Sync with 1.7.12.1
[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
19 #ifdef NO_FAST_WORKING_DIRECTORY
20 #define FAST_WORKING_DIRECTORY 0
21 #else
22 #define FAST_WORKING_DIRECTORY 1
23 #endif
24
25 static int diff_detect_rename_default;
26 static int diff_rename_limit_default = 400;
27 static int diff_suppress_blank_empty;
28 int diff_use_color_default = -1;
29 static const char *diff_word_regex_cfg;
30 static const char *external_diff_cmd_cfg;
31 int diff_auto_refresh_index = 1;
32 static int diff_mnemonic_prefix;
33 static int diff_no_prefix;
34 static int diff_stat_graph_width;
35 static int diff_dirstat_permille_default = 30;
36 static struct diff_options default_diff_options;
37
38 static char diff_colors[][COLOR_MAXLEN] = {
39         GIT_COLOR_RESET,
40         GIT_COLOR_NORMAL,       /* PLAIN */
41         GIT_COLOR_BOLD,         /* METAINFO */
42         GIT_COLOR_CYAN,         /* FRAGINFO */
43         GIT_COLOR_RED,          /* OLD */
44         GIT_COLOR_GREEN,        /* NEW */
45         GIT_COLOR_YELLOW,       /* COMMIT */
46         GIT_COLOR_BG_RED,       /* WHITESPACE */
47         GIT_COLOR_NORMAL,       /* FUNCINFO */
48 };
49
50 static int parse_diff_color_slot(const char *var, int ofs)
51 {
52         if (!strcasecmp(var+ofs, "plain"))
53                 return DIFF_PLAIN;
54         if (!strcasecmp(var+ofs, "meta"))
55                 return DIFF_METAINFO;
56         if (!strcasecmp(var+ofs, "frag"))
57                 return DIFF_FRAGINFO;
58         if (!strcasecmp(var+ofs, "old"))
59                 return DIFF_FILE_OLD;
60         if (!strcasecmp(var+ofs, "new"))
61                 return DIFF_FILE_NEW;
62         if (!strcasecmp(var+ofs, "commit"))
63                 return DIFF_COMMIT;
64         if (!strcasecmp(var+ofs, "whitespace"))
65                 return DIFF_WHITESPACE;
66         if (!strcasecmp(var+ofs, "func"))
67                 return DIFF_FUNCINFO;
68         return -1;
69 }
70
71 static int parse_dirstat_params(struct diff_options *options, const char *params,
72                                 struct strbuf *errmsg)
73 {
74         const char *p = params;
75         int p_len, ret = 0;
76
77         while (*p) {
78                 p_len = strchrnul(p, ',') - p;
79                 if (!memcmp(p, "changes", p_len)) {
80                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
81                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
82                 } else if (!memcmp(p, "lines", p_len)) {
83                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
84                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
85                 } else if (!memcmp(p, "files", p_len)) {
86                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
87                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
88                 } else if (!memcmp(p, "noncumulative", p_len)) {
89                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
90                 } else if (!memcmp(p, "cumulative", p_len)) {
91                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
92                 } else if (isdigit(*p)) {
93                         char *end;
94                         int permille = strtoul(p, &end, 10) * 10;
95                         if (*end == '.' && isdigit(*++end)) {
96                                 /* only use first digit */
97                                 permille += *end - '0';
98                                 /* .. and ignore any further digits */
99                                 while (isdigit(*++end))
100                                         ; /* nothing */
101                         }
102                         if (end - p == p_len)
103                                 options->dirstat_permille = permille;
104                         else {
105                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%.*s'\n"),
106                                             p_len, p);
107                                 ret++;
108                         }
109                 } else {
110                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%.*s'\n"),
111                                     p_len, p);
112                         ret++;
113                 }
114
115                 p += p_len;
116
117                 if (*p)
118                         p++; /* more parameters, swallow separator */
119         }
120         return ret;
121 }
122
123 static int git_config_rename(const char *var, const char *value)
124 {
125         if (!value)
126                 return DIFF_DETECT_RENAME;
127         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
128                 return  DIFF_DETECT_COPY;
129         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
130 }
131
132 /*
133  * These are to give UI layer defaults.
134  * The core-level commands such as git-diff-files should
135  * never be affected by the setting of diff.renames
136  * the user happens to have in the configuration file.
137  */
138 int git_diff_ui_config(const char *var, const char *value, void *cb)
139 {
140         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
141                 diff_use_color_default = git_config_colorbool(var, value);
142                 return 0;
143         }
144         if (!strcmp(var, "diff.renames")) {
145                 diff_detect_rename_default = git_config_rename(var, value);
146                 return 0;
147         }
148         if (!strcmp(var, "diff.autorefreshindex")) {
149                 diff_auto_refresh_index = git_config_bool(var, value);
150                 return 0;
151         }
152         if (!strcmp(var, "diff.mnemonicprefix")) {
153                 diff_mnemonic_prefix = git_config_bool(var, value);
154                 return 0;
155         }
156         if (!strcmp(var, "diff.noprefix")) {
157                 diff_no_prefix = git_config_bool(var, value);
158                 return 0;
159         }
160         if (!strcmp(var, "diff.statgraphwidth")) {
161                 diff_stat_graph_width = git_config_int(var, value);
162                 return 0;
163         }
164         if (!strcmp(var, "diff.external"))
165                 return git_config_string(&external_diff_cmd_cfg, var, value);
166         if (!strcmp(var, "diff.wordregex"))
167                 return git_config_string(&diff_word_regex_cfg, var, value);
168
169         if (!strcmp(var, "diff.ignoresubmodules"))
170                 handle_ignore_submodules_arg(&default_diff_options, value);
171
172         if (git_color_config(var, value, cb) < 0)
173                 return -1;
174
175         return git_diff_basic_config(var, value, cb);
176 }
177
178 int git_diff_basic_config(const char *var, const char *value, void *cb)
179 {
180         if (!strcmp(var, "diff.renamelimit")) {
181                 diff_rename_limit_default = git_config_int(var, value);
182                 return 0;
183         }
184
185         if (userdiff_config(var, value) < 0)
186                 return -1;
187
188         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
189                 int slot = parse_diff_color_slot(var, 11);
190                 if (slot < 0)
191                         return 0;
192                 if (!value)
193                         return config_error_nonbool(var);
194                 color_parse(value, var, diff_colors[slot]);
195                 return 0;
196         }
197
198         /* like GNU diff's --suppress-blank-empty option  */
199         if (!strcmp(var, "diff.suppressblankempty") ||
200                         /* for backwards compatibility */
201                         !strcmp(var, "diff.suppress-blank-empty")) {
202                 diff_suppress_blank_empty = git_config_bool(var, value);
203                 return 0;
204         }
205
206         if (!strcmp(var, "diff.dirstat")) {
207                 struct strbuf errmsg = STRBUF_INIT;
208                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
209                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
210                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
211                                 errmsg.buf);
212                 strbuf_release(&errmsg);
213                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
214                 return 0;
215         }
216
217         if (!prefixcmp(var, "submodule."))
218                 return parse_submodule_config_option(var, value);
219
220         return git_default_config(var, value, cb);
221 }
222
223 static char *quote_two(const char *one, const char *two)
224 {
225         int need_one = quote_c_style(one, NULL, NULL, 1);
226         int need_two = quote_c_style(two, NULL, NULL, 1);
227         struct strbuf res = STRBUF_INIT;
228
229         if (need_one + need_two) {
230                 strbuf_addch(&res, '"');
231                 quote_c_style(one, &res, NULL, 1);
232                 quote_c_style(two, &res, NULL, 1);
233                 strbuf_addch(&res, '"');
234         } else {
235                 strbuf_addstr(&res, one);
236                 strbuf_addstr(&res, two);
237         }
238         return strbuf_detach(&res, NULL);
239 }
240
241 static const char *external_diff(void)
242 {
243         static const char *external_diff_cmd = NULL;
244         static int done_preparing = 0;
245
246         if (done_preparing)
247                 return external_diff_cmd;
248         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
249         if (!external_diff_cmd)
250                 external_diff_cmd = external_diff_cmd_cfg;
251         done_preparing = 1;
252         return external_diff_cmd;
253 }
254
255 static struct diff_tempfile {
256         const char *name; /* filename external diff should read from */
257         char hex[41];
258         char mode[10];
259         char tmp_path[PATH_MAX];
260 } diff_temp[2];
261
262 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
263
264 struct emit_callback {
265         int color_diff;
266         unsigned ws_rule;
267         int blank_at_eof_in_preimage;
268         int blank_at_eof_in_postimage;
269         int lno_in_preimage;
270         int lno_in_postimage;
271         sane_truncate_fn truncate;
272         const char **label_path;
273         struct diff_words_data *diff_words;
274         struct diff_options *opt;
275         int *found_changesp;
276         struct strbuf *header;
277 };
278
279 static int count_lines(const char *data, int size)
280 {
281         int count, ch, completely_empty = 1, nl_just_seen = 0;
282         count = 0;
283         while (0 < size--) {
284                 ch = *data++;
285                 if (ch == '\n') {
286                         count++;
287                         nl_just_seen = 1;
288                         completely_empty = 0;
289                 }
290                 else {
291                         nl_just_seen = 0;
292                         completely_empty = 0;
293                 }
294         }
295         if (completely_empty)
296                 return 0;
297         if (!nl_just_seen)
298                 count++; /* no trailing newline */
299         return count;
300 }
301
302 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
303 {
304         if (!DIFF_FILE_VALID(one)) {
305                 mf->ptr = (char *)""; /* does not matter */
306                 mf->size = 0;
307                 return 0;
308         }
309         else if (diff_populate_filespec(one, 0))
310                 return -1;
311
312         mf->ptr = one->data;
313         mf->size = one->size;
314         return 0;
315 }
316
317 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
318 static unsigned long diff_filespec_size(struct diff_filespec *one)
319 {
320         if (!DIFF_FILE_VALID(one))
321                 return 0;
322         diff_populate_filespec(one, 1);
323         return one->size;
324 }
325
326 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
327 {
328         char *ptr = mf->ptr;
329         long size = mf->size;
330         int cnt = 0;
331
332         if (!size)
333                 return cnt;
334         ptr += size - 1; /* pointing at the very end */
335         if (*ptr != '\n')
336                 ; /* incomplete line */
337         else
338                 ptr--; /* skip the last LF */
339         while (mf->ptr < ptr) {
340                 char *prev_eol;
341                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
342                         if (*prev_eol == '\n')
343                                 break;
344                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
345                         break;
346                 cnt++;
347                 ptr = prev_eol - 1;
348         }
349         return cnt;
350 }
351
352 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
353                                struct emit_callback *ecbdata)
354 {
355         int l1, l2, at;
356         unsigned ws_rule = ecbdata->ws_rule;
357         l1 = count_trailing_blank(mf1, ws_rule);
358         l2 = count_trailing_blank(mf2, ws_rule);
359         if (l2 <= l1) {
360                 ecbdata->blank_at_eof_in_preimage = 0;
361                 ecbdata->blank_at_eof_in_postimage = 0;
362                 return;
363         }
364         at = count_lines(mf1->ptr, mf1->size);
365         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
366
367         at = count_lines(mf2->ptr, mf2->size);
368         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
369 }
370
371 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
372                         int first, const char *line, int len)
373 {
374         int has_trailing_newline, has_trailing_carriage_return;
375         int nofirst;
376         FILE *file = o->file;
377
378         if (o->output_prefix) {
379                 struct strbuf *msg = NULL;
380                 msg = o->output_prefix(o, o->output_prefix_data);
381                 assert(msg);
382                 fwrite(msg->buf, msg->len, 1, file);
383         }
384
385         if (len == 0) {
386                 has_trailing_newline = (first == '\n');
387                 has_trailing_carriage_return = (!has_trailing_newline &&
388                                                 (first == '\r'));
389                 nofirst = has_trailing_newline || has_trailing_carriage_return;
390         } else {
391                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
392                 if (has_trailing_newline)
393                         len--;
394                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
395                 if (has_trailing_carriage_return)
396                         len--;
397                 nofirst = 0;
398         }
399
400         if (len || !nofirst) {
401                 fputs(set, file);
402                 if (!nofirst)
403                         fputc(first, file);
404                 fwrite(line, len, 1, file);
405                 fputs(reset, file);
406         }
407         if (has_trailing_carriage_return)
408                 fputc('\r', file);
409         if (has_trailing_newline)
410                 fputc('\n', file);
411 }
412
413 static void emit_line(struct diff_options *o, const char *set, const char *reset,
414                       const char *line, int len)
415 {
416         emit_line_0(o, set, reset, line[0], line+1, len-1);
417 }
418
419 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
420 {
421         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
422               ecbdata->blank_at_eof_in_preimage &&
423               ecbdata->blank_at_eof_in_postimage &&
424               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
425               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
426                 return 0;
427         return ws_blank_line(line, len, ecbdata->ws_rule);
428 }
429
430 static void emit_add_line(const char *reset,
431                           struct emit_callback *ecbdata,
432                           const char *line, int len)
433 {
434         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
435         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
436
437         if (!*ws)
438                 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
439         else if (new_blank_line_at_eof(ecbdata, line, len))
440                 /* Blank line at EOF - paint '+' as well */
441                 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
442         else {
443                 /* Emit just the prefix, then the rest. */
444                 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
445                 ws_check_emit(line, len, ecbdata->ws_rule,
446                               ecbdata->opt->file, set, reset, ws);
447         }
448 }
449
450 static void emit_hunk_header(struct emit_callback *ecbdata,
451                              const char *line, int len)
452 {
453         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
454         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
455         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
456         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
457         static const char atat[2] = { '@', '@' };
458         const char *cp, *ep;
459         struct strbuf msgbuf = STRBUF_INIT;
460         int org_len = len;
461         int i = 1;
462
463         /*
464          * As a hunk header must begin with "@@ -<old>, +<new> @@",
465          * it always is at least 10 bytes long.
466          */
467         if (len < 10 ||
468             memcmp(line, atat, 2) ||
469             !(ep = memmem(line + 2, len - 2, atat, 2))) {
470                 emit_line(ecbdata->opt, plain, reset, line, len);
471                 return;
472         }
473         ep += 2; /* skip over @@ */
474
475         /* The hunk header in fraginfo color */
476         strbuf_add(&msgbuf, frag, strlen(frag));
477         strbuf_add(&msgbuf, line, ep - line);
478         strbuf_add(&msgbuf, reset, strlen(reset));
479
480         /*
481          * trailing "\r\n"
482          */
483         for ( ; i < 3; i++)
484                 if (line[len - i] == '\r' || line[len - i] == '\n')
485                         len--;
486
487         /* blank before the func header */
488         for (cp = ep; ep - line < len; ep++)
489                 if (*ep != ' ' && *ep != '\t')
490                         break;
491         if (ep != cp) {
492                 strbuf_add(&msgbuf, plain, strlen(plain));
493                 strbuf_add(&msgbuf, cp, ep - cp);
494                 strbuf_add(&msgbuf, reset, strlen(reset));
495         }
496
497         if (ep < line + len) {
498                 strbuf_add(&msgbuf, func, strlen(func));
499                 strbuf_add(&msgbuf, ep, line + len - ep);
500                 strbuf_add(&msgbuf, reset, strlen(reset));
501         }
502
503         strbuf_add(&msgbuf, line + len, org_len - len);
504         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
505         strbuf_release(&msgbuf);
506 }
507
508 static struct diff_tempfile *claim_diff_tempfile(void) {
509         int i;
510         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
511                 if (!diff_temp[i].name)
512                         return diff_temp + i;
513         die("BUG: diff is failing to clean up its tempfiles");
514 }
515
516 static int remove_tempfile_installed;
517
518 static void remove_tempfile(void)
519 {
520         int i;
521         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
522                 if (diff_temp[i].name == diff_temp[i].tmp_path)
523                         unlink_or_warn(diff_temp[i].name);
524                 diff_temp[i].name = NULL;
525         }
526 }
527
528 static void remove_tempfile_on_signal(int signo)
529 {
530         remove_tempfile();
531         sigchain_pop(signo);
532         raise(signo);
533 }
534
535 static void print_line_count(FILE *file, int count)
536 {
537         switch (count) {
538         case 0:
539                 fprintf(file, "0,0");
540                 break;
541         case 1:
542                 fprintf(file, "1");
543                 break;
544         default:
545                 fprintf(file, "1,%d", count);
546                 break;
547         }
548 }
549
550 static void emit_rewrite_lines(struct emit_callback *ecb,
551                                int prefix, const char *data, int size)
552 {
553         const char *endp = NULL;
554         static const char *nneof = " No newline at end of file\n";
555         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
556         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
557
558         while (0 < size) {
559                 int len;
560
561                 endp = memchr(data, '\n', size);
562                 len = endp ? (endp - data + 1) : size;
563                 if (prefix != '+') {
564                         ecb->lno_in_preimage++;
565                         emit_line_0(ecb->opt, old, reset, '-',
566                                     data, len);
567                 } else {
568                         ecb->lno_in_postimage++;
569                         emit_add_line(reset, ecb, data, len);
570                 }
571                 size -= len;
572                 data += len;
573         }
574         if (!endp) {
575                 const char *plain = diff_get_color(ecb->color_diff,
576                                                    DIFF_PLAIN);
577                 putc('\n', ecb->opt->file);
578                 emit_line_0(ecb->opt, plain, reset, '\\',
579                             nneof, strlen(nneof));
580         }
581 }
582
583 static void emit_rewrite_diff(const char *name_a,
584                               const char *name_b,
585                               struct diff_filespec *one,
586                               struct diff_filespec *two,
587                               struct userdiff_driver *textconv_one,
588                               struct userdiff_driver *textconv_two,
589                               struct diff_options *o)
590 {
591         int lc_a, lc_b;
592         const char *name_a_tab, *name_b_tab;
593         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
594         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
595         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
596         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
597         const char *a_prefix, *b_prefix;
598         char *data_one, *data_two;
599         size_t size_one, size_two;
600         struct emit_callback ecbdata;
601         char *line_prefix = "";
602         struct strbuf *msgbuf;
603
604         if (o && o->output_prefix) {
605                 msgbuf = o->output_prefix(o, o->output_prefix_data);
606                 line_prefix = msgbuf->buf;
607         }
608
609         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
610                 a_prefix = o->b_prefix;
611                 b_prefix = o->a_prefix;
612         } else {
613                 a_prefix = o->a_prefix;
614                 b_prefix = o->b_prefix;
615         }
616
617         name_a += (*name_a == '/');
618         name_b += (*name_b == '/');
619         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
620         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
621
622         strbuf_reset(&a_name);
623         strbuf_reset(&b_name);
624         quote_two_c_style(&a_name, a_prefix, name_a, 0);
625         quote_two_c_style(&b_name, b_prefix, name_b, 0);
626
627         size_one = fill_textconv(textconv_one, one, &data_one);
628         size_two = fill_textconv(textconv_two, two, &data_two);
629
630         memset(&ecbdata, 0, sizeof(ecbdata));
631         ecbdata.color_diff = want_color(o->use_color);
632         ecbdata.found_changesp = &o->found_changes;
633         ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
634         ecbdata.opt = o;
635         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
636                 mmfile_t mf1, mf2;
637                 mf1.ptr = (char *)data_one;
638                 mf2.ptr = (char *)data_two;
639                 mf1.size = size_one;
640                 mf2.size = size_two;
641                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
642         }
643         ecbdata.lno_in_preimage = 1;
644         ecbdata.lno_in_postimage = 1;
645
646         lc_a = count_lines(data_one, size_one);
647         lc_b = count_lines(data_two, size_two);
648         fprintf(o->file,
649                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
650                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
651                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
652                 line_prefix, fraginfo);
653         if (!o->irreversible_delete)
654                 print_line_count(o->file, lc_a);
655         else
656                 fprintf(o->file, "?,?");
657         fprintf(o->file, " +");
658         print_line_count(o->file, lc_b);
659         fprintf(o->file, " @@%s\n", reset);
660         if (lc_a && !o->irreversible_delete)
661                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
662         if (lc_b)
663                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
664         if (textconv_one)
665                 free((char *)data_one);
666         if (textconv_two)
667                 free((char *)data_two);
668 }
669
670 struct diff_words_buffer {
671         mmfile_t text;
672         long alloc;
673         struct diff_words_orig {
674                 const char *begin, *end;
675         } *orig;
676         int orig_nr, orig_alloc;
677 };
678
679 static void diff_words_append(char *line, unsigned long len,
680                 struct diff_words_buffer *buffer)
681 {
682         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
683         line++;
684         len--;
685         memcpy(buffer->text.ptr + buffer->text.size, line, len);
686         buffer->text.size += len;
687         buffer->text.ptr[buffer->text.size] = '\0';
688 }
689
690 struct diff_words_style_elem {
691         const char *prefix;
692         const char *suffix;
693         const char *color; /* NULL; filled in by the setup code if
694                             * color is enabled */
695 };
696
697 struct diff_words_style {
698         enum diff_words_type type;
699         struct diff_words_style_elem new, old, ctx;
700         const char *newline;
701 };
702
703 static struct diff_words_style diff_words_styles[] = {
704         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
705         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
706         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
707 };
708
709 struct diff_words_data {
710         struct diff_words_buffer minus, plus;
711         const char *current_plus;
712         int last_minus;
713         struct diff_options *opt;
714         regex_t *word_regex;
715         enum diff_words_type type;
716         struct diff_words_style *style;
717 };
718
719 static int fn_out_diff_words_write_helper(FILE *fp,
720                                           struct diff_words_style_elem *st_el,
721                                           const char *newline,
722                                           size_t count, const char *buf,
723                                           const char *line_prefix)
724 {
725         int print = 0;
726
727         while (count) {
728                 char *p = memchr(buf, '\n', count);
729                 if (print)
730                         fputs(line_prefix, fp);
731                 if (p != buf) {
732                         if (st_el->color && fputs(st_el->color, fp) < 0)
733                                 return -1;
734                         if (fputs(st_el->prefix, fp) < 0 ||
735                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
736                             fputs(st_el->suffix, fp) < 0)
737                                 return -1;
738                         if (st_el->color && *st_el->color
739                             && fputs(GIT_COLOR_RESET, fp) < 0)
740                                 return -1;
741                 }
742                 if (!p)
743                         return 0;
744                 if (fputs(newline, fp) < 0)
745                         return -1;
746                 count -= p + 1 - buf;
747                 buf = p + 1;
748                 print = 1;
749         }
750         return 0;
751 }
752
753 /*
754  * '--color-words' algorithm can be described as:
755  *
756  *   1. collect a the minus/plus lines of a diff hunk, divided into
757  *      minus-lines and plus-lines;
758  *
759  *   2. break both minus-lines and plus-lines into words and
760  *      place them into two mmfile_t with one word for each line;
761  *
762  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
763  *
764  * And for the common parts of the both file, we output the plus side text.
765  * diff_words->current_plus is used to trace the current position of the plus file
766  * which printed. diff_words->last_minus is used to trace the last minus word
767  * printed.
768  *
769  * For '--graph' to work with '--color-words', we need to output the graph prefix
770  * on each line of color words output. Generally, there are two conditions on
771  * which we should output the prefix.
772  *
773  *   1. diff_words->last_minus == 0 &&
774  *      diff_words->current_plus == diff_words->plus.text.ptr
775  *
776  *      that is: the plus text must start as a new line, and if there is no minus
777  *      word printed, a graph prefix must be printed.
778  *
779  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
780  *      *(diff_words->current_plus - 1) == '\n'
781  *
782  *      that is: a graph prefix must be printed following a '\n'
783  */
784 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
785 {
786         if ((diff_words->last_minus == 0 &&
787                 diff_words->current_plus == diff_words->plus.text.ptr) ||
788                 (diff_words->current_plus > diff_words->plus.text.ptr &&
789                 *(diff_words->current_plus - 1) == '\n')) {
790                 return 1;
791         } else {
792                 return 0;
793         }
794 }
795
796 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
797 {
798         struct diff_words_data *diff_words = priv;
799         struct diff_words_style *style = diff_words->style;
800         int minus_first, minus_len, plus_first, plus_len;
801         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
802         struct diff_options *opt = diff_words->opt;
803         struct strbuf *msgbuf;
804         char *line_prefix = "";
805
806         if (line[0] != '@' || parse_hunk_header(line, len,
807                         &minus_first, &minus_len, &plus_first, &plus_len))
808                 return;
809
810         assert(opt);
811         if (opt->output_prefix) {
812                 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
813                 line_prefix = msgbuf->buf;
814         }
815
816         /* POSIX requires that first be decremented by one if len == 0... */
817         if (minus_len) {
818                 minus_begin = diff_words->minus.orig[minus_first].begin;
819                 minus_end =
820                         diff_words->minus.orig[minus_first + minus_len - 1].end;
821         } else
822                 minus_begin = minus_end =
823                         diff_words->minus.orig[minus_first].end;
824
825         if (plus_len) {
826                 plus_begin = diff_words->plus.orig[plus_first].begin;
827                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
828         } else
829                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
830
831         if (color_words_output_graph_prefix(diff_words)) {
832                 fputs(line_prefix, diff_words->opt->file);
833         }
834         if (diff_words->current_plus != plus_begin) {
835                 fn_out_diff_words_write_helper(diff_words->opt->file,
836                                 &style->ctx, style->newline,
837                                 plus_begin - diff_words->current_plus,
838                                 diff_words->current_plus, line_prefix);
839                 if (*(plus_begin - 1) == '\n')
840                         fputs(line_prefix, diff_words->opt->file);
841         }
842         if (minus_begin != minus_end) {
843                 fn_out_diff_words_write_helper(diff_words->opt->file,
844                                 &style->old, style->newline,
845                                 minus_end - minus_begin, minus_begin,
846                                 line_prefix);
847         }
848         if (plus_begin != plus_end) {
849                 fn_out_diff_words_write_helper(diff_words->opt->file,
850                                 &style->new, style->newline,
851                                 plus_end - plus_begin, plus_begin,
852                                 line_prefix);
853         }
854
855         diff_words->current_plus = plus_end;
856         diff_words->last_minus = minus_first;
857 }
858
859 /* This function starts looking at *begin, and returns 0 iff a word was found. */
860 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
861                 int *begin, int *end)
862 {
863         if (word_regex && *begin < buffer->size) {
864                 regmatch_t match[1];
865                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
866                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
867                                         '\n', match[0].rm_eo - match[0].rm_so);
868                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
869                         *begin += match[0].rm_so;
870                         return *begin >= *end;
871                 }
872                 return -1;
873         }
874
875         /* find the next word */
876         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
877                 (*begin)++;
878         if (*begin >= buffer->size)
879                 return -1;
880
881         /* find the end of the word */
882         *end = *begin + 1;
883         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
884                 (*end)++;
885
886         return 0;
887 }
888
889 /*
890  * This function splits the words in buffer->text, stores the list with
891  * newline separator into out, and saves the offsets of the original words
892  * in buffer->orig.
893  */
894 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
895                 regex_t *word_regex)
896 {
897         int i, j;
898         long alloc = 0;
899
900         out->size = 0;
901         out->ptr = NULL;
902
903         /* fake an empty "0th" word */
904         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
905         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
906         buffer->orig_nr = 1;
907
908         for (i = 0; i < buffer->text.size; i++) {
909                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
910                         return;
911
912                 /* store original boundaries */
913                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
914                                 buffer->orig_alloc);
915                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
916                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
917                 buffer->orig_nr++;
918
919                 /* store one word */
920                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
921                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
922                 out->ptr[out->size + j - i] = '\n';
923                 out->size += j - i + 1;
924
925                 i = j - 1;
926         }
927 }
928
929 /* this executes the word diff on the accumulated buffers */
930 static void diff_words_show(struct diff_words_data *diff_words)
931 {
932         xpparam_t xpp;
933         xdemitconf_t xecfg;
934         mmfile_t minus, plus;
935         struct diff_words_style *style = diff_words->style;
936
937         struct diff_options *opt = diff_words->opt;
938         struct strbuf *msgbuf;
939         char *line_prefix = "";
940
941         assert(opt);
942         if (opt->output_prefix) {
943                 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
944                 line_prefix = msgbuf->buf;
945         }
946
947         /* special case: only removal */
948         if (!diff_words->plus.text.size) {
949                 fputs(line_prefix, diff_words->opt->file);
950                 fn_out_diff_words_write_helper(diff_words->opt->file,
951                         &style->old, style->newline,
952                         diff_words->minus.text.size,
953                         diff_words->minus.text.ptr, line_prefix);
954                 diff_words->minus.text.size = 0;
955                 return;
956         }
957
958         diff_words->current_plus = diff_words->plus.text.ptr;
959         diff_words->last_minus = 0;
960
961         memset(&xpp, 0, sizeof(xpp));
962         memset(&xecfg, 0, sizeof(xecfg));
963         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
964         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
965         xpp.flags = 0;
966         /* as only the hunk header will be parsed, we need a 0-context */
967         xecfg.ctxlen = 0;
968         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
969                       &xpp, &xecfg);
970         free(minus.ptr);
971         free(plus.ptr);
972         if (diff_words->current_plus != diff_words->plus.text.ptr +
973                         diff_words->plus.text.size) {
974                 if (color_words_output_graph_prefix(diff_words))
975                         fputs(line_prefix, diff_words->opt->file);
976                 fn_out_diff_words_write_helper(diff_words->opt->file,
977                         &style->ctx, style->newline,
978                         diff_words->plus.text.ptr + diff_words->plus.text.size
979                         - diff_words->current_plus, diff_words->current_plus,
980                         line_prefix);
981         }
982         diff_words->minus.text.size = diff_words->plus.text.size = 0;
983 }
984
985 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
986 static void diff_words_flush(struct emit_callback *ecbdata)
987 {
988         if (ecbdata->diff_words->minus.text.size ||
989             ecbdata->diff_words->plus.text.size)
990                 diff_words_show(ecbdata->diff_words);
991 }
992
993 static void diff_filespec_load_driver(struct diff_filespec *one)
994 {
995         /* Use already-loaded driver */
996         if (one->driver)
997                 return;
998
999         if (S_ISREG(one->mode))
1000                 one->driver = userdiff_find_by_path(one->path);
1001
1002         /* Fallback to default settings */
1003         if (!one->driver)
1004                 one->driver = userdiff_find_by_name("default");
1005 }
1006
1007 static const char *userdiff_word_regex(struct diff_filespec *one)
1008 {
1009         diff_filespec_load_driver(one);
1010         return one->driver->word_regex;
1011 }
1012
1013 static void init_diff_words_data(struct emit_callback *ecbdata,
1014                                  struct diff_options *orig_opts,
1015                                  struct diff_filespec *one,
1016                                  struct diff_filespec *two)
1017 {
1018         int i;
1019         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1020         memcpy(o, orig_opts, sizeof(struct diff_options));
1021
1022         ecbdata->diff_words =
1023                 xcalloc(1, sizeof(struct diff_words_data));
1024         ecbdata->diff_words->type = o->word_diff;
1025         ecbdata->diff_words->opt = o;
1026         if (!o->word_regex)
1027                 o->word_regex = userdiff_word_regex(one);
1028         if (!o->word_regex)
1029                 o->word_regex = userdiff_word_regex(two);
1030         if (!o->word_regex)
1031                 o->word_regex = diff_word_regex_cfg;
1032         if (o->word_regex) {
1033                 ecbdata->diff_words->word_regex = (regex_t *)
1034                         xmalloc(sizeof(regex_t));
1035                 if (regcomp(ecbdata->diff_words->word_regex,
1036                             o->word_regex,
1037                             REG_EXTENDED | REG_NEWLINE))
1038                         die ("Invalid regular expression: %s",
1039                              o->word_regex);
1040         }
1041         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1042                 if (o->word_diff == diff_words_styles[i].type) {
1043                         ecbdata->diff_words->style =
1044                                 &diff_words_styles[i];
1045                         break;
1046                 }
1047         }
1048         if (want_color(o->use_color)) {
1049                 struct diff_words_style *st = ecbdata->diff_words->style;
1050                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1051                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1052                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1053         }
1054 }
1055
1056 static void free_diff_words_data(struct emit_callback *ecbdata)
1057 {
1058         if (ecbdata->diff_words) {
1059                 diff_words_flush(ecbdata);
1060                 free (ecbdata->diff_words->opt);
1061                 free (ecbdata->diff_words->minus.text.ptr);
1062                 free (ecbdata->diff_words->minus.orig);
1063                 free (ecbdata->diff_words->plus.text.ptr);
1064                 free (ecbdata->diff_words->plus.orig);
1065                 if (ecbdata->diff_words->word_regex) {
1066                         regfree(ecbdata->diff_words->word_regex);
1067                         free(ecbdata->diff_words->word_regex);
1068                 }
1069                 free(ecbdata->diff_words);
1070                 ecbdata->diff_words = NULL;
1071         }
1072 }
1073
1074 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1075 {
1076         if (want_color(diff_use_color))
1077                 return diff_colors[ix];
1078         return "";
1079 }
1080
1081 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1082 {
1083         const char *cp;
1084         unsigned long allot;
1085         size_t l = len;
1086
1087         if (ecb->truncate)
1088                 return ecb->truncate(line, len);
1089         cp = line;
1090         allot = l;
1091         while (0 < l) {
1092                 (void) utf8_width(&cp, &l);
1093                 if (!cp)
1094                         break; /* truncated in the middle? */
1095         }
1096         return allot - l;
1097 }
1098
1099 static void find_lno(const char *line, struct emit_callback *ecbdata)
1100 {
1101         const char *p;
1102         ecbdata->lno_in_preimage = 0;
1103         ecbdata->lno_in_postimage = 0;
1104         p = strchr(line, '-');
1105         if (!p)
1106                 return; /* cannot happen */
1107         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1108         p = strchr(p, '+');
1109         if (!p)
1110                 return; /* cannot happen */
1111         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1112 }
1113
1114 static void fn_out_consume(void *priv, char *line, unsigned long len)
1115 {
1116         struct emit_callback *ecbdata = priv;
1117         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1118         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1119         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1120         struct diff_options *o = ecbdata->opt;
1121         char *line_prefix = "";
1122         struct strbuf *msgbuf;
1123
1124         if (o && o->output_prefix) {
1125                 msgbuf = o->output_prefix(o, o->output_prefix_data);
1126                 line_prefix = msgbuf->buf;
1127         }
1128
1129         if (ecbdata->header) {
1130                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1131                 strbuf_reset(ecbdata->header);
1132                 ecbdata->header = NULL;
1133         }
1134         *(ecbdata->found_changesp) = 1;
1135
1136         if (ecbdata->label_path[0]) {
1137                 const char *name_a_tab, *name_b_tab;
1138
1139                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1140                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1141
1142                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1143                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1144                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1145                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1146                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1147         }
1148
1149         if (diff_suppress_blank_empty
1150             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1151                 line[0] = '\n';
1152                 len = 1;
1153         }
1154
1155         if (line[0] == '@') {
1156                 if (ecbdata->diff_words)
1157                         diff_words_flush(ecbdata);
1158                 len = sane_truncate_line(ecbdata, line, len);
1159                 find_lno(line, ecbdata);
1160                 emit_hunk_header(ecbdata, line, len);
1161                 if (line[len-1] != '\n')
1162                         putc('\n', ecbdata->opt->file);
1163                 return;
1164         }
1165
1166         if (len < 1) {
1167                 emit_line(ecbdata->opt, reset, reset, line, len);
1168                 if (ecbdata->diff_words
1169                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1170                         fputs("~\n", ecbdata->opt->file);
1171                 return;
1172         }
1173
1174         if (ecbdata->diff_words) {
1175                 if (line[0] == '-') {
1176                         diff_words_append(line, len,
1177                                           &ecbdata->diff_words->minus);
1178                         return;
1179                 } else if (line[0] == '+') {
1180                         diff_words_append(line, len,
1181                                           &ecbdata->diff_words->plus);
1182                         return;
1183                 } else if (!prefixcmp(line, "\\ ")) {
1184                         /*
1185                          * Eat the "no newline at eof" marker as if we
1186                          * saw a "+" or "-" line with nothing on it,
1187                          * and return without diff_words_flush() to
1188                          * defer processing. If this is the end of
1189                          * preimage, more "+" lines may come after it.
1190                          */
1191                         return;
1192                 }
1193                 diff_words_flush(ecbdata);
1194                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1195                         emit_line(ecbdata->opt, plain, reset, line, len);
1196                         fputs("~\n", ecbdata->opt->file);
1197                 } else {
1198                         /*
1199                          * Skip the prefix character, if any.  With
1200                          * diff_suppress_blank_empty, there may be
1201                          * none.
1202                          */
1203                         if (line[0] != '\n') {
1204                               line++;
1205                               len--;
1206                         }
1207                         emit_line(ecbdata->opt, plain, reset, line, len);
1208                 }
1209                 return;
1210         }
1211
1212         if (line[0] != '+') {
1213                 const char *color =
1214                         diff_get_color(ecbdata->color_diff,
1215                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1216                 ecbdata->lno_in_preimage++;
1217                 if (line[0] == ' ')
1218                         ecbdata->lno_in_postimage++;
1219                 emit_line(ecbdata->opt, color, reset, line, len);
1220         } else {
1221                 ecbdata->lno_in_postimage++;
1222                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1223         }
1224 }
1225
1226 static char *pprint_rename(const char *a, const char *b)
1227 {
1228         const char *old = a;
1229         const char *new = b;
1230         struct strbuf name = STRBUF_INIT;
1231         int pfx_length, sfx_length;
1232         int len_a = strlen(a);
1233         int len_b = strlen(b);
1234         int a_midlen, b_midlen;
1235         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1236         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1237
1238         if (qlen_a || qlen_b) {
1239                 quote_c_style(a, &name, NULL, 0);
1240                 strbuf_addstr(&name, " => ");
1241                 quote_c_style(b, &name, NULL, 0);
1242                 return strbuf_detach(&name, NULL);
1243         }
1244
1245         /* Find common prefix */
1246         pfx_length = 0;
1247         while (*old && *new && *old == *new) {
1248                 if (*old == '/')
1249                         pfx_length = old - a + 1;
1250                 old++;
1251                 new++;
1252         }
1253
1254         /* Find common suffix */
1255         old = a + len_a;
1256         new = b + len_b;
1257         sfx_length = 0;
1258         while (a <= old && b <= new && *old == *new) {
1259                 if (*old == '/')
1260                         sfx_length = len_a - (old - a);
1261                 old--;
1262                 new--;
1263         }
1264
1265         /*
1266          * pfx{mid-a => mid-b}sfx
1267          * {pfx-a => pfx-b}sfx
1268          * pfx{sfx-a => sfx-b}
1269          * name-a => name-b
1270          */
1271         a_midlen = len_a - pfx_length - sfx_length;
1272         b_midlen = len_b - pfx_length - sfx_length;
1273         if (a_midlen < 0)
1274                 a_midlen = 0;
1275         if (b_midlen < 0)
1276                 b_midlen = 0;
1277
1278         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1279         if (pfx_length + sfx_length) {
1280                 strbuf_add(&name, a, pfx_length);
1281                 strbuf_addch(&name, '{');
1282         }
1283         strbuf_add(&name, a + pfx_length, a_midlen);
1284         strbuf_addstr(&name, " => ");
1285         strbuf_add(&name, b + pfx_length, b_midlen);
1286         if (pfx_length + sfx_length) {
1287                 strbuf_addch(&name, '}');
1288                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1289         }
1290         return strbuf_detach(&name, NULL);
1291 }
1292
1293 struct diffstat_t {
1294         int nr;
1295         int alloc;
1296         struct diffstat_file {
1297                 char *from_name;
1298                 char *name;
1299                 char *print_name;
1300                 unsigned is_unmerged:1;
1301                 unsigned is_binary:1;
1302                 unsigned is_renamed:1;
1303                 uintmax_t added, deleted;
1304         } **files;
1305 };
1306
1307 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1308                                           const char *name_a,
1309                                           const char *name_b)
1310 {
1311         struct diffstat_file *x;
1312         x = xcalloc(sizeof (*x), 1);
1313         if (diffstat->nr == diffstat->alloc) {
1314                 diffstat->alloc = alloc_nr(diffstat->alloc);
1315                 diffstat->files = xrealloc(diffstat->files,
1316                                 diffstat->alloc * sizeof(x));
1317         }
1318         diffstat->files[diffstat->nr++] = x;
1319         if (name_b) {
1320                 x->from_name = xstrdup(name_a);
1321                 x->name = xstrdup(name_b);
1322                 x->is_renamed = 1;
1323         }
1324         else {
1325                 x->from_name = NULL;
1326                 x->name = xstrdup(name_a);
1327         }
1328         return x;
1329 }
1330
1331 static void diffstat_consume(void *priv, char *line, unsigned long len)
1332 {
1333         struct diffstat_t *diffstat = priv;
1334         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1335
1336         if (line[0] == '+')
1337                 x->added++;
1338         else if (line[0] == '-')
1339                 x->deleted++;
1340 }
1341
1342 const char mime_boundary_leader[] = "------------";
1343
1344 static int scale_linear(int it, int width, int max_change)
1345 {
1346         if (!it)
1347                 return 0;
1348         /*
1349          * make sure that at least one '-' or '+' is printed if
1350          * there is any change to this path. The easiest way is to
1351          * scale linearly as if the alloted width is one column shorter
1352          * than it is, and then add 1 to the result.
1353          */
1354         return 1 + (it * (width - 1) / max_change);
1355 }
1356
1357 static void show_name(FILE *file,
1358                       const char *prefix, const char *name, int len)
1359 {
1360         fprintf(file, " %s%-*s |", prefix, len, name);
1361 }
1362
1363 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1364 {
1365         if (cnt <= 0)
1366                 return;
1367         fprintf(file, "%s", set);
1368         while (cnt--)
1369                 putc(ch, file);
1370         fprintf(file, "%s", reset);
1371 }
1372
1373 static void fill_print_name(struct diffstat_file *file)
1374 {
1375         char *pname;
1376
1377         if (file->print_name)
1378                 return;
1379
1380         if (!file->is_renamed) {
1381                 struct strbuf buf = STRBUF_INIT;
1382                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1383                         pname = strbuf_detach(&buf, NULL);
1384                 } else {
1385                         pname = file->name;
1386                         strbuf_release(&buf);
1387                 }
1388         } else {
1389                 pname = pprint_rename(file->from_name, file->name);
1390         }
1391         file->print_name = pname;
1392 }
1393
1394 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1395 {
1396         struct strbuf sb = STRBUF_INIT;
1397         int ret;
1398
1399         if (!files) {
1400                 assert(insertions == 0 && deletions == 0);
1401                 return fprintf(fp, "%s\n", " 0 files changed");
1402         }
1403
1404         strbuf_addf(&sb,
1405                     (files == 1) ? " %d file changed" : " %d files changed",
1406                     files);
1407
1408         /*
1409          * For binary diff, the caller may want to print "x files
1410          * changed" with insertions == 0 && deletions == 0.
1411          *
1412          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1413          * is probably less confusing (i.e skip over "2 files changed
1414          * but nothing about added/removed lines? Is this a bug in Git?").
1415          */
1416         if (insertions || deletions == 0) {
1417                 /*
1418                  * TRANSLATORS: "+" in (+) is a line addition marker;
1419                  * do not translate it.
1420                  */
1421                 strbuf_addf(&sb,
1422                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1423                             insertions);
1424         }
1425
1426         if (deletions || insertions == 0) {
1427                 /*
1428                  * TRANSLATORS: "-" in (-) is a line removal marker;
1429                  * do not translate it.
1430                  */
1431                 strbuf_addf(&sb,
1432                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1433                             deletions);
1434         }
1435         strbuf_addch(&sb, '\n');
1436         ret = fputs(sb.buf, fp);
1437         strbuf_release(&sb);
1438         return ret;
1439 }
1440
1441 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1442 {
1443         int i, len, add, del, adds = 0, dels = 0;
1444         uintmax_t max_change = 0, max_len = 0;
1445         int total_files = data->nr, count;
1446         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1447         const char *reset, *add_c, *del_c;
1448         const char *line_prefix = "";
1449         int extra_shown = 0;
1450         struct strbuf *msg = NULL;
1451
1452         if (data->nr == 0)
1453                 return;
1454
1455         if (options->output_prefix) {
1456                 msg = options->output_prefix(options, options->output_prefix_data);
1457                 line_prefix = msg->buf;
1458         }
1459
1460         count = options->stat_count ? options->stat_count : data->nr;
1461
1462         reset = diff_get_color_opt(options, DIFF_RESET);
1463         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1464         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1465
1466         /*
1467          * Find the longest filename and max number of changes
1468          */
1469         for (i = 0; (i < count) && (i < data->nr); i++) {
1470                 struct diffstat_file *file = data->files[i];
1471                 uintmax_t change = file->added + file->deleted;
1472                 if (!data->files[i]->is_renamed &&
1473                          (change == 0)) {
1474                         count++; /* not shown == room for one more */
1475                         continue;
1476                 }
1477                 fill_print_name(file);
1478                 len = strlen(file->print_name);
1479                 if (max_len < len)
1480                         max_len = len;
1481
1482                 if (file->is_unmerged) {
1483                         /* "Unmerged" is 8 characters */
1484                         bin_width = bin_width < 8 ? 8 : bin_width;
1485                         continue;
1486                 }
1487                 if (file->is_binary) {
1488                         /* "Bin XXX -> YYY bytes" */
1489                         int w = 14 + decimal_width(file->added)
1490                                 + decimal_width(file->deleted);
1491                         bin_width = bin_width < w ? w : bin_width;
1492                         /* Display change counts aligned with "Bin" */
1493                         number_width = 3;
1494                         continue;
1495                 }
1496
1497                 if (max_change < change)
1498                         max_change = change;
1499         }
1500         count = i; /* min(count, data->nr) */
1501
1502         /*
1503          * We have width = stat_width or term_columns() columns total.
1504          * We want a maximum of min(max_len, stat_name_width) for the name part.
1505          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1506          * We also need 1 for " " and 4 + decimal_width(max_change)
1507          * for " | NNNN " and one the empty column at the end, altogether
1508          * 6 + decimal_width(max_change).
1509          *
1510          * If there's not enough space, we will use the smaller of
1511          * stat_name_width (if set) and 5/8*width for the filename,
1512          * and the rest for constant elements + graph part, but no more
1513          * than stat_graph_width for the graph part.
1514          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1515          * for the standard terminal size).
1516          *
1517          * In other words: stat_width limits the maximum width, and
1518          * stat_name_width fixes the maximum width of the filename,
1519          * and is also used to divide available columns if there
1520          * aren't enough.
1521          *
1522          * Binary files are displayed with "Bin XXX -> YYY bytes"
1523          * instead of the change count and graph. This part is treated
1524          * similarly to the graph part, except that it is not
1525          * "scaled". If total width is too small to accomodate the
1526          * guaranteed minimum width of the filename part and the
1527          * separators and this message, this message will "overflow"
1528          * making the line longer than the maximum width.
1529          */
1530
1531         if (options->stat_width == -1)
1532                 width = term_columns() - options->output_prefix_length;
1533         else
1534                 width = options->stat_width ? options->stat_width : 80;
1535         number_width = decimal_width(max_change) > number_width ?
1536                 decimal_width(max_change) : number_width;
1537
1538         if (options->stat_graph_width == -1)
1539                 options->stat_graph_width = diff_stat_graph_width;
1540
1541         /*
1542          * Guarantee 3/8*16==6 for the graph part
1543          * and 5/8*16==10 for the filename part
1544          */
1545         if (width < 16 + 6 + number_width)
1546                 width = 16 + 6 + number_width;
1547
1548         /*
1549          * First assign sizes that are wanted, ignoring available width.
1550          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1551          * starting from "XXX" should fit in graph_width.
1552          */
1553         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1554         if (options->stat_graph_width &&
1555             options->stat_graph_width < graph_width)
1556                 graph_width = options->stat_graph_width;
1557
1558         name_width = (options->stat_name_width > 0 &&
1559                       options->stat_name_width < max_len) ?
1560                 options->stat_name_width : max_len;
1561
1562         /*
1563          * Adjust adjustable widths not to exceed maximum width
1564          */
1565         if (name_width + number_width + 6 + graph_width > width) {
1566                 if (graph_width > width * 3/8 - number_width - 6) {
1567                         graph_width = width * 3/8 - number_width - 6;
1568                         if (graph_width < 6)
1569                                 graph_width = 6;
1570                 }
1571
1572                 if (options->stat_graph_width &&
1573                     graph_width > options->stat_graph_width)
1574                         graph_width = options->stat_graph_width;
1575                 if (name_width > width - number_width - 6 - graph_width)
1576                         name_width = width - number_width - 6 - graph_width;
1577                 else
1578                         graph_width = width - number_width - 6 - name_width;
1579         }
1580
1581         /*
1582          * From here name_width is the width of the name area,
1583          * and graph_width is the width of the graph area.
1584          * max_change is used to scale graph properly.
1585          */
1586         for (i = 0; i < count; i++) {
1587                 const char *prefix = "";
1588                 char *name = data->files[i]->print_name;
1589                 uintmax_t added = data->files[i]->added;
1590                 uintmax_t deleted = data->files[i]->deleted;
1591                 int name_len;
1592
1593                 if (!data->files[i]->is_renamed &&
1594                          (added + deleted == 0)) {
1595                         total_files--;
1596                         continue;
1597                 }
1598                 /*
1599                  * "scale" the filename
1600                  */
1601                 len = name_width;
1602                 name_len = strlen(name);
1603                 if (name_width < name_len) {
1604                         char *slash;
1605                         prefix = "...";
1606                         len -= 3;
1607                         name += name_len - len;
1608                         slash = strchr(name, '/');
1609                         if (slash)
1610                                 name = slash;
1611                 }
1612
1613                 if (data->files[i]->is_binary) {
1614                         fprintf(options->file, "%s", line_prefix);
1615                         show_name(options->file, prefix, name, len);
1616                         fprintf(options->file, " %*s", number_width, "Bin");
1617                         if (!added && !deleted) {
1618                                 putc('\n', options->file);
1619                                 continue;
1620                         }
1621                         fprintf(options->file, " %s%"PRIuMAX"%s",
1622                                 del_c, deleted, reset);
1623                         fprintf(options->file, " -> ");
1624                         fprintf(options->file, "%s%"PRIuMAX"%s",
1625                                 add_c, added, reset);
1626                         fprintf(options->file, " bytes");
1627                         fprintf(options->file, "\n");
1628                         continue;
1629                 }
1630                 else if (data->files[i]->is_unmerged) {
1631                         fprintf(options->file, "%s", line_prefix);
1632                         show_name(options->file, prefix, name, len);
1633                         fprintf(options->file, " Unmerged\n");
1634                         continue;
1635                 }
1636
1637                 /*
1638                  * scale the add/delete
1639                  */
1640                 add = added;
1641                 del = deleted;
1642                 adds += add;
1643                 dels += del;
1644
1645                 if (graph_width <= max_change) {
1646                         int total = add + del;
1647
1648                         total = scale_linear(add + del, graph_width, max_change);
1649                         if (total < 2 && add && del)
1650                                 /* width >= 2 due to the sanity check */
1651                                 total = 2;
1652                         if (add < del) {
1653                                 add = scale_linear(add, graph_width, max_change);
1654                                 del = total - add;
1655                         } else {
1656                                 del = scale_linear(del, graph_width, max_change);
1657                                 add = total - del;
1658                         }
1659                 }
1660                 fprintf(options->file, "%s", line_prefix);
1661                 show_name(options->file, prefix, name, len);
1662                 fprintf(options->file, " %*"PRIuMAX"%s",
1663                         number_width, added + deleted,
1664                         added + deleted ? " " : "");
1665                 show_graph(options->file, '+', add, add_c, reset);
1666                 show_graph(options->file, '-', del, del_c, reset);
1667                 fprintf(options->file, "\n");
1668         }
1669         for (i = count; i < data->nr; i++) {
1670                 uintmax_t added = data->files[i]->added;
1671                 uintmax_t deleted = data->files[i]->deleted;
1672                 if (!data->files[i]->is_renamed &&
1673                          (added + deleted == 0)) {
1674                         total_files--;
1675                         continue;
1676                 }
1677                 adds += added;
1678                 dels += deleted;
1679                 if (!extra_shown)
1680                         fprintf(options->file, "%s ...\n", line_prefix);
1681                 extra_shown = 1;
1682         }
1683         fprintf(options->file, "%s", line_prefix);
1684         print_stat_summary(options->file, total_files, adds, dels);
1685 }
1686
1687 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1688 {
1689         int i, adds = 0, dels = 0, total_files = data->nr;
1690
1691         if (data->nr == 0)
1692                 return;
1693
1694         for (i = 0; i < data->nr; i++) {
1695                 int added = data->files[i]->added;
1696                 int deleted= data->files[i]->deleted;
1697
1698                 if (data->files[i]->is_unmerged)
1699                         continue;
1700                 if (!data->files[i]->is_renamed && (added + deleted == 0)) {
1701                         total_files--;
1702                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1703                         adds += added;
1704                         dels += deleted;
1705                 }
1706         }
1707         if (options->output_prefix) {
1708                 struct strbuf *msg = NULL;
1709                 msg = options->output_prefix(options,
1710                                 options->output_prefix_data);
1711                 fprintf(options->file, "%s", msg->buf);
1712         }
1713         print_stat_summary(options->file, total_files, adds, dels);
1714 }
1715
1716 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1717 {
1718         int i;
1719
1720         if (data->nr == 0)
1721                 return;
1722
1723         for (i = 0; i < data->nr; i++) {
1724                 struct diffstat_file *file = data->files[i];
1725
1726                 if (options->output_prefix) {
1727                         struct strbuf *msg = NULL;
1728                         msg = options->output_prefix(options,
1729                                         options->output_prefix_data);
1730                         fprintf(options->file, "%s", msg->buf);
1731                 }
1732
1733                 if (file->is_binary)
1734                         fprintf(options->file, "-\t-\t");
1735                 else
1736                         fprintf(options->file,
1737                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1738                                 file->added, file->deleted);
1739                 if (options->line_termination) {
1740                         fill_print_name(file);
1741                         if (!file->is_renamed)
1742                                 write_name_quoted(file->name, options->file,
1743                                                   options->line_termination);
1744                         else {
1745                                 fputs(file->print_name, options->file);
1746                                 putc(options->line_termination, options->file);
1747                         }
1748                 } else {
1749                         if (file->is_renamed) {
1750                                 putc('\0', options->file);
1751                                 write_name_quoted(file->from_name, options->file, '\0');
1752                         }
1753                         write_name_quoted(file->name, options->file, '\0');
1754                 }
1755         }
1756 }
1757
1758 struct dirstat_file {
1759         const char *name;
1760         unsigned long changed;
1761 };
1762
1763 struct dirstat_dir {
1764         struct dirstat_file *files;
1765         int alloc, nr, permille, cumulative;
1766 };
1767
1768 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1769                 unsigned long changed, const char *base, int baselen)
1770 {
1771         unsigned long this_dir = 0;
1772         unsigned int sources = 0;
1773         const char *line_prefix = "";
1774         struct strbuf *msg = NULL;
1775
1776         if (opt->output_prefix) {
1777                 msg = opt->output_prefix(opt, opt->output_prefix_data);
1778                 line_prefix = msg->buf;
1779         }
1780
1781         while (dir->nr) {
1782                 struct dirstat_file *f = dir->files;
1783                 int namelen = strlen(f->name);
1784                 unsigned long this;
1785                 char *slash;
1786
1787                 if (namelen < baselen)
1788                         break;
1789                 if (memcmp(f->name, base, baselen))
1790                         break;
1791                 slash = strchr(f->name + baselen, '/');
1792                 if (slash) {
1793                         int newbaselen = slash + 1 - f->name;
1794                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1795                         sources++;
1796                 } else {
1797                         this = f->changed;
1798                         dir->files++;
1799                         dir->nr--;
1800                         sources += 2;
1801                 }
1802                 this_dir += this;
1803         }
1804
1805         /*
1806          * We don't report dirstat's for
1807          *  - the top level
1808          *  - or cases where everything came from a single directory
1809          *    under this directory (sources == 1).
1810          */
1811         if (baselen && sources != 1) {
1812                 if (this_dir) {
1813                         int permille = this_dir * 1000 / changed;
1814                         if (permille >= dir->permille) {
1815                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1816                                         permille / 10, permille % 10, baselen, base);
1817                                 if (!dir->cumulative)
1818                                         return 0;
1819                         }
1820                 }
1821         }
1822         return this_dir;
1823 }
1824
1825 static int dirstat_compare(const void *_a, const void *_b)
1826 {
1827         const struct dirstat_file *a = _a;
1828         const struct dirstat_file *b = _b;
1829         return strcmp(a->name, b->name);
1830 }
1831
1832 static void show_dirstat(struct diff_options *options)
1833 {
1834         int i;
1835         unsigned long changed;
1836         struct dirstat_dir dir;
1837         struct diff_queue_struct *q = &diff_queued_diff;
1838
1839         dir.files = NULL;
1840         dir.alloc = 0;
1841         dir.nr = 0;
1842         dir.permille = options->dirstat_permille;
1843         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1844
1845         changed = 0;
1846         for (i = 0; i < q->nr; i++) {
1847                 struct diff_filepair *p = q->queue[i];
1848                 const char *name;
1849                 unsigned long copied, added, damage;
1850                 int content_changed;
1851
1852                 name = p->two->path ? p->two->path : p->one->path;
1853
1854                 if (p->one->sha1_valid && p->two->sha1_valid)
1855                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1856                 else
1857                         content_changed = 1;
1858
1859                 if (!content_changed) {
1860                         /*
1861                          * The SHA1 has not changed, so pre-/post-content is
1862                          * identical. We can therefore skip looking at the
1863                          * file contents altogether.
1864                          */
1865                         damage = 0;
1866                         goto found_damage;
1867                 }
1868
1869                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1870                         /*
1871                          * In --dirstat-by-file mode, we don't really need to
1872                          * look at the actual file contents at all.
1873                          * The fact that the SHA1 changed is enough for us to
1874                          * add this file to the list of results
1875                          * (with each file contributing equal damage).
1876                          */
1877                         damage = 1;
1878                         goto found_damage;
1879                 }
1880
1881                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1882                         diff_populate_filespec(p->one, 0);
1883                         diff_populate_filespec(p->two, 0);
1884                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1885                                                &copied, &added);
1886                         diff_free_filespec_data(p->one);
1887                         diff_free_filespec_data(p->two);
1888                 } else if (DIFF_FILE_VALID(p->one)) {
1889                         diff_populate_filespec(p->one, 1);
1890                         copied = added = 0;
1891                         diff_free_filespec_data(p->one);
1892                 } else if (DIFF_FILE_VALID(p->two)) {
1893                         diff_populate_filespec(p->two, 1);
1894                         copied = 0;
1895                         added = p->two->size;
1896                         diff_free_filespec_data(p->two);
1897                 } else
1898                         continue;
1899
1900                 /*
1901                  * Original minus copied is the removed material,
1902                  * added is the new material.  They are both damages
1903                  * made to the preimage.
1904                  * If the resulting damage is zero, we know that
1905                  * diffcore_count_changes() considers the two entries to
1906                  * be identical, but since content_changed is true, we
1907                  * know that there must have been _some_ kind of change,
1908                  * so we force all entries to have damage > 0.
1909                  */
1910                 damage = (p->one->size - copied) + added;
1911                 if (!damage)
1912                         damage = 1;
1913
1914 found_damage:
1915                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1916                 dir.files[dir.nr].name = name;
1917                 dir.files[dir.nr].changed = damage;
1918                 changed += damage;
1919                 dir.nr++;
1920         }
1921
1922         /* This can happen even with many files, if everything was renames */
1923         if (!changed)
1924                 return;
1925
1926         /* Show all directories with more than x% of the changes */
1927         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1928         gather_dirstat(options, &dir, changed, "", 0);
1929 }
1930
1931 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1932 {
1933         int i;
1934         unsigned long changed;
1935         struct dirstat_dir dir;
1936
1937         if (data->nr == 0)
1938                 return;
1939
1940         dir.files = NULL;
1941         dir.alloc = 0;
1942         dir.nr = 0;
1943         dir.permille = options->dirstat_permille;
1944         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1945
1946         changed = 0;
1947         for (i = 0; i < data->nr; i++) {
1948                 struct diffstat_file *file = data->files[i];
1949                 unsigned long damage = file->added + file->deleted;
1950                 if (file->is_binary)
1951                         /*
1952                          * binary files counts bytes, not lines. Must find some
1953                          * way to normalize binary bytes vs. textual lines.
1954                          * The following heuristic assumes that there are 64
1955                          * bytes per "line".
1956                          * This is stupid and ugly, but very cheap...
1957                          */
1958                         damage = (damage + 63) / 64;
1959                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1960                 dir.files[dir.nr].name = file->name;
1961                 dir.files[dir.nr].changed = damage;
1962                 changed += damage;
1963                 dir.nr++;
1964         }
1965
1966         /* This can happen even with many files, if everything was renames */
1967         if (!changed)
1968                 return;
1969
1970         /* Show all directories with more than x% of the changes */
1971         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1972         gather_dirstat(options, &dir, changed, "", 0);
1973 }
1974
1975 static void free_diffstat_info(struct diffstat_t *diffstat)
1976 {
1977         int i;
1978         for (i = 0; i < diffstat->nr; i++) {
1979                 struct diffstat_file *f = diffstat->files[i];
1980                 if (f->name != f->print_name)
1981                         free(f->print_name);
1982                 free(f->name);
1983                 free(f->from_name);
1984                 free(f);
1985         }
1986         free(diffstat->files);
1987 }
1988
1989 struct checkdiff_t {
1990         const char *filename;
1991         int lineno;
1992         int conflict_marker_size;
1993         struct diff_options *o;
1994         unsigned ws_rule;
1995         unsigned status;
1996 };
1997
1998 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1999 {
2000         char firstchar;
2001         int cnt;
2002
2003         if (len < marker_size + 1)
2004                 return 0;
2005         firstchar = line[0];
2006         switch (firstchar) {
2007         case '=': case '>': case '<': case '|':
2008                 break;
2009         default:
2010                 return 0;
2011         }
2012         for (cnt = 1; cnt < marker_size; cnt++)
2013                 if (line[cnt] != firstchar)
2014                         return 0;
2015         /* line[1] thru line[marker_size-1] are same as firstchar */
2016         if (len < marker_size + 1 || !isspace(line[marker_size]))
2017                 return 0;
2018         return 1;
2019 }
2020
2021 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2022 {
2023         struct checkdiff_t *data = priv;
2024         int marker_size = data->conflict_marker_size;
2025         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2026         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2027         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2028         char *err;
2029         char *line_prefix = "";
2030         struct strbuf *msgbuf;
2031
2032         assert(data->o);
2033         if (data->o->output_prefix) {
2034                 msgbuf = data->o->output_prefix(data->o,
2035                         data->o->output_prefix_data);
2036                 line_prefix = msgbuf->buf;
2037         }
2038
2039         if (line[0] == '+') {
2040                 unsigned bad;
2041                 data->lineno++;
2042                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2043                         data->status |= 1;
2044                         fprintf(data->o->file,
2045                                 "%s%s:%d: leftover conflict marker\n",
2046                                 line_prefix, data->filename, data->lineno);
2047                 }
2048                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2049                 if (!bad)
2050                         return;
2051                 data->status |= bad;
2052                 err = whitespace_error_string(bad);
2053                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2054                         line_prefix, data->filename, data->lineno, err);
2055                 free(err);
2056                 emit_line(data->o, set, reset, line, 1);
2057                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2058                               data->o->file, set, reset, ws);
2059         } else if (line[0] == ' ') {
2060                 data->lineno++;
2061         } else if (line[0] == '@') {
2062                 char *plus = strchr(line, '+');
2063                 if (plus)
2064                         data->lineno = strtol(plus, NULL, 10) - 1;
2065                 else
2066                         die("invalid diff");
2067         }
2068 }
2069
2070 static unsigned char *deflate_it(char *data,
2071                                  unsigned long size,
2072                                  unsigned long *result_size)
2073 {
2074         int bound;
2075         unsigned char *deflated;
2076         git_zstream stream;
2077
2078         memset(&stream, 0, sizeof(stream));
2079         git_deflate_init(&stream, zlib_compression_level);
2080         bound = git_deflate_bound(&stream, size);
2081         deflated = xmalloc(bound);
2082         stream.next_out = deflated;
2083         stream.avail_out = bound;
2084
2085         stream.next_in = (unsigned char *)data;
2086         stream.avail_in = size;
2087         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2088                 ; /* nothing */
2089         git_deflate_end(&stream);
2090         *result_size = stream.total_out;
2091         return deflated;
2092 }
2093
2094 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2095 {
2096         void *cp;
2097         void *delta;
2098         void *deflated;
2099         void *data;
2100         unsigned long orig_size;
2101         unsigned long delta_size;
2102         unsigned long deflate_size;
2103         unsigned long data_size;
2104
2105         /* We could do deflated delta, or we could do just deflated two,
2106          * whichever is smaller.
2107          */
2108         delta = NULL;
2109         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2110         if (one->size && two->size) {
2111                 delta = diff_delta(one->ptr, one->size,
2112                                    two->ptr, two->size,
2113                                    &delta_size, deflate_size);
2114                 if (delta) {
2115                         void *to_free = delta;
2116                         orig_size = delta_size;
2117                         delta = deflate_it(delta, delta_size, &delta_size);
2118                         free(to_free);
2119                 }
2120         }
2121
2122         if (delta && delta_size < deflate_size) {
2123                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2124                 free(deflated);
2125                 data = delta;
2126                 data_size = delta_size;
2127         }
2128         else {
2129                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2130                 free(delta);
2131                 data = deflated;
2132                 data_size = deflate_size;
2133         }
2134
2135         /* emit data encoded in base85 */
2136         cp = data;
2137         while (data_size) {
2138                 int bytes = (52 < data_size) ? 52 : data_size;
2139                 char line[70];
2140                 data_size -= bytes;
2141                 if (bytes <= 26)
2142                         line[0] = bytes + 'A' - 1;
2143                 else
2144                         line[0] = bytes - 26 + 'a' - 1;
2145                 encode_85(line + 1, cp, bytes);
2146                 cp = (char *) cp + bytes;
2147                 fprintf(file, "%s", prefix);
2148                 fputs(line, file);
2149                 fputc('\n', file);
2150         }
2151         fprintf(file, "%s\n", prefix);
2152         free(data);
2153 }
2154
2155 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2156 {
2157         fprintf(file, "%sGIT binary patch\n", prefix);
2158         emit_binary_diff_body(file, one, two, prefix);
2159         emit_binary_diff_body(file, two, one, prefix);
2160 }
2161
2162 int diff_filespec_is_binary(struct diff_filespec *one)
2163 {
2164         if (one->is_binary == -1) {
2165                 diff_filespec_load_driver(one);
2166                 if (one->driver->binary != -1)
2167                         one->is_binary = one->driver->binary;
2168                 else {
2169                         if (!one->data && DIFF_FILE_VALID(one))
2170                                 diff_populate_filespec(one, 0);
2171                         if (one->data)
2172                                 one->is_binary = buffer_is_binary(one->data,
2173                                                 one->size);
2174                         if (one->is_binary == -1)
2175                                 one->is_binary = 0;
2176                 }
2177         }
2178         return one->is_binary;
2179 }
2180
2181 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2182 {
2183         diff_filespec_load_driver(one);
2184         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2185 }
2186
2187 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2188 {
2189         if (!options->a_prefix)
2190                 options->a_prefix = a;
2191         if (!options->b_prefix)
2192                 options->b_prefix = b;
2193 }
2194
2195 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2196 {
2197         if (!DIFF_FILE_VALID(one))
2198                 return NULL;
2199
2200         diff_filespec_load_driver(one);
2201         return userdiff_get_textconv(one->driver);
2202 }
2203
2204 static void builtin_diff(const char *name_a,
2205                          const char *name_b,
2206                          struct diff_filespec *one,
2207                          struct diff_filespec *two,
2208                          const char *xfrm_msg,
2209                          int must_show_header,
2210                          struct diff_options *o,
2211                          int complete_rewrite)
2212 {
2213         mmfile_t mf1, mf2;
2214         const char *lbl[2];
2215         char *a_one, *b_two;
2216         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
2217         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2218         const char *a_prefix, *b_prefix;
2219         struct userdiff_driver *textconv_one = NULL;
2220         struct userdiff_driver *textconv_two = NULL;
2221         struct strbuf header = STRBUF_INIT;
2222         struct strbuf *msgbuf;
2223         char *line_prefix = "";
2224
2225         if (o->output_prefix) {
2226                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2227                 line_prefix = msgbuf->buf;
2228         }
2229
2230         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2231                         (!one->mode || S_ISGITLINK(one->mode)) &&
2232                         (!two->mode || S_ISGITLINK(two->mode))) {
2233                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2234                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2235                 show_submodule_summary(o->file, one ? one->path : two->path,
2236                                 one->sha1, two->sha1, two->dirty_submodule,
2237                                 del, add, reset);
2238                 return;
2239         }
2240
2241         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2242                 textconv_one = get_textconv(one);
2243                 textconv_two = get_textconv(two);
2244         }
2245
2246         diff_set_mnemonic_prefix(o, "a/", "b/");
2247         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2248                 a_prefix = o->b_prefix;
2249                 b_prefix = o->a_prefix;
2250         } else {
2251                 a_prefix = o->a_prefix;
2252                 b_prefix = o->b_prefix;
2253         }
2254
2255         /* Never use a non-valid filename anywhere if at all possible */
2256         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2257         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2258
2259         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2260         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2261         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2262         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2263         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
2264         if (lbl[0][0] == '/') {
2265                 /* /dev/null */
2266                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
2267                 if (xfrm_msg)
2268                         strbuf_addstr(&header, xfrm_msg);
2269                 must_show_header = 1;
2270         }
2271         else if (lbl[1][0] == '/') {
2272                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
2273                 if (xfrm_msg)
2274                         strbuf_addstr(&header, xfrm_msg);
2275                 must_show_header = 1;
2276         }
2277         else {
2278                 if (one->mode != two->mode) {
2279                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
2280                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
2281                         must_show_header = 1;
2282                 }
2283                 if (xfrm_msg)
2284                         strbuf_addstr(&header, xfrm_msg);
2285
2286                 /*
2287                  * we do not run diff between different kind
2288                  * of objects.
2289                  */
2290                 if ((one->mode ^ two->mode) & S_IFMT)
2291                         goto free_ab_and_return;
2292                 if (complete_rewrite &&
2293                     (textconv_one || !diff_filespec_is_binary(one)) &&
2294                     (textconv_two || !diff_filespec_is_binary(two))) {
2295                         fprintf(o->file, "%s", header.buf);
2296                         strbuf_reset(&header);
2297                         emit_rewrite_diff(name_a, name_b, one, two,
2298                                                 textconv_one, textconv_two, o);
2299                         o->found_changes = 1;
2300                         goto free_ab_and_return;
2301                 }
2302         }
2303
2304         if (o->irreversible_delete && lbl[1][0] == '/') {
2305                 fprintf(o->file, "%s", header.buf);
2306                 strbuf_reset(&header);
2307                 goto free_ab_and_return;
2308         } else if (!DIFF_OPT_TST(o, TEXT) &&
2309             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2310               (!textconv_two && diff_filespec_is_binary(two)) )) {
2311                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2312                         die("unable to read files to diff");
2313                 /* Quite common confusing case */
2314                 if (mf1.size == mf2.size &&
2315                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2316                         if (must_show_header)
2317                                 fprintf(o->file, "%s", header.buf);
2318                         goto free_ab_and_return;
2319                 }
2320                 fprintf(o->file, "%s", header.buf);
2321                 strbuf_reset(&header);
2322                 if (DIFF_OPT_TST(o, BINARY))
2323                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2324                 else
2325                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2326                                 line_prefix, lbl[0], lbl[1]);
2327                 o->found_changes = 1;
2328         } else {
2329                 /* Crazy xdl interfaces.. */
2330                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2331                 xpparam_t xpp;
2332                 xdemitconf_t xecfg;
2333                 struct emit_callback ecbdata;
2334                 const struct userdiff_funcname *pe;
2335
2336                 if (must_show_header) {
2337                         fprintf(o->file, "%s", header.buf);
2338                         strbuf_reset(&header);
2339                 }
2340
2341                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2342                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2343
2344                 pe = diff_funcname_pattern(one);
2345                 if (!pe)
2346                         pe = diff_funcname_pattern(two);
2347
2348                 memset(&xpp, 0, sizeof(xpp));
2349                 memset(&xecfg, 0, sizeof(xecfg));
2350                 memset(&ecbdata, 0, sizeof(ecbdata));
2351                 ecbdata.label_path = lbl;
2352                 ecbdata.color_diff = want_color(o->use_color);
2353                 ecbdata.found_changesp = &o->found_changes;
2354                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
2355                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2356                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2357                 ecbdata.opt = o;
2358                 ecbdata.header = header.len ? &header : NULL;
2359                 xpp.flags = o->xdl_opts;
2360                 xecfg.ctxlen = o->context;
2361                 xecfg.interhunkctxlen = o->interhunkcontext;
2362                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2363                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2364                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2365                 if (pe)
2366                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2367                 if (!diffopts)
2368                         ;
2369                 else if (!prefixcmp(diffopts, "--unified="))
2370                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2371                 else if (!prefixcmp(diffopts, "-u"))
2372                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2373                 if (o->word_diff)
2374                         init_diff_words_data(&ecbdata, o, one, two);
2375                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2376                               &xpp, &xecfg);
2377                 if (o->word_diff)
2378                         free_diff_words_data(&ecbdata);
2379                 if (textconv_one)
2380                         free(mf1.ptr);
2381                 if (textconv_two)
2382                         free(mf2.ptr);
2383                 xdiff_clear_find_func(&xecfg);
2384         }
2385
2386  free_ab_and_return:
2387         strbuf_release(&header);
2388         diff_free_filespec_data(one);
2389         diff_free_filespec_data(two);
2390         free(a_one);
2391         free(b_two);
2392         return;
2393 }
2394
2395 static void builtin_diffstat(const char *name_a, const char *name_b,
2396                              struct diff_filespec *one,
2397                              struct diff_filespec *two,
2398                              struct diffstat_t *diffstat,
2399                              struct diff_options *o,
2400                              int complete_rewrite)
2401 {
2402         mmfile_t mf1, mf2;
2403         struct diffstat_file *data;
2404         int same_contents;
2405
2406         data = diffstat_add(diffstat, name_a, name_b);
2407
2408         if (!one || !two) {
2409                 data->is_unmerged = 1;
2410                 return;
2411         }
2412
2413         same_contents = !hashcmp(one->sha1, two->sha1);
2414
2415         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2416                 data->is_binary = 1;
2417                 if (same_contents) {
2418                         data->added = 0;
2419                         data->deleted = 0;
2420                 } else {
2421                         data->added = diff_filespec_size(two);
2422                         data->deleted = diff_filespec_size(one);
2423                 }
2424         }
2425
2426         else if (complete_rewrite) {
2427                 diff_populate_filespec(one, 0);
2428                 diff_populate_filespec(two, 0);
2429                 data->deleted = count_lines(one->data, one->size);
2430                 data->added = count_lines(two->data, two->size);
2431         }
2432
2433         else if (!same_contents) {
2434                 /* Crazy xdl interfaces.. */
2435                 xpparam_t xpp;
2436                 xdemitconf_t xecfg;
2437
2438                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2439                         die("unable to read files to diff");
2440
2441                 memset(&xpp, 0, sizeof(xpp));
2442                 memset(&xecfg, 0, sizeof(xecfg));
2443                 xpp.flags = o->xdl_opts;
2444                 xecfg.ctxlen = o->context;
2445                 xecfg.interhunkctxlen = o->interhunkcontext;
2446                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2447                               &xpp, &xecfg);
2448         }
2449
2450         diff_free_filespec_data(one);
2451         diff_free_filespec_data(two);
2452 }
2453
2454 static void builtin_checkdiff(const char *name_a, const char *name_b,
2455                               const char *attr_path,
2456                               struct diff_filespec *one,
2457                               struct diff_filespec *two,
2458                               struct diff_options *o)
2459 {
2460         mmfile_t mf1, mf2;
2461         struct checkdiff_t data;
2462
2463         if (!two)
2464                 return;
2465
2466         memset(&data, 0, sizeof(data));
2467         data.filename = name_b ? name_b : name_a;
2468         data.lineno = 0;
2469         data.o = o;
2470         data.ws_rule = whitespace_rule(attr_path);
2471         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2472
2473         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2474                 die("unable to read files to diff");
2475
2476         /*
2477          * All the other codepaths check both sides, but not checking
2478          * the "old" side here is deliberate.  We are checking the newly
2479          * introduced changes, and as long as the "new" side is text, we
2480          * can and should check what it introduces.
2481          */
2482         if (diff_filespec_is_binary(two))
2483                 goto free_and_return;
2484         else {
2485                 /* Crazy xdl interfaces.. */
2486                 xpparam_t xpp;
2487                 xdemitconf_t xecfg;
2488
2489                 memset(&xpp, 0, sizeof(xpp));
2490                 memset(&xecfg, 0, sizeof(xecfg));
2491                 xecfg.ctxlen = 1; /* at least one context line */
2492                 xpp.flags = 0;
2493                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2494                               &xpp, &xecfg);
2495
2496                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2497                         struct emit_callback ecbdata;
2498                         int blank_at_eof;
2499
2500                         ecbdata.ws_rule = data.ws_rule;
2501                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2502                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2503
2504                         if (blank_at_eof) {
2505                                 static char *err;
2506                                 if (!err)
2507                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2508                                 fprintf(o->file, "%s:%d: %s.\n",
2509                                         data.filename, blank_at_eof, err);
2510                                 data.status = 1; /* report errors */
2511                         }
2512                 }
2513         }
2514  free_and_return:
2515         diff_free_filespec_data(one);
2516         diff_free_filespec_data(two);
2517         if (data.status)
2518                 DIFF_OPT_SET(o, CHECK_FAILED);
2519 }
2520
2521 struct diff_filespec *alloc_filespec(const char *path)
2522 {
2523         int namelen = strlen(path);
2524         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2525
2526         memset(spec, 0, sizeof(*spec));
2527         spec->path = (char *)(spec + 1);
2528         memcpy(spec->path, path, namelen+1);
2529         spec->count = 1;
2530         spec->is_binary = -1;
2531         return spec;
2532 }
2533
2534 void free_filespec(struct diff_filespec *spec)
2535 {
2536         if (!--spec->count) {
2537                 diff_free_filespec_data(spec);
2538                 free(spec);
2539         }
2540 }
2541
2542 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2543                    int sha1_valid, unsigned short mode)
2544 {
2545         if (mode) {
2546                 spec->mode = canon_mode(mode);
2547                 hashcpy(spec->sha1, sha1);
2548                 spec->sha1_valid = sha1_valid;
2549         }
2550 }
2551
2552 /*
2553  * Given a name and sha1 pair, if the index tells us the file in
2554  * the work tree has that object contents, return true, so that
2555  * prepare_temp_file() does not have to inflate and extract.
2556  */
2557 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2558 {
2559         struct cache_entry *ce;
2560         struct stat st;
2561         int pos, len;
2562
2563         /*
2564          * We do not read the cache ourselves here, because the
2565          * benchmark with my previous version that always reads cache
2566          * shows that it makes things worse for diff-tree comparing
2567          * two linux-2.6 kernel trees in an already checked out work
2568          * tree.  This is because most diff-tree comparisons deal with
2569          * only a small number of files, while reading the cache is
2570          * expensive for a large project, and its cost outweighs the
2571          * savings we get by not inflating the object to a temporary
2572          * file.  Practically, this code only helps when we are used
2573          * by diff-cache --cached, which does read the cache before
2574          * calling us.
2575          */
2576         if (!active_cache)
2577                 return 0;
2578
2579         /* We want to avoid the working directory if our caller
2580          * doesn't need the data in a normal file, this system
2581          * is rather slow with its stat/open/mmap/close syscalls,
2582          * and the object is contained in a pack file.  The pack
2583          * is probably already open and will be faster to obtain
2584          * the data through than the working directory.  Loose
2585          * objects however would tend to be slower as they need
2586          * to be individually opened and inflated.
2587          */
2588         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2589                 return 0;
2590
2591         len = strlen(name);
2592         pos = cache_name_pos(name, len);
2593         if (pos < 0)
2594                 return 0;
2595         ce = active_cache[pos];
2596
2597         /*
2598          * This is not the sha1 we are looking for, or
2599          * unreusable because it is not a regular file.
2600          */
2601         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2602                 return 0;
2603
2604         /*
2605          * If ce is marked as "assume unchanged", there is no
2606          * guarantee that work tree matches what we are looking for.
2607          */
2608         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2609                 return 0;
2610
2611         /*
2612          * If ce matches the file in the work tree, we can reuse it.
2613          */
2614         if (ce_uptodate(ce) ||
2615             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2616                 return 1;
2617
2618         return 0;
2619 }
2620
2621 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2622 {
2623         int len;
2624         char *data = xmalloc(100), *dirty = "";
2625
2626         /* Are we looking at the work tree? */
2627         if (s->dirty_submodule)
2628                 dirty = "-dirty";
2629
2630         len = snprintf(data, 100,
2631                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2632         s->data = data;
2633         s->size = len;
2634         s->should_free = 1;
2635         if (size_only) {
2636                 s->data = NULL;
2637                 free(data);
2638         }
2639         return 0;
2640 }
2641
2642 /*
2643  * While doing rename detection and pickaxe operation, we may need to
2644  * grab the data for the blob (or file) for our own in-core comparison.
2645  * diff_filespec has data and size fields for this purpose.
2646  */
2647 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2648 {
2649         int err = 0;
2650         if (!DIFF_FILE_VALID(s))
2651                 die("internal error: asking to populate invalid file.");
2652         if (S_ISDIR(s->mode))
2653                 return -1;
2654
2655         if (s->data)
2656                 return 0;
2657
2658         if (size_only && 0 < s->size)
2659                 return 0;
2660
2661         if (S_ISGITLINK(s->mode))
2662                 return diff_populate_gitlink(s, size_only);
2663
2664         if (!s->sha1_valid ||
2665             reuse_worktree_file(s->path, s->sha1, 0)) {
2666                 struct strbuf buf = STRBUF_INIT;
2667                 struct stat st;
2668                 int fd;
2669
2670                 if (lstat(s->path, &st) < 0) {
2671                         if (errno == ENOENT) {
2672                         err_empty:
2673                                 err = -1;
2674                         empty:
2675                                 s->data = (char *)"";
2676                                 s->size = 0;
2677                                 return err;
2678                         }
2679                 }
2680                 s->size = xsize_t(st.st_size);
2681                 if (!s->size)
2682                         goto empty;
2683                 if (S_ISLNK(st.st_mode)) {
2684                         struct strbuf sb = STRBUF_INIT;
2685
2686                         if (strbuf_readlink(&sb, s->path, s->size))
2687                                 goto err_empty;
2688                         s->size = sb.len;
2689                         s->data = strbuf_detach(&sb, NULL);
2690                         s->should_free = 1;
2691                         return 0;
2692                 }
2693                 if (size_only)
2694                         return 0;
2695                 fd = open(s->path, O_RDONLY);
2696                 if (fd < 0)
2697                         goto err_empty;
2698                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2699                 close(fd);
2700                 s->should_munmap = 1;
2701
2702                 /*
2703                  * Convert from working tree format to canonical git format
2704                  */
2705                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2706                         size_t size = 0;
2707                         munmap(s->data, s->size);
2708                         s->should_munmap = 0;
2709                         s->data = strbuf_detach(&buf, &size);
2710                         s->size = size;
2711                         s->should_free = 1;
2712                 }
2713         }
2714         else {
2715                 enum object_type type;
2716                 if (size_only) {
2717                         type = sha1_object_info(s->sha1, &s->size);
2718                         if (type < 0)
2719                                 die("unable to read %s", sha1_to_hex(s->sha1));
2720                 } else {
2721                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2722                         if (!s->data)
2723                                 die("unable to read %s", sha1_to_hex(s->sha1));
2724                         s->should_free = 1;
2725                 }
2726         }
2727         return 0;
2728 }
2729
2730 void diff_free_filespec_blob(struct diff_filespec *s)
2731 {
2732         if (s->should_free)
2733                 free(s->data);
2734         else if (s->should_munmap)
2735                 munmap(s->data, s->size);
2736
2737         if (s->should_free || s->should_munmap) {
2738                 s->should_free = s->should_munmap = 0;
2739                 s->data = NULL;
2740         }
2741 }
2742
2743 void diff_free_filespec_data(struct diff_filespec *s)
2744 {
2745         diff_free_filespec_blob(s);
2746         free(s->cnt_data);
2747         s->cnt_data = NULL;
2748 }
2749
2750 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2751                            void *blob,
2752                            unsigned long size,
2753                            const unsigned char *sha1,
2754                            int mode)
2755 {
2756         int fd;
2757         struct strbuf buf = STRBUF_INIT;
2758         struct strbuf template = STRBUF_INIT;
2759         char *path_dup = xstrdup(path);
2760         const char *base = basename(path_dup);
2761
2762         /* Generate "XXXXXX_basename.ext" */
2763         strbuf_addstr(&template, "XXXXXX_");
2764         strbuf_addstr(&template, base);
2765
2766         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2767                         strlen(base) + 1);
2768         if (fd < 0)
2769                 die_errno("unable to create temp-file");
2770         if (convert_to_working_tree(path,
2771                         (const char *)blob, (size_t)size, &buf)) {
2772                 blob = buf.buf;
2773                 size = buf.len;
2774         }
2775         if (write_in_full(fd, blob, size) != size)
2776                 die_errno("unable to write temp-file");
2777         close(fd);
2778         temp->name = temp->tmp_path;
2779         strcpy(temp->hex, sha1_to_hex(sha1));
2780         temp->hex[40] = 0;
2781         sprintf(temp->mode, "%06o", mode);
2782         strbuf_release(&buf);
2783         strbuf_release(&template);
2784         free(path_dup);
2785 }
2786
2787 static struct diff_tempfile *prepare_temp_file(const char *name,
2788                 struct diff_filespec *one)
2789 {
2790         struct diff_tempfile *temp = claim_diff_tempfile();
2791
2792         if (!DIFF_FILE_VALID(one)) {
2793         not_a_valid_file:
2794                 /* A '-' entry produces this for file-2, and
2795                  * a '+' entry produces this for file-1.
2796                  */
2797                 temp->name = "/dev/null";
2798                 strcpy(temp->hex, ".");
2799                 strcpy(temp->mode, ".");
2800                 return temp;
2801         }
2802
2803         if (!remove_tempfile_installed) {
2804                 atexit(remove_tempfile);
2805                 sigchain_push_common(remove_tempfile_on_signal);
2806                 remove_tempfile_installed = 1;
2807         }
2808
2809         if (!one->sha1_valid ||
2810             reuse_worktree_file(name, one->sha1, 1)) {
2811                 struct stat st;
2812                 if (lstat(name, &st) < 0) {
2813                         if (errno == ENOENT)
2814                                 goto not_a_valid_file;
2815                         die_errno("stat(%s)", name);
2816                 }
2817                 if (S_ISLNK(st.st_mode)) {
2818                         struct strbuf sb = STRBUF_INIT;
2819                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2820                                 die_errno("readlink(%s)", name);
2821                         prep_temp_blob(name, temp, sb.buf, sb.len,
2822                                        (one->sha1_valid ?
2823                                         one->sha1 : null_sha1),
2824                                        (one->sha1_valid ?
2825                                         one->mode : S_IFLNK));
2826                         strbuf_release(&sb);
2827                 }
2828                 else {
2829                         /* we can borrow from the file in the work tree */
2830                         temp->name = name;
2831                         if (!one->sha1_valid)
2832                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2833                         else
2834                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2835                         /* Even though we may sometimes borrow the
2836                          * contents from the work tree, we always want
2837                          * one->mode.  mode is trustworthy even when
2838                          * !(one->sha1_valid), as long as
2839                          * DIFF_FILE_VALID(one).
2840                          */
2841                         sprintf(temp->mode, "%06o", one->mode);
2842                 }
2843                 return temp;
2844         }
2845         else {
2846                 if (diff_populate_filespec(one, 0))
2847                         die("cannot read data blob for %s", one->path);
2848                 prep_temp_blob(name, temp, one->data, one->size,
2849                                one->sha1, one->mode);
2850         }
2851         return temp;
2852 }
2853
2854 /* An external diff command takes:
2855  *
2856  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2857  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2858  *
2859  */
2860 static void run_external_diff(const char *pgm,
2861                               const char *name,
2862                               const char *other,
2863                               struct diff_filespec *one,
2864                               struct diff_filespec *two,
2865                               const char *xfrm_msg,
2866                               int complete_rewrite)
2867 {
2868         const char *spawn_arg[10];
2869         int retval;
2870         const char **arg = &spawn_arg[0];
2871
2872         if (one && two) {
2873                 struct diff_tempfile *temp_one, *temp_two;
2874                 const char *othername = (other ? other : name);
2875                 temp_one = prepare_temp_file(name, one);
2876                 temp_two = prepare_temp_file(othername, two);
2877                 *arg++ = pgm;
2878                 *arg++ = name;
2879                 *arg++ = temp_one->name;
2880                 *arg++ = temp_one->hex;
2881                 *arg++ = temp_one->mode;
2882                 *arg++ = temp_two->name;
2883                 *arg++ = temp_two->hex;
2884                 *arg++ = temp_two->mode;
2885                 if (other) {
2886                         *arg++ = other;
2887                         *arg++ = xfrm_msg;
2888                 }
2889         } else {
2890                 *arg++ = pgm;
2891                 *arg++ = name;
2892         }
2893         *arg = NULL;
2894         fflush(NULL);
2895         retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2896         remove_tempfile();
2897         if (retval) {
2898                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2899                 exit(1);
2900         }
2901 }
2902
2903 static int similarity_index(struct diff_filepair *p)
2904 {
2905         return p->score * 100 / MAX_SCORE;
2906 }
2907
2908 static void fill_metainfo(struct strbuf *msg,
2909                           const char *name,
2910                           const char *other,
2911                           struct diff_filespec *one,
2912                           struct diff_filespec *two,
2913                           struct diff_options *o,
2914                           struct diff_filepair *p,
2915                           int *must_show_header,
2916                           int use_color)
2917 {
2918         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2919         const char *reset = diff_get_color(use_color, DIFF_RESET);
2920         struct strbuf *msgbuf;
2921         char *line_prefix = "";
2922
2923         *must_show_header = 1;
2924         if (o->output_prefix) {
2925                 msgbuf = o->output_prefix(o, o->output_prefix_data);
2926                 line_prefix = msgbuf->buf;
2927         }
2928         strbuf_init(msg, PATH_MAX * 2 + 300);
2929         switch (p->status) {
2930         case DIFF_STATUS_COPIED:
2931                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2932                             line_prefix, set, similarity_index(p));
2933                 strbuf_addf(msg, "%s\n%s%scopy from ",
2934                             reset,  line_prefix, set);
2935                 quote_c_style(name, msg, NULL, 0);
2936                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2937                 quote_c_style(other, msg, NULL, 0);
2938                 strbuf_addf(msg, "%s\n", reset);
2939                 break;
2940         case DIFF_STATUS_RENAMED:
2941                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2942                             line_prefix, set, similarity_index(p));
2943                 strbuf_addf(msg, "%s\n%s%srename from ",
2944                             reset, line_prefix, set);
2945                 quote_c_style(name, msg, NULL, 0);
2946                 strbuf_addf(msg, "%s\n%s%srename to ",
2947                             reset, line_prefix, set);
2948                 quote_c_style(other, msg, NULL, 0);
2949                 strbuf_addf(msg, "%s\n", reset);
2950                 break;
2951         case DIFF_STATUS_MODIFIED:
2952                 if (p->score) {
2953                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2954                                     line_prefix,
2955                                     set, similarity_index(p), reset);
2956                         break;
2957                 }
2958                 /* fallthru */
2959         default:
2960                 *must_show_header = 0;
2961         }
2962         if (one && two && hashcmp(one->sha1, two->sha1)) {
2963                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2964
2965                 if (DIFF_OPT_TST(o, BINARY)) {
2966                         mmfile_t mf;
2967                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2968                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2969                                 abbrev = 40;
2970                 }
2971                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2972                             find_unique_abbrev(one->sha1, abbrev));
2973                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2974                 if (one->mode == two->mode)
2975                         strbuf_addf(msg, " %06o", one->mode);
2976                 strbuf_addf(msg, "%s\n", reset);
2977         }
2978 }
2979
2980 static void run_diff_cmd(const char *pgm,
2981                          const char *name,
2982                          const char *other,
2983                          const char *attr_path,
2984                          struct diff_filespec *one,
2985                          struct diff_filespec *two,
2986                          struct strbuf *msg,
2987                          struct diff_options *o,
2988                          struct diff_filepair *p)
2989 {
2990         const char *xfrm_msg = NULL;
2991         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2992         int must_show_header = 0;
2993
2994
2995         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
2996                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2997                 if (drv && drv->external)
2998                         pgm = drv->external;
2999         }
3000
3001         if (msg) {
3002                 /*
3003                  * don't use colors when the header is intended for an
3004                  * external diff driver
3005                  */
3006                 fill_metainfo(msg, name, other, one, two, o, p,
3007                               &must_show_header,
3008                               want_color(o->use_color) && !pgm);
3009                 xfrm_msg = msg->len ? msg->buf : NULL;
3010         }
3011
3012         if (pgm) {
3013                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3014                                   complete_rewrite);
3015                 return;
3016         }
3017         if (one && two)
3018                 builtin_diff(name, other ? other : name,
3019                              one, two, xfrm_msg, must_show_header,
3020                              o, complete_rewrite);
3021         else
3022                 fprintf(o->file, "* Unmerged path %s\n", name);
3023 }
3024
3025 static void diff_fill_sha1_info(struct diff_filespec *one)
3026 {
3027         if (DIFF_FILE_VALID(one)) {
3028                 if (!one->sha1_valid) {
3029                         struct stat st;
3030                         if (one->is_stdin) {
3031                                 hashcpy(one->sha1, null_sha1);
3032                                 return;
3033                         }
3034                         if (lstat(one->path, &st) < 0)
3035                                 die_errno("stat '%s'", one->path);
3036                         if (index_path(one->sha1, one->path, &st, 0))
3037                                 die("cannot hash %s", one->path);
3038                 }
3039         }
3040         else
3041                 hashclr(one->sha1);
3042 }
3043
3044 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3045 {
3046         /* Strip the prefix but do not molest /dev/null and absolute paths */
3047         if (*namep && **namep != '/') {
3048                 *namep += prefix_length;
3049                 if (**namep == '/')
3050                         ++*namep;
3051         }
3052         if (*otherp && **otherp != '/') {
3053                 *otherp += prefix_length;
3054                 if (**otherp == '/')
3055                         ++*otherp;
3056         }
3057 }
3058
3059 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3060 {
3061         const char *pgm = external_diff();
3062         struct strbuf msg;
3063         struct diff_filespec *one = p->one;
3064         struct diff_filespec *two = p->two;
3065         const char *name;
3066         const char *other;
3067         const char *attr_path;
3068
3069         name  = p->one->path;
3070         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3071         attr_path = name;
3072         if (o->prefix_length)
3073                 strip_prefix(o->prefix_length, &name, &other);
3074
3075         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3076                 pgm = NULL;
3077
3078         if (DIFF_PAIR_UNMERGED(p)) {
3079                 run_diff_cmd(pgm, name, NULL, attr_path,
3080                              NULL, NULL, NULL, o, p);
3081                 return;
3082         }
3083
3084         diff_fill_sha1_info(one);
3085         diff_fill_sha1_info(two);
3086
3087         if (!pgm &&
3088             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3089             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3090                 /*
3091                  * a filepair that changes between file and symlink
3092                  * needs to be split into deletion and creation.
3093                  */
3094                 struct diff_filespec *null = alloc_filespec(two->path);
3095                 run_diff_cmd(NULL, name, other, attr_path,
3096                              one, null, &msg, o, p);
3097                 free(null);
3098                 strbuf_release(&msg);
3099
3100                 null = alloc_filespec(one->path);
3101                 run_diff_cmd(NULL, name, other, attr_path,
3102                              null, two, &msg, o, p);
3103                 free(null);
3104         }
3105         else
3106                 run_diff_cmd(pgm, name, other, attr_path,
3107                              one, two, &msg, o, p);
3108
3109         strbuf_release(&msg);
3110 }
3111
3112 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3113                          struct diffstat_t *diffstat)
3114 {
3115         const char *name;
3116         const char *other;
3117         int complete_rewrite = 0;
3118
3119         if (DIFF_PAIR_UNMERGED(p)) {
3120                 /* unmerged */
3121                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
3122                 return;
3123         }
3124
3125         name = p->one->path;
3126         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3127
3128         if (o->prefix_length)
3129                 strip_prefix(o->prefix_length, &name, &other);
3130
3131         diff_fill_sha1_info(p->one);
3132         diff_fill_sha1_info(p->two);
3133
3134         if (p->status == DIFF_STATUS_MODIFIED && p->score)
3135                 complete_rewrite = 1;
3136         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
3137 }
3138
3139 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3140 {
3141         const char *name;
3142         const char *other;
3143         const char *attr_path;
3144
3145         if (DIFF_PAIR_UNMERGED(p)) {
3146                 /* unmerged */
3147                 return;
3148         }
3149
3150         name = p->one->path;
3151         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3152         attr_path = other ? other : name;
3153
3154         if (o->prefix_length)
3155                 strip_prefix(o->prefix_length, &name, &other);
3156
3157         diff_fill_sha1_info(p->one);
3158         diff_fill_sha1_info(p->two);
3159
3160         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3161 }
3162
3163 void diff_setup(struct diff_options *options)
3164 {
3165         memcpy(options, &default_diff_options, sizeof(*options));
3166
3167         options->file = stdout;
3168
3169         options->line_termination = '\n';
3170         options->break_opt = -1;
3171         options->rename_limit = -1;
3172         options->dirstat_permille = diff_dirstat_permille_default;
3173         options->context = 3;
3174         DIFF_OPT_SET(options, RENAME_EMPTY);
3175
3176         options->change = diff_change;
3177         options->add_remove = diff_addremove;
3178         options->use_color = diff_use_color_default;
3179         options->detect_rename = diff_detect_rename_default;
3180
3181         if (diff_no_prefix) {
3182                 options->a_prefix = options->b_prefix = "";
3183         } else if (!diff_mnemonic_prefix) {
3184                 options->a_prefix = "a/";
3185                 options->b_prefix = "b/";
3186         }
3187 }
3188
3189 void diff_setup_done(struct diff_options *options)
3190 {
3191         int count = 0;
3192
3193         if (options->output_format & DIFF_FORMAT_NAME)
3194                 count++;
3195         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3196                 count++;
3197         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3198                 count++;
3199         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3200                 count++;
3201         if (count > 1)
3202                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3203
3204         /*
3205          * Most of the time we can say "there are changes"
3206          * only by checking if there are changed paths, but
3207          * --ignore-whitespace* options force us to look
3208          * inside contents.
3209          */
3210
3211         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3212             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3213             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3214                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3215         else
3216                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3217
3218         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3219                 options->detect_rename = DIFF_DETECT_COPY;
3220
3221         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3222                 options->prefix = NULL;
3223         if (options->prefix)
3224                 options->prefix_length = strlen(options->prefix);
3225         else
3226                 options->prefix_length = 0;
3227
3228         if (options->output_format & (DIFF_FORMAT_NAME |
3229                                       DIFF_FORMAT_NAME_STATUS |
3230                                       DIFF_FORMAT_CHECKDIFF |
3231                                       DIFF_FORMAT_NO_OUTPUT))
3232                 options->output_format &= ~(DIFF_FORMAT_RAW |
3233                                             DIFF_FORMAT_NUMSTAT |
3234                                             DIFF_FORMAT_DIFFSTAT |
3235                                             DIFF_FORMAT_SHORTSTAT |
3236                                             DIFF_FORMAT_DIRSTAT |
3237                                             DIFF_FORMAT_SUMMARY |
3238                                             DIFF_FORMAT_PATCH);
3239
3240         /*
3241          * These cases always need recursive; we do not drop caller-supplied
3242          * recursive bits for other formats here.
3243          */
3244         if (options->output_format & (DIFF_FORMAT_PATCH |
3245                                       DIFF_FORMAT_NUMSTAT |
3246                                       DIFF_FORMAT_DIFFSTAT |
3247                                       DIFF_FORMAT_SHORTSTAT |
3248                                       DIFF_FORMAT_DIRSTAT |
3249                                       DIFF_FORMAT_SUMMARY |
3250                                       DIFF_FORMAT_CHECKDIFF))
3251                 DIFF_OPT_SET(options, RECURSIVE);
3252         /*
3253          * Also pickaxe would not work very well if you do not say recursive
3254          */
3255         if (options->pickaxe)
3256                 DIFF_OPT_SET(options, RECURSIVE);
3257         /*
3258          * When patches are generated, submodules diffed against the work tree
3259          * must be checked for dirtiness too so it can be shown in the output
3260          */
3261         if (options->output_format & DIFF_FORMAT_PATCH)
3262                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3263
3264         if (options->detect_rename && options->rename_limit < 0)
3265                 options->rename_limit = diff_rename_limit_default;
3266         if (options->setup & DIFF_SETUP_USE_CACHE) {
3267                 if (!active_cache)
3268                         /* read-cache does not die even when it fails
3269                          * so it is safe for us to do this here.  Also
3270                          * it does not smudge active_cache or active_nr
3271                          * when it fails, so we do not have to worry about
3272                          * cleaning it up ourselves either.
3273                          */
3274                         read_cache();
3275         }
3276         if (options->abbrev <= 0 || 40 < options->abbrev)
3277                 options->abbrev = 40; /* full */
3278
3279         /*
3280          * It does not make sense to show the first hit we happened
3281          * to have found.  It does not make sense not to return with
3282          * exit code in such a case either.
3283          */
3284         if (DIFF_OPT_TST(options, QUICK)) {
3285                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3286                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3287         }
3288 }
3289
3290 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3291 {
3292         char c, *eq;
3293         int len;
3294
3295         if (*arg != '-')
3296                 return 0;
3297         c = *++arg;
3298         if (!c)
3299                 return 0;
3300         if (c == arg_short) {
3301                 c = *++arg;
3302                 if (!c)
3303                         return 1;
3304                 if (val && isdigit(c)) {
3305                         char *end;
3306                         int n = strtoul(arg, &end, 10);
3307                         if (*end)
3308                                 return 0;
3309                         *val = n;
3310                         return 1;
3311                 }
3312                 return 0;
3313         }
3314         if (c != '-')
3315                 return 0;
3316         arg++;
3317         eq = strchr(arg, '=');
3318         if (eq)
3319                 len = eq - arg;
3320         else
3321                 len = strlen(arg);
3322         if (!len || strncmp(arg, arg_long, len))
3323                 return 0;
3324         if (eq) {
3325                 int n;
3326                 char *end;
3327                 if (!isdigit(*++eq))
3328                         return 0;
3329                 n = strtoul(eq, &end, 10);
3330                 if (*end)
3331                         return 0;
3332                 *val = n;
3333         }
3334         return 1;
3335 }
3336
3337 static int diff_scoreopt_parse(const char *opt);
3338
3339 static inline int short_opt(char opt, const char **argv,
3340                             const char **optarg)
3341 {
3342         const char *arg = argv[0];
3343         if (arg[0] != '-' || arg[1] != opt)
3344                 return 0;
3345         if (arg[2] != '\0') {
3346                 *optarg = arg + 2;
3347                 return 1;
3348         }
3349         if (!argv[1])
3350                 die("Option '%c' requires a value", opt);
3351         *optarg = argv[1];
3352         return 2;
3353 }
3354
3355 int parse_long_opt(const char *opt, const char **argv,
3356                    const char **optarg)
3357 {
3358         const char *arg = argv[0];
3359         if (arg[0] != '-' || arg[1] != '-')
3360                 return 0;
3361         arg += strlen("--");
3362         if (prefixcmp(arg, opt))
3363                 return 0;
3364         arg += strlen(opt);
3365         if (*arg == '=') { /* sticked form: --option=value */
3366                 *optarg = arg + 1;
3367                 return 1;
3368         }
3369         if (*arg != '\0')
3370                 return 0;
3371         /* separate form: --option value */
3372         if (!argv[1])
3373                 die("Option '--%s' requires a value", opt);
3374         *optarg = argv[1];
3375         return 2;
3376 }
3377
3378 static int stat_opt(struct diff_options *options, const char **av)
3379 {
3380         const char *arg = av[0];
3381         char *end;
3382         int width = options->stat_width;
3383         int name_width = options->stat_name_width;
3384         int graph_width = options->stat_graph_width;
3385         int count = options->stat_count;
3386         int argcount = 1;
3387
3388         arg += strlen("--stat");
3389         end = (char *)arg;
3390
3391         switch (*arg) {
3392         case '-':
3393                 if (!prefixcmp(arg, "-width")) {
3394                         arg += strlen("-width");
3395                         if (*arg == '=')
3396                                 width = strtoul(arg + 1, &end, 10);
3397                         else if (!*arg && !av[1])
3398                                 die("Option '--stat-width' requires a value");
3399                         else if (!*arg) {
3400                                 width = strtoul(av[1], &end, 10);
3401                                 argcount = 2;
3402                         }
3403                 } else if (!prefixcmp(arg, "-name-width")) {
3404                         arg += strlen("-name-width");
3405                         if (*arg == '=')
3406                                 name_width = strtoul(arg + 1, &end, 10);
3407                         else if (!*arg && !av[1])
3408                                 die("Option '--stat-name-width' requires a value");
3409                         else if (!*arg) {
3410                                 name_width = strtoul(av[1], &end, 10);
3411                                 argcount = 2;
3412                         }
3413                 } else if (!prefixcmp(arg, "-graph-width")) {
3414                         arg += strlen("-graph-width");
3415                         if (*arg == '=')
3416                                 graph_width = strtoul(arg + 1, &end, 10);
3417                         else if (!*arg && !av[1])
3418                                 die("Option '--stat-graph-width' requires a value");
3419                         else if (!*arg) {
3420                                 graph_width = strtoul(av[1], &end, 10);
3421                                 argcount = 2;
3422                         }
3423                 } else if (!prefixcmp(arg, "-count")) {
3424                         arg += strlen("-count");
3425                         if (*arg == '=')
3426                                 count = strtoul(arg + 1, &end, 10);
3427                         else if (!*arg && !av[1])
3428                                 die("Option '--stat-count' requires a value");
3429                         else if (!*arg) {
3430                                 count = strtoul(av[1], &end, 10);
3431                                 argcount = 2;
3432                         }
3433                 }
3434                 break;
3435         case '=':
3436                 width = strtoul(arg+1, &end, 10);
3437                 if (*end == ',')
3438                         name_width = strtoul(end+1, &end, 10);
3439                 if (*end == ',')
3440                         count = strtoul(end+1, &end, 10);
3441         }
3442
3443         /* Important! This checks all the error cases! */
3444         if (*end)
3445                 return 0;
3446         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3447         options->stat_name_width = name_width;
3448         options->stat_graph_width = graph_width;
3449         options->stat_width = width;
3450         options->stat_count = count;
3451         return argcount;
3452 }
3453
3454 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3455 {
3456         struct strbuf errmsg = STRBUF_INIT;
3457         if (parse_dirstat_params(options, params, &errmsg))
3458                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3459                     errmsg.buf);
3460         strbuf_release(&errmsg);
3461         /*
3462          * The caller knows a dirstat-related option is given from the command
3463          * line; allow it to say "return this_function();"
3464          */
3465         options->output_format |= DIFF_FORMAT_DIRSTAT;
3466         return 1;
3467 }
3468
3469 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3470 {
3471         const char *arg = av[0];
3472         const char *optarg;
3473         int argcount;
3474
3475         /* Output format options */
3476         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3477                 options->output_format |= DIFF_FORMAT_PATCH;
3478         else if (opt_arg(arg, 'U', "unified", &options->context))
3479                 options->output_format |= DIFF_FORMAT_PATCH;
3480         else if (!strcmp(arg, "--raw"))
3481                 options->output_format |= DIFF_FORMAT_RAW;
3482         else if (!strcmp(arg, "--patch-with-raw"))
3483                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3484         else if (!strcmp(arg, "--numstat"))
3485                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3486         else if (!strcmp(arg, "--shortstat"))
3487                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3488         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3489                 return parse_dirstat_opt(options, "");
3490         else if (!prefixcmp(arg, "-X"))
3491                 return parse_dirstat_opt(options, arg + 2);
3492         else if (!prefixcmp(arg, "--dirstat="))
3493                 return parse_dirstat_opt(options, arg + 10);
3494         else if (!strcmp(arg, "--cumulative"))
3495                 return parse_dirstat_opt(options, "cumulative");
3496         else if (!strcmp(arg, "--dirstat-by-file"))
3497                 return parse_dirstat_opt(options, "files");
3498         else if (!prefixcmp(arg, "--dirstat-by-file=")) {
3499                 parse_dirstat_opt(options, "files");
3500                 return parse_dirstat_opt(options, arg + 18);
3501         }
3502         else if (!strcmp(arg, "--check"))
3503                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3504         else if (!strcmp(arg, "--summary"))
3505                 options->output_format |= DIFF_FORMAT_SUMMARY;
3506         else if (!strcmp(arg, "--patch-with-stat"))
3507                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3508         else if (!strcmp(arg, "--name-only"))
3509                 options->output_format |= DIFF_FORMAT_NAME;
3510         else if (!strcmp(arg, "--name-status"))
3511                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3512         else if (!strcmp(arg, "-s"))
3513                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3514         else if (!prefixcmp(arg, "--stat"))
3515                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3516                 return stat_opt(options, av);
3517
3518         /* renames options */
3519         else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3520                  !strcmp(arg, "--break-rewrites")) {
3521                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3522                         return error("invalid argument to -B: %s", arg+2);
3523         }
3524         else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3525                  !strcmp(arg, "--find-renames")) {
3526                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3527                         return error("invalid argument to -M: %s", arg+2);
3528                 options->detect_rename = DIFF_DETECT_RENAME;
3529         }
3530         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3531                 options->irreversible_delete = 1;
3532         }
3533         else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3534                  !strcmp(arg, "--find-copies")) {
3535                 if (options->detect_rename == DIFF_DETECT_COPY)
3536                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3537                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3538                         return error("invalid argument to -C: %s", arg+2);
3539                 options->detect_rename = DIFF_DETECT_COPY;
3540         }
3541         else if (!strcmp(arg, "--no-renames"))
3542                 options->detect_rename = 0;
3543         else if (!strcmp(arg, "--rename-empty"))
3544                 DIFF_OPT_SET(options, RENAME_EMPTY);
3545         else if (!strcmp(arg, "--no-rename-empty"))
3546                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3547         else if (!strcmp(arg, "--relative"))
3548                 DIFF_OPT_SET(options, RELATIVE_NAME);
3549         else if (!prefixcmp(arg, "--relative=")) {
3550                 DIFF_OPT_SET(options, RELATIVE_NAME);
3551                 options->prefix = arg + 11;
3552         }
3553
3554         /* xdiff options */
3555         else if (!strcmp(arg, "--minimal"))
3556                 DIFF_XDL_SET(options, NEED_MINIMAL);
3557         else if (!strcmp(arg, "--no-minimal"))
3558                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3559         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3560                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3561         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3562                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3563         else if (!strcmp(arg, "--ignore-space-at-eol"))
3564                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3565         else if (!strcmp(arg, "--patience"))
3566                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3567         else if (!strcmp(arg, "--histogram"))
3568                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3569
3570         /* flags options */
3571         else if (!strcmp(arg, "--binary")) {
3572                 options->output_format |= DIFF_FORMAT_PATCH;
3573                 DIFF_OPT_SET(options, BINARY);
3574         }
3575         else if (!strcmp(arg, "--full-index"))
3576                 DIFF_OPT_SET(options, FULL_INDEX);
3577         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3578                 DIFF_OPT_SET(options, TEXT);
3579         else if (!strcmp(arg, "-R"))
3580                 DIFF_OPT_SET(options, REVERSE_DIFF);
3581         else if (!strcmp(arg, "--find-copies-harder"))
3582                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3583         else if (!strcmp(arg, "--follow"))
3584                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3585         else if (!strcmp(arg, "--color"))
3586                 options->use_color = 1;
3587         else if (!prefixcmp(arg, "--color=")) {
3588                 int value = git_config_colorbool(NULL, arg+8);
3589                 if (value < 0)
3590                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3591                 options->use_color = value;
3592         }
3593         else if (!strcmp(arg, "--no-color"))
3594                 options->use_color = 0;
3595         else if (!strcmp(arg, "--color-words")) {
3596                 options->use_color = 1;
3597                 options->word_diff = DIFF_WORDS_COLOR;
3598         }
3599         else if (!prefixcmp(arg, "--color-words=")) {
3600                 options->use_color = 1;
3601                 options->word_diff = DIFF_WORDS_COLOR;
3602                 options->word_regex = arg + 14;
3603         }
3604         else if (!strcmp(arg, "--word-diff")) {
3605                 if (options->word_diff == DIFF_WORDS_NONE)
3606                         options->word_diff = DIFF_WORDS_PLAIN;
3607         }
3608         else if (!prefixcmp(arg, "--word-diff=")) {
3609                 const char *type = arg + 12;
3610                 if (!strcmp(type, "plain"))
3611                         options->word_diff = DIFF_WORDS_PLAIN;
3612                 else if (!strcmp(type, "color")) {
3613                         options->use_color = 1;
3614                         options->word_diff = DIFF_WORDS_COLOR;
3615                 }
3616                 else if (!strcmp(type, "porcelain"))
3617                         options->word_diff = DIFF_WORDS_PORCELAIN;
3618                 else if (!strcmp(type, "none"))
3619                         options->word_diff = DIFF_WORDS_NONE;
3620                 else
3621                         die("bad --word-diff argument: %s", type);
3622         }
3623         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3624                 if (options->word_diff == DIFF_WORDS_NONE)
3625                         options->word_diff = DIFF_WORDS_PLAIN;
3626                 options->word_regex = optarg;
3627                 return argcount;
3628         }
3629         else if (!strcmp(arg, "--exit-code"))
3630                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3631         else if (!strcmp(arg, "--quiet"))
3632                 DIFF_OPT_SET(options, QUICK);
3633         else if (!strcmp(arg, "--ext-diff"))
3634                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3635         else if (!strcmp(arg, "--no-ext-diff"))
3636                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3637         else if (!strcmp(arg, "--textconv"))
3638                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3639         else if (!strcmp(arg, "--no-textconv"))
3640                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3641         else if (!strcmp(arg, "--ignore-submodules")) {
3642                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3643                 handle_ignore_submodules_arg(options, "all");
3644         } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3645                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3646                 handle_ignore_submodules_arg(options, arg + 20);
3647         } else if (!strcmp(arg, "--submodule"))
3648                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3649         else if (!prefixcmp(arg, "--submodule=")) {
3650                 if (!strcmp(arg + 12, "log"))
3651                         DIFF_OPT_SET(options, SUBMODULE_LOG);
3652         }
3653
3654         /* misc options */
3655         else if (!strcmp(arg, "-z"))
3656                 options->line_termination = 0;
3657         else if ((argcount = short_opt('l', av, &optarg))) {
3658                 options->rename_limit = strtoul(optarg, NULL, 10);
3659                 return argcount;
3660         }
3661         else if ((argcount = short_opt('S', av, &optarg))) {
3662                 options->pickaxe = optarg;
3663                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3664                 return argcount;
3665         } else if ((argcount = short_opt('G', av, &optarg))) {
3666                 options->pickaxe = optarg;
3667                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3668                 return argcount;
3669         }
3670         else if (!strcmp(arg, "--pickaxe-all"))
3671                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3672         else if (!strcmp(arg, "--pickaxe-regex"))
3673                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3674         else if ((argcount = short_opt('O', av, &optarg))) {
3675                 options->orderfile = optarg;
3676                 return argcount;
3677         }
3678         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3679                 options->filter = optarg;
3680                 return argcount;
3681         }
3682         else if (!strcmp(arg, "--abbrev"))
3683                 options->abbrev = DEFAULT_ABBREV;
3684         else if (!prefixcmp(arg, "--abbrev=")) {
3685                 options->abbrev = strtoul(arg + 9, NULL, 10);
3686                 if (options->abbrev < MINIMUM_ABBREV)
3687                         options->abbrev = MINIMUM_ABBREV;
3688                 else if (40 < options->abbrev)
3689                         options->abbrev = 40;
3690         }
3691         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3692                 options->a_prefix = optarg;
3693                 return argcount;
3694         }
3695         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3696                 options->b_prefix = optarg;
3697                 return argcount;
3698         }
3699         else if (!strcmp(arg, "--no-prefix"))
3700                 options->a_prefix = options->b_prefix = "";
3701         else if (opt_arg(arg, '\0', "inter-hunk-context",
3702                          &options->interhunkcontext))
3703                 ;
3704         else if (!strcmp(arg, "-W"))
3705                 DIFF_OPT_SET(options, FUNCCONTEXT);
3706         else if (!strcmp(arg, "--function-context"))
3707                 DIFF_OPT_SET(options, FUNCCONTEXT);
3708         else if (!strcmp(arg, "--no-function-context"))
3709                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3710         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3711                 options->file = fopen(optarg, "w");
3712                 if (!options->file)
3713                         die_errno("Could not open '%s'", optarg);
3714                 options->close_file = 1;
3715                 return argcount;
3716         } else
3717                 return 0;
3718         return 1;
3719 }
3720
3721 int parse_rename_score(const char **cp_p)
3722 {
3723         unsigned long num, scale;
3724         int ch, dot;
3725         const char *cp = *cp_p;
3726
3727         num = 0;
3728         scale = 1;
3729         dot = 0;
3730         for (;;) {
3731                 ch = *cp;
3732                 if ( !dot && ch == '.' ) {
3733                         scale = 1;
3734                         dot = 1;
3735                 } else if ( ch == '%' ) {
3736                         scale = dot ? scale*100 : 100;
3737                         cp++;   /* % is always at the end */
3738                         break;
3739                 } else if ( ch >= '0' && ch <= '9' ) {
3740                         if ( scale < 100000 ) {
3741                                 scale *= 10;
3742                                 num = (num*10) + (ch-'0');
3743                         }
3744                 } else {
3745                         break;
3746                 }
3747                 cp++;
3748         }
3749         *cp_p = cp;
3750
3751         /* user says num divided by scale and we say internally that
3752          * is MAX_SCORE * num / scale.
3753          */
3754         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3755 }
3756
3757 static int diff_scoreopt_parse(const char *opt)
3758 {
3759         int opt1, opt2, cmd;
3760
3761         if (*opt++ != '-')
3762                 return -1;
3763         cmd = *opt++;
3764         if (cmd == '-') {
3765                 /* convert the long-form arguments into short-form versions */
3766                 if (!prefixcmp(opt, "break-rewrites")) {
3767                         opt += strlen("break-rewrites");
3768                         if (*opt == 0 || *opt++ == '=')
3769                                 cmd = 'B';
3770                 } else if (!prefixcmp(opt, "find-copies")) {
3771                         opt += strlen("find-copies");
3772                         if (*opt == 0 || *opt++ == '=')
3773                                 cmd = 'C';
3774                 } else if (!prefixcmp(opt, "find-renames")) {
3775                         opt += strlen("find-renames");
3776                         if (*opt == 0 || *opt++ == '=')
3777                                 cmd = 'M';
3778                 }
3779         }
3780         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3781                 return -1; /* that is not a -M, -C nor -B option */
3782
3783         opt1 = parse_rename_score(&opt);
3784         if (cmd != 'B')
3785                 opt2 = 0;
3786         else {
3787                 if (*opt == 0)
3788                         opt2 = 0;
3789                 else if (*opt != '/')
3790                         return -1; /* we expect -B80/99 or -B80 */
3791                 else {
3792                         opt++;
3793                         opt2 = parse_rename_score(&opt);
3794                 }
3795         }
3796         if (*opt != 0)
3797                 return -1;
3798         return opt1 | (opt2 << 16);
3799 }
3800
3801 struct diff_queue_struct diff_queued_diff;
3802
3803 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3804 {
3805         if (queue->alloc <= queue->nr) {
3806                 queue->alloc = alloc_nr(queue->alloc);
3807                 queue->queue = xrealloc(queue->queue,
3808                                         sizeof(dp) * queue->alloc);
3809         }
3810         queue->queue[queue->nr++] = dp;
3811 }
3812
3813 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3814                                  struct diff_filespec *one,
3815                                  struct diff_filespec *two)
3816 {
3817         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3818         dp->one = one;
3819         dp->two = two;
3820         if (queue)
3821                 diff_q(queue, dp);
3822         return dp;
3823 }
3824
3825 void diff_free_filepair(struct diff_filepair *p)
3826 {
3827         free_filespec(p->one);
3828         free_filespec(p->two);
3829         free(p);
3830 }
3831
3832 /* This is different from find_unique_abbrev() in that
3833  * it stuffs the result with dots for alignment.
3834  */
3835 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3836 {
3837         int abblen;
3838         const char *abbrev;
3839         if (len == 40)
3840                 return sha1_to_hex(sha1);
3841
3842         abbrev = find_unique_abbrev(sha1, len);
3843         abblen = strlen(abbrev);
3844         if (abblen < 37) {
3845                 static char hex[41];
3846                 if (len < abblen && abblen <= len + 2)
3847                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3848                 else
3849                         sprintf(hex, "%s...", abbrev);
3850                 return hex;
3851         }
3852         return sha1_to_hex(sha1);
3853 }
3854
3855 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3856 {
3857         int line_termination = opt->line_termination;
3858         int inter_name_termination = line_termination ? '\t' : '\0';
3859         if (opt->output_prefix) {
3860                 struct strbuf *msg = NULL;
3861                 msg = opt->output_prefix(opt, opt->output_prefix_data);
3862                 fprintf(opt->file, "%s", msg->buf);
3863         }
3864
3865         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3866                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3867                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3868                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3869         }
3870         if (p->score) {
3871                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3872                         inter_name_termination);
3873         } else {
3874                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3875         }
3876
3877         if (p->status == DIFF_STATUS_COPIED ||
3878             p->status == DIFF_STATUS_RENAMED) {
3879                 const char *name_a, *name_b;
3880                 name_a = p->one->path;
3881                 name_b = p->two->path;
3882                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3883                 write_name_quoted(name_a, opt->file, inter_name_termination);
3884                 write_name_quoted(name_b, opt->file, line_termination);
3885         } else {
3886                 const char *name_a, *name_b;
3887                 name_a = p->one->mode ? p->one->path : p->two->path;
3888                 name_b = NULL;
3889                 strip_prefix(opt->prefix_length, &name_a, &name_b);
3890                 write_name_quoted(name_a, opt->file, line_termination);
3891         }
3892 }
3893
3894 int diff_unmodified_pair(struct diff_filepair *p)
3895 {
3896         /* This function is written stricter than necessary to support
3897          * the currently implemented transformers, but the idea is to
3898          * let transformers to produce diff_filepairs any way they want,
3899          * and filter and clean them up here before producing the output.
3900          */
3901         struct diff_filespec *one = p->one, *two = p->two;
3902
3903         if (DIFF_PAIR_UNMERGED(p))
3904                 return 0; /* unmerged is interesting */
3905
3906         /* deletion, addition, mode or type change
3907          * and rename are all interesting.
3908          */
3909         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3910             DIFF_PAIR_MODE_CHANGED(p) ||
3911             strcmp(one->path, two->path))
3912                 return 0;
3913
3914         /* both are valid and point at the same path.  that is, we are
3915          * dealing with a change.
3916          */
3917         if (one->sha1_valid && two->sha1_valid &&
3918             !hashcmp(one->sha1, two->sha1) &&
3919             !one->dirty_submodule && !two->dirty_submodule)
3920                 return 1; /* no change */
3921         if (!one->sha1_valid && !two->sha1_valid)
3922                 return 1; /* both look at the same file on the filesystem. */
3923         return 0;
3924 }
3925
3926 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3927 {
3928         if (diff_unmodified_pair(p))
3929                 return;
3930
3931         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3932             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3933                 return; /* no tree diffs in patch format */
3934
3935         run_diff(p, o);
3936 }
3937
3938 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3939                             struct diffstat_t *diffstat)
3940 {
3941         if (diff_unmodified_pair(p))
3942                 return;
3943
3944         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3945             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3946                 return; /* no useful stat for tree diffs */
3947
3948         run_diffstat(p, o, diffstat);
3949 }
3950
3951 static void diff_flush_checkdiff(struct diff_filepair *p,
3952                 struct diff_options *o)
3953 {
3954         if (diff_unmodified_pair(p))
3955                 return;
3956
3957         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3958             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3959                 return; /* nothing to check in tree diffs */
3960
3961         run_checkdiff(p, o);
3962 }
3963
3964 int diff_queue_is_empty(void)
3965 {
3966         struct diff_queue_struct *q = &diff_queued_diff;
3967         int i;
3968         for (i = 0; i < q->nr; i++)
3969                 if (!diff_unmodified_pair(q->queue[i]))
3970                         return 0;
3971         return 1;
3972 }
3973
3974 #if DIFF_DEBUG
3975 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3976 {
3977         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3978                 x, one ? one : "",
3979                 s->path,
3980                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3981                 s->mode,
3982                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3983         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3984                 x, one ? one : "",
3985                 s->size, s->xfrm_flags);
3986 }
3987
3988 void diff_debug_filepair(const struct diff_filepair *p, int i)
3989 {
3990         diff_debug_filespec(p->one, i, "one");
3991         diff_debug_filespec(p->two, i, "two");
3992         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3993                 p->score, p->status ? p->status : '?',
3994                 p->one->rename_used, p->broken_pair);
3995 }
3996
3997 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3998 {
3999         int i;
4000         if (msg)
4001                 fprintf(stderr, "%s\n", msg);
4002         fprintf(stderr, "q->nr = %d\n", q->nr);
4003         for (i = 0; i < q->nr; i++) {
4004                 struct diff_filepair *p = q->queue[i];
4005                 diff_debug_filepair(p, i);
4006         }
4007 }
4008 #endif
4009
4010 static void diff_resolve_rename_copy(void)
4011 {
4012         int i;
4013         struct diff_filepair *p;
4014         struct diff_queue_struct *q = &diff_queued_diff;
4015
4016         diff_debug_queue("resolve-rename-copy", q);
4017
4018         for (i = 0; i < q->nr; i++) {
4019                 p = q->queue[i];
4020                 p->status = 0; /* undecided */
4021                 if (DIFF_PAIR_UNMERGED(p))
4022                         p->status = DIFF_STATUS_UNMERGED;
4023                 else if (!DIFF_FILE_VALID(p->one))
4024                         p->status = DIFF_STATUS_ADDED;
4025                 else if (!DIFF_FILE_VALID(p->two))
4026                         p->status = DIFF_STATUS_DELETED;
4027                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4028                         p->status = DIFF_STATUS_TYPE_CHANGED;
4029
4030                 /* from this point on, we are dealing with a pair
4031                  * whose both sides are valid and of the same type, i.e.
4032                  * either in-place edit or rename/copy edit.
4033                  */
4034                 else if (DIFF_PAIR_RENAME(p)) {
4035                         /*
4036                          * A rename might have re-connected a broken
4037                          * pair up, causing the pathnames to be the
4038                          * same again. If so, that's not a rename at
4039                          * all, just a modification..
4040                          *
4041                          * Otherwise, see if this source was used for
4042                          * multiple renames, in which case we decrement
4043                          * the count, and call it a copy.
4044                          */
4045                         if (!strcmp(p->one->path, p->two->path))
4046                                 p->status = DIFF_STATUS_MODIFIED;
4047                         else if (--p->one->rename_used > 0)
4048                                 p->status = DIFF_STATUS_COPIED;
4049                         else
4050                                 p->status = DIFF_STATUS_RENAMED;
4051                 }
4052                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4053                          p->one->mode != p->two->mode ||
4054                          p->one->dirty_submodule ||
4055                          p->two->dirty_submodule ||
4056                          is_null_sha1(p->one->sha1))
4057                         p->status = DIFF_STATUS_MODIFIED;
4058                 else {
4059                         /* This is a "no-change" entry and should not
4060                          * happen anymore, but prepare for broken callers.
4061                          */
4062                         error("feeding unmodified %s to diffcore",
4063                               p->one->path);
4064                         p->status = DIFF_STATUS_UNKNOWN;
4065                 }
4066         }
4067         diff_debug_queue("resolve-rename-copy done", q);
4068 }
4069
4070 static int check_pair_status(struct diff_filepair *p)
4071 {
4072         switch (p->status) {
4073         case DIFF_STATUS_UNKNOWN:
4074                 return 0;
4075         case 0:
4076                 die("internal error in diff-resolve-rename-copy");
4077         default:
4078                 return 1;
4079         }
4080 }
4081
4082 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4083 {
4084         int fmt = opt->output_format;
4085
4086         if (fmt & DIFF_FORMAT_CHECKDIFF)
4087                 diff_flush_checkdiff(p, opt);
4088         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4089                 diff_flush_raw(p, opt);
4090         else if (fmt & DIFF_FORMAT_NAME) {
4091                 const char *name_a, *name_b;
4092                 name_a = p->two->path;
4093                 name_b = NULL;
4094                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4095                 write_name_quoted(name_a, opt->file, opt->line_termination);
4096         }
4097 }
4098
4099 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4100 {
4101         if (fs->mode)
4102                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4103         else
4104                 fprintf(file, " %s ", newdelete);
4105         write_name_quoted(fs->path, file, '\n');
4106 }
4107
4108
4109 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4110                 const char *line_prefix)
4111 {
4112         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4113                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4114                         p->two->mode, show_name ? ' ' : '\n');
4115                 if (show_name) {
4116                         write_name_quoted(p->two->path, file, '\n');
4117                 }
4118         }
4119 }
4120
4121 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4122                         const char *line_prefix)
4123 {
4124         char *names = pprint_rename(p->one->path, p->two->path);
4125
4126         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4127         free(names);
4128         show_mode_change(file, p, 0, line_prefix);
4129 }
4130
4131 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4132 {
4133         FILE *file = opt->file;
4134         char *line_prefix = "";
4135
4136         if (opt->output_prefix) {
4137                 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
4138                 line_prefix = buf->buf;
4139         }
4140
4141         switch(p->status) {
4142         case DIFF_STATUS_DELETED:
4143                 fputs(line_prefix, file);
4144                 show_file_mode_name(file, "delete", p->one);
4145                 break;
4146         case DIFF_STATUS_ADDED:
4147                 fputs(line_prefix, file);
4148                 show_file_mode_name(file, "create", p->two);
4149                 break;
4150         case DIFF_STATUS_COPIED:
4151                 fputs(line_prefix, file);
4152                 show_rename_copy(file, "copy", p, line_prefix);
4153                 break;
4154         case DIFF_STATUS_RENAMED:
4155                 fputs(line_prefix, file);
4156                 show_rename_copy(file, "rename", p, line_prefix);
4157                 break;
4158         default:
4159                 if (p->score) {
4160                         fprintf(file, "%s rewrite ", line_prefix);
4161                         write_name_quoted(p->two->path, file, ' ');
4162                         fprintf(file, "(%d%%)\n", similarity_index(p));
4163                 }
4164                 show_mode_change(file, p, !p->score, line_prefix);
4165                 break;
4166         }
4167 }
4168
4169 struct patch_id_t {
4170         git_SHA_CTX *ctx;
4171         int patchlen;
4172 };
4173
4174 static int remove_space(char *line, int len)
4175 {
4176         int i;
4177         char *dst = line;
4178         unsigned char c;
4179
4180         for (i = 0; i < len; i++)
4181                 if (!isspace((c = line[i])))
4182                         *dst++ = c;
4183
4184         return dst - line;
4185 }
4186
4187 static void patch_id_consume(void *priv, char *line, unsigned long len)
4188 {
4189         struct patch_id_t *data = priv;
4190         int new_len;
4191
4192         /* Ignore line numbers when computing the SHA1 of the patch */
4193         if (!prefixcmp(line, "@@ -"))
4194                 return;
4195
4196         new_len = remove_space(line, len);
4197
4198         git_SHA1_Update(data->ctx, line, new_len);
4199         data->patchlen += new_len;
4200 }
4201
4202 /* returns 0 upon success, and writes result into sha1 */
4203 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4204 {
4205         struct diff_queue_struct *q = &diff_queued_diff;
4206         int i;
4207         git_SHA_CTX ctx;
4208         struct patch_id_t data;
4209         char buffer[PATH_MAX * 4 + 20];
4210
4211         git_SHA1_Init(&ctx);
4212         memset(&data, 0, sizeof(struct patch_id_t));
4213         data.ctx = &ctx;
4214
4215         for (i = 0; i < q->nr; i++) {
4216                 xpparam_t xpp;
4217                 xdemitconf_t xecfg;
4218                 mmfile_t mf1, mf2;
4219                 struct diff_filepair *p = q->queue[i];
4220                 int len1, len2;
4221
4222                 memset(&xpp, 0, sizeof(xpp));
4223                 memset(&xecfg, 0, sizeof(xecfg));
4224                 if (p->status == 0)
4225                         return error("internal diff status error");
4226                 if (p->status == DIFF_STATUS_UNKNOWN)
4227                         continue;
4228                 if (diff_unmodified_pair(p))
4229                         continue;
4230                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4231                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4232                         continue;
4233                 if (DIFF_PAIR_UNMERGED(p))
4234                         continue;
4235
4236                 diff_fill_sha1_info(p->one);
4237                 diff_fill_sha1_info(p->two);
4238                 if (fill_mmfile(&mf1, p->one) < 0 ||
4239                                 fill_mmfile(&mf2, p->two) < 0)
4240                         return error("unable to read files to diff");
4241
4242                 len1 = remove_space(p->one->path, strlen(p->one->path));
4243                 len2 = remove_space(p->two->path, strlen(p->two->path));
4244                 if (p->one->mode == 0)
4245                         len1 = snprintf(buffer, sizeof(buffer),
4246                                         "diff--gita/%.*sb/%.*s"
4247                                         "newfilemode%06o"
4248                                         "---/dev/null"
4249                                         "+++b/%.*s",
4250                                         len1, p->one->path,
4251                                         len2, p->two->path,
4252                                         p->two->mode,
4253                                         len2, p->two->path);
4254                 else if (p->two->mode == 0)
4255                         len1 = snprintf(buffer, sizeof(buffer),
4256                                         "diff--gita/%.*sb/%.*s"
4257                                         "deletedfilemode%06o"
4258                                         "---a/%.*s"
4259                                         "+++/dev/null",
4260                                         len1, p->one->path,
4261                                         len2, p->two->path,
4262                                         p->one->mode,
4263                                         len1, p->one->path);
4264                 else
4265                         len1 = snprintf(buffer, sizeof(buffer),
4266                                         "diff--gita/%.*sb/%.*s"
4267                                         "---a/%.*s"
4268                                         "+++b/%.*s",
4269                                         len1, p->one->path,
4270                                         len2, p->two->path,
4271                                         len1, p->one->path,
4272                                         len2, p->two->path);
4273                 git_SHA1_Update(&ctx, buffer, len1);
4274
4275                 if (diff_filespec_is_binary(p->one) ||
4276                     diff_filespec_is_binary(p->two)) {
4277                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4278                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4279                         continue;
4280                 }
4281
4282                 xpp.flags = 0;
4283                 xecfg.ctxlen = 3;
4284                 xecfg.flags = 0;
4285                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4286                               &xpp, &xecfg);
4287         }
4288
4289         git_SHA1_Final(sha1, &ctx);
4290         return 0;
4291 }
4292
4293 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4294 {
4295         struct diff_queue_struct *q = &diff_queued_diff;
4296         int i;
4297         int result = diff_get_patch_id(options, sha1);
4298
4299         for (i = 0; i < q->nr; i++)
4300                 diff_free_filepair(q->queue[i]);
4301
4302         free(q->queue);
4303         DIFF_QUEUE_CLEAR(q);
4304
4305         return result;
4306 }
4307
4308 static int is_summary_empty(const struct diff_queue_struct *q)
4309 {
4310         int i;
4311
4312         for (i = 0; i < q->nr; i++) {
4313                 const struct diff_filepair *p = q->queue[i];
4314
4315                 switch (p->status) {
4316                 case DIFF_STATUS_DELETED:
4317                 case DIFF_STATUS_ADDED:
4318                 case DIFF_STATUS_COPIED:
4319                 case DIFF_STATUS_RENAMED:
4320                         return 0;
4321                 default:
4322                         if (p->score)
4323                                 return 0;
4324                         if (p->one->mode && p->two->mode &&
4325                             p->one->mode != p->two->mode)
4326                                 return 0;
4327                         break;
4328                 }
4329         }
4330         return 1;
4331 }
4332
4333 static const char rename_limit_warning[] =
4334 "inexact rename detection was skipped due to too many files.";
4335
4336 static const char degrade_cc_to_c_warning[] =
4337 "only found copies from modified paths due to too many files.";
4338
4339 static const char rename_limit_advice[] =
4340 "you may want to set your %s variable to at least "
4341 "%d and retry the command.";
4342
4343 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4344 {
4345         if (degraded_cc)
4346                 warning(degrade_cc_to_c_warning);
4347         else if (needed)
4348                 warning(rename_limit_warning);
4349         else
4350                 return;
4351         if (0 < needed && needed < 32767)
4352                 warning(rename_limit_advice, varname, needed);
4353 }
4354
4355 void diff_flush(struct diff_options *options)
4356 {
4357         struct diff_queue_struct *q = &diff_queued_diff;
4358         int i, output_format = options->output_format;
4359         int separator = 0;
4360         int dirstat_by_line = 0;
4361
4362         /*
4363          * Order: raw, stat, summary, patch
4364          * or:    name/name-status/checkdiff (other bits clear)
4365          */
4366         if (!q->nr)
4367                 goto free_queue;
4368
4369         if (output_format & (DIFF_FORMAT_RAW |
4370                              DIFF_FORMAT_NAME |
4371                              DIFF_FORMAT_NAME_STATUS |
4372                              DIFF_FORMAT_CHECKDIFF)) {
4373                 for (i = 0; i < q->nr; i++) {
4374                         struct diff_filepair *p = q->queue[i];
4375                         if (check_pair_status(p))
4376                                 flush_one_pair(p, options);
4377                 }
4378                 separator++;
4379         }
4380
4381         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4382                 dirstat_by_line = 1;
4383
4384         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4385             dirstat_by_line) {
4386                 struct diffstat_t diffstat;
4387
4388                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4389                 for (i = 0; i < q->nr; i++) {
4390                         struct diff_filepair *p = q->queue[i];
4391                         if (check_pair_status(p))
4392                                 diff_flush_stat(p, options, &diffstat);
4393                 }
4394                 if (output_format & DIFF_FORMAT_NUMSTAT)
4395                         show_numstat(&diffstat, options);
4396                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4397                         show_stats(&diffstat, options);
4398                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4399                         show_shortstats(&diffstat, options);
4400                 if (output_format & DIFF_FORMAT_DIRSTAT)
4401                         show_dirstat_by_line(&diffstat, options);
4402                 free_diffstat_info(&diffstat);
4403                 separator++;
4404         }
4405         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4406                 show_dirstat(options);
4407
4408         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4409                 for (i = 0; i < q->nr; i++) {
4410                         diff_summary(options, q->queue[i]);
4411                 }
4412                 separator++;
4413         }
4414
4415         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4416             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4417             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4418                 /*
4419                  * run diff_flush_patch for the exit status. setting
4420                  * options->file to /dev/null should be safe, becaue we
4421                  * aren't supposed to produce any output anyway.
4422                  */
4423                 if (options->close_file)
4424                         fclose(options->file);
4425                 options->file = fopen("/dev/null", "w");
4426                 if (!options->file)
4427                         die_errno("Could not open /dev/null");
4428                 options->close_file = 1;
4429                 for (i = 0; i < q->nr; i++) {
4430                         struct diff_filepair *p = q->queue[i];
4431                         if (check_pair_status(p))
4432                                 diff_flush_patch(p, options);
4433                         if (options->found_changes)
4434                                 break;
4435                 }
4436         }
4437
4438         if (output_format & DIFF_FORMAT_PATCH) {
4439                 if (separator) {
4440                         if (options->output_prefix) {
4441                                 struct strbuf *msg = NULL;
4442                                 msg = options->output_prefix(options,
4443                                         options->output_prefix_data);
4444                                 fwrite(msg->buf, msg->len, 1, stdout);
4445                         }
4446                         putc(options->line_termination, options->file);
4447                         if (options->stat_sep) {
4448                                 /* attach patch instead of inline */
4449                                 fputs(options->stat_sep, options->file);
4450                         }
4451                 }
4452
4453                 for (i = 0; i < q->nr; i++) {
4454                         struct diff_filepair *p = q->queue[i];
4455                         if (check_pair_status(p))
4456                                 diff_flush_patch(p, options);
4457                 }
4458         }
4459
4460         if (output_format & DIFF_FORMAT_CALLBACK)
4461                 options->format_callback(q, options, options->format_callback_data);
4462
4463         for (i = 0; i < q->nr; i++)
4464                 diff_free_filepair(q->queue[i]);
4465 free_queue:
4466         free(q->queue);
4467         DIFF_QUEUE_CLEAR(q);
4468         if (options->close_file)
4469                 fclose(options->file);
4470
4471         /*
4472          * Report the content-level differences with HAS_CHANGES;
4473          * diff_addremove/diff_change does not set the bit when
4474          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4475          */
4476         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4477                 if (options->found_changes)
4478                         DIFF_OPT_SET(options, HAS_CHANGES);
4479                 else
4480                         DIFF_OPT_CLR(options, HAS_CHANGES);
4481         }
4482 }
4483
4484 static void diffcore_apply_filter(const char *filter)
4485 {
4486         int i;
4487         struct diff_queue_struct *q = &diff_queued_diff;
4488         struct diff_queue_struct outq;
4489         DIFF_QUEUE_CLEAR(&outq);
4490
4491         if (!filter)
4492                 return;
4493
4494         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4495                 int found;
4496                 for (i = found = 0; !found && i < q->nr; i++) {
4497                         struct diff_filepair *p = q->queue[i];
4498                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4499                              ((p->score &&
4500                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4501                               (!p->score &&
4502                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4503                             ((p->status != DIFF_STATUS_MODIFIED) &&
4504                              strchr(filter, p->status)))
4505                                 found++;
4506                 }
4507                 if (found)
4508                         return;
4509
4510                 /* otherwise we will clear the whole queue
4511                  * by copying the empty outq at the end of this
4512                  * function, but first clear the current entries
4513                  * in the queue.
4514                  */
4515                 for (i = 0; i < q->nr; i++)
4516                         diff_free_filepair(q->queue[i]);
4517         }
4518         else {
4519                 /* Only the matching ones */
4520                 for (i = 0; i < q->nr; i++) {
4521                         struct diff_filepair *p = q->queue[i];
4522
4523                         if (((p->status == DIFF_STATUS_MODIFIED) &&
4524                              ((p->score &&
4525                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4526                               (!p->score &&
4527                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4528                             ((p->status != DIFF_STATUS_MODIFIED) &&
4529                              strchr(filter, p->status)))
4530                                 diff_q(&outq, p);
4531                         else
4532                                 diff_free_filepair(p);
4533                 }
4534         }
4535         free(q->queue);
4536         *q = outq;
4537 }
4538
4539 /* Check whether two filespecs with the same mode and size are identical */
4540 static int diff_filespec_is_identical(struct diff_filespec *one,
4541                                       struct diff_filespec *two)
4542 {
4543         if (S_ISGITLINK(one->mode))
4544                 return 0;
4545         if (diff_populate_filespec(one, 0))
4546                 return 0;
4547         if (diff_populate_filespec(two, 0))
4548                 return 0;
4549         return !memcmp(one->data, two->data, one->size);
4550 }
4551
4552 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4553 {
4554         int i;
4555         struct diff_queue_struct *q = &diff_queued_diff;
4556         struct diff_queue_struct outq;
4557         DIFF_QUEUE_CLEAR(&outq);
4558
4559         for (i = 0; i < q->nr; i++) {
4560                 struct diff_filepair *p = q->queue[i];
4561
4562                 /*
4563                  * 1. Entries that come from stat info dirtiness
4564                  *    always have both sides (iow, not create/delete),
4565                  *    one side of the object name is unknown, with
4566                  *    the same mode and size.  Keep the ones that
4567                  *    do not match these criteria.  They have real
4568                  *    differences.
4569                  *
4570                  * 2. At this point, the file is known to be modified,
4571                  *    with the same mode and size, and the object
4572                  *    name of one side is unknown.  Need to inspect
4573                  *    the identical contents.
4574                  */
4575                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4576                     !DIFF_FILE_VALID(p->two) ||
4577                     (p->one->sha1_valid && p->two->sha1_valid) ||
4578                     (p->one->mode != p->two->mode) ||
4579                     diff_populate_filespec(p->one, 1) ||
4580                     diff_populate_filespec(p->two, 1) ||
4581                     (p->one->size != p->two->size) ||
4582                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4583                         diff_q(&outq, p);
4584                 else {
4585                         /*
4586                          * The caller can subtract 1 from skip_stat_unmatch
4587                          * to determine how many paths were dirty only
4588                          * due to stat info mismatch.
4589                          */
4590                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4591                                 diffopt->skip_stat_unmatch++;
4592                         diff_free_filepair(p);
4593                 }
4594         }
4595         free(q->queue);
4596         *q = outq;
4597 }
4598
4599 static int diffnamecmp(const void *a_, const void *b_)
4600 {
4601         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4602         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4603         const char *name_a, *name_b;
4604
4605         name_a = a->one ? a->one->path : a->two->path;
4606         name_b = b->one ? b->one->path : b->two->path;
4607         return strcmp(name_a, name_b);
4608 }
4609
4610 void diffcore_fix_diff_index(struct diff_options *options)
4611 {
4612         struct diff_queue_struct *q = &diff_queued_diff;
4613         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4614 }
4615
4616 void diffcore_std(struct diff_options *options)
4617 {
4618         if (options->skip_stat_unmatch)
4619                 diffcore_skip_stat_unmatch(options);
4620         if (!options->found_follow) {
4621                 /* See try_to_follow_renames() in tree-diff.c */
4622                 if (options->break_opt != -1)
4623                         diffcore_break(options->break_opt);
4624                 if (options->detect_rename)
4625                         diffcore_rename(options);
4626                 if (options->break_opt != -1)
4627                         diffcore_merge_broken();
4628         }
4629         if (options->pickaxe)
4630                 diffcore_pickaxe(options);
4631         if (options->orderfile)
4632                 diffcore_order(options->orderfile);
4633         if (!options->found_follow)
4634                 /* See try_to_follow_renames() in tree-diff.c */
4635                 diff_resolve_rename_copy();
4636         diffcore_apply_filter(options->filter);
4637
4638         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4639                 DIFF_OPT_SET(options, HAS_CHANGES);
4640         else
4641                 DIFF_OPT_CLR(options, HAS_CHANGES);
4642
4643         options->found_follow = 0;
4644 }
4645
4646 int diff_result_code(struct diff_options *opt, int status)
4647 {
4648         int result = 0;
4649
4650         diff_warn_rename_limit("diff.renamelimit",
4651                                opt->needed_rename_limit,
4652                                opt->degraded_cc_to_c);
4653         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4654             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4655                 return status;
4656         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4657             DIFF_OPT_TST(opt, HAS_CHANGES))
4658                 result |= 01;
4659         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4660             DIFF_OPT_TST(opt, CHECK_FAILED))
4661                 result |= 02;
4662         return result;
4663 }
4664
4665 int diff_can_quit_early(struct diff_options *opt)
4666 {
4667         return (DIFF_OPT_TST(opt, QUICK) &&
4668                 !opt->filter &&
4669                 DIFF_OPT_TST(opt, HAS_CHANGES));
4670 }
4671
4672 /*
4673  * Shall changes to this submodule be ignored?
4674  *
4675  * Submodule changes can be configured to be ignored separately for each path,
4676  * but that configuration can be overridden from the command line.
4677  */
4678 static int is_submodule_ignored(const char *path, struct diff_options *options)
4679 {
4680         int ignored = 0;
4681         unsigned orig_flags = options->flags;
4682         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4683                 set_diffopt_flags_from_submodule_config(options, path);
4684         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4685                 ignored = 1;
4686         options->flags = orig_flags;
4687         return ignored;
4688 }
4689
4690 void diff_addremove(struct diff_options *options,
4691                     int addremove, unsigned mode,
4692                     const unsigned char *sha1,
4693                     int sha1_valid,
4694                     const char *concatpath, unsigned dirty_submodule)
4695 {
4696         struct diff_filespec *one, *two;
4697
4698         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4699                 return;
4700
4701         /* This may look odd, but it is a preparation for
4702          * feeding "there are unchanged files which should
4703          * not produce diffs, but when you are doing copy
4704          * detection you would need them, so here they are"
4705          * entries to the diff-core.  They will be prefixed
4706          * with something like '=' or '*' (I haven't decided
4707          * which but should not make any difference).
4708          * Feeding the same new and old to diff_change()
4709          * also has the same effect.
4710          * Before the final output happens, they are pruned after
4711          * merged into rename/copy pairs as appropriate.
4712          */
4713         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4714                 addremove = (addremove == '+' ? '-' :
4715                              addremove == '-' ? '+' : addremove);
4716
4717         if (options->prefix &&
4718             strncmp(concatpath, options->prefix, options->prefix_length))
4719                 return;
4720
4721         one = alloc_filespec(concatpath);
4722         two = alloc_filespec(concatpath);
4723
4724         if (addremove != '+')
4725                 fill_filespec(one, sha1, sha1_valid, mode);
4726         if (addremove != '-') {
4727                 fill_filespec(two, sha1, sha1_valid, mode);
4728                 two->dirty_submodule = dirty_submodule;
4729         }
4730
4731         diff_queue(&diff_queued_diff, one, two);
4732         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4733                 DIFF_OPT_SET(options, HAS_CHANGES);
4734 }
4735
4736 void diff_change(struct diff_options *options,
4737                  unsigned old_mode, unsigned new_mode,
4738                  const unsigned char *old_sha1,
4739                  const unsigned char *new_sha1,
4740                  int old_sha1_valid, int new_sha1_valid,
4741                  const char *concatpath,
4742                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4743 {
4744         struct diff_filespec *one, *two;
4745
4746         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4747             is_submodule_ignored(concatpath, options))
4748                 return;
4749
4750         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4751                 unsigned tmp;
4752                 const unsigned char *tmp_c;
4753                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4754                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4755                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4756                         new_sha1_valid = tmp;
4757                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4758                         new_dirty_submodule = tmp;
4759         }
4760
4761         if (options->prefix &&
4762             strncmp(concatpath, options->prefix, options->prefix_length))
4763                 return;
4764
4765         one = alloc_filespec(concatpath);
4766         two = alloc_filespec(concatpath);
4767         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4768         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4769         one->dirty_submodule = old_dirty_submodule;
4770         two->dirty_submodule = new_dirty_submodule;
4771
4772         diff_queue(&diff_queued_diff, one, two);
4773         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4774                 DIFF_OPT_SET(options, HAS_CHANGES);
4775 }
4776
4777 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4778 {
4779         struct diff_filepair *pair;
4780         struct diff_filespec *one, *two;
4781
4782         if (options->prefix &&
4783             strncmp(path, options->prefix, options->prefix_length))
4784                 return NULL;
4785
4786         one = alloc_filespec(path);
4787         two = alloc_filespec(path);
4788         pair = diff_queue(&diff_queued_diff, one, two);
4789         pair->is_unmerged = 1;
4790         return pair;
4791 }
4792
4793 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4794                 size_t *outsize)
4795 {
4796         struct diff_tempfile *temp;
4797         const char *argv[3];
4798         const char **arg = argv;
4799         struct child_process child;
4800         struct strbuf buf = STRBUF_INIT;
4801         int err = 0;
4802
4803         temp = prepare_temp_file(spec->path, spec);
4804         *arg++ = pgm;
4805         *arg++ = temp->name;
4806         *arg = NULL;
4807
4808         memset(&child, 0, sizeof(child));
4809         child.use_shell = 1;
4810         child.argv = argv;
4811         child.out = -1;
4812         if (start_command(&child)) {
4813                 remove_tempfile();
4814                 return NULL;
4815         }
4816
4817         if (strbuf_read(&buf, child.out, 0) < 0)
4818                 err = error("error reading from textconv command '%s'", pgm);
4819         close(child.out);
4820
4821         if (finish_command(&child) || err) {
4822                 strbuf_release(&buf);
4823                 remove_tempfile();
4824                 return NULL;
4825         }
4826         remove_tempfile();
4827
4828         return strbuf_detach(&buf, outsize);
4829 }
4830
4831 size_t fill_textconv(struct userdiff_driver *driver,
4832                      struct diff_filespec *df,
4833                      char **outbuf)
4834 {
4835         size_t size;
4836
4837         if (!driver || !driver->textconv) {
4838                 if (!DIFF_FILE_VALID(df)) {
4839                         *outbuf = "";
4840                         return 0;
4841                 }
4842                 if (diff_populate_filespec(df, 0))
4843                         die("unable to read files to diff");
4844                 *outbuf = df->data;
4845                 return df->size;
4846         }
4847
4848         if (driver->textconv_cache && df->sha1_valid) {
4849                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4850                                           &size);
4851                 if (*outbuf)
4852                         return size;
4853         }
4854
4855         *outbuf = run_textconv(driver->textconv, df, &size);
4856         if (!*outbuf)
4857                 die("unable to read files to diff");
4858
4859         if (driver->textconv_cache && df->sha1_valid) {
4860                 /* ignore errors, as we might be in a readonly repository */
4861                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4862                                 size);
4863                 /*
4864                  * we could save up changes and flush them all at the end,
4865                  * but we would need an extra call after all diffing is done.
4866                  * Since generating a cache entry is the slow path anyway,
4867                  * this extra overhead probably isn't a big deal.
4868                  */
4869                 notes_cache_write(driver->textconv_cache);
4870         }
4871
4872         return size;
4873 }