the output.
4049          */
4050         struct diff_filespec *one = p->one, *two = p->two;
4051
4052         if (DIFF_PAIR_UNMERGED(p))
4053                 return 0; /* unmerged is interesting */
4054
4055         /* deletion, addition, mode or type change
4056          * and rename are all interesting.
4057          */
4058         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4059             DIFF_PAIR_MODE_CHANGED(p) ||
4060             strcmp(one->path, two->path))
4061                 return 0;
4062
4063         /* both are valid and point at the same path.  that is, we are
4064          * dealing with a change.
4065          */
4066         if (one->sha1_valid && two->sha1_valid &&
4067             !hashcmp(one->sha1, two->sha1) &&
4068             !one->dirty_submodule && !two->dirty_submodule)
4069                 return 1; /* no change */
4070         if (!one->sha1_valid && !two->sha1_valid)
4071                 return 1; /* both look at the same file on the filesystem. */
4072         return 0;
4073 }
4074
4075 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4076 {
4077         if (diff_unmodified_pair(p))
4078                 return;
4079
4080         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4081             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4082                 return; /* no tree diffs in patch format */
4083
4084         run_diff(p, o);
4085 }
4086
4087 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4088                             struct diffstat_t *diffstat)
4089 {
4090         if (diff_unmodified_pair(p))
4091                 return;
4092
4093         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4094             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4095                 return; /* no useful stat for tree diffs */
4096
4097         run_diffstat(p, o, diffstat);
4098 }
4099
4100 static void diff_flush_checkdiff(struct diff_filepair *p,
4101                 struct diff_options *o)
4102 {
4103         if (diff_unmodified_pair(p))
4104                 return;
4105
4106         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4107             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4108                 return; /* nothing to check in tree diffs */
4109
4110         run_checkdiff(p, o);
4111 }
4112
4113 int diff_queue_is_empty(void)
4114 {
4115         struct diff_queue_struct *q = &diff_queued_diff;
4116         int i;
4117         for (i = 0; i < q->nr; i++)
4118                 if (!diff_unmodified_pair(q->queue[i]))
4119                         return 0;
4120         return 1;
4121 }
4122
4123 #if DIFF_DEBUG
4124 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4125 {
4126         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4127                 x, one ? one : "",
4128                 s->path,
4129                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4130                 s->mode,
4131                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4132         fprintf(stderr, "queue[%d] %s size %lu\n",
4133                 x, one ? one : "",
4134                 s->size);
4135 }
4136
4137 void diff_debug_filepair(const struct diff_filepair *p, int i)
4138 {
4139         diff_debug_filespec(p->one, i, "one");
4140         diff_debug_filespec(p->two, i, "two");
4141         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4142                 p->score, p->status ? p->status : '?',
4143                 p->one->rename_used, p->broken_pair);
4144 }
4145
4146 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4147 {
4148         int i;
4149         if (msg)
4150                 fprintf(stderr, "%s\n", msg);
4151         fprintf(stderr, "q->nr = %d\n", q->nr);
4152         for (i = 0; i < q->nr; i++) {
4153                 struct diff_filepair *p = q->queue[i];
4154                 diff_debug_filepair(p, i);
4155         }
4156 }
4157 #endif
4158
4159 static void diff_resolve_rename_copy(void)
4160 {
4161         int i;
4162         struct diff_filepair *p;
4163         struct diff_queue_struct *q = &diff_queued_diff;
4164
4165         diff_debug_queue("resolve-rename-copy", q);
4166
4167         for (i = 0; i < q->nr; i++) {
4168                 p = q->queue[i];
4169                 p->status = 0; /* undecided */
4170                 if (DIFF_PAIR_UNMERGED(p))
4171                         p->status = DIFF_STATUS_UNMERGED;
4172                 else if (!DIFF_FILE_VALID(p->one))
4173                         p->status = DIFF_STATUS_ADDED;
4174                 else if (!DIFF_FILE_VALID(p->two))
4175                         p->status = DIFF_STATUS_DELETED;
4176                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4177                         p->status = DIFF_STATUS_TYPE_CHANGED;
4178
4179                 /* from this point on, we are dealing with a pair
4180                  * whose both sides are valid and of the same type, i.e.
4181                  * either in-place edit or rename/copy edit.
4182                  */
4183                 else if (DIFF_PAIR_RENAME(p)) {
4184                         /*
4185                          * A rename might have re-connected a broken
4186                          * pair up, causing the pathnames to be the
4187                          * same again. If so, that's not a rename at
4188                          * all, just a modification..
4189                          *
4190                          * Otherwise, see if this source was used for
4191                          * multiple renames, in which case we decrement
4192                          * the count, and call it a copy.
4193                          */
4194                         if (!strcmp(p->one->path, p->two->path))
4195                                 p->status = DIFF_STATUS_MODIFIED;
4196                         else if (--p->one->rename_used > 0)
4197                                 p->status = DIFF_STATUS_COPIED;
4198                         else
4199                                 p->status = DIFF_STATUS_RENAMED;
4200                 }
4201                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4202                          p->one->mode != p->two->mode ||
4203                          p->one->dirty_submodule ||
4204                          p->two->dirty_submodule ||
4205                          is_null_sha1(p->one->sha1))
4206                         p->status = DIFF_STATUS_MODIFIED;
4207                 else {
4208                         /* This is a "no-change" entry and should not
4209                          * happen anymore, but prepare for broken callers.
4210                          */
4211                         error("feeding unmodified %s to diffcore",
4212                               p->one->path);
4213                         p->status = DIFF_STATUS_UNKNOWN;
4214                 }
4215         }
4216         diff_debug_queue("resolve-rename-copy done", q);
4217 }
4218
4219 static int check_pair_status(struct diff_filepair *p)
4220 {
4221         switch (p->status) {
4222         case DIFF_STATUS_UNKNOWN:
4223                 return 0;
4224         case 0:
4225                 die("internal error in diff-resolve-rename-copy");
4226         default:
4227                 return 1;
4228         }
4229 }
4230
4231 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4232 {
4233         int fmt = opt->output_format;
4234
4235         if (fmt & DIFF_FORMAT_CHECKDIFF)
4236                 diff_flush_checkdiff(p, opt);
4237         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4238                 diff_flush_raw(p, opt);
4239         else if (fmt & DIFF_FORMAT_NAME) {
4240                 const char *name_a, *name_b;
4241                 name_a = p->two->path;
4242                 name_b = NULL;
4243                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4244                 write_name_quoted(name_a, opt->file, opt->line_termination);
4245         }
4246 }
4247
4248 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4249 {
4250         if (fs->mode)
4251                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4252         else
4253                 fprintf(file, " %s ", newdelete);
4254         write_name_quoted(fs->path, file, '\n');
4255 }
4256
4257
4258 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4259                 const char *line_prefix)
4260 {
4261         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4262                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4263                         p->two->mode, show_name ? ' ' : '\n');
4264                 if (show_name) {
4265                         write_name_quoted(p->two->path, file, '\n');
4266                 }
4267         }
4268 }
4269
4270 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4271                         const char *line_prefix)
4272 {
4273         char *names = pprint_rename(p->one->path, p->two->path);
4274
4275         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4276         free(names);
4277         show_mode_change(file, p, 0, line_prefix);
4278 }
4279
4280 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4281 {
4282         FILE *file = opt->file;
4283         const char *line_prefix = diff_line_prefix(opt);
4284
4285         switch(p->status) {
4286         case DIFF_STATUS_DELETED:
4287                 fputs(line_prefix, file);
4288                 show_file_mode_name(file, "delete", p->one);
4289                 break;
4290         case DIFF_STATUS_ADDED:
4291                 fputs(line_prefix, file);
4292                 show_file_mode_name(file, "create", p->two);
4293                 break;
4294         case DIFF_STATUS_COPIED:
4295                 fputs(line_prefix, file);
4296                 show_rename_copy(file, "copy", p, line_prefix);
4297                 break;
4298         case DIFF_STATUS_RENAMED:
4299                 fputs(line_prefix, file);
4300                 show_rename_copy(file, "rename", p, line_prefix);
4301                 break;
4302         default:
4303                 if (p->score) {
4304                         fprintf(file, "%s rewrite ", line_prefix);
4305                         write_name_quoted(p->two->path, file, ' ');
4306                         fprintf(file, "(%d%%)\n", similarity_index(p));
4307                 }
4308                 show_mode_change(file, p, !p->score, line_prefix);
4309                 break;
4310         }
4311 }
4312
4313 struct patch_id_t {
4314         git_SHA_CTX *ctx;
4315         int patchlen;
4316 };
4317
4318 static int remove_space(char *line, int len)
4319 {
4320         int i;
4321         char *dst = line;
4322         unsigned char c;
4323
4324         for (i = 0; i < len; i++)
4325                 if (!isspace((c = line[i])))
4326                         *dst++ = c;
4327
4328         return dst - line;
4329 }
4330
4331 static void patch_id_consume(void *priv, char *line, unsigned long len)
4332 {
4333         struct patch_id_t *data = priv;
4334         int new_len;
4335
4336         /* Ignore line numbers when computing the SHA1 of the patch */
4337         if (starts_with(line, "@@ -"))
4338                 return;
4339
4340         new_len = remove_space(line, len);
4341
4342         git_SHA1_Update(data->ctx, line, new_len);
4343         data->patchlen += new_len;
4344 }
4345
4346 /* returns 0 upon success, and writes result into sha1 */
4347 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4348 {
4349         struct diff_queue_struct *q = &diff_queued_diff;
4350         int i;
4351         git_SHA_CTX ctx;
4352         struct patch_id_t data;
4353         char buffer[PATH_MAX * 4 + 20];
4354
4355         git_SHA1_Init(&ctx);
4356         memset(&data, 0, sizeof(struct patch_id_t));
4357         data.ctx = &ctx;
4358
4359         for (i = 0; i < q->nr; i++) {
4360                 xpparam_t xpp;
4361                 xdemitconf_t xecfg;
4362                 mmfile_t mf1, mf2;
4363                 struct diff_filepair *p = q->queue[i];
4364                 int len1, len2;
4365
4366                 memset(&xpp, 0, sizeof(xpp));
4367                 memset(&xecfg, 0, sizeof(xecfg));
4368                 if (p->status == 0)
4369                         return error("internal diff status error");
4370                 if (p->status == DIFF_STATUS_UNKNOWN)
4371                         continue;
4372                 if (diff_unmodified_pair(p))
4373                         continue;
4374                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4375                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4376                         continue;
4377                 if (DIFF_PAIR_UNMERGED(p))
4378                         continue;
4379
4380                 diff_fill_sha1_info(p->one);
4381                 diff_fill_sha1_info(p->two);
4382                 if (fill_mmfile(&mf1, p->one) < 0 ||
4383                                 fill_mmfile(&mf2, p->two) < 0)
4384                         return error("unable to read files to diff");
4385
4386                 len1 = remove_space(p->one->path, strlen(p->one->path));
4387                 len2 = remove_space(p->two->path, strlen(p->two->path));
4388                 if (p->one->mode == 0)
4389                         len1 = snprintf(buffer, sizeof(buffer),
4390                                         "diff--gita/%.*sb/%.*s"
4391                                         "newfilemode%06o"
4392                                         "---/dev/null"
4393                                         "+++b/%.*s",
4394                                         len1, p->one->path,
4395                                         len2, p->two->path,
4396                                         p->two->mode,
4397                                         len2, p->two->path);
4398                 else if (p->two->mode == 0)
4399                         len1 = snprintf(buffer, sizeof(buffer),
4400                                         "diff--gita/%.*sb/%.*s"
4401                                         "deletedfilemode%06o"
4402                                         "---a/%.*s"
4403                                         "+++/dev/null",
4404                                         len1, p->one->path,
4405                                         len2, p->two->path,
4406                                         p->one->mode,
4407                                         len1, p->one->path);
4408                 else
4409                         len1 = snprintf(buffer, sizeof(buffer),
4410                                         "diff--gita/%.*sb/%.*s"
4411                                         "---a/%.*s"
4412                                         "+++b/%.*s",
4413                                         len1, p->one->path,
4414                                         len2, p->two->path,
4415                                         len1, p->one->path,
4416                                         len2, p->two->path);
4417                 git_SHA1_Update(&ctx, buffer, len1);
4418
4419                 if (diff_filespec_is_binary(p->one) ||
4420                     diff_filespec_is_binary(p->two)) {
4421                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4422                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4423                         continue;
4424                 }
4425
4426                 xpp.flags = 0;
4427                 xecfg.ctxlen = 3;
4428                 xecfg.flags = 0;
4429                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4430                               &xpp, &xecfg);
4431         }
4432
4433         git_SHA1_Final(sha1, &ctx);
4434         return 0;
4435 }
4436
4437 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4438 {
4439         struct diff_queue_struct *q = &diff_queued_diff;
4440         int i;
4441         int result = diff_get_patch_id(options, sha1);
4442
4443         for (i = 0; i < q->nr; i++)
4444                 diff_free_filepair(q->queue[i]);
4445
4446         free(q->queue);
4447         DIFF_QUEUE_CLEAR(q);
4448
4449         return result;
4450 }
4451
4452 static int is_summary_empty(const struct diff_queue_struct *q)
4453 {
4454         int i;
4455
4456         for (i = 0; i < q->nr; i++) {
4457                 const struct diff_filepair *p = q->queue[i];
4458
4459                 switch (p->status) {
4460                 case DIFF_STATUS_DELETED:
4461                 case DIFF_STATUS_ADDED:
4462                 case DIFF_STATUS_COPIED:
4463                 case DIFF_STATUS_RENAMED:
4464                         return 0;
4465                 default:
4466                         if (p->score)
4467                                 return 0;
4468                         if (p->one->mode && p->two->mode &&
4469                             p->one->mode != p->two->mode)
4470                                 return 0;
4471                         break;
4472                 }
4473         }
4474         return 1;
4475 }
4476
4477 static const char rename_limit_warning[] =
4478 "inexact rename detection was skipped due to too many files.";
4479
4480 static const char degrade_cc_to_c_warning[] =
4481 "only found copies from modified paths due to too many files.";
4482
4483 static const char rename_limit_advice[] =
4484 "you may want to set your %s variable to at least "
4485 "%d and retry the command.";
4486
4487 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4488 {
4489         if (degraded_cc)
4490                 warning(degrade_cc_to_c_warning);
4491         else if (needed)
4492                 warning(rename_limit_warning);
4493         else
4494                 return;
4495         if (0 < needed && needed < 32767)
4496                 warning(rename_limit_advice, varname, needed);
4497 }
4498
4499 void diff_flush(struct diff_options *options)
4500 {
4501         struct diff_queue_struct *q = &diff_queued_diff;
4502         int i, output_format = options->output_format;
4503         int separator = 0;
4504         int dirstat_by_line = 0;
4505
4506         /*
4507          * Order: raw, stat, summary, patch
4508          * or:    name/name-status/checkdiff (other bits clear)
4509          */
4510         if (!q->nr)
4511                 goto free_queue;
4512
4513         if (output_format & (DIFF_FORMAT_RAW |
4514                              DIFF_FORMAT_NAME |
4515                              DIFF_FORMAT_NAME_STATUS |
4516                              DIFF_FORMAT_CHECKDIFF)) {
4517                 for (i = 0; i < q->nr; i++) {
4518                         struct diff_filepair *p = q->queue[i];
4519                         if (check_pair_status(p))
4520                                 flush_one_pair(p, options);
4521                 }
4522                 separator++;
4523         }
4524
4525         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4526                 dirstat_by_line = 1;
4527
4528         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4529             dirstat_by_line) {
4530                 struct diffstat_t diffstat;
4531
4532                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4533                 for (i = 0; i < q->nr; i++) {
4534                         struct diff_filepair *p = q->queue[i];
4535                         if (check_pair_status(p))
4536                                 diff_flush_stat(p, options, &diffstat);
4537                 }
4538                 if (output_format & DIFF_FORMAT_NUMSTAT)
4539                         show_numstat(&diffstat, options);
4540                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4541                         show_stats(&diffstat, options);
4542                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4543                         show_shortstats(&diffstat, options);
4544                 if (output_format & DIFF_FORMAT_DIRSTAT)
4545                         show_dirstat_by_line(&diffstat, options);
4546                 free_diffstat_info(&diffstat);
4547                 separator++;
4548         }
4549         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4550                 show_dirstat(options);
4551
4552         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4553                 for (i = 0; i < q->nr; i++) {
4554                         diff_summary(options, q->queue[i]);
4555                 }
4556                 separator++;
4557         }
4558
4559         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4560             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4561             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4562                 /*
4563                  * run diff_flush_patch for the exit status. setting
4564                  * options->file to /dev/null should be safe, because we
4565                  * aren't supposed to produce any output anyway.
4566                  */
4567                 if (options->close_file)
4568                         fclose(options->file);
4569                 options->file = fopen("/dev/null", "w");
4570                 if (!options->file)
4571                         die_errno("Could not open /dev/null");
4572                 options->close_file = 1;
4573                 for (i = 0; i < q->nr; i++) {
4574                         struct diff_filepair *p = q->queue[i];
4575                         if (check_pair_status(p))
4576                                 diff_flush_patch(p, options);
4577                         if (options->found_changes)
4578                                 break;
4579                 }
4580         }
4581
4582         if (output_format & DIFF_FORMAT_PATCH) {
4583                 if (separator) {
4584                         fprintf(options->file, "%s%c",
4585                                 diff_line_prefix(options),
4586                                 options->line_termination);
4587                         if (options->stat_sep) {
4588                                 /* attach patch instead of inline */
4589                                 fputs(options->stat_sep, options->file);
4590                         }
4591                 }
4592
4593                 for (i = 0; i < q->nr; i++) {
4594                         struct diff_filepair *p = q->queue[i];
4595                         if (check_pair_status(p))
4596                                 diff_flush_patch(p, options);
4597                 }
4598         }
4599
4600         if (output_format & DIFF_FORMAT_CALLBACK)
4601                 options->format_callback(q, options, options->format_callback_data);
4602
4603         for (i = 0; i < q->nr; i++)
4604                 diff_free_filepair(q->queue[i]);
4605 free_queue:
4606         free(q->queue);
4607         DIFF_QUEUE_CLEAR(q);
4608         if (options->close_file)
4609                 fclose(options->file);
4610
4611         /*
4612          * Report the content-level differences with HAS_CHANGES;
4613          * diff_addremove/diff_change does not set the bit when
4614          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4615          */
4616         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4617                 if (options->found_changes)
4618                         DIFF_OPT_SET(options, HAS_CHANGES);
4619                 else
4620                         DIFF_OPT_CLR(options, HAS_CHANGES);
4621         }
4622 }
4623
4624 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4625 {
4626         return (((p->status == DIFF_STATUS_MODIFIED) &&
4627                  ((p->score &&
4628                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4629                   (!p->score &&
4630                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4631                 ((p->status != DIFF_STATUS_MODIFIED) &&
4632                  filter_bit_tst(p->status, options)));
4633 }
4634
4635 static void diffcore_apply_filter(struct diff_options *options)
4636 {
4637         int i;
4638         struct diff_queue_struct *q = &diff_queued_diff;
4639         struct diff_queue_struct outq;
4640
4641         DIFF_QUEUE_CLEAR(&outq);
4642
4643         if (!options->filter)
4644                 return;
4645
4646         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4647                 int found;
4648                 for (i = found = 0; !found && i < q->nr; i++) {
4649                         if (match_filter(options, q->queue[i]))
4650                                 found++;
4651                 }
4652                 if (found)
4653                         return;
4654
4655                 /* otherwise we will clear the whole queue
4656                  * by copying the empty outq at the end of this
4657                  * function, but first clear the current entries
4658                  * in the queue.
4659                  */
4660                 for (i = 0; i < q->nr; i++)
4661                         diff_free_filepair(q->queue[i]);
4662         }
4663         else {
4664                 /* Only the matching ones */
4665                 for (i = 0; i < q->nr; i++) {
4666                         struct diff_filepair *p = q->queue[i];
4667                         if (match_filter(options, p))
4668                                 diff_q(&outq, p);
4669                         else
4670                                 diff_free_filepair(p);
4671                 }
4672         }
4673         free(q->queue);
4674         *q = outq;
4675 }
4676
4677 /* Check whether two filespecs with the same mode and size are identical */
4678 static int diff_filespec_is_identical(struct diff_filespec *one,
4679                                       struct diff_filespec *two)
4680 {
4681         if (S_ISGITLINK(one->mode))
4682                 return 0;
4683         if (diff_populate_filespec(one, 0))
4684                 return 0;
4685         if (diff_populate_filespec(two, 0))
4686                 return 0;
4687         return !memcmp(one->data, two->data, one->size);
4688 }
4689
4690 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4691 {
4692         if (p->done_skip_stat_unmatch)
4693                 return p->skip_stat_unmatch_result;
4694
4695         p->done_skip_stat_unmatch = 1;
4696         p->skip_stat_unmatch_result = 0;
4697         /*
4698          * 1. Entries that come from stat info dirtiness
4699          *    always have both sides (iow, not create/delete),
4700          *    one side of the object name is unknown, with
4701          *    the same mode and size.  Keep the ones that
4702          *    do not match these criteria.  They have real
4703          *    differences.
4704          *
4705          * 2. At this point, the file is known to be modified,
4706          *    with the same mode and size, and the object
4707          *    name of one side is unknown.  Need to inspect
4708          *    the identical contents.
4709          */
4710         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4711             !DIFF_FILE_VALID(p->two) ||
4712             (p->one->sha1_valid && p->two->sha1_valid) ||
4713             (p->one->mode != p->two->mode) ||
4714             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4715             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4716             (p->one->size != p->two->size) ||
4717             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4718                 p->skip_stat_unmatch_result = 1;
4719         return p->skip_stat_unmatch_result;
4720 }
4721
4722 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4723 {
4724         int i;
4725         struct diff_queue_struct *q = &diff_queued_diff;
4726         struct diff_queue_struct outq;
4727         DIFF_QUEUE_CLEAR(&outq);
4728
4729         for (i = 0; i < q->nr; i++) {
4730                 struct diff_filepair *p = q->queue[i];
4731
4732                 if (diff_filespec_check_stat_unmatch(p))
4733                         diff_q(&outq, p);
4734                 else {
4735                         /*
4736                          * The caller can subtract 1 from skip_stat_unmatch
4737                          * to determine how many paths were dirty only
4738                          * due to stat info mismatch.
4739                          */
4740                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4741                                 diffopt->skip_stat_unmatch++;
4742                         diff_free_filepair(p);
4743                 }
4744         }
4745         free(q->queue);
4746         *q = outq;
4747 }
4748
4749 static int diffnamecmp(const void *a_, const void *b_)
4750 {
4751         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4752         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4753         const char *name_a, *name_b;
4754
4755         name_a = a->one ? a->one->path : a->two->path;
4756         name_b = b->one ? b->one->path : b->two->path;
4757         return strcmp(name_a, name_b);
4758 }
4759
4760 void diffcore_fix_diff_index(struct diff_options *options)
4761 {
4762         struct diff_queue_struct *q = &diff_queued_diff;
4763         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4764 }
4765
4766 void diffcore_std(struct diff_options *options)
4767 {
4768         /* NOTE please keep the following in sync with diff_tree_combined() */
4769         if (options->skip_stat_unmatch)
4770                 diffcore_skip_stat_unmatch(options);
4771         if (!options->found_follow) {
4772                 /* See try_to_follow_renames() in tree-diff.c */
4773                 if (options->break_opt != -1)
4774                         diffcore_break(options->break_opt);
4775                 if (options->detect_rename)
4776                         diffcore_rename(options);
4777                 if (options->break_opt != -1)
4778                         diffcore_merge_broken();
4779         }
4780         if (options->pickaxe)
4781                 diffcore_pickaxe(options);
4782         if (options->orderfile)
4783                 diffcore_order(options->orderfile);
4784         if (!options->found_follow)
4785                 /* See try_to_follow_renames() in tree-diff.c */
4786                 diff_resolve_rename_copy();
4787         diffcore_apply_filter(options);
4788
4789         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4790                 DIFF_OPT_SET(options, HAS_CHANGES);
4791         else
4792                 DIFF_OPT_CLR(options, HAS_CHANGES);
4793
4794         options->found_follow = 0;
4795 }
4796
4797 int diff_result_code(struct diff_options *opt, int status)
4798 {
4799         int result = 0;
4800
4801         diff_warn_rename_limit("diff.renameLimit",
4802                                opt->needed_rename_limit,
4803                                opt->degraded_cc_to_c);
4804         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4805             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4806                 return status;
4807         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4808             DIFF_OPT_TST(opt, HAS_CHANGES))
4809                 result |= 01;
4810         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4811             DIFF_OPT_TST(opt, CHECK_FAILED))
4812                 result |= 02;
4813         return result;
4814 }
4815
4816 int diff_can_quit_early(struct diff_options *opt)
4817 {
4818         return (DIFF_OPT_TST(opt, QUICK) &&
4819                 !opt->filter &&
4820                 DIFF_OPT_TST(opt, HAS_CHANGES));
4821 }
4822
4823 /*
4824  * Shall changes to this submodule be ignored?
4825  *
4826  * Submodule changes can be configured to be ignored separately for each path,
4827  * but that configuration can be overridden from the command line.
4828  */
4829 static int is_submodule_ignored(const char *path, struct diff_options *options)
4830 {
4831         int ignored = 0;
4832         unsigned orig_flags = options->flags;
4833         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4834                 set_diffopt_flags_from_submodule_config(options, path);
4835         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4836                 ignored = 1;
4837         options->flags = orig_flags;
4838         return ignored;
4839 }
4840
4841 void diff_addremove(struct diff_options *options,
4842                     int addremove, unsigned mode,
4843                     const unsigned char *sha1,
4844                     int sha1_valid,
4845                     const char *concatpath, unsigned dirty_submodule)
4846 {
4847         struct diff_filespec *one, *two;
4848
4849         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4850                 return;
4851
4852         /* This may look odd, but it is a preparation for
4853          * feeding "there are unchanged files which should
4854          * not produce diffs, but when you are doing copy
4855          * detection you would need them, so here they are"
4856          * entries to the diff-core.  They will be prefixed
4857          * with something like '=' or '*' (I haven't decided
4858          * which but should not make any difference).
4859          * Feeding the same new and old to diff_change()
4860          * also has the same effect.
4861          * Before the final output happens, they are pruned after
4862          * merged into rename/copy pairs as appropriate.
4863          */
4864         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4865                 addremove = (addremove == '+' ? '-' :
4866                              addremove == '-' ? '+' : addremove);
4867
4868         if (options->prefix &&
4869             strncmp(concatpath, options->prefix, options->prefix_length))
4870                 return;
4871
4872         one = alloc_filespec(concatpath);
4873         two = alloc_filespec(concatpath);
4874
4875         if (addremove != '+')
4876                 fill_filespec(one, sha1, sha1_valid, mode);
4877         if (addremove != '-') {
4878                 fill_filespec(two, sha1, sha1_valid, mode);
4879                 two->dirty_submodule = dirty_submodule;
4880         }
4881
4882         diff_queue(&diff_queued_diff, one, two);
4883         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4884                 DIFF_OPT_SET(options, HAS_CHANGES);
4885 }
4886
4887 void diff_change(struct diff_options *options,
4888                  unsigned old_mode, unsigned new_mode,
4889                  const unsigned char *old_sha1,
4890                  const unsigned char *new_sha1,
4891                  int old_sha1_valid, int new_sha1_valid,
4892                  const char *concatpath,
4893                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4894 {
4895         struct diff_filespec *one, *two;
4896         struct diff_filepair *p;
4897
4898         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4899             is_submodule_ignored(concatpath, options))
4900                 return;
4901
4902         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4903                 unsigned tmp;
4904                 const unsigned char *tmp_c;
4905                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4906                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4907                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4908                         new_sha1_valid = tmp;
4909                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4910                         new_dirty_submodule = tmp;
4911         }
4912
4913         if (options->prefix &&
4914             strncmp(concatpath, options->prefix, options->prefix_length))
4915                 return;
4916
4917         one = alloc_filespec(concatpath);
4918         two = alloc_filespec(concatpath);
4919         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4920         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4921         one->dirty_submodule = old_dirty_submodule;
4922         two->dirty_submodule = new_dirty_submodule;
4923         p = diff_queue(&diff_queued_diff, one, two);
4924
4925         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4926                 return;
4927
4928         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
4929             !diff_filespec_check_stat_unmatch(p))
4930                 return;
4931
4932         DIFF_OPT_SET(options, HAS_CHANGES);
4933 }
4934
4935 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4936 {
4937         struct diff_filepair *pair;
4938         struct diff_filespec *one, *two;
4939
4940         if (options->prefix &&
4941             strncmp(path, options->prefix, options->prefix_length))
4942                 return NULL;
4943
4944         one = alloc_filespec(path);
4945         two = alloc_filespec(path);
4946         pair = diff_queue(&diff_queued_diff, one, two);
4947         pair->is_unmerged = 1;
4948         return pair;
4949 }
4950
4951 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4952                 size_t *outsize)
4953 {
4954         struct diff_tempfile *temp;
4955         const char *argv[3];
4956         const char **arg = argv;
4957         struct child_process child = CHILD_PROCESS_INIT;
4958         struct strbuf buf = STRBUF_INIT;
4959         int err = 0;
4960
4961         temp = prepare_temp_file(spec->path, spec);
4962         *arg++ = pgm;
4963         *arg++ = temp->name;
4964         *arg = NULL;
4965
4966         child.use_shell = 1;
4967         child.argv = argv;
4968         child.out = -1;
4969         if (start_command(&child)) {
4970                 remove_tempfile();
4971                 return NULL;
4972         }
4973
4974         if (strbuf_read(&buf, child.out, 0) < 0)
4975                 err = error("error reading from textconv command '%s'", pgm);
4976         close(child.out);
4977
4978         if (finish_command(&child) || err) {
4979                 strbuf_release(&buf);
4980                 remove_tempfile();
4981                 return NULL;
4982         }
4983         remove_tempfile();
4984
4985         return strbuf_detach(&buf, outsize);
4986 }
4987
4988 size_t fill_textconv(struct userdiff_driver *driver,
4989                      struct diff_filespec *df,
4990                      char **outbuf)
4991 {
4992         size_t size;
4993
4994         if (!driver || !driver->textconv) {
4995                 if (!DIFF_FILE_VALID(df)) {
4996                         *outbuf = "";
4997                         return 0;
4998                 }
4999                 if (diff_populate_filespec(df, 0))
5000                         die("unable to read files to diff");
5001                 *outbuf = df->data;
5002                 return df->size;
5003         }
5004
5005         if (driver->textconv_cache && df->sha1_valid) {
5006                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5007                                           &size);
5008                 if (*outbuf)
5009                         return size;
5010         }
5011
5012         *outbuf = run_textconv(driver->textconv, df, &size);
5013         if (!*outbuf)
5014                 die("unable to read files to diff");
5015
5016         if (driver->textconv_cache && df->sha1_valid) {
5017                 /* ignore errors, as we might be in a readonly repository */
5018                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5019                                 size);
5020                 /*
5021                  * we could save up changes and flush them all at the end,
5022                  * but we would need an extra call after all diffing is done.
5023                  * Since generating a cache entry is the slow path anyway,
5024                  * this extra overhead probably isn't a big deal.
5025                  */
5026                 notes_cache_write(driver->textconv_cache);
5027         }
5028
5029         return size;
5030 }
5031
5032 void setup_diff_pager(struct diff_options *opt)
5033 {
5034         /*
5035          * If the user asked for our exit code, then either they want --quiet
5036          * or --exit-code. We should definitely not bother with a pager in the
5037          * former case, as we will generate no output. Since we still properly
5038          * report our exit code even when a pager is run, we _could_ run a
5039          * pager with --exit-code. But since we have not done so historically,
5040          * and because it is easy to find people oneline advising "git diff
5041          * --exit-code" in hooks and other scripts, we do not do so.
5042          */
5043         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5044             check_pager_config("diff") != 0)
5045                 setup_pager();
5046 }