First batch for post 2.8 cycle
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "tempfile.h"
6 #include "quote.h"
7 #include "diff.h"
8 #include "diffcore.h"
9 #include "delta.h"
10 #include "xdiff-interface.h"
11 #include "color.h"
12 #include "attr.h"
13 #include "run-command.h"
14 #include "utf8.h"
15 #include "userdiff.h"
16 #include "submodule-config.h"
17 #include "submodule.h"
18 #include "ll-merge.h"
19 #include "string-list.h"
20 #include "argv-array.h"
21
22 #ifdef NO_FAST_WORKING_DIRECTORY
23 #define FAST_WORKING_DIRECTORY 0
24 #else
25 #define FAST_WORKING_DIRECTORY 1
26 #endif
27
28 static int diff_detect_rename_default;
29 static int diff_rename_limit_default = 400;
30 static int diff_suppress_blank_empty;
31 static int diff_use_color_default = -1;
32 static int diff_context_default = 3;
33 static const char *diff_word_regex_cfg;
34 static const char *external_diff_cmd_cfg;
35 static const char *diff_order_file_cfg;
36 int diff_auto_refresh_index = 1;
37 static int diff_mnemonic_prefix;
38 static int diff_no_prefix;
39 static int diff_stat_graph_width;
40 static int diff_dirstat_permille_default = 30;
41 static struct diff_options default_diff_options;
42 static long diff_algorithm;
43
44 static char diff_colors[][COLOR_MAXLEN] = {
45         GIT_COLOR_RESET,
46         GIT_COLOR_NORMAL,       /* CONTEXT */
47         GIT_COLOR_BOLD,         /* METAINFO */
48         GIT_COLOR_CYAN,         /* FRAGINFO */
49         GIT_COLOR_RED,          /* OLD */
50         GIT_COLOR_GREEN,        /* NEW */
51         GIT_COLOR_YELLOW,       /* COMMIT */
52         GIT_COLOR_BG_RED,       /* WHITESPACE */
53         GIT_COLOR_NORMAL,       /* FUNCINFO */
54 };
55
56 static int parse_diff_color_slot(const char *var)
57 {
58         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
59                 return DIFF_CONTEXT;
60         if (!strcasecmp(var, "meta"))
61                 return DIFF_METAINFO;
62         if (!strcasecmp(var, "frag"))
63                 return DIFF_FRAGINFO;
64         if (!strcasecmp(var, "old"))
65                 return DIFF_FILE_OLD;
66         if (!strcasecmp(var, "new"))
67                 return DIFF_FILE_NEW;
68         if (!strcasecmp(var, "commit"))
69                 return DIFF_COMMIT;
70         if (!strcasecmp(var, "whitespace"))
71                 return DIFF_WHITESPACE;
72         if (!strcasecmp(var, "func"))
73                 return DIFF_FUNCINFO;
74         return -1;
75 }
76
77 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
78                                 struct strbuf *errmsg)
79 {
80         char *params_copy = xstrdup(params_string);
81         struct string_list params = STRING_LIST_INIT_NODUP;
82         int ret = 0;
83         int i;
84
85         if (*params_copy)
86                 string_list_split_in_place(&params, params_copy, ',', -1);
87         for (i = 0; i < params.nr; i++) {
88                 const char *p = params.items[i].string;
89                 if (!strcmp(p, "changes")) {
90                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
91                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
92                 } else if (!strcmp(p, "lines")) {
93                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
94                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
95                 } else if (!strcmp(p, "files")) {
96                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
97                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
98                 } else if (!strcmp(p, "noncumulative")) {
99                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
100                 } else if (!strcmp(p, "cumulative")) {
101                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
102                 } else if (isdigit(*p)) {
103                         char *end;
104                         int permille = strtoul(p, &end, 10) * 10;
105                         if (*end == '.' && isdigit(*++end)) {
106                                 /* only use first digit */
107                                 permille += *end - '0';
108                                 /* .. and ignore any further digits */
109                                 while (isdigit(*++end))
110                                         ; /* nothing */
111                         }
112                         if (!*end)
113                                 options->dirstat_permille = permille;
114                         else {
115                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
116                                             p);
117                                 ret++;
118                         }
119                 } else {
120                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
121                         ret++;
122                 }
123
124         }
125         string_list_clear(&params, 0);
126         free(params_copy);
127         return ret;
128 }
129
130 static int parse_submodule_params(struct diff_options *options, const char *value)
131 {
132         if (!strcmp(value, "log"))
133                 DIFF_OPT_SET(options, SUBMODULE_LOG);
134         else if (!strcmp(value, "short"))
135                 DIFF_OPT_CLR(options, SUBMODULE_LOG);
136         else
137                 return -1;
138         return 0;
139 }
140
141 static int git_config_rename(const char *var, const char *value)
142 {
143         if (!value)
144                 return DIFF_DETECT_RENAME;
145         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
146                 return  DIFF_DETECT_COPY;
147         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
148 }
149
150 long parse_algorithm_value(const char *value)
151 {
152         if (!value)
153                 return -1;
154         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
155                 return 0;
156         else if (!strcasecmp(value, "minimal"))
157                 return XDF_NEED_MINIMAL;
158         else if (!strcasecmp(value, "patience"))
159                 return XDF_PATIENCE_DIFF;
160         else if (!strcasecmp(value, "histogram"))
161                 return XDF_HISTOGRAM_DIFF;
162         return -1;
163 }
164
165 /*
166  * These are to give UI layer defaults.
167  * The core-level commands such as git-diff-files should
168  * never be affected by the setting of diff.renames
169  * the user happens to have in the configuration file.
170  */
171 void init_diff_ui_defaults(void)
172 {
173         diff_detect_rename_default = 1;
174 }
175
176 int git_diff_ui_config(const char *var, const char *value, void *cb)
177 {
178         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
179                 diff_use_color_default = git_config_colorbool(var, value);
180                 return 0;
181         }
182         if (!strcmp(var, "diff.context")) {
183                 diff_context_default = git_config_int(var, value);
184                 if (diff_context_default < 0)
185                         return -1;
186                 return 0;
187         }
188         if (!strcmp(var, "diff.renames")) {
189                 diff_detect_rename_default = git_config_rename(var, value);
190                 return 0;
191         }
192         if (!strcmp(var, "diff.autorefreshindex")) {
193                 diff_auto_refresh_index = git_config_bool(var, value);
194                 return 0;
195         }
196         if (!strcmp(var, "diff.mnemonicprefix")) {
197                 diff_mnemonic_prefix = git_config_bool(var, value);
198                 return 0;
199         }
200         if (!strcmp(var, "diff.noprefix")) {
201                 diff_no_prefix = git_config_bool(var, value);
202                 return 0;
203         }
204         if (!strcmp(var, "diff.statgraphwidth")) {
205                 diff_stat_graph_width = git_config_int(var, value);
206                 return 0;
207         }
208         if (!strcmp(var, "diff.external"))
209                 return git_config_string(&external_diff_cmd_cfg, var, value);
210         if (!strcmp(var, "diff.wordregex"))
211                 return git_config_string(&diff_word_regex_cfg, var, value);
212         if (!strcmp(var, "diff.orderfile"))
213                 return git_config_pathname(&diff_order_file_cfg, var, value);
214
215         if (!strcmp(var, "diff.ignoresubmodules"))
216                 handle_ignore_submodules_arg(&default_diff_options, value);
217
218         if (!strcmp(var, "diff.submodule")) {
219                 if (parse_submodule_params(&default_diff_options, value))
220                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
221                                 value);
222                 return 0;
223         }
224
225         if (!strcmp(var, "diff.algorithm")) {
226                 diff_algorithm = parse_algorithm_value(value);
227                 if (diff_algorithm < 0)
228                         return -1;
229                 return 0;
230         }
231
232         if (git_color_config(var, value, cb) < 0)
233                 return -1;
234
235         return git_diff_basic_config(var, value, cb);
236 }
237
238 int git_diff_basic_config(const char *var, const char *value, void *cb)
239 {
240         const char *name;
241
242         if (!strcmp(var, "diff.renamelimit")) {
243                 diff_rename_limit_default = git_config_int(var, value);
244                 return 0;
245         }
246
247         if (userdiff_config(var, value) < 0)
248                 return -1;
249
250         if (skip_prefix(var, "diff.color.", &name) ||
251             skip_prefix(var, "color.diff.", &name)) {
252                 int slot = parse_diff_color_slot(name);
253                 if (slot < 0)
254                         return 0;
255                 if (!value)
256                         return config_error_nonbool(var);
257                 return color_parse(value, diff_colors[slot]);
258         }
259
260         /* like GNU diff's --suppress-blank-empty option  */
261         if (!strcmp(var, "diff.suppressblankempty") ||
262                         /* for backwards compatibility */
263                         !strcmp(var, "diff.suppress-blank-empty")) {
264                 diff_suppress_blank_empty = git_config_bool(var, value);
265                 return 0;
266         }
267
268         if (!strcmp(var, "diff.dirstat")) {
269                 struct strbuf errmsg = STRBUF_INIT;
270                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
271                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
272                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
273                                 errmsg.buf);
274                 strbuf_release(&errmsg);
275                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
276                 return 0;
277         }
278
279         if (starts_with(var, "submodule."))
280                 return parse_submodule_config_option(var, value);
281
282         return git_default_config(var, value, cb);
283 }
284
285 static char *quote_two(const char *one, const char *two)
286 {
287         int need_one = quote_c_style(one, NULL, NULL, 1);
288         int need_two = quote_c_style(two, NULL, NULL, 1);
289         struct strbuf res = STRBUF_INIT;
290
291         if (need_one + need_two) {
292                 strbuf_addch(&res, '"');
293                 quote_c_style(one, &res, NULL, 1);
294                 quote_c_style(two, &res, NULL, 1);
295                 strbuf_addch(&res, '"');
296         } else {
297                 strbuf_addstr(&res, one);
298                 strbuf_addstr(&res, two);
299         }
300         return strbuf_detach(&res, NULL);
301 }
302
303 static const char *external_diff(void)
304 {
305         static const char *external_diff_cmd = NULL;
306         static int done_preparing = 0;
307
308         if (done_preparing)
309                 return external_diff_cmd;
310         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
311         if (!external_diff_cmd)
312                 external_diff_cmd = external_diff_cmd_cfg;
313         done_preparing = 1;
314         return external_diff_cmd;
315 }
316
317 /*
318  * Keep track of files used for diffing. Sometimes such an entry
319  * refers to a temporary file, sometimes to an existing file, and
320  * sometimes to "/dev/null".
321  */
322 static struct diff_tempfile {
323         /*
324          * filename external diff should read from, or NULL if this
325          * entry is currently not in use:
326          */
327         const char *name;
328
329         char hex[GIT_SHA1_HEXSZ + 1];
330         char mode[10];
331
332         /*
333          * If this diff_tempfile instance refers to a temporary file,
334          * this tempfile object is used to manage its lifetime.
335          */
336         struct tempfile tempfile;
337 } diff_temp[2];
338
339 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
340
341 struct emit_callback {
342         int color_diff;
343         unsigned ws_rule;
344         int blank_at_eof_in_preimage;
345         int blank_at_eof_in_postimage;
346         int lno_in_preimage;
347         int lno_in_postimage;
348         sane_truncate_fn truncate;
349         const char **label_path;
350         struct diff_words_data *diff_words;
351         struct diff_options *opt;
352         int *found_changesp;
353         struct strbuf *header;
354 };
355
356 static int count_lines(const char *data, int size)
357 {
358         int count, ch, completely_empty = 1, nl_just_seen = 0;
359         count = 0;
360         while (0 < size--) {
361                 ch = *data++;
362                 if (ch == '\n') {
363                         count++;
364                         nl_just_seen = 1;
365                         completely_empty = 0;
366                 }
367                 else {
368                         nl_just_seen = 0;
369                         completely_empty = 0;
370                 }
371         }
372         if (completely_empty)
373                 return 0;
374         if (!nl_just_seen)
375                 count++; /* no trailing newline */
376         return count;
377 }
378
379 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
380 {
381         if (!DIFF_FILE_VALID(one)) {
382                 mf->ptr = (char *)""; /* does not matter */
383                 mf->size = 0;
384                 return 0;
385         }
386         else if (diff_populate_filespec(one, 0))
387                 return -1;
388
389         mf->ptr = one->data;
390         mf->size = one->size;
391         return 0;
392 }
393
394 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
395 static unsigned long diff_filespec_size(struct diff_filespec *one)
396 {
397         if (!DIFF_FILE_VALID(one))
398                 return 0;
399         diff_populate_filespec(one, CHECK_SIZE_ONLY);
400         return one->size;
401 }
402
403 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
404 {
405         char *ptr = mf->ptr;
406         long size = mf->size;
407         int cnt = 0;
408
409         if (!size)
410                 return cnt;
411         ptr += size - 1; /* pointing at the very end */
412         if (*ptr != '\n')
413                 ; /* incomplete line */
414         else
415                 ptr--; /* skip the last LF */
416         while (mf->ptr < ptr) {
417                 char *prev_eol;
418                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
419                         if (*prev_eol == '\n')
420                                 break;
421                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
422                         break;
423                 cnt++;
424                 ptr = prev_eol - 1;
425         }
426         return cnt;
427 }
428
429 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
430                                struct emit_callback *ecbdata)
431 {
432         int l1, l2, at;
433         unsigned ws_rule = ecbdata->ws_rule;
434         l1 = count_trailing_blank(mf1, ws_rule);
435         l2 = count_trailing_blank(mf2, ws_rule);
436         if (l2 <= l1) {
437                 ecbdata->blank_at_eof_in_preimage = 0;
438                 ecbdata->blank_at_eof_in_postimage = 0;
439                 return;
440         }
441         at = count_lines(mf1->ptr, mf1->size);
442         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
443
444         at = count_lines(mf2->ptr, mf2->size);
445         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
446 }
447
448 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
449                         int first, const char *line, int len)
450 {
451         int has_trailing_newline, has_trailing_carriage_return;
452         int nofirst;
453         FILE *file = o->file;
454
455         fputs(diff_line_prefix(o), file);
456
457         if (len == 0) {
458                 has_trailing_newline = (first == '\n');
459                 has_trailing_carriage_return = (!has_trailing_newline &&
460                                                 (first == '\r'));
461                 nofirst = has_trailing_newline || has_trailing_carriage_return;
462         } else {
463                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
464                 if (has_trailing_newline)
465                         len--;
466                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
467                 if (has_trailing_carriage_return)
468                         len--;
469                 nofirst = 0;
470         }
471
472         if (len || !nofirst) {
473                 fputs(set, file);
474                 if (!nofirst)
475                         fputc(first, file);
476                 fwrite(line, len, 1, file);
477                 fputs(reset, file);
478         }
479         if (has_trailing_carriage_return)
480                 fputc('\r', file);
481         if (has_trailing_newline)
482                 fputc('\n', file);
483 }
484
485 static void emit_line(struct diff_options *o, const char *set, const char *reset,
486                       const char *line, int len)
487 {
488         emit_line_0(o, set, reset, line[0], line+1, len-1);
489 }
490
491 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
492 {
493         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
494               ecbdata->blank_at_eof_in_preimage &&
495               ecbdata->blank_at_eof_in_postimage &&
496               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
497               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
498                 return 0;
499         return ws_blank_line(line, len, ecbdata->ws_rule);
500 }
501
502 static void emit_line_checked(const char *reset,
503                               struct emit_callback *ecbdata,
504                               const char *line, int len,
505                               enum color_diff color,
506                               unsigned ws_error_highlight,
507                               char sign)
508 {
509         const char *set = diff_get_color(ecbdata->color_diff, color);
510         const char *ws = NULL;
511
512         if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
513                 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
514                 if (!*ws)
515                         ws = NULL;
516         }
517
518         if (!ws)
519                 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
520         else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
521                 /* Blank line at EOF - paint '+' as well */
522                 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
523         else {
524                 /* Emit just the prefix, then the rest. */
525                 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
526                 ws_check_emit(line, len, ecbdata->ws_rule,
527                               ecbdata->opt->file, set, reset, ws);
528         }
529 }
530
531 static void emit_add_line(const char *reset,
532                           struct emit_callback *ecbdata,
533                           const char *line, int len)
534 {
535         emit_line_checked(reset, ecbdata, line, len,
536                           DIFF_FILE_NEW, WSEH_NEW, '+');
537 }
538
539 static void emit_del_line(const char *reset,
540                           struct emit_callback *ecbdata,
541                           const char *line, int len)
542 {
543         emit_line_checked(reset, ecbdata, line, len,
544                           DIFF_FILE_OLD, WSEH_OLD, '-');
545 }
546
547 static void emit_context_line(const char *reset,
548                               struct emit_callback *ecbdata,
549                               const char *line, int len)
550 {
551         emit_line_checked(reset, ecbdata, line, len,
552                           DIFF_CONTEXT, WSEH_CONTEXT, ' ');
553 }
554
555 static void emit_hunk_header(struct emit_callback *ecbdata,
556                              const char *line, int len)
557 {
558         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
559         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
560         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
561         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
562         static const char atat[2] = { '@', '@' };
563         const char *cp, *ep;
564         struct strbuf msgbuf = STRBUF_INIT;
565         int org_len = len;
566         int i = 1;
567
568         /*
569          * As a hunk header must begin with "@@ -<old>, +<new> @@",
570          * it always is at least 10 bytes long.
571          */
572         if (len < 10 ||
573             memcmp(line, atat, 2) ||
574             !(ep = memmem(line + 2, len - 2, atat, 2))) {
575                 emit_line(ecbdata->opt, context, reset, line, len);
576                 return;
577         }
578         ep += 2; /* skip over @@ */
579
580         /* The hunk header in fraginfo color */
581         strbuf_addstr(&msgbuf, frag);
582         strbuf_add(&msgbuf, line, ep - line);
583         strbuf_addstr(&msgbuf, reset);
584
585         /*
586          * trailing "\r\n"
587          */
588         for ( ; i < 3; i++)
589                 if (line[len - i] == '\r' || line[len - i] == '\n')
590                         len--;
591
592         /* blank before the func header */
593         for (cp = ep; ep - line < len; ep++)
594                 if (*ep != ' ' && *ep != '\t')
595                         break;
596         if (ep != cp) {
597                 strbuf_addstr(&msgbuf, context);
598                 strbuf_add(&msgbuf, cp, ep - cp);
599                 strbuf_addstr(&msgbuf, reset);
600         }
601
602         if (ep < line + len) {
603                 strbuf_addstr(&msgbuf, func);
604                 strbuf_add(&msgbuf, ep, line + len - ep);
605                 strbuf_addstr(&msgbuf, reset);
606         }
607
608         strbuf_add(&msgbuf, line + len, org_len - len);
609         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
610         strbuf_release(&msgbuf);
611 }
612
613 static struct diff_tempfile *claim_diff_tempfile(void) {
614         int i;
615         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
616                 if (!diff_temp[i].name)
617                         return diff_temp + i;
618         die("BUG: diff is failing to clean up its tempfiles");
619 }
620
621 static void remove_tempfile(void)
622 {
623         int i;
624         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
625                 if (is_tempfile_active(&diff_temp[i].tempfile))
626                         delete_tempfile(&diff_temp[i].tempfile);
627                 diff_temp[i].name = NULL;
628         }
629 }
630
631 static void print_line_count(FILE *file, int count)
632 {
633         switch (count) {
634         case 0:
635                 fprintf(file, "0,0");
636                 break;
637         case 1:
638                 fprintf(file, "1");
639                 break;
640         default:
641                 fprintf(file, "1,%d", count);
642                 break;
643         }
644 }
645
646 static void emit_rewrite_lines(struct emit_callback *ecb,
647                                int prefix, const char *data, int size)
648 {
649         const char *endp = NULL;
650         static const char *nneof = " No newline at end of file\n";
651         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
652
653         while (0 < size) {
654                 int len;
655
656                 endp = memchr(data, '\n', size);
657                 len = endp ? (endp - data + 1) : size;
658                 if (prefix != '+') {
659                         ecb->lno_in_preimage++;
660                         emit_del_line(reset, ecb, data, len);
661                 } else {
662                         ecb->lno_in_postimage++;
663                         emit_add_line(reset, ecb, data, len);
664                 }
665                 size -= len;
666                 data += len;
667         }
668         if (!endp) {
669                 const char *context = diff_get_color(ecb->color_diff,
670                                                      DIFF_CONTEXT);
671                 putc('\n', ecb->opt->file);
672                 emit_line_0(ecb->opt, context, reset, '\\',
673                             nneof, strlen(nneof));
674         }
675 }
676
677 static void emit_rewrite_diff(const char *name_a,
678                               const char *name_b,
679                               struct diff_filespec *one,
680                               struct diff_filespec *two,
681                               struct userdiff_driver *textconv_one,
682                               struct userdiff_driver *textconv_two,
683                               struct diff_options *o)
684 {
685         int lc_a, lc_b;
686         const char *name_a_tab, *name_b_tab;
687         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
688         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
689         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
690         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
691         const char *a_prefix, *b_prefix;
692         char *data_one, *data_two;
693         size_t size_one, size_two;
694         struct emit_callback ecbdata;
695         const char *line_prefix = diff_line_prefix(o);
696
697         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
698                 a_prefix = o->b_prefix;
699                 b_prefix = o->a_prefix;
700         } else {
701                 a_prefix = o->a_prefix;
702                 b_prefix = o->b_prefix;
703         }
704
705         name_a += (*name_a == '/');
706         name_b += (*name_b == '/');
707         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
708         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
709
710         strbuf_reset(&a_name);
711         strbuf_reset(&b_name);
712         quote_two_c_style(&a_name, a_prefix, name_a, 0);
713         quote_two_c_style(&b_name, b_prefix, name_b, 0);
714
715         size_one = fill_textconv(textconv_one, one, &data_one);
716         size_two = fill_textconv(textconv_two, two, &data_two);
717
718         memset(&ecbdata, 0, sizeof(ecbdata));
719         ecbdata.color_diff = want_color(o->use_color);
720         ecbdata.found_changesp = &o->found_changes;
721         ecbdata.ws_rule = whitespace_rule(name_b);
722         ecbdata.opt = o;
723         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
724                 mmfile_t mf1, mf2;
725                 mf1.ptr = (char *)data_one;
726                 mf2.ptr = (char *)data_two;
727                 mf1.size = size_one;
728                 mf2.size = size_two;
729                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
730         }
731         ecbdata.lno_in_preimage = 1;
732         ecbdata.lno_in_postimage = 1;
733
734         lc_a = count_lines(data_one, size_one);
735         lc_b = count_lines(data_two, size_two);
736         fprintf(o->file,
737                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
738                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
739                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
740                 line_prefix, fraginfo);
741         if (!o->irreversible_delete)
742                 print_line_count(o->file, lc_a);
743         else
744                 fprintf(o->file, "?,?");
745         fprintf(o->file, " +");
746         print_line_count(o->file, lc_b);
747         fprintf(o->file, " @@%s\n", reset);
748         if (lc_a && !o->irreversible_delete)
749                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
750         if (lc_b)
751                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
752         if (textconv_one)
753                 free((char *)data_one);
754         if (textconv_two)
755                 free((char *)data_two);
756 }
757
758 struct diff_words_buffer {
759         mmfile_t text;
760         long alloc;
761         struct diff_words_orig {
762                 const char *begin, *end;
763         } *orig;
764         int orig_nr, orig_alloc;
765 };
766
767 static void diff_words_append(char *line, unsigned long len,
768                 struct diff_words_buffer *buffer)
769 {
770         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
771         line++;
772         len--;
773         memcpy(buffer->text.ptr + buffer->text.size, line, len);
774         buffer->text.size += len;
775         buffer->text.ptr[buffer->text.size] = '\0';
776 }
777
778 struct diff_words_style_elem {
779         const char *prefix;
780         const char *suffix;
781         const char *color; /* NULL; filled in by the setup code if
782                             * color is enabled */
783 };
784
785 struct diff_words_style {
786         enum diff_words_type type;
787         struct diff_words_style_elem new, old, ctx;
788         const char *newline;
789 };
790
791 static struct diff_words_style diff_words_styles[] = {
792         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
793         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
794         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
795 };
796
797 struct diff_words_data {
798         struct diff_words_buffer minus, plus;
799         const char *current_plus;
800         int last_minus;
801         struct diff_options *opt;
802         regex_t *word_regex;
803         enum diff_words_type type;
804         struct diff_words_style *style;
805 };
806
807 static int fn_out_diff_words_write_helper(FILE *fp,
808                                           struct diff_words_style_elem *st_el,
809                                           const char *newline,
810                                           size_t count, const char *buf,
811                                           const char *line_prefix)
812 {
813         int print = 0;
814
815         while (count) {
816                 char *p = memchr(buf, '\n', count);
817                 if (print)
818                         fputs(line_prefix, fp);
819                 if (p != buf) {
820                         if (st_el->color && fputs(st_el->color, fp) < 0)
821                                 return -1;
822                         if (fputs(st_el->prefix, fp) < 0 ||
823                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
824                             fputs(st_el->suffix, fp) < 0)
825                                 return -1;
826                         if (st_el->color && *st_el->color
827                             && fputs(GIT_COLOR_RESET, fp) < 0)
828                                 return -1;
829                 }
830                 if (!p)
831                         return 0;
832                 if (fputs(newline, fp) < 0)
833                         return -1;
834                 count -= p + 1 - buf;
835                 buf = p + 1;
836                 print = 1;
837         }
838         return 0;
839 }
840
841 /*
842  * '--color-words' algorithm can be described as:
843  *
844  *   1. collect a the minus/plus lines of a diff hunk, divided into
845  *      minus-lines and plus-lines;
846  *
847  *   2. break both minus-lines and plus-lines into words and
848  *      place them into two mmfile_t with one word for each line;
849  *
850  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
851  *
852  * And for the common parts of the both file, we output the plus side text.
853  * diff_words->current_plus is used to trace the current position of the plus file
854  * which printed. diff_words->last_minus is used to trace the last minus word
855  * printed.
856  *
857  * For '--graph' to work with '--color-words', we need to output the graph prefix
858  * on each line of color words output. Generally, there are two conditions on
859  * which we should output the prefix.
860  *
861  *   1. diff_words->last_minus == 0 &&
862  *      diff_words->current_plus == diff_words->plus.text.ptr
863  *
864  *      that is: the plus text must start as a new line, and if there is no minus
865  *      word printed, a graph prefix must be printed.
866  *
867  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
868  *      *(diff_words->current_plus - 1) == '\n'
869  *
870  *      that is: a graph prefix must be printed following a '\n'
871  */
872 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
873 {
874         if ((diff_words->last_minus == 0 &&
875                 diff_words->current_plus == diff_words->plus.text.ptr) ||
876                 (diff_words->current_plus > diff_words->plus.text.ptr &&
877                 *(diff_words->current_plus - 1) == '\n')) {
878                 return 1;
879         } else {
880                 return 0;
881         }
882 }
883
884 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
885 {
886         struct diff_words_data *diff_words = priv;
887         struct diff_words_style *style = diff_words->style;
888         int minus_first, minus_len, plus_first, plus_len;
889         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
890         struct diff_options *opt = diff_words->opt;
891         const char *line_prefix;
892
893         if (line[0] != '@' || parse_hunk_header(line, len,
894                         &minus_first, &minus_len, &plus_first, &plus_len))
895                 return;
896
897         assert(opt);
898         line_prefix = diff_line_prefix(opt);
899
900         /* POSIX requires that first be decremented by one if len == 0... */
901         if (minus_len) {
902                 minus_begin = diff_words->minus.orig[minus_first].begin;
903                 minus_end =
904                         diff_words->minus.orig[minus_first + minus_len - 1].end;
905         } else
906                 minus_begin = minus_end =
907                         diff_words->minus.orig[minus_first].end;
908
909         if (plus_len) {
910                 plus_begin = diff_words->plus.orig[plus_first].begin;
911                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
912         } else
913                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
914
915         if (color_words_output_graph_prefix(diff_words)) {
916                 fputs(line_prefix, diff_words->opt->file);
917         }
918         if (diff_words->current_plus != plus_begin) {
919                 fn_out_diff_words_write_helper(diff_words->opt->file,
920                                 &style->ctx, style->newline,
921                                 plus_begin - diff_words->current_plus,
922                                 diff_words->current_plus, line_prefix);
923                 if (*(plus_begin - 1) == '\n')
924                         fputs(line_prefix, diff_words->opt->file);
925         }
926         if (minus_begin != minus_end) {
927                 fn_out_diff_words_write_helper(diff_words->opt->file,
928                                 &style->old, style->newline,
929                                 minus_end - minus_begin, minus_begin,
930                                 line_prefix);
931         }
932         if (plus_begin != plus_end) {
933                 fn_out_diff_words_write_helper(diff_words->opt->file,
934                                 &style->new, style->newline,
935                                 plus_end - plus_begin, plus_begin,
936                                 line_prefix);
937         }
938
939         diff_words->current_plus = plus_end;
940         diff_words->last_minus = minus_first;
941 }
942
943 /* This function starts looking at *begin, and returns 0 iff a word was found. */
944 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
945                 int *begin, int *end)
946 {
947         if (word_regex && *begin < buffer->size) {
948                 regmatch_t match[1];
949                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
950                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
951                                         '\n', match[0].rm_eo - match[0].rm_so);
952                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
953                         *begin += match[0].rm_so;
954                         return *begin >= *end;
955                 }
956                 return -1;
957         }
958
959         /* find the next word */
960         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
961                 (*begin)++;
962         if (*begin >= buffer->size)
963                 return -1;
964
965         /* find the end of the word */
966         *end = *begin + 1;
967         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
968                 (*end)++;
969
970         return 0;
971 }
972
973 /*
974  * This function splits the words in buffer->text, stores the list with
975  * newline separator into out, and saves the offsets of the original words
976  * in buffer->orig.
977  */
978 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
979                 regex_t *word_regex)
980 {
981         int i, j;
982         long alloc = 0;
983
984         out->size = 0;
985         out->ptr = NULL;
986
987         /* fake an empty "0th" word */
988         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
989         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
990         buffer->orig_nr = 1;
991
992         for (i = 0; i < buffer->text.size; i++) {
993                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
994                         return;
995
996                 /* store original boundaries */
997                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
998                                 buffer->orig_alloc);
999                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1000                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1001                 buffer->orig_nr++;
1002
1003                 /* store one word */
1004                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1005                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1006                 out->ptr[out->size + j - i] = '\n';
1007                 out->size += j - i + 1;
1008
1009                 i = j - 1;
1010         }
1011 }
1012
1013 /* this executes the word diff on the accumulated buffers */
1014 static void diff_words_show(struct diff_words_data *diff_words)
1015 {
1016         xpparam_t xpp;
1017         xdemitconf_t xecfg;
1018         mmfile_t minus, plus;
1019         struct diff_words_style *style = diff_words->style;
1020
1021         struct diff_options *opt = diff_words->opt;
1022         const char *line_prefix;
1023
1024         assert(opt);
1025         line_prefix = diff_line_prefix(opt);
1026
1027         /* special case: only removal */
1028         if (!diff_words->plus.text.size) {
1029                 fputs(line_prefix, diff_words->opt->file);
1030                 fn_out_diff_words_write_helper(diff_words->opt->file,
1031                         &style->old, style->newline,
1032                         diff_words->minus.text.size,
1033                         diff_words->minus.text.ptr, line_prefix);
1034                 diff_words->minus.text.size = 0;
1035                 return;
1036         }
1037
1038         diff_words->current_plus = diff_words->plus.text.ptr;
1039         diff_words->last_minus = 0;
1040
1041         memset(&xpp, 0, sizeof(xpp));
1042         memset(&xecfg, 0, sizeof(xecfg));
1043         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1044         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1045         xpp.flags = 0;
1046         /* as only the hunk header will be parsed, we need a 0-context */
1047         xecfg.ctxlen = 0;
1048         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1049                           &xpp, &xecfg))
1050                 die("unable to generate word diff");
1051         free(minus.ptr);
1052         free(plus.ptr);
1053         if (diff_words->current_plus != diff_words->plus.text.ptr +
1054                         diff_words->plus.text.size) {
1055                 if (color_words_output_graph_prefix(diff_words))
1056                         fputs(line_prefix, diff_words->opt->file);
1057                 fn_out_diff_words_write_helper(diff_words->opt->file,
1058                         &style->ctx, style->newline,
1059                         diff_words->plus.text.ptr + diff_words->plus.text.size
1060                         - diff_words->current_plus, diff_words->current_plus,
1061                         line_prefix);
1062         }
1063         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1064 }
1065
1066 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1067 static void diff_words_flush(struct emit_callback *ecbdata)
1068 {
1069         if (ecbdata->diff_words->minus.text.size ||
1070             ecbdata->diff_words->plus.text.size)
1071                 diff_words_show(ecbdata->diff_words);
1072 }
1073
1074 static void diff_filespec_load_driver(struct diff_filespec *one)
1075 {
1076         /* Use already-loaded driver */
1077         if (one->driver)
1078                 return;
1079
1080         if (S_ISREG(one->mode))
1081                 one->driver = userdiff_find_by_path(one->path);
1082
1083         /* Fallback to default settings */
1084         if (!one->driver)
1085                 one->driver = userdiff_find_by_name("default");
1086 }
1087
1088 static const char *userdiff_word_regex(struct diff_filespec *one)
1089 {
1090         diff_filespec_load_driver(one);
1091         return one->driver->word_regex;
1092 }
1093
1094 static void init_diff_words_data(struct emit_callback *ecbdata,
1095                                  struct diff_options *orig_opts,
1096                                  struct diff_filespec *one,
1097                                  struct diff_filespec *two)
1098 {
1099         int i;
1100         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1101         memcpy(o, orig_opts, sizeof(struct diff_options));
1102
1103         ecbdata->diff_words =
1104                 xcalloc(1, sizeof(struct diff_words_data));
1105         ecbdata->diff_words->type = o->word_diff;
1106         ecbdata->diff_words->opt = o;
1107         if (!o->word_regex)
1108                 o->word_regex = userdiff_word_regex(one);
1109         if (!o->word_regex)
1110                 o->word_regex = userdiff_word_regex(two);
1111         if (!o->word_regex)
1112                 o->word_regex = diff_word_regex_cfg;
1113         if (o->word_regex) {
1114                 ecbdata->diff_words->word_regex = (regex_t *)
1115                         xmalloc(sizeof(regex_t));
1116                 if (regcomp(ecbdata->diff_words->word_regex,
1117                             o->word_regex,
1118                             REG_EXTENDED | REG_NEWLINE))
1119                         die ("Invalid regular expression: %s",
1120                              o->word_regex);
1121         }
1122         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1123                 if (o->word_diff == diff_words_styles[i].type) {
1124                         ecbdata->diff_words->style =
1125                                 &diff_words_styles[i];
1126                         break;
1127                 }
1128         }
1129         if (want_color(o->use_color)) {
1130                 struct diff_words_style *st = ecbdata->diff_words->style;
1131                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1132                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1133                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1134         }
1135 }
1136
1137 static void free_diff_words_data(struct emit_callback *ecbdata)
1138 {
1139         if (ecbdata->diff_words) {
1140                 diff_words_flush(ecbdata);
1141                 free (ecbdata->diff_words->opt);
1142                 free (ecbdata->diff_words->minus.text.ptr);
1143                 free (ecbdata->diff_words->minus.orig);
1144                 free (ecbdata->diff_words->plus.text.ptr);
1145                 free (ecbdata->diff_words->plus.orig);
1146                 if (ecbdata->diff_words->word_regex) {
1147                         regfree(ecbdata->diff_words->word_regex);
1148                         free(ecbdata->diff_words->word_regex);
1149                 }
1150                 free(ecbdata->diff_words);
1151                 ecbdata->diff_words = NULL;
1152         }
1153 }
1154
1155 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1156 {
1157         if (want_color(diff_use_color))
1158                 return diff_colors[ix];
1159         return "";
1160 }
1161
1162 const char *diff_line_prefix(struct diff_options *opt)
1163 {
1164         struct strbuf *msgbuf;
1165         if (!opt->output_prefix)
1166                 return "";
1167
1168         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1169         return msgbuf->buf;
1170 }
1171
1172 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1173 {
1174         const char *cp;
1175         unsigned long allot;
1176         size_t l = len;
1177
1178         if (ecb->truncate)
1179                 return ecb->truncate(line, len);
1180         cp = line;
1181         allot = l;
1182         while (0 < l) {
1183                 (void) utf8_width(&cp, &l);
1184                 if (!cp)
1185                         break; /* truncated in the middle? */
1186         }
1187         return allot - l;
1188 }
1189
1190 static void find_lno(const char *line, struct emit_callback *ecbdata)
1191 {
1192         const char *p;
1193         ecbdata->lno_in_preimage = 0;
1194         ecbdata->lno_in_postimage = 0;
1195         p = strchr(line, '-');
1196         if (!p)
1197                 return; /* cannot happen */
1198         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1199         p = strchr(p, '+');
1200         if (!p)
1201                 return; /* cannot happen */
1202         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1203 }
1204
1205 static void fn_out_consume(void *priv, char *line, unsigned long len)
1206 {
1207         struct emit_callback *ecbdata = priv;
1208         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1209         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1210         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1211         struct diff_options *o = ecbdata->opt;
1212         const char *line_prefix = diff_line_prefix(o);
1213
1214         if (ecbdata->header) {
1215                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1216                 strbuf_reset(ecbdata->header);
1217                 ecbdata->header = NULL;
1218         }
1219         *(ecbdata->found_changesp) = 1;
1220
1221         if (ecbdata->label_path[0]) {
1222                 const char *name_a_tab, *name_b_tab;
1223
1224                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1225                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1226
1227                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1228                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1229                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1230                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1231                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1232         }
1233
1234         if (diff_suppress_blank_empty
1235             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1236                 line[0] = '\n';
1237                 len = 1;
1238         }
1239
1240         if (line[0] == '@') {
1241                 if (ecbdata->diff_words)
1242                         diff_words_flush(ecbdata);
1243                 len = sane_truncate_line(ecbdata, line, len);
1244                 find_lno(line, ecbdata);
1245                 emit_hunk_header(ecbdata, line, len);
1246                 if (line[len-1] != '\n')
1247                         putc('\n', ecbdata->opt->file);
1248                 return;
1249         }
1250
1251         if (len < 1) {
1252                 emit_line(ecbdata->opt, reset, reset, line, len);
1253                 if (ecbdata->diff_words
1254                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1255                         fputs("~\n", ecbdata->opt->file);
1256                 return;
1257         }
1258
1259         if (ecbdata->diff_words) {
1260                 if (line[0] == '-') {
1261                         diff_words_append(line, len,
1262                                           &ecbdata->diff_words->minus);
1263                         return;
1264                 } else if (line[0] == '+') {
1265                         diff_words_append(line, len,
1266                                           &ecbdata->diff_words->plus);
1267                         return;
1268                 } else if (starts_with(line, "\\ ")) {
1269                         /*
1270                          * Eat the "no newline at eof" marker as if we
1271                          * saw a "+" or "-" line with nothing on it,
1272                          * and return without diff_words_flush() to
1273                          * defer processing. If this is the end of
1274                          * preimage, more "+" lines may come after it.
1275                          */
1276                         return;
1277                 }
1278                 diff_words_flush(ecbdata);
1279                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1280                         emit_line(ecbdata->opt, context, reset, line, len);
1281                         fputs("~\n", ecbdata->opt->file);
1282                 } else {
1283                         /*
1284                          * Skip the prefix character, if any.  With
1285                          * diff_suppress_blank_empty, there may be
1286                          * none.
1287                          */
1288                         if (line[0] != '\n') {
1289                               line++;
1290                               len--;
1291                         }
1292                         emit_line(ecbdata->opt, context, reset, line, len);
1293                 }
1294                 return;
1295         }
1296
1297         switch (line[0]) {
1298         case '+':
1299                 ecbdata->lno_in_postimage++;
1300                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1301                 break;
1302         case '-':
1303                 ecbdata->lno_in_preimage++;
1304                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1305                 break;
1306         case ' ':
1307                 ecbdata->lno_in_postimage++;
1308                 ecbdata->lno_in_preimage++;
1309                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1310                 break;
1311         default:
1312                 /* incomplete line at the end */
1313                 ecbdata->lno_in_preimage++;
1314                 emit_line(ecbdata->opt,
1315                           diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1316                           reset, line, len);
1317                 break;
1318         }
1319 }
1320
1321 static char *pprint_rename(const char *a, const char *b)
1322 {
1323         const char *old = a;
1324         const char *new = b;
1325         struct strbuf name = STRBUF_INIT;
1326         int pfx_length, sfx_length;
1327         int pfx_adjust_for_slash;
1328         int len_a = strlen(a);
1329         int len_b = strlen(b);
1330         int a_midlen, b_midlen;
1331         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1332         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1333
1334         if (qlen_a || qlen_b) {
1335                 quote_c_style(a, &name, NULL, 0);
1336                 strbuf_addstr(&name, " => ");
1337                 quote_c_style(b, &name, NULL, 0);
1338                 return strbuf_detach(&name, NULL);
1339         }
1340
1341         /* Find common prefix */
1342         pfx_length = 0;
1343         while (*old && *new && *old == *new) {
1344                 if (*old == '/')
1345                         pfx_length = old - a + 1;
1346                 old++;
1347                 new++;
1348         }
1349
1350         /* Find common suffix */
1351         old = a + len_a;
1352         new = b + len_b;
1353         sfx_length = 0;
1354         /*
1355          * If there is a common prefix, it must end in a slash.  In
1356          * that case we let this loop run 1 into the prefix to see the
1357          * same slash.
1358          *
1359          * If there is no common prefix, we cannot do this as it would
1360          * underrun the input strings.
1361          */
1362         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1363         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1364                b + pfx_length - pfx_adjust_for_slash <= new &&
1365                *old == *new) {
1366                 if (*old == '/')
1367                         sfx_length = len_a - (old - a);
1368                 old--;
1369                 new--;
1370         }
1371
1372         /*
1373          * pfx{mid-a => mid-b}sfx
1374          * {pfx-a => pfx-b}sfx
1375          * pfx{sfx-a => sfx-b}
1376          * name-a => name-b
1377          */
1378         a_midlen = len_a - pfx_length - sfx_length;
1379         b_midlen = len_b - pfx_length - sfx_length;
1380         if (a_midlen < 0)
1381                 a_midlen = 0;
1382         if (b_midlen < 0)
1383                 b_midlen = 0;
1384
1385         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1386         if (pfx_length + sfx_length) {
1387                 strbuf_add(&name, a, pfx_length);
1388                 strbuf_addch(&name, '{');
1389         }
1390         strbuf_add(&name, a + pfx_length, a_midlen);
1391         strbuf_addstr(&name, " => ");
1392         strbuf_add(&name, b + pfx_length, b_midlen);
1393         if (pfx_length + sfx_length) {
1394                 strbuf_addch(&name, '}');
1395                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1396         }
1397         return strbuf_detach(&name, NULL);
1398 }
1399
1400 struct diffstat_t {
1401         int nr;
1402         int alloc;
1403         struct diffstat_file {
1404                 char *from_name;
1405                 char *name;
1406                 char *print_name;
1407                 unsigned is_unmerged:1;
1408                 unsigned is_binary:1;
1409                 unsigned is_renamed:1;
1410                 unsigned is_interesting:1;
1411                 uintmax_t added, deleted;
1412         } **files;
1413 };
1414
1415 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1416                                           const char *name_a,
1417                                           const char *name_b)
1418 {
1419         struct diffstat_file *x;
1420         x = xcalloc(1, sizeof(*x));
1421         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1422         diffstat->files[diffstat->nr++] = x;
1423         if (name_b) {
1424                 x->from_name = xstrdup(name_a);
1425                 x->name = xstrdup(name_b);
1426                 x->is_renamed = 1;
1427         }
1428         else {
1429                 x->from_name = NULL;
1430                 x->name = xstrdup(name_a);
1431         }
1432         return x;
1433 }
1434
1435 static void diffstat_consume(void *priv, char *line, unsigned long len)
1436 {
1437         struct diffstat_t *diffstat = priv;
1438         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1439
1440         if (line[0] == '+')
1441                 x->added++;
1442         else if (line[0] == '-')
1443                 x->deleted++;
1444 }
1445
1446 const char mime_boundary_leader[] = "------------";
1447
1448 static int scale_linear(int it, int width, int max_change)
1449 {
1450         if (!it)
1451                 return 0;
1452         /*
1453          * make sure that at least one '-' or '+' is printed if
1454          * there is any change to this path. The easiest way is to
1455          * scale linearly as if the alloted width is one column shorter
1456          * than it is, and then add 1 to the result.
1457          */
1458         return 1 + (it * (width - 1) / max_change);
1459 }
1460
1461 static void show_name(FILE *file,
1462                       const char *prefix, const char *name, int len)
1463 {
1464         fprintf(file, " %s%-*s |", prefix, len, name);
1465 }
1466
1467 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1468 {
1469         if (cnt <= 0)
1470                 return;
1471         fprintf(file, "%s", set);
1472         while (cnt--)
1473                 putc(ch, file);
1474         fprintf(file, "%s", reset);
1475 }
1476
1477 static void fill_print_name(struct diffstat_file *file)
1478 {
1479         char *pname;
1480
1481         if (file->print_name)
1482                 return;
1483
1484         if (!file->is_renamed) {
1485                 struct strbuf buf = STRBUF_INIT;
1486                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1487                         pname = strbuf_detach(&buf, NULL);
1488                 } else {
1489                         pname = file->name;
1490                         strbuf_release(&buf);
1491                 }
1492         } else {
1493                 pname = pprint_rename(file->from_name, file->name);
1494         }
1495         file->print_name = pname;
1496 }
1497
1498 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1499 {
1500         struct strbuf sb = STRBUF_INIT;
1501         int ret;
1502
1503         if (!files) {
1504                 assert(insertions == 0 && deletions == 0);
1505                 return fprintf(fp, "%s\n", " 0 files changed");
1506         }
1507
1508         strbuf_addf(&sb,
1509                     (files == 1) ? " %d file changed" : " %d files changed",
1510                     files);
1511
1512         /*
1513          * For binary diff, the caller may want to print "x files
1514          * changed" with insertions == 0 && deletions == 0.
1515          *
1516          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1517          * is probably less confusing (i.e skip over "2 files changed
1518          * but nothing about added/removed lines? Is this a bug in Git?").
1519          */
1520         if (insertions || deletions == 0) {
1521                 strbuf_addf(&sb,
1522                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1523                             insertions);
1524         }
1525
1526         if (deletions || insertions == 0) {
1527                 strbuf_addf(&sb,
1528                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1529                             deletions);
1530         }
1531         strbuf_addch(&sb, '\n');
1532         ret = fputs(sb.buf, fp);
1533         strbuf_release(&sb);
1534         return ret;
1535 }
1536
1537 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1538 {
1539         int i, len, add, del, adds = 0, dels = 0;
1540         uintmax_t max_change = 0, max_len = 0;
1541         int total_files = data->nr, count;
1542         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1543         const char *reset, *add_c, *del_c;
1544         const char *line_prefix = "";
1545         int extra_shown = 0;
1546
1547         if (data->nr == 0)
1548                 return;
1549
1550         line_prefix = diff_line_prefix(options);
1551         count = options->stat_count ? options->stat_count : data->nr;
1552
1553         reset = diff_get_color_opt(options, DIFF_RESET);
1554         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1555         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1556
1557         /*
1558          * Find the longest filename and max number of changes
1559          */
1560         for (i = 0; (i < count) && (i < data->nr); i++) {
1561                 struct diffstat_file *file = data->files[i];
1562                 uintmax_t change = file->added + file->deleted;
1563
1564                 if (!file->is_interesting && (change == 0)) {
1565                         count++; /* not shown == room for one more */
1566                         continue;
1567                 }
1568                 fill_print_name(file);
1569                 len = strlen(file->print_name);
1570                 if (max_len < len)
1571                         max_len = len;
1572
1573                 if (file->is_unmerged) {
1574                         /* "Unmerged" is 8 characters */
1575                         bin_width = bin_width < 8 ? 8 : bin_width;
1576                         continue;
1577                 }
1578                 if (file->is_binary) {
1579                         /* "Bin XXX -> YYY bytes" */
1580                         int w = 14 + decimal_width(file->added)
1581                                 + decimal_width(file->deleted);
1582                         bin_width = bin_width < w ? w : bin_width;
1583                         /* Display change counts aligned with "Bin" */
1584                         number_width = 3;
1585                         continue;
1586                 }
1587
1588                 if (max_change < change)
1589                         max_change = change;
1590         }
1591         count = i; /* where we can stop scanning in data->files[] */
1592
1593         /*
1594          * We have width = stat_width or term_columns() columns total.
1595          * We want a maximum of min(max_len, stat_name_width) for the name part.
1596          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1597          * We also need 1 for " " and 4 + decimal_width(max_change)
1598          * for " | NNNN " and one the empty column at the end, altogether
1599          * 6 + decimal_width(max_change).
1600          *
1601          * If there's not enough space, we will use the smaller of
1602          * stat_name_width (if set) and 5/8*width for the filename,
1603          * and the rest for constant elements + graph part, but no more
1604          * than stat_graph_width for the graph part.
1605          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1606          * for the standard terminal size).
1607          *
1608          * In other words: stat_width limits the maximum width, and
1609          * stat_name_width fixes the maximum width of the filename,
1610          * and is also used to divide available columns if there
1611          * aren't enough.
1612          *
1613          * Binary files are displayed with "Bin XXX -> YYY bytes"
1614          * instead of the change count and graph. This part is treated
1615          * similarly to the graph part, except that it is not
1616          * "scaled". If total width is too small to accommodate the
1617          * guaranteed minimum width of the filename part and the
1618          * separators and this message, this message will "overflow"
1619          * making the line longer than the maximum width.
1620          */
1621
1622         if (options->stat_width == -1)
1623                 width = term_columns() - options->output_prefix_length;
1624         else
1625                 width = options->stat_width ? options->stat_width : 80;
1626         number_width = decimal_width(max_change) > number_width ?
1627                 decimal_width(max_change) : number_width;
1628
1629         if (options->stat_graph_width == -1)
1630                 options->stat_graph_width = diff_stat_graph_width;
1631
1632         /*
1633          * Guarantee 3/8*16==6 for the graph part
1634          * and 5/8*16==10 for the filename part
1635          */
1636         if (width < 16 + 6 + number_width)
1637                 width = 16 + 6 + number_width;
1638
1639         /*
1640          * First assign sizes that are wanted, ignoring available width.
1641          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1642          * starting from "XXX" should fit in graph_width.
1643          */
1644         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1645         if (options->stat_graph_width &&
1646             options->stat_graph_width < graph_width)
1647                 graph_width = options->stat_graph_width;
1648
1649         name_width = (options->stat_name_width > 0 &&
1650                       options->stat_name_width < max_len) ?
1651                 options->stat_name_width : max_len;
1652
1653         /*
1654          * Adjust adjustable widths not to exceed maximum width
1655          */
1656         if (name_width + number_width + 6 + graph_width > width) {
1657                 if (graph_width > width * 3/8 - number_width - 6) {
1658                         graph_width = width * 3/8 - number_width - 6;
1659                         if (graph_width < 6)
1660                                 graph_width = 6;
1661                 }
1662
1663                 if (options->stat_graph_width &&
1664                     graph_width > options->stat_graph_width)
1665                         graph_width = options->stat_graph_width;
1666                 if (name_width > width - number_width - 6 - graph_width)
1667                         name_width = width - number_width - 6 - graph_width;
1668                 else
1669                         graph_width = width - number_width - 6 - name_width;
1670         }
1671
1672         /*
1673          * From here name_width is the width of the name area,
1674          * and graph_width is the width of the graph area.
1675          * max_change is used to scale graph properly.
1676          */
1677         for (i = 0; i < count; i++) {
1678                 const char *prefix = "";
1679                 struct diffstat_file *file = data->files[i];
1680                 char *name = file->print_name;
1681                 uintmax_t added = file->added;
1682                 uintmax_t deleted = file->deleted;
1683                 int name_len;
1684
1685                 if (!file->is_interesting && (added + deleted == 0))
1686                         continue;
1687
1688                 /*
1689                  * "scale" the filename
1690                  */
1691                 len = name_width;
1692                 name_len = strlen(name);
1693                 if (name_width < name_len) {
1694                         char *slash;
1695                         prefix = "...";
1696                         len -= 3;
1697                         name += name_len - len;
1698                         slash = strchr(name, '/');
1699                         if (slash)
1700                                 name = slash;
1701                 }
1702
1703                 if (file->is_binary) {
1704                         fprintf(options->file, "%s", line_prefix);
1705                         show_name(options->file, prefix, name, len);
1706                         fprintf(options->file, " %*s", number_width, "Bin");
1707                         if (!added && !deleted) {
1708                                 putc('\n', options->file);
1709                                 continue;
1710                         }
1711                         fprintf(options->file, " %s%"PRIuMAX"%s",
1712                                 del_c, deleted, reset);
1713                         fprintf(options->file, " -> ");
1714                         fprintf(options->file, "%s%"PRIuMAX"%s",
1715                                 add_c, added, reset);
1716                         fprintf(options->file, " bytes");
1717                         fprintf(options->file, "\n");
1718                         continue;
1719                 }
1720                 else if (file->is_unmerged) {
1721                         fprintf(options->file, "%s", line_prefix);
1722                         show_name(options->file, prefix, name, len);
1723                         fprintf(options->file, " Unmerged\n");
1724                         continue;
1725                 }
1726
1727                 /*
1728                  * scale the add/delete
1729                  */
1730                 add = added;
1731                 del = deleted;
1732
1733                 if (graph_width <= max_change) {
1734                         int total = scale_linear(add + del, graph_width, max_change);
1735                         if (total < 2 && add && del)
1736                                 /* width >= 2 due to the sanity check */
1737                                 total = 2;
1738                         if (add < del) {
1739                                 add = scale_linear(add, graph_width, max_change);
1740                                 del = total - add;
1741                         } else {
1742                                 del = scale_linear(del, graph_width, max_change);
1743                                 add = total - del;
1744                         }
1745                 }
1746                 fprintf(options->file, "%s", line_prefix);
1747                 show_name(options->file, prefix, name, len);
1748                 fprintf(options->file, " %*"PRIuMAX"%s",
1749                         number_width, added + deleted,
1750                         added + deleted ? " " : "");
1751                 show_graph(options->file, '+', add, add_c, reset);
1752                 show_graph(options->file, '-', del, del_c, reset);
1753                 fprintf(options->file, "\n");
1754         }
1755
1756         for (i = 0; i < data->nr; i++) {
1757                 struct diffstat_file *file = data->files[i];
1758                 uintmax_t added = file->added;
1759                 uintmax_t deleted = file->deleted;
1760
1761                 if (file->is_unmerged ||
1762                     (!file->is_interesting && (added + deleted == 0))) {
1763                         total_files--;
1764                         continue;
1765                 }
1766
1767                 if (!file->is_binary) {
1768                         adds += added;
1769                         dels += deleted;
1770                 }
1771                 if (i < count)
1772                         continue;
1773                 if (!extra_shown)
1774                         fprintf(options->file, "%s ...\n", line_prefix);
1775                 extra_shown = 1;
1776         }
1777         fprintf(options->file, "%s", line_prefix);
1778         print_stat_summary(options->file, total_files, adds, dels);
1779 }
1780
1781 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1782 {
1783         int i, adds = 0, dels = 0, total_files = data->nr;
1784
1785         if (data->nr == 0)
1786                 return;
1787
1788         for (i = 0; i < data->nr; i++) {
1789                 int added = data->files[i]->added;
1790                 int deleted= data->files[i]->deleted;
1791
1792                 if (data->files[i]->is_unmerged ||
1793                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1794                         total_files--;
1795                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1796                         adds += added;
1797                         dels += deleted;
1798                 }
1799         }
1800         fprintf(options->file, "%s", diff_line_prefix(options));
1801         print_stat_summary(options->file, total_files, adds, dels);
1802 }
1803
1804 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1805 {
1806         int i;
1807
1808         if (data->nr == 0)
1809                 return;
1810
1811         for (i = 0; i < data->nr; i++) {
1812                 struct diffstat_file *file = data->files[i];
1813
1814                 fprintf(options->file, "%s", diff_line_prefix(options));
1815
1816                 if (file->is_binary)
1817                         fprintf(options->file, "-\t-\t");
1818                 else
1819                         fprintf(options->file,
1820                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1821                                 file->added, file->deleted);
1822                 if (options->line_termination) {
1823                         fill_print_name(file);
1824                         if (!file->is_renamed)
1825                                 write_name_quoted(file->name, options->file,
1826                                                   options->line_termination);
1827                         else {
1828                                 fputs(file->print_name, options->file);
1829                                 putc(options->line_termination, options->file);
1830                         }
1831                 } else {
1832                         if (file->is_renamed) {
1833                                 putc('\0', options->file);
1834                                 write_name_quoted(file->from_name, options->file, '\0');
1835                         }
1836                         write_name_quoted(file->name, options->file, '\0');
1837                 }
1838         }
1839 }
1840
1841 struct dirstat_file {
1842         const char *name;
1843         unsigned long changed;
1844 };
1845
1846 struct dirstat_dir {
1847         struct dirstat_file *files;
1848         int alloc, nr, permille, cumulative;
1849 };
1850
1851 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1852                 unsigned long changed, const char *base, int baselen)
1853 {
1854         unsigned long this_dir = 0;
1855         unsigned int sources = 0;
1856         const char *line_prefix = diff_line_prefix(opt);
1857
1858         while (dir->nr) {
1859                 struct dirstat_file *f = dir->files;
1860                 int namelen = strlen(f->name);
1861                 unsigned long this;
1862                 char *slash;
1863
1864                 if (namelen < baselen)
1865                         break;
1866                 if (memcmp(f->name, base, baselen))
1867                         break;
1868                 slash = strchr(f->name + baselen, '/');
1869                 if (slash) {
1870                         int newbaselen = slash + 1 - f->name;
1871                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1872                         sources++;
1873                 } else {
1874                         this = f->changed;
1875                         dir->files++;
1876                         dir->nr--;
1877                         sources += 2;
1878                 }
1879                 this_dir += this;
1880         }
1881
1882         /*
1883          * We don't report dirstat's for
1884          *  - the top level
1885          *  - or cases where everything came from a single directory
1886          *    under this directory (sources == 1).
1887          */
1888         if (baselen && sources != 1) {
1889                 if (this_dir) {
1890                         int permille = this_dir * 1000 / changed;
1891                         if (permille >= dir->permille) {
1892                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1893                                         permille / 10, permille % 10, baselen, base);
1894                                 if (!dir->cumulative)
1895                                         return 0;
1896                         }
1897                 }
1898         }
1899         return this_dir;
1900 }
1901
1902 static int dirstat_compare(const void *_a, const void *_b)
1903 {
1904         const struct dirstat_file *a = _a;
1905         const struct dirstat_file *b = _b;
1906         return strcmp(a->name, b->name);
1907 }
1908
1909 static void show_dirstat(struct diff_options *options)
1910 {
1911         int i;
1912         unsigned long changed;
1913         struct dirstat_dir dir;
1914         struct diff_queue_struct *q = &diff_queued_diff;
1915
1916         dir.files = NULL;
1917         dir.alloc = 0;
1918         dir.nr = 0;
1919         dir.permille = options->dirstat_permille;
1920         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1921
1922         changed = 0;
1923         for (i = 0; i < q->nr; i++) {
1924                 struct diff_filepair *p = q->queue[i];
1925                 const char *name;
1926                 unsigned long copied, added, damage;
1927                 int content_changed;
1928
1929                 name = p->two->path ? p->two->path : p->one->path;
1930
1931                 if (p->one->sha1_valid && p->two->sha1_valid)
1932                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1933                 else
1934                         content_changed = 1;
1935
1936                 if (!content_changed) {
1937                         /*
1938                          * The SHA1 has not changed, so pre-/post-content is
1939                          * identical. We can therefore skip looking at the
1940                          * file contents altogether.
1941                          */
1942                         damage = 0;
1943                         goto found_damage;
1944                 }
1945
1946                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1947                         /*
1948                          * In --dirstat-by-file mode, we don't really need to
1949                          * look at the actual file contents at all.
1950                          * The fact that the SHA1 changed is enough for us to
1951                          * add this file to the list of results
1952                          * (with each file contributing equal damage).
1953                          */
1954                         damage = 1;
1955                         goto found_damage;
1956                 }
1957
1958                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1959                         diff_populate_filespec(p->one, 0);
1960                         diff_populate_filespec(p->two, 0);
1961                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1962                                                &copied, &added);
1963                         diff_free_filespec_data(p->one);
1964                         diff_free_filespec_data(p->two);
1965                 } else if (DIFF_FILE_VALID(p->one)) {
1966                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1967                         copied = added = 0;
1968                         diff_free_filespec_data(p->one);
1969                 } else if (DIFF_FILE_VALID(p->two)) {
1970                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1971                         copied = 0;
1972                         added = p->two->size;
1973                         diff_free_filespec_data(p->two);
1974                 } else
1975                         continue;
1976
1977                 /*
1978                  * Original minus copied is the removed material,
1979                  * added is the new material.  They are both damages
1980                  * made to the preimage.
1981                  * If the resulting damage is zero, we know that
1982                  * diffcore_count_changes() considers the two entries to
1983                  * be identical, but since content_changed is true, we
1984                  * know that there must have been _some_ kind of change,
1985                  * so we force all entries to have damage > 0.
1986                  */
1987                 damage = (p->one->size - copied) + added;
1988                 if (!damage)
1989                         damage = 1;
1990
1991 found_damage:
1992                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1993                 dir.files[dir.nr].name = name;
1994                 dir.files[dir.nr].changed = damage;
1995                 changed += damage;
1996                 dir.nr++;
1997         }
1998
1999         /* This can happen even with many files, if everything was renames */
2000         if (!changed)
2001                 return;
2002
2003         /* Show all directories with more than x% of the changes */
2004         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2005         gather_dirstat(options, &dir, changed, "", 0);
2006 }
2007
2008 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2009 {
2010         int i;
2011         unsigned long changed;
2012         struct dirstat_dir dir;
2013
2014         if (data->nr == 0)
2015                 return;
2016
2017         dir.files = NULL;
2018         dir.alloc = 0;
2019         dir.nr = 0;
2020         dir.permille = options->dirstat_permille;
2021         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2022
2023         changed = 0;
2024         for (i = 0; i < data->nr; i++) {
2025                 struct diffstat_file *file = data->files[i];
2026                 unsigned long damage = file->added + file->deleted;
2027                 if (file->is_binary)
2028                         /*
2029                          * binary files counts bytes, not lines. Must find some
2030                          * way to normalize binary bytes vs. textual lines.
2031                          * The following heuristic assumes that there are 64
2032                          * bytes per "line".
2033                          * This is stupid and ugly, but very cheap...
2034                          */
2035                         damage = (damage + 63) / 64;
2036                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2037                 dir.files[dir.nr].name = file->name;
2038                 dir.files[dir.nr].changed = damage;
2039                 changed += damage;
2040                 dir.nr++;
2041         }
2042
2043         /* This can happen even with many files, if everything was renames */
2044         if (!changed)
2045                 return;
2046
2047         /* Show all directories with more than x% of the changes */
2048         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2049         gather_dirstat(options, &dir, changed, "", 0);
2050 }
2051
2052 static void free_diffstat_info(struct diffstat_t *diffstat)
2053 {
2054         int i;
2055         for (i = 0; i < diffstat->nr; i++) {
2056                 struct diffstat_file *f = diffstat->files[i];
2057                 if (f->name != f->print_name)
2058                         free(f->print_name);
2059                 free(f->name);
2060                 free(f->from_name);
2061                 free(f);
2062         }
2063         free(diffstat->files);
2064 }
2065
2066 struct checkdiff_t {
2067         const char *filename;
2068         int lineno;
2069         int conflict_marker_size;
2070         struct diff_options *o;
2071         unsigned ws_rule;
2072         unsigned status;
2073 };
2074
2075 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2076 {
2077         char firstchar;
2078         int cnt;
2079
2080         if (len < marker_size + 1)
2081                 return 0;
2082         firstchar = line[0];
2083         switch (firstchar) {
2084         case '=': case '>': case '<': case '|':
2085                 break;
2086         default:
2087                 return 0;
2088         }
2089         for (cnt = 1; cnt < marker_size; cnt++)
2090                 if (line[cnt] != firstchar)
2091                         return 0;
2092         /* line[1] thru line[marker_size-1] are same as firstchar */
2093         if (len < marker_size + 1 || !isspace(line[marker_size]))
2094                 return 0;
2095         return 1;
2096 }
2097
2098 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2099 {
2100         struct checkdiff_t *data = priv;
2101         int marker_size = data->conflict_marker_size;
2102         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2103         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2104         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2105         char *err;
2106         const char *line_prefix;
2107
2108         assert(data->o);
2109         line_prefix = diff_line_prefix(data->o);
2110
2111         if (line[0] == '+') {
2112                 unsigned bad;
2113                 data->lineno++;
2114                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2115                         data->status |= 1;
2116                         fprintf(data->o->file,
2117                                 "%s%s:%d: leftover conflict marker\n",
2118                                 line_prefix, data->filename, data->lineno);
2119                 }
2120                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2121                 if (!bad)
2122                         return;
2123                 data->status |= bad;
2124                 err = whitespace_error_string(bad);
2125                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2126                         line_prefix, data->filename, data->lineno, err);
2127                 free(err);
2128                 emit_line(data->o, set, reset, line, 1);
2129                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2130                               data->o->file, set, reset, ws);
2131         } else if (line[0] == ' ') {
2132                 data->lineno++;
2133         } else if (line[0] == '@') {
2134                 char *plus = strchr(line, '+');
2135                 if (plus)
2136                         data->lineno = strtol(plus, NULL, 10) - 1;
2137                 else
2138                         die("invalid diff");
2139         }
2140 }
2141
2142 static unsigned char *deflate_it(char *data,
2143                                  unsigned long size,
2144                                  unsigned long *result_size)
2145 {
2146         int bound;
2147         unsigned char *deflated;
2148         git_zstream stream;
2149
2150         git_deflate_init(&stream, zlib_compression_level);
2151         bound = git_deflate_bound(&stream, size);
2152         deflated = xmalloc(bound);
2153         stream.next_out = deflated;
2154         stream.avail_out = bound;
2155
2156         stream.next_in = (unsigned char *)data;
2157         stream.avail_in = size;
2158         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2159                 ; /* nothing */
2160         git_deflate_end(&stream);
2161         *result_size = stream.total_out;
2162         return deflated;
2163 }
2164
2165 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2166                                   const char *prefix)
2167 {
2168         void *cp;
2169         void *delta;
2170         void *deflated;
2171         void *data;
2172         unsigned long orig_size;
2173         unsigned long delta_size;
2174         unsigned long deflate_size;
2175         unsigned long data_size;
2176
2177         /* We could do deflated delta, or we could do just deflated two,
2178          * whichever is smaller.
2179          */
2180         delta = NULL;
2181         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2182         if (one->size && two->size) {
2183                 delta = diff_delta(one->ptr, one->size,
2184                                    two->ptr, two->size,
2185                                    &delta_size, deflate_size);
2186                 if (delta) {
2187                         void *to_free = delta;
2188                         orig_size = delta_size;
2189                         delta = deflate_it(delta, delta_size, &delta_size);
2190                         free(to_free);
2191                 }
2192         }
2193
2194         if (delta && delta_size < deflate_size) {
2195                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2196                 free(deflated);
2197                 data = delta;
2198                 data_size = delta_size;
2199         }
2200         else {
2201                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2202                 free(delta);
2203                 data = deflated;
2204                 data_size = deflate_size;
2205         }
2206
2207         /* emit data encoded in base85 */
2208         cp = data;
2209         while (data_size) {
2210                 int bytes = (52 < data_size) ? 52 : data_size;
2211                 char line[70];
2212                 data_size -= bytes;
2213                 if (bytes <= 26)
2214                         line[0] = bytes + 'A' - 1;
2215                 else
2216                         line[0] = bytes - 26 + 'a' - 1;
2217                 encode_85(line + 1, cp, bytes);
2218                 cp = (char *) cp + bytes;
2219                 fprintf(file, "%s", prefix);
2220                 fputs(line, file);
2221                 fputc('\n', file);
2222         }
2223         fprintf(file, "%s\n", prefix);
2224         free(data);
2225 }
2226
2227 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2228                              const char *prefix)
2229 {
2230         fprintf(file, "%sGIT binary patch\n", prefix);
2231         emit_binary_diff_body(file, one, two, prefix);
2232         emit_binary_diff_body(file, two, one, prefix);
2233 }
2234
2235 int diff_filespec_is_binary(struct diff_filespec *one)
2236 {
2237         if (one->is_binary == -1) {
2238                 diff_filespec_load_driver(one);
2239                 if (one->driver->binary != -1)
2240                         one->is_binary = one->driver->binary;
2241                 else {
2242                         if (!one->data && DIFF_FILE_VALID(one))
2243                                 diff_populate_filespec(one, CHECK_BINARY);
2244                         if (one->is_binary == -1 && one->data)
2245                                 one->is_binary = buffer_is_binary(one->data,
2246                                                 one->size);
2247                         if (one->is_binary == -1)
2248                                 one->is_binary = 0;
2249                 }
2250         }
2251         return one->is_binary;
2252 }
2253
2254 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2255 {
2256         diff_filespec_load_driver(one);
2257         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2258 }
2259
2260 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2261 {
2262         if (!options->a_prefix)
2263                 options->a_prefix = a;
2264         if (!options->b_prefix)
2265                 options->b_prefix = b;
2266 }
2267
2268 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2269 {
2270         if (!DIFF_FILE_VALID(one))
2271                 return NULL;
2272
2273         diff_filespec_load_driver(one);
2274         return userdiff_get_textconv(one->driver);
2275 }
2276
2277 static void builtin_diff(const char *name_a,
2278                          const char *name_b,
2279                          struct diff_filespec *one,
2280                          struct diff_filespec *two,
2281                          const char *xfrm_msg,
2282                          int must_show_header,
2283                          struct diff_options *o,
2284                          int complete_rewrite)
2285 {
2286         mmfile_t mf1, mf2;
2287         const char *lbl[2];
2288         char *a_one, *b_two;
2289         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2290         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2291         const char *a_prefix, *b_prefix;
2292         struct userdiff_driver *textconv_one = NULL;
2293         struct userdiff_driver *textconv_two = NULL;
2294         struct strbuf header = STRBUF_INIT;
2295         const char *line_prefix = diff_line_prefix(o);
2296
2297         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2298                         (!one->mode || S_ISGITLINK(one->mode)) &&
2299                         (!two->mode || S_ISGITLINK(two->mode))) {
2300                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2301                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2302                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2303                                 line_prefix,
2304                                 one->sha1, two->sha1, two->dirty_submodule,
2305                                 meta, del, add, reset);
2306                 return;
2307         }
2308
2309         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2310                 textconv_one = get_textconv(one);
2311                 textconv_two = get_textconv(two);
2312         }
2313
2314         diff_set_mnemonic_prefix(o, "a/", "b/");
2315         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2316                 a_prefix = o->b_prefix;
2317                 b_prefix = o->a_prefix;
2318         } else {
2319                 a_prefix = o->a_prefix;
2320                 b_prefix = o->b_prefix;
2321         }
2322
2323         /* Never use a non-valid filename anywhere if at all possible */
2324         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2325         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2326
2327         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2328         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2329         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2330         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2331         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2332         if (lbl[0][0] == '/') {
2333                 /* /dev/null */
2334                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2335                 if (xfrm_msg)
2336                         strbuf_addstr(&header, xfrm_msg);
2337                 must_show_header = 1;
2338         }
2339         else if (lbl[1][0] == '/') {
2340                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2341                 if (xfrm_msg)
2342                         strbuf_addstr(&header, xfrm_msg);
2343                 must_show_header = 1;
2344         }
2345         else {
2346                 if (one->mode != two->mode) {
2347                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2348                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2349                         must_show_header = 1;
2350                 }
2351                 if (xfrm_msg)
2352                         strbuf_addstr(&header, xfrm_msg);
2353
2354                 /*
2355                  * we do not run diff between different kind
2356                  * of objects.
2357                  */
2358                 if ((one->mode ^ two->mode) & S_IFMT)
2359                         goto free_ab_and_return;
2360                 if (complete_rewrite &&
2361                     (textconv_one || !diff_filespec_is_binary(one)) &&
2362                     (textconv_two || !diff_filespec_is_binary(two))) {
2363                         fprintf(o->file, "%s", header.buf);
2364                         strbuf_reset(&header);
2365                         emit_rewrite_diff(name_a, name_b, one, two,
2366                                                 textconv_one, textconv_two, o);
2367                         o->found_changes = 1;
2368                         goto free_ab_and_return;
2369                 }
2370         }
2371
2372         if (o->irreversible_delete && lbl[1][0] == '/') {
2373                 fprintf(o->file, "%s", header.buf);
2374                 strbuf_reset(&header);
2375                 goto free_ab_and_return;
2376         } else if (!DIFF_OPT_TST(o, TEXT) &&
2377             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2378               (!textconv_two && diff_filespec_is_binary(two)) )) {
2379                 if (!one->data && !two->data &&
2380                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2381                     !DIFF_OPT_TST(o, BINARY)) {
2382                         if (!hashcmp(one->sha1, two->sha1)) {
2383                                 if (must_show_header)
2384                                         fprintf(o->file, "%s", header.buf);
2385                                 goto free_ab_and_return;
2386                         }
2387                         fprintf(o->file, "%s", header.buf);
2388                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2389                                 line_prefix, lbl[0], lbl[1]);
2390                         goto free_ab_and_return;
2391                 }
2392                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2393                         die("unable to read files to diff");
2394                 /* Quite common confusing case */
2395                 if (mf1.size == mf2.size &&
2396                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2397                         if (must_show_header)
2398                                 fprintf(o->file, "%s", header.buf);
2399                         goto free_ab_and_return;
2400                 }
2401                 fprintf(o->file, "%s", header.buf);
2402                 strbuf_reset(&header);
2403                 if (DIFF_OPT_TST(o, BINARY))
2404                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2405                 else
2406                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2407                                 line_prefix, lbl[0], lbl[1]);
2408                 o->found_changes = 1;
2409         } else {
2410                 /* Crazy xdl interfaces.. */
2411                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2412                 const char *v;
2413                 xpparam_t xpp;
2414                 xdemitconf_t xecfg;
2415                 struct emit_callback ecbdata;
2416                 const struct userdiff_funcname *pe;
2417
2418                 if (must_show_header) {
2419                         fprintf(o->file, "%s", header.buf);
2420                         strbuf_reset(&header);
2421                 }
2422
2423                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2424                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2425
2426                 pe = diff_funcname_pattern(one);
2427                 if (!pe)
2428                         pe = diff_funcname_pattern(two);
2429
2430                 memset(&xpp, 0, sizeof(xpp));
2431                 memset(&xecfg, 0, sizeof(xecfg));
2432                 memset(&ecbdata, 0, sizeof(ecbdata));
2433                 ecbdata.label_path = lbl;
2434                 ecbdata.color_diff = want_color(o->use_color);
2435                 ecbdata.found_changesp = &o->found_changes;
2436                 ecbdata.ws_rule = whitespace_rule(name_b);
2437                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2438                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2439                 ecbdata.opt = o;
2440                 ecbdata.header = header.len ? &header : NULL;
2441                 xpp.flags = o->xdl_opts;
2442                 xecfg.ctxlen = o->context;
2443                 xecfg.interhunkctxlen = o->interhunkcontext;
2444                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2445                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2446                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2447                 if (pe)
2448                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2449                 if (!diffopts)
2450                         ;
2451                 else if (skip_prefix(diffopts, "--unified=", &v))
2452                         xecfg.ctxlen = strtoul(v, NULL, 10);
2453                 else if (skip_prefix(diffopts, "-u", &v))
2454                         xecfg.ctxlen = strtoul(v, NULL, 10);
2455                 if (o->word_diff)
2456                         init_diff_words_data(&ecbdata, o, one, two);
2457                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2458                                   &xpp, &xecfg))
2459                         die("unable to generate diff for %s", one->path);
2460                 if (o->word_diff)
2461                         free_diff_words_data(&ecbdata);
2462                 if (textconv_one)
2463                         free(mf1.ptr);
2464                 if (textconv_two)
2465                         free(mf2.ptr);
2466                 xdiff_clear_find_func(&xecfg);
2467         }
2468
2469  free_ab_and_return:
2470         strbuf_release(&header);
2471         diff_free_filespec_data(one);
2472         diff_free_filespec_data(two);
2473         free(a_one);
2474         free(b_two);
2475         return;
2476 }
2477
2478 static void builtin_diffstat(const char *name_a, const char *name_b,
2479                              struct diff_filespec *one,
2480                              struct diff_filespec *two,
2481                              struct diffstat_t *diffstat,
2482                              struct diff_options *o,
2483                              struct diff_filepair *p)
2484 {
2485         mmfile_t mf1, mf2;
2486         struct diffstat_file *data;
2487         int same_contents;
2488         int complete_rewrite = 0;
2489
2490         if (!DIFF_PAIR_UNMERGED(p)) {
2491                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2492                         complete_rewrite = 1;
2493         }
2494
2495         data = diffstat_add(diffstat, name_a, name_b);
2496         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2497
2498         if (!one || !two) {
2499                 data->is_unmerged = 1;
2500                 return;
2501         }
2502
2503         same_contents = !hashcmp(one->sha1, two->sha1);
2504
2505         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2506                 data->is_binary = 1;
2507                 if (same_contents) {
2508                         data->added = 0;
2509                         data->deleted = 0;
2510                 } else {
2511                         data->added = diff_filespec_size(two);
2512                         data->deleted = diff_filespec_size(one);
2513                 }
2514         }
2515
2516         else if (complete_rewrite) {
2517                 diff_populate_filespec(one, 0);
2518                 diff_populate_filespec(two, 0);
2519                 data->deleted = count_lines(one->data, one->size);
2520                 data->added = count_lines(two->data, two->size);
2521         }
2522
2523         else if (!same_contents) {
2524                 /* Crazy xdl interfaces.. */
2525                 xpparam_t xpp;
2526                 xdemitconf_t xecfg;
2527
2528                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2529                         die("unable to read files to diff");
2530
2531                 memset(&xpp, 0, sizeof(xpp));
2532                 memset(&xecfg, 0, sizeof(xecfg));
2533                 xpp.flags = o->xdl_opts;
2534                 xecfg.ctxlen = o->context;
2535                 xecfg.interhunkctxlen = o->interhunkcontext;
2536                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2537                                   &xpp, &xecfg))
2538                         die("unable to generate diffstat for %s", one->path);
2539         }
2540
2541         diff_free_filespec_data(one);
2542         diff_free_filespec_data(two);
2543 }
2544
2545 static void builtin_checkdiff(const char *name_a, const char *name_b,
2546                               const char *attr_path,
2547                               struct diff_filespec *one,
2548                               struct diff_filespec *two,
2549                               struct diff_options *o)
2550 {
2551         mmfile_t mf1, mf2;
2552         struct checkdiff_t data;
2553
2554         if (!two)
2555                 return;
2556
2557         memset(&data, 0, sizeof(data));
2558         data.filename = name_b ? name_b : name_a;
2559         data.lineno = 0;
2560         data.o = o;
2561         data.ws_rule = whitespace_rule(attr_path);
2562         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2563
2564         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2565                 die("unable to read files to diff");
2566
2567         /*
2568          * All the other codepaths check both sides, but not checking
2569          * the "old" side here is deliberate.  We are checking the newly
2570          * introduced changes, and as long as the "new" side is text, we
2571          * can and should check what it introduces.
2572          */
2573         if (diff_filespec_is_binary(two))
2574                 goto free_and_return;
2575         else {
2576                 /* Crazy xdl interfaces.. */
2577                 xpparam_t xpp;
2578                 xdemitconf_t xecfg;
2579
2580                 memset(&xpp, 0, sizeof(xpp));
2581                 memset(&xecfg, 0, sizeof(xecfg));
2582                 xecfg.ctxlen = 1; /* at least one context line */
2583                 xpp.flags = 0;
2584                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2585                                   &xpp, &xecfg))
2586                         die("unable to generate checkdiff for %s", one->path);
2587
2588                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2589                         struct emit_callback ecbdata;
2590                         int blank_at_eof;
2591
2592                         ecbdata.ws_rule = data.ws_rule;
2593                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2594                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2595
2596                         if (blank_at_eof) {
2597                                 static char *err;
2598                                 if (!err)
2599                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2600                                 fprintf(o->file, "%s:%d: %s.\n",
2601                                         data.filename, blank_at_eof, err);
2602                                 data.status = 1; /* report errors */
2603                         }
2604                 }
2605         }
2606  free_and_return:
2607         diff_free_filespec_data(one);
2608         diff_free_filespec_data(two);
2609         if (data.status)
2610                 DIFF_OPT_SET(o, CHECK_FAILED);
2611 }
2612
2613 struct diff_filespec *alloc_filespec(const char *path)
2614 {
2615         struct diff_filespec *spec;
2616
2617         FLEXPTR_ALLOC_STR(spec, path, path);
2618         spec->count = 1;
2619         spec->is_binary = -1;
2620         return spec;
2621 }
2622
2623 void free_filespec(struct diff_filespec *spec)
2624 {
2625         if (!--spec->count) {
2626                 diff_free_filespec_data(spec);
2627                 free(spec);
2628         }
2629 }
2630
2631 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2632                    int sha1_valid, unsigned short mode)
2633 {
2634         if (mode) {
2635                 spec->mode = canon_mode(mode);
2636                 hashcpy(spec->sha1, sha1);
2637                 spec->sha1_valid = sha1_valid;
2638         }
2639 }
2640
2641 /*
2642  * Given a name and sha1 pair, if the index tells us the file in
2643  * the work tree has that object contents, return true, so that
2644  * prepare_temp_file() does not have to inflate and extract.
2645  */
2646 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2647 {
2648         const struct cache_entry *ce;
2649         struct stat st;
2650         int pos, len;
2651
2652         /*
2653          * We do not read the cache ourselves here, because the
2654          * benchmark with my previous version that always reads cache
2655          * shows that it makes things worse for diff-tree comparing
2656          * two linux-2.6 kernel trees in an already checked out work
2657          * tree.  This is because most diff-tree comparisons deal with
2658          * only a small number of files, while reading the cache is
2659          * expensive for a large project, and its cost outweighs the
2660          * savings we get by not inflating the object to a temporary
2661          * file.  Practically, this code only helps when we are used
2662          * by diff-cache --cached, which does read the cache before
2663          * calling us.
2664          */
2665         if (!active_cache)
2666                 return 0;
2667
2668         /* We want to avoid the working directory if our caller
2669          * doesn't need the data in a normal file, this system
2670          * is rather slow with its stat/open/mmap/close syscalls,
2671          * and the object is contained in a pack file.  The pack
2672          * is probably already open and will be faster to obtain
2673          * the data through than the working directory.  Loose
2674          * objects however would tend to be slower as they need
2675          * to be individually opened and inflated.
2676          */
2677         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2678                 return 0;
2679
2680         len = strlen(name);
2681         pos = cache_name_pos(name, len);
2682         if (pos < 0)
2683                 return 0;
2684         ce = active_cache[pos];
2685
2686         /*
2687          * This is not the sha1 we are looking for, or
2688          * unreusable because it is not a regular file.
2689          */
2690         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2691                 return 0;
2692
2693         /*
2694          * If ce is marked as "assume unchanged", there is no
2695          * guarantee that work tree matches what we are looking for.
2696          */
2697         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2698                 return 0;
2699
2700         /*
2701          * If ce matches the file in the work tree, we can reuse it.
2702          */
2703         if (ce_uptodate(ce) ||
2704             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2705                 return 1;
2706
2707         return 0;
2708 }
2709
2710 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2711 {
2712         struct strbuf buf = STRBUF_INIT;
2713         char *dirty = "";
2714
2715         /* Are we looking at the work tree? */
2716         if (s->dirty_submodule)
2717                 dirty = "-dirty";
2718
2719         strbuf_addf(&buf, "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2720         s->size = buf.len;
2721         if (size_only) {
2722                 s->data = NULL;
2723                 strbuf_release(&buf);
2724         } else {
2725                 s->data = strbuf_detach(&buf, NULL);
2726                 s->should_free = 1;
2727         }
2728         return 0;
2729 }
2730
2731 /*
2732  * While doing rename detection and pickaxe operation, we may need to
2733  * grab the data for the blob (or file) for our own in-core comparison.
2734  * diff_filespec has data and size fields for this purpose.
2735  */
2736 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2737 {
2738         int size_only = flags & CHECK_SIZE_ONLY;
2739         int err = 0;
2740         /*
2741          * demote FAIL to WARN to allow inspecting the situation
2742          * instead of refusing.
2743          */
2744         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2745                                     ? SAFE_CRLF_WARN
2746                                     : safe_crlf);
2747
2748         if (!DIFF_FILE_VALID(s))
2749                 die("internal error: asking to populate invalid file.");
2750         if (S_ISDIR(s->mode))
2751                 return -1;
2752
2753         if (s->data)
2754                 return 0;
2755
2756         if (size_only && 0 < s->size)
2757                 return 0;
2758
2759         if (S_ISGITLINK(s->mode))
2760                 return diff_populate_gitlink(s, size_only);
2761
2762         if (!s->sha1_valid ||
2763             reuse_worktree_file(s->path, s->sha1, 0)) {
2764                 struct strbuf buf = STRBUF_INIT;
2765                 struct stat st;
2766                 int fd;
2767
2768                 if (lstat(s->path, &st) < 0) {
2769                         if (errno == ENOENT) {
2770                         err_empty:
2771                                 err = -1;
2772                         empty:
2773                                 s->data = (char *)"";
2774                                 s->size = 0;
2775                                 return err;
2776                         }
2777                 }
2778                 s->size = xsize_t(st.st_size);
2779                 if (!s->size)
2780                         goto empty;
2781                 if (S_ISLNK(st.st_mode)) {
2782                         struct strbuf sb = STRBUF_INIT;
2783
2784                         if (strbuf_readlink(&sb, s->path, s->size))
2785                                 goto err_empty;
2786                         s->size = sb.len;
2787                         s->data = strbuf_detach(&sb, NULL);
2788                         s->should_free = 1;
2789                         return 0;
2790                 }
2791                 if (size_only)
2792                         return 0;
2793                 if ((flags & CHECK_BINARY) &&
2794                     s->size > big_file_threshold && s->is_binary == -1) {
2795                         s->is_binary = 1;
2796                         return 0;
2797                 }
2798                 fd = open(s->path, O_RDONLY);
2799                 if (fd < 0)
2800                         goto err_empty;
2801                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2802                 close(fd);
2803                 s->should_munmap = 1;
2804
2805                 /*
2806                  * Convert from working tree format to canonical git format
2807                  */
2808                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2809                         size_t size = 0;
2810                         munmap(s->data, s->size);
2811                         s->should_munmap = 0;
2812                         s->data = strbuf_detach(&buf, &size);
2813                         s->size = size;
2814                         s->should_free = 1;
2815                 }
2816         }
2817         else {
2818                 enum object_type type;
2819                 if (size_only || (flags & CHECK_BINARY)) {
2820                         type = sha1_object_info(s->sha1, &s->size);
2821                         if (type < 0)
2822                                 die("unable to read %s", sha1_to_hex(s->sha1));
2823                         if (size_only)
2824                                 return 0;
2825                         if (s->size > big_file_threshold && s->is_binary == -1) {
2826                                 s->is_binary = 1;
2827                                 return 0;
2828                         }
2829                 }
2830                 s->data = read_sha1_file(s->sha1, &type, &s->size);
2831                 if (!s->data)
2832                         die("unable to read %s", sha1_to_hex(s->sha1));
2833                 s->should_free = 1;
2834         }
2835         return 0;
2836 }
2837
2838 void diff_free_filespec_blob(struct diff_filespec *s)
2839 {
2840         if (s->should_free)
2841                 free(s->data);
2842         else if (s->should_munmap)
2843                 munmap(s->data, s->size);
2844
2845         if (s->should_free || s->should_munmap) {
2846                 s->should_free = s->should_munmap = 0;
2847                 s->data = NULL;
2848         }
2849 }
2850
2851 void diff_free_filespec_data(struct diff_filespec *s)
2852 {
2853         diff_free_filespec_blob(s);
2854         free(s->cnt_data);
2855         s->cnt_data = NULL;
2856 }
2857
2858 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2859                            void *blob,
2860                            unsigned long size,
2861                            const unsigned char *sha1,
2862                            int mode)
2863 {
2864         int fd;
2865         struct strbuf buf = STRBUF_INIT;
2866         struct strbuf template = STRBUF_INIT;
2867         char *path_dup = xstrdup(path);
2868         const char *base = basename(path_dup);
2869
2870         /* Generate "XXXXXX_basename.ext" */
2871         strbuf_addstr(&template, "XXXXXX_");
2872         strbuf_addstr(&template, base);
2873
2874         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2875         if (fd < 0)
2876                 die_errno("unable to create temp-file");
2877         if (convert_to_working_tree(path,
2878                         (const char *)blob, (size_t)size, &buf)) {
2879                 blob = buf.buf;
2880                 size = buf.len;
2881         }
2882         if (write_in_full(fd, blob, size) != size)
2883                 die_errno("unable to write temp-file");
2884         close_tempfile(&temp->tempfile);
2885         temp->name = get_tempfile_path(&temp->tempfile);
2886         sha1_to_hex_r(temp->hex, sha1);
2887         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
2888         strbuf_release(&buf);
2889         strbuf_release(&template);
2890         free(path_dup);
2891 }
2892
2893 static struct diff_tempfile *prepare_temp_file(const char *name,
2894                 struct diff_filespec *one)
2895 {
2896         struct diff_tempfile *temp = claim_diff_tempfile();
2897
2898         if (!DIFF_FILE_VALID(one)) {
2899         not_a_valid_file:
2900                 /* A '-' entry produces this for file-2, and
2901                  * a '+' entry produces this for file-1.
2902                  */
2903                 temp->name = "/dev/null";
2904                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
2905                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
2906                 return temp;
2907         }
2908
2909         if (!S_ISGITLINK(one->mode) &&
2910             (!one->sha1_valid ||
2911              reuse_worktree_file(name, one->sha1, 1))) {
2912                 struct stat st;
2913                 if (lstat(name, &st) < 0) {
2914                         if (errno == ENOENT)
2915                                 goto not_a_valid_file;
2916                         die_errno("stat(%s)", name);
2917                 }
2918                 if (S_ISLNK(st.st_mode)) {
2919                         struct strbuf sb = STRBUF_INIT;
2920                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2921                                 die_errno("readlink(%s)", name);
2922                         prep_temp_blob(name, temp, sb.buf, sb.len,
2923                                        (one->sha1_valid ?
2924                                         one->sha1 : null_sha1),
2925                                        (one->sha1_valid ?
2926                                         one->mode : S_IFLNK));
2927                         strbuf_release(&sb);
2928                 }
2929                 else {
2930                         /* we can borrow from the file in the work tree */
2931                         temp->name = name;
2932                         if (!one->sha1_valid)
2933                                 sha1_to_hex_r(temp->hex, null_sha1);
2934                         else
2935                                 sha1_to_hex_r(temp->hex, one->sha1);
2936                         /* Even though we may sometimes borrow the
2937                          * contents from the work tree, we always want
2938                          * one->mode.  mode is trustworthy even when
2939                          * !(one->sha1_valid), as long as
2940                          * DIFF_FILE_VALID(one).
2941                          */
2942                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
2943                 }
2944                 return temp;
2945         }
2946         else {
2947                 if (diff_populate_filespec(one, 0))
2948                         die("cannot read data blob for %s", one->path);
2949                 prep_temp_blob(name, temp, one->data, one->size,
2950                                one->sha1, one->mode);
2951         }
2952         return temp;
2953 }
2954
2955 static void add_external_diff_name(struct argv_array *argv,
2956                                    const char *name,
2957                                    struct diff_filespec *df)
2958 {
2959         struct diff_tempfile *temp = prepare_temp_file(name, df);
2960         argv_array_push(argv, temp->name);
2961         argv_array_push(argv, temp->hex);
2962         argv_array_push(argv, temp->mode);
2963 }
2964
2965 /* An external diff command takes:
2966  *
2967  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2968  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2969  *
2970  */
2971 static void run_external_diff(const char *pgm,
2972                               const char *name,
2973                               const char *other,
2974                               struct diff_filespec *one,
2975                               struct diff_filespec *two,
2976                               const char *xfrm_msg,
2977                               int complete_rewrite,
2978                               struct diff_options *o)
2979 {
2980         struct argv_array argv = ARGV_ARRAY_INIT;
2981         struct argv_array env = ARGV_ARRAY_INIT;
2982         struct diff_queue_struct *q = &diff_queued_diff;
2983
2984         argv_array_push(&argv, pgm);
2985         argv_array_push(&argv, name);
2986
2987         if (one && two) {
2988                 add_external_diff_name(&argv, name, one);
2989                 if (!other)
2990                         add_external_diff_name(&argv, name, two);
2991                 else {
2992                         add_external_diff_name(&argv, other, two);
2993                         argv_array_push(&argv, other);
2994                         argv_array_push(&argv, xfrm_msg);
2995                 }
2996         }
2997
2998         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
2999         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3000
3001         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3002                 die(_("external diff died, stopping at %s"), name);
3003
3004         remove_tempfile();
3005         argv_array_clear(&argv);
3006         argv_array_clear(&env);
3007 }
3008
3009 static int similarity_index(struct diff_filepair *p)
3010 {
3011         return p->score * 100 / MAX_SCORE;
3012 }
3013
3014 static void fill_metainfo(struct strbuf *msg,
3015                           const char *name,
3016                           const char *other,
3017                           struct diff_filespec *one,
3018                           struct diff_filespec *two,
3019                           struct diff_options *o,
3020                           struct diff_filepair *p,
3021                           int *must_show_header,
3022                           int use_color)
3023 {
3024         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3025         const char *reset = diff_get_color(use_color, DIFF_RESET);
3026         const char *line_prefix = diff_line_prefix(o);
3027
3028         *must_show_header = 1;
3029         strbuf_init(msg, PATH_MAX * 2 + 300);
3030         switch (p->status) {
3031         case DIFF_STATUS_COPIED:
3032                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3033                             line_prefix, set, similarity_index(p));
3034                 strbuf_addf(msg, "%s\n%s%scopy from ",
3035                             reset,  line_prefix, set);
3036                 quote_c_style(name, msg, NULL, 0);
3037                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3038                 quote_c_style(other, msg, NULL, 0);
3039                 strbuf_addf(msg, "%s\n", reset);
3040                 break;
3041         case DIFF_STATUS_RENAMED:
3042                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3043                             line_prefix, set, similarity_index(p));
3044                 strbuf_addf(msg, "%s\n%s%srename from ",
3045                             reset, line_prefix, set);
3046                 quote_c_style(name, msg, NULL, 0);
3047                 strbuf_addf(msg, "%s\n%s%srename to ",
3048                             reset, line_prefix, set);
3049                 quote_c_style(other, msg, NULL, 0);
3050                 strbuf_addf(msg, "%s\n", reset);
3051                 break;
3052         case DIFF_STATUS_MODIFIED:
3053                 if (p->score) {
3054                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3055                                     line_prefix,
3056                                     set, similarity_index(p), reset);
3057                         break;
3058                 }
3059                 /* fallthru */
3060         default:
3061                 *must_show_header = 0;
3062         }
3063         if (one && two && hashcmp(one->sha1, two->sha1)) {
3064                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3065
3066                 if (DIFF_OPT_TST(o, BINARY)) {
3067                         mmfile_t mf;
3068                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3069                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3070                                 abbrev = 40;
3071                 }
3072                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3073                             find_unique_abbrev(one->sha1, abbrev));
3074                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3075                 if (one->mode == two->mode)
3076                         strbuf_addf(msg, " %06o", one->mode);
3077                 strbuf_addf(msg, "%s\n", reset);
3078         }
3079 }
3080
3081 static void run_diff_cmd(const char *pgm,
3082                          const char *name,
3083                          const char *other,
3084                          const char *attr_path,
3085                          struct diff_filespec *one,
3086                          struct diff_filespec *two,
3087                          struct strbuf *msg,
3088                          struct diff_options *o,
3089                          struct diff_filepair *p)
3090 {
3091         const char *xfrm_msg = NULL;
3092         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3093         int must_show_header = 0;
3094
3095
3096         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3097                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3098                 if (drv && drv->external)
3099                         pgm = drv->external;
3100         }
3101
3102         if (msg) {
3103                 /*
3104                  * don't use colors when the header is intended for an
3105                  * external diff driver
3106                  */
3107                 fill_metainfo(msg, name, other, one, two, o, p,
3108                               &must_show_header,
3109                               want_color(o->use_color) && !pgm);
3110                 xfrm_msg = msg->len ? msg->buf : NULL;
3111         }
3112
3113         if (pgm) {
3114                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3115                                   complete_rewrite, o);
3116                 return;
3117         }
3118         if (one && two)
3119                 builtin_diff(name, other ? other : name,
3120                              one, two, xfrm_msg, must_show_header,
3121                              o, complete_rewrite);
3122         else
3123                 fprintf(o->file, "* Unmerged path %s\n", name);
3124 }
3125
3126 static void diff_fill_sha1_info(struct diff_filespec *one)
3127 {
3128         if (DIFF_FILE_VALID(one)) {
3129                 if (!one->sha1_valid) {
3130                         struct stat st;
3131                         if (one->is_stdin) {
3132                                 hashcpy(one->sha1, null_sha1);
3133                                 return;
3134                         }
3135                         if (lstat(one->path, &st) < 0)
3136                                 die_errno("stat '%s'", one->path);
3137                         if (index_path(one->sha1, one->path, &st, 0))
3138                                 die("cannot hash %s", one->path);
3139                 }
3140         }
3141         else
3142                 hashclr(one->sha1);
3143 }
3144
3145 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3146 {
3147         /* Strip the prefix but do not molest /dev/null and absolute paths */
3148         if (*namep && **namep != '/') {
3149                 *namep += prefix_length;
3150                 if (**namep == '/')
3151                         ++*namep;
3152         }
3153         if (*otherp && **otherp != '/') {
3154                 *otherp += prefix_length;
3155                 if (**otherp == '/')
3156                         ++*otherp;
3157         }
3158 }
3159
3160 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3161 {
3162         const char *pgm = external_diff();
3163         struct strbuf msg;
3164         struct diff_filespec *one = p->one;
3165         struct diff_filespec *two = p->two;
3166         const char *name;
3167         const char *other;
3168         const char *attr_path;
3169
3170         name  = p->one->path;
3171         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3172         attr_path = name;
3173         if (o->prefix_length)
3174                 strip_prefix(o->prefix_length, &name, &other);
3175
3176         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3177                 pgm = NULL;
3178
3179         if (DIFF_PAIR_UNMERGED(p)) {
3180                 run_diff_cmd(pgm, name, NULL, attr_path,
3181                              NULL, NULL, NULL, o, p);
3182                 return;
3183         }
3184
3185         diff_fill_sha1_info(one);
3186         diff_fill_sha1_info(two);
3187
3188         if (!pgm &&
3189             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3190             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3191                 /*
3192                  * a filepair that changes between file and symlink
3193                  * needs to be split into deletion and creation.
3194                  */
3195                 struct diff_filespec *null = alloc_filespec(two->path);
3196                 run_diff_cmd(NULL, name, other, attr_path,
3197                              one, null, &msg, o, p);
3198                 free(null);
3199                 strbuf_release(&msg);
3200
3201                 null = alloc_filespec(one->path);
3202                 run_diff_cmd(NULL, name, other, attr_path,
3203                              null, two, &msg, o, p);
3204                 free(null);
3205         }
3206         else
3207                 run_diff_cmd(pgm, name, other, attr_path,
3208                              one, two, &msg, o, p);
3209
3210         strbuf_release(&msg);
3211 }
3212
3213 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3214                          struct diffstat_t *diffstat)
3215 {
3216         const char *name;
3217         const char *other;
3218
3219         if (DIFF_PAIR_UNMERGED(p)) {
3220                 /* unmerged */
3221                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3222                 return;
3223         }
3224
3225         name = p->one->path;
3226         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3227
3228         if (o->prefix_length)
3229                 strip_prefix(o->prefix_length, &name, &other);
3230
3231         diff_fill_sha1_info(p->one);
3232         diff_fill_sha1_info(p->two);
3233
3234         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3235 }
3236
3237 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3238 {
3239         const char *name;
3240         const char *other;
3241         const char *attr_path;
3242
3243         if (DIFF_PAIR_UNMERGED(p)) {
3244                 /* unmerged */
3245                 return;
3246         }
3247
3248         name = p->one->path;
3249         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3250         attr_path = other ? other : name;
3251
3252         if (o->prefix_length)
3253                 strip_prefix(o->prefix_length, &name, &other);
3254
3255         diff_fill_sha1_info(p->one);
3256         diff_fill_sha1_info(p->two);
3257
3258         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3259 }
3260
3261 void diff_setup(struct diff_options *options)
3262 {
3263         memcpy(options, &default_diff_options, sizeof(*options));
3264
3265         options->file = stdout;
3266
3267         options->line_termination = '\n';
3268         options->break_opt = -1;
3269         options->rename_limit = -1;
3270         options->dirstat_permille = diff_dirstat_permille_default;
3271         options->context = diff_context_default;
3272         options->ws_error_highlight = WSEH_NEW;
3273         DIFF_OPT_SET(options, RENAME_EMPTY);
3274
3275         /* pathchange left =NULL by default */
3276         options->change = diff_change;
3277         options->add_remove = diff_addremove;
3278         options->use_color = diff_use_color_default;
3279         options->detect_rename = diff_detect_rename_default;
3280         options->xdl_opts |= diff_algorithm;
3281
3282         options->orderfile = diff_order_file_cfg;
3283
3284         if (diff_no_prefix) {
3285                 options->a_prefix = options->b_prefix = "";
3286         } else if (!diff_mnemonic_prefix) {
3287                 options->a_prefix = "a/";
3288                 options->b_prefix = "b/";
3289         }
3290 }
3291
3292 void diff_setup_done(struct diff_options *options)
3293 {
3294         int count = 0;
3295
3296         if (options->set_default)
3297                 options->set_default(options);
3298
3299         if (options->output_format & DIFF_FORMAT_NAME)
3300                 count++;
3301         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3302                 count++;
3303         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3304                 count++;
3305         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3306                 count++;
3307         if (count > 1)
3308                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3309
3310         /*
3311          * Most of the time we can say "there are changes"
3312          * only by checking if there are changed paths, but
3313          * --ignore-whitespace* options force us to look
3314          * inside contents.
3315          */
3316
3317         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3318             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3319             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3320                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3321         else
3322                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3323
3324         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3325                 options->detect_rename = DIFF_DETECT_COPY;
3326
3327         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3328                 options->prefix = NULL;
3329         if (options->prefix)
3330                 options->prefix_length = strlen(options->prefix);
3331         else
3332                 options->prefix_length = 0;
3333
3334         if (options->output_format & (DIFF_FORMAT_NAME |
3335                                       DIFF_FORMAT_NAME_STATUS |
3336                                       DIFF_FORMAT_CHECKDIFF |
3337                                       DIFF_FORMAT_NO_OUTPUT))
3338                 options->output_format &= ~(DIFF_FORMAT_RAW |
3339                                             DIFF_FORMAT_NUMSTAT |
3340                                             DIFF_FORMAT_DIFFSTAT |
3341                                             DIFF_FORMAT_SHORTSTAT |
3342                                             DIFF_FORMAT_DIRSTAT |
3343                                             DIFF_FORMAT_SUMMARY |
3344                                             DIFF_FORMAT_PATCH);
3345
3346         /*
3347          * These cases always need recursive; we do not drop caller-supplied
3348          * recursive bits for other formats here.
3349          */
3350         if (options->output_format & (DIFF_FORMAT_PATCH |
3351                                       DIFF_FORMAT_NUMSTAT |
3352                                       DIFF_FORMAT_DIFFSTAT |
3353                                       DIFF_FORMAT_SHORTSTAT |
3354                                       DIFF_FORMAT_DIRSTAT |
3355                                       DIFF_FORMAT_SUMMARY |
3356                                       DIFF_FORMAT_CHECKDIFF))
3357                 DIFF_OPT_SET(options, RECURSIVE);
3358         /*
3359          * Also pickaxe would not work very well if you do not say recursive
3360          */
3361         if (options->pickaxe)
3362                 DIFF_OPT_SET(options, RECURSIVE);
3363         /*
3364          * When patches are generated, submodules diffed against the work tree
3365          * must be checked for dirtiness too so it can be shown in the output
3366          */
3367         if (options->output_format & DIFF_FORMAT_PATCH)
3368                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3369
3370         if (options->detect_rename && options->rename_limit < 0)
3371                 options->rename_limit = diff_rename_limit_default;
3372         if (options->setup & DIFF_SETUP_USE_CACHE) {
3373                 if (!active_cache)
3374                         /* read-cache does not die even when it fails
3375                          * so it is safe for us to do this here.  Also
3376                          * it does not smudge active_cache or active_nr
3377                          * when it fails, so we do not have to worry about
3378                          * cleaning it up ourselves either.
3379                          */
3380                         read_cache();
3381         }
3382         if (options->abbrev <= 0 || 40 < options->abbrev)
3383                 options->abbrev = 40; /* full */
3384
3385         /*
3386          * It does not make sense to show the first hit we happened
3387          * to have found.  It does not make sense not to return with
3388          * exit code in such a case either.
3389          */
3390         if (DIFF_OPT_TST(options, QUICK)) {
3391                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3392                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3393         }
3394
3395         options->diff_path_counter = 0;
3396
3397         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3398                 die(_("--follow requires exactly one pathspec"));
3399 }
3400
3401 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3402 {
3403         char c, *eq;
3404         int len;
3405
3406         if (*arg != '-')
3407                 return 0;
3408         c = *++arg;
3409         if (!c)
3410                 return 0;
3411         if (c == arg_short) {
3412                 c = *++arg;
3413                 if (!c)
3414                         return 1;
3415                 if (val && isdigit(c)) {
3416                         char *end;
3417                         int n = strtoul(arg, &end, 10);
3418                         if (*end)
3419                                 return 0;
3420                         *val = n;
3421                         return 1;
3422                 }
3423                 return 0;
3424         }
3425         if (c != '-')
3426                 return 0;
3427         arg++;
3428         eq = strchrnul(arg, '=');
3429         len = eq - arg;
3430         if (!len || strncmp(arg, arg_long, len))
3431                 return 0;
3432         if (*eq) {
3433                 int n;
3434                 char *end;
3435                 if (!isdigit(*++eq))
3436                         return 0;
3437                 n = strtoul(eq, &end, 10);
3438                 if (*end)
3439                         return 0;
3440                 *val = n;
3441         }
3442         return 1;
3443 }
3444
3445 static int diff_scoreopt_parse(const char *opt);
3446
3447 static inline int short_opt(char opt, const char **argv,
3448                             const char **optarg)
3449 {
3450         const char *arg = argv[0];
3451         if (arg[0] != '-' || arg[1] != opt)
3452                 return 0;
3453         if (arg[2] != '\0') {
3454                 *optarg = arg + 2;
3455                 return 1;
3456         }
3457         if (!argv[1])
3458                 die("Option '%c' requires a value", opt);
3459         *optarg = argv[1];
3460         return 2;
3461 }
3462
3463 int parse_long_opt(const char *opt, const char **argv,
3464                    const char **optarg)
3465 {
3466         const char *arg = argv[0];
3467         if (!skip_prefix(arg, "--", &arg))
3468                 return 0;
3469         if (!skip_prefix(arg, opt, &arg))
3470                 return 0;
3471         if (*arg == '=') { /* stuck form: --option=value */
3472                 *optarg = arg + 1;
3473                 return 1;
3474         }
3475         if (*arg != '\0')
3476                 return 0;
3477         /* separate form: --option value */
3478         if (!argv[1])
3479                 die("Option '--%s' requires a value", opt);
3480         *optarg = argv[1];
3481         return 2;
3482 }
3483
3484 static int stat_opt(struct diff_options *options, const char **av)
3485 {
3486         const char *arg = av[0];
3487         char *end;
3488         int width = options->stat_width;
3489         int name_width = options->stat_name_width;
3490         int graph_width = options->stat_graph_width;
3491         int count = options->stat_count;
3492         int argcount = 1;
3493
3494         if (!skip_prefix(arg, "--stat", &arg))
3495                 die("BUG: stat option does not begin with --stat: %s", arg);
3496         end = (char *)arg;
3497
3498         switch (*arg) {
3499         case '-':
3500                 if (skip_prefix(arg, "-width", &arg)) {
3501                         if (*arg == '=')
3502                                 width = strtoul(arg + 1, &end, 10);
3503                         else if (!*arg && !av[1])
3504                                 die("Option '--stat-width' requires a value");
3505                         else if (!*arg) {
3506                                 width = strtoul(av[1], &end, 10);
3507                                 argcount = 2;
3508                         }
3509                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3510                         if (*arg == '=')
3511                                 name_width = strtoul(arg + 1, &end, 10);
3512                         else if (!*arg && !av[1])
3513                                 die("Option '--stat-name-width' requires a value");
3514                         else if (!*arg) {
3515                                 name_width = strtoul(av[1], &end, 10);
3516                                 argcount = 2;
3517                         }
3518                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3519                         if (*arg == '=')
3520                                 graph_width = strtoul(arg + 1, &end, 10);
3521                         else if (!*arg && !av[1])
3522                                 die("Option '--stat-graph-width' requires a value");
3523                         else if (!*arg) {
3524                                 graph_width = strtoul(av[1], &end, 10);
3525                                 argcount = 2;
3526                         }
3527                 } else if (skip_prefix(arg, "-count", &arg)) {
3528                         if (*arg == '=')
3529                                 count = strtoul(arg + 1, &end, 10);
3530                         else if (!*arg && !av[1])
3531                                 die("Option '--stat-count' requires a value");
3532                         else if (!*arg) {
3533                                 count = strtoul(av[1], &end, 10);
3534                                 argcount = 2;
3535                         }
3536                 }
3537                 break;
3538         case '=':
3539                 width = strtoul(arg+1, &end, 10);
3540                 if (*end == ',')
3541                         name_width = strtoul(end+1, &end, 10);
3542                 if (*end == ',')
3543                         count = strtoul(end+1, &end, 10);
3544         }
3545
3546         /* Important! This checks all the error cases! */
3547         if (*end)
3548                 return 0;
3549         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3550         options->stat_name_width = name_width;
3551         options->stat_graph_width = graph_width;
3552         options->stat_width = width;
3553         options->stat_count = count;
3554         return argcount;
3555 }
3556
3557 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3558 {
3559         struct strbuf errmsg = STRBUF_INIT;
3560         if (parse_dirstat_params(options, params, &errmsg))
3561                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3562                     errmsg.buf);
3563         strbuf_release(&errmsg);
3564         /*
3565          * The caller knows a dirstat-related option is given from the command
3566          * line; allow it to say "return this_function();"
3567          */
3568         options->output_format |= DIFF_FORMAT_DIRSTAT;
3569         return 1;
3570 }
3571
3572 static int parse_submodule_opt(struct diff_options *options, const char *value)
3573 {
3574         if (parse_submodule_params(options, value))
3575                 die(_("Failed to parse --submodule option parameter: '%s'"),
3576                         value);
3577         return 1;
3578 }
3579
3580 static const char diff_status_letters[] = {
3581         DIFF_STATUS_ADDED,
3582         DIFF_STATUS_COPIED,
3583         DIFF_STATUS_DELETED,
3584         DIFF_STATUS_MODIFIED,
3585         DIFF_STATUS_RENAMED,
3586         DIFF_STATUS_TYPE_CHANGED,
3587         DIFF_STATUS_UNKNOWN,
3588         DIFF_STATUS_UNMERGED,
3589         DIFF_STATUS_FILTER_AON,
3590         DIFF_STATUS_FILTER_BROKEN,
3591         '\0',
3592 };
3593
3594 static unsigned int filter_bit['Z' + 1];
3595
3596 static void prepare_filter_bits(void)
3597 {
3598         int i;
3599
3600         if (!filter_bit[DIFF_STATUS_ADDED]) {
3601                 for (i = 0; diff_status_letters[i]; i++)
3602                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3603         }
3604 }
3605
3606 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3607 {
3608         return opt->filter & filter_bit[(int) status];
3609 }
3610
3611 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3612 {
3613         int i, optch;
3614
3615         prepare_filter_bits();
3616
3617         /*
3618          * If there is a negation e.g. 'd' in the input, and we haven't
3619          * initialized the filter field with another --diff-filter, start
3620          * from full set of bits, except for AON.
3621          */
3622         if (!opt->filter) {
3623                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3624                         if (optch < 'a' || 'z' < optch)
3625                                 continue;
3626                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3627                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3628                         break;
3629                 }
3630         }
3631
3632         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3633                 unsigned int bit;
3634                 int negate;
3635
3636                 if ('a' <= optch && optch <= 'z') {
3637                         negate = 1;
3638                         optch = toupper(optch);
3639                 } else {
3640                         negate = 0;
3641                 }
3642
3643                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3644                 if (!bit)
3645                         return optarg[i];
3646                 if (negate)
3647                         opt->filter &= ~bit;
3648                 else
3649                         opt->filter |= bit;
3650         }
3651         return 0;
3652 }
3653
3654 static void enable_patch_output(int *fmt) {
3655         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3656         *fmt |= DIFF_FORMAT_PATCH;
3657 }
3658
3659 static int parse_one_token(const char **arg, const char *token)
3660 {
3661         const char *rest;
3662         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
3663                 *arg = rest;
3664                 return 1;
3665         }
3666         return 0;
3667 }
3668
3669 static int parse_ws_error_highlight(struct diff_options *opt, const char *arg)
3670 {
3671         const char *orig_arg = arg;
3672         unsigned val = 0;
3673         while (*arg) {
3674                 if (parse_one_token(&arg, "none"))
3675                         val = 0;
3676                 else if (parse_one_token(&arg, "default"))
3677                         val = WSEH_NEW;
3678                 else if (parse_one_token(&arg, "all"))
3679                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
3680                 else if (parse_one_token(&arg, "new"))
3681                         val |= WSEH_NEW;
3682                 else if (parse_one_token(&arg, "old"))
3683                         val |= WSEH_OLD;
3684                 else if (parse_one_token(&arg, "context"))
3685                         val |= WSEH_CONTEXT;
3686                 else {
3687                         error("unknown value after ws-error-highlight=%.*s",
3688                               (int)(arg - orig_arg), orig_arg);
3689                         return 0;
3690                 }
3691                 if (*arg)
3692                         arg++;
3693         }
3694         opt->ws_error_highlight = val;
3695         return 1;
3696 }
3697
3698 int diff_opt_parse(struct diff_options *options,
3699                    const char **av, int ac, const char *prefix)
3700 {
3701         const char *arg = av[0];
3702         const char *optarg;
3703         int argcount;
3704
3705         if (!prefix)
3706                 prefix = "";
3707
3708         /* Output format options */
3709         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3710             || opt_arg(arg, 'U', "unified", &options->context))
3711                 enable_patch_output(&options->output_format);
3712         else if (!strcmp(arg, "--raw"))
3713                 options->output_format |= DIFF_FORMAT_RAW;
3714         else if (!strcmp(arg, "--patch-with-raw")) {
3715                 enable_patch_output(&options->output_format);
3716                 options->output_format |= DIFF_FORMAT_RAW;
3717         } else if (!strcmp(arg, "--numstat"))
3718                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3719         else if (!strcmp(arg, "--shortstat"))
3720                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3721         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3722                 return parse_dirstat_opt(options, "");
3723         else if (skip_prefix(arg, "-X", &arg))
3724                 return parse_dirstat_opt(options, arg);
3725         else if (skip_prefix(arg, "--dirstat=", &arg))
3726                 return parse_dirstat_opt(options, arg);
3727         else if (!strcmp(arg, "--cumulative"))
3728                 return parse_dirstat_opt(options, "cumulative");
3729         else if (!strcmp(arg, "--dirstat-by-file"))
3730                 return parse_dirstat_opt(options, "files");
3731         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3732                 parse_dirstat_opt(options, "files");
3733                 return parse_dirstat_opt(options, arg);
3734         }
3735         else if (!strcmp(arg, "--check"))
3736                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3737         else if (!strcmp(arg, "--summary"))
3738                 options->output_format |= DIFF_FORMAT_SUMMARY;
3739         else if (!strcmp(arg, "--patch-with-stat")) {
3740                 enable_patch_output(&options->output_format);
3741                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3742         } else if (!strcmp(arg, "--name-only"))
3743                 options->output_format |= DIFF_FORMAT_NAME;
3744         else if (!strcmp(arg, "--name-status"))
3745                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3746         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3747                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3748         else if (starts_with(arg, "--stat"))
3749                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3750                 return stat_opt(options, av);
3751
3752         /* renames options */
3753         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3754                  !strcmp(arg, "--break-rewrites")) {
3755                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3756                         return error("invalid argument to -B: %s", arg+2);
3757         }
3758         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3759                  !strcmp(arg, "--find-renames")) {
3760                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3761                         return error("invalid argument to -M: %s", arg+2);
3762                 options->detect_rename = DIFF_DETECT_RENAME;
3763         }
3764         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3765                 options->irreversible_delete = 1;
3766         }
3767         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3768                  !strcmp(arg, "--find-copies")) {
3769                 if (options->detect_rename == DIFF_DETECT_COPY)
3770                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3771                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3772                         return error("invalid argument to -C: %s", arg+2);
3773                 options->detect_rename = DIFF_DETECT_COPY;
3774         }
3775         else if (!strcmp(arg, "--no-renames"))
3776                 options->detect_rename = 0;
3777         else if (!strcmp(arg, "--rename-empty"))
3778                 DIFF_OPT_SET(options, RENAME_EMPTY);
3779         else if (!strcmp(arg, "--no-rename-empty"))
3780                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3781         else if (!strcmp(arg, "--relative"))
3782                 DIFF_OPT_SET(options, RELATIVE_NAME);
3783         else if (skip_prefix(arg, "--relative=", &arg)) {
3784                 DIFF_OPT_SET(options, RELATIVE_NAME);
3785                 options->prefix = arg;
3786         }
3787
3788         /* xdiff options */
3789         else if (!strcmp(arg, "--minimal"))
3790                 DIFF_XDL_SET(options, NEED_MINIMAL);
3791         else if (!strcmp(arg, "--no-minimal"))
3792                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3793         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3794                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3795         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3796                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3797         else if (!strcmp(arg, "--ignore-space-at-eol"))
3798                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3799         else if (!strcmp(arg, "--ignore-blank-lines"))
3800                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3801         else if (!strcmp(arg, "--patience"))
3802                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3803         else if (!strcmp(arg, "--histogram"))
3804                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3805         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3806                 long value = parse_algorithm_value(optarg);
3807                 if (value < 0)
3808                         return error("option diff-algorithm accepts \"myers\", "
3809                                      "\"minimal\", \"patience\" and \"histogram\"");
3810                 /* clear out previous settings */
3811                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3812                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3813                 options->xdl_opts |= value;
3814                 return argcount;
3815         }
3816
3817         /* flags options */
3818         else if (!strcmp(arg, "--binary")) {
3819                 enable_patch_output(&options->output_format);
3820                 DIFF_OPT_SET(options, BINARY);
3821         }
3822         else if (!strcmp(arg, "--full-index"))
3823                 DIFF_OPT_SET(options, FULL_INDEX);
3824         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3825                 DIFF_OPT_SET(options, TEXT);
3826         else if (!strcmp(arg, "-R"))
3827                 DIFF_OPT_SET(options, REVERSE_DIFF);
3828         else if (!strcmp(arg, "--find-copies-harder"))
3829                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3830         else if (!strcmp(arg, "--follow"))
3831                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3832         else if (!strcmp(arg, "--no-follow")) {
3833                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3834                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3835         } else if (!strcmp(arg, "--color"))
3836                 options->use_color = 1;
3837         else if (skip_prefix(arg, "--color=", &arg)) {
3838                 int value = git_config_colorbool(NULL, arg);
3839                 if (value < 0)
3840                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3841                 options->use_color = value;
3842         }
3843         else if (!strcmp(arg, "--no-color"))
3844                 options->use_color = 0;
3845         else if (!strcmp(arg, "--color-words")) {
3846                 options->use_color = 1;
3847                 options->word_diff = DIFF_WORDS_COLOR;
3848         }
3849         else if (skip_prefix(arg, "--color-words=", &arg)) {
3850                 options->use_color = 1;
3851                 options->word_diff = DIFF_WORDS_COLOR;
3852                 options->word_regex = arg;
3853         }
3854         else if (!strcmp(arg, "--word-diff")) {
3855                 if (options->word_diff == DIFF_WORDS_NONE)
3856                         options->word_diff = DIFF_WORDS_PLAIN;
3857         }
3858         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3859                 if (!strcmp(arg, "plain"))
3860                         options->word_diff = DIFF_WORDS_PLAIN;
3861                 else if (!strcmp(arg, "color")) {
3862                         options->use_color = 1;
3863                         options->word_diff = DIFF_WORDS_COLOR;
3864                 }
3865                 else if (!strcmp(arg, "porcelain"))
3866                         options->word_diff = DIFF_WORDS_PORCELAIN;
3867                 else if (!strcmp(arg, "none"))
3868                         options->word_diff = DIFF_WORDS_NONE;
3869                 else
3870                         die("bad --word-diff argument: %s", arg);
3871         }
3872         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3873                 if (options->word_diff == DIFF_WORDS_NONE)
3874                         options->word_diff = DIFF_WORDS_PLAIN;
3875                 options->word_regex = optarg;
3876                 return argcount;
3877         }
3878         else if (!strcmp(arg, "--exit-code"))
3879                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3880         else if (!strcmp(arg, "--quiet"))
3881                 DIFF_OPT_SET(options, QUICK);
3882         else if (!strcmp(arg, "--ext-diff"))
3883                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3884         else if (!strcmp(arg, "--no-ext-diff"))
3885                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3886         else if (!strcmp(arg, "--textconv"))
3887                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3888         else if (!strcmp(arg, "--no-textconv"))
3889                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3890         else if (!strcmp(arg, "--ignore-submodules")) {
3891                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3892                 handle_ignore_submodules_arg(options, "all");
3893         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3894                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3895                 handle_ignore_submodules_arg(options, arg);
3896         } else if (!strcmp(arg, "--submodule"))
3897                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3898         else if (skip_prefix(arg, "--submodule=", &arg))
3899                 return parse_submodule_opt(options, arg);
3900         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3901                 return parse_ws_error_highlight(options, arg);
3902
3903         /* misc options */
3904         else if (!strcmp(arg, "-z"))
3905                 options->line_termination = 0;
3906         else if ((argcount = short_opt('l', av, &optarg))) {
3907                 options->rename_limit = strtoul(optarg, NULL, 10);
3908                 return argcount;
3909         }
3910         else if ((argcount = short_opt('S', av, &optarg))) {
3911                 options->pickaxe = optarg;
3912                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3913                 return argcount;
3914         } else if ((argcount = short_opt('G', av, &optarg))) {
3915                 options->pickaxe = optarg;
3916                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3917                 return argcount;
3918         }
3919         else if (!strcmp(arg, "--pickaxe-all"))
3920                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3921         else if (!strcmp(arg, "--pickaxe-regex"))
3922                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3923         else if ((argcount = short_opt('O', av, &optarg))) {
3924                 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
3925                 options->orderfile = xstrdup(path);
3926                 return argcount;
3927         }
3928         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3929                 int offending = parse_diff_filter_opt(optarg, options);
3930                 if (offending)
3931                         die("unknown change class '%c' in --diff-filter=%s",
3932                             offending, optarg);
3933                 return argcount;
3934         }
3935         else if (!strcmp(arg, "--abbrev"))
3936                 options->abbrev = DEFAULT_ABBREV;
3937         else if (skip_prefix(arg, "--abbrev=", &arg)) {
3938                 options->abbrev = strtoul(arg, NULL, 10);
3939                 if (options->abbrev < MINIMUM_ABBREV)
3940                         options->abbrev = MINIMUM_ABBREV;
3941                 else if (40 < options->abbrev)
3942                         options->abbrev = 40;
3943         }
3944         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3945                 options->a_prefix = optarg;
3946                 return argcount;
3947         }
3948         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3949                 options->b_prefix = optarg;
3950                 return argcount;
3951         }
3952         else if (!strcmp(arg, "--no-prefix"))
3953                 options->a_prefix = options->b_prefix = "";
3954         else if (opt_arg(arg, '\0', "inter-hunk-context",
3955                          &options->interhunkcontext))
3956                 ;
3957         else if (!strcmp(arg, "-W"))
3958                 DIFF_OPT_SET(options, FUNCCONTEXT);
3959         else if (!strcmp(arg, "--function-context"))
3960                 DIFF_OPT_SET(options, FUNCCONTEXT);
3961         else if (!strcmp(arg, "--no-function-context"))
3962                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3963         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3964                 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
3965                 options->file = fopen(path, "w");
3966                 if (!options->file)
3967                         die_errno("Could not open '%s'", path);
3968                 options->close_file = 1;
3969                 return argcount;
3970         } else
3971                 return 0;
3972         return 1;
3973 }
3974
3975 int parse_rename_score(const char **cp_p)
3976 {
3977         unsigned long num, scale;
3978         int ch, dot;
3979         const char *cp = *cp_p;
3980
3981         num = 0;
3982         scale = 1;
3983         dot = 0;
3984         for (;;) {
3985                 ch = *cp;
3986                 if ( !dot && ch == '.' ) {
3987                         scale = 1;
3988                         dot = 1;
3989                 } else if ( ch == '%' ) {
3990                         scale = dot ? scale*100 : 100;
3991                         cp++;   /* % is always at the end */
3992                         break;
3993                 } else if ( ch >= '0' && ch <= '9' ) {
3994                         if ( scale < 100000 ) {
3995                                 scale *= 10;
3996                                 num = (num*10) + (ch-'0');
3997                         }
3998                 } else {
3999                         break;
4000                 }
4001                 cp++;
4002         }
4003         *cp_p = cp;
4004
4005         /* user says num divided by scale and we say internally that
4006          * is MAX_SCORE * num / scale.
4007          */
4008         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4009 }
4010
4011 static int diff_scoreopt_parse(const char *opt)
4012 {
4013         int opt1, opt2, cmd;
4014
4015         if (*opt++ != '-')
4016                 return -1;
4017         cmd = *opt++;
4018         if (cmd == '-') {
4019                 /* convert the long-form arguments into short-form versions */
4020                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4021                         if (*opt == 0 || *opt++ == '=')
4022                                 cmd = 'B';
4023                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4024                         if (*opt == 0 || *opt++ == '=')
4025                                 cmd = 'C';
4026                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4027                         if (*opt == 0 || *opt++ == '=')
4028                                 cmd = 'M';
4029                 }
4030         }
4031         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4032                 return -1; /* that is not a -M, -C, or -B option */
4033
4034         opt1 = parse_rename_score(&opt);
4035         if (cmd != 'B')
4036                 opt2 = 0;
4037         else {
4038                 if (*opt == 0)
4039                         opt2 = 0;
4040                 else if (*opt != '/')
4041                         return -1; /* we expect -B80/99 or -B80 */
4042                 else {
4043                         opt++;
4044                         opt2 = parse_rename_score(&opt);
4045                 }
4046         }
4047         if (*opt != 0)
4048                 return -1;
4049         return opt1 | (opt2 << 16);
4050 }
4051
4052 struct diff_queue_struct diff_queued_diff;
4053
4054 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4055 {
4056         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4057         queue->queue[queue->nr++] = dp;
4058 }
4059
4060 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4061                                  struct diff_filespec *one,
4062                                  struct diff_filespec *two)
4063 {
4064         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4065         dp->one = one;
4066         dp->two = two;
4067         if (queue)
4068                 diff_q(queue, dp);
4069         return dp;
4070 }
4071
4072 void diff_free_filepair(struct diff_filepair *p)
4073 {
4074         free_filespec(p->one);
4075         free_filespec(p->two);
4076         free(p);
4077 }
4078
4079 /* This is different from find_unique_abbrev() in that
4080  * it stuffs the result with dots for alignment.
4081  */
4082 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
4083 {
4084         int abblen;
4085         const char *abbrev;
4086         if (len == 40)
4087                 return sha1_to_hex(sha1);
4088
4089         abbrev = find_unique_abbrev(sha1, len);
4090         abblen = strlen(abbrev);
4091         if (abblen < 37) {
4092                 static char hex[41];
4093                 if (len < abblen && abblen <= len + 2)
4094                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4095                 else
4096                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4097                 return hex;
4098         }
4099         return sha1_to_hex(sha1);
4100 }
4101
4102 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4103 {
4104         int line_termination = opt->line_termination;
4105         int inter_name_termination = line_termination ? '\t' : '\0';
4106
4107         fprintf(opt->file, "%s", diff_line_prefix(opt));
4108         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4109                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4110                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
4111                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4112         }
4113         if (p->score) {
4114                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4115                         inter_name_termination);
4116         } else {
4117                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4118         }
4119
4120         if (p->status == DIFF_STATUS_COPIED ||
4121             p->status == DIFF_STATUS_RENAMED) {
4122                 const char *name_a, *name_b;
4123                 name_a = p->one->path;
4124                 name_b = p->two->path;
4125                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4126                 write_name_quoted(name_a, opt->file, inter_name_termination);
4127                 write_name_quoted(name_b, opt->file, line_termination);
4128         } else {
4129                 const char *name_a, *name_b;
4130                 name_a = p->one->mode ? p->one->path : p->two->path;
4131                 name_b = NULL;
4132                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4133                 write_name_quoted(name_a, opt->file, line_termination);
4134         }
4135 }
4136
4137 int diff_unmodified_pair(struct diff_filepair *p)
4138 {
4139         /* This function is written stricter than necessary to support
4140          * the currently implemented transformers, but the idea is to
4141          * let transformers to produce diff_filepairs any way they want,
4142          * and filter and clean them up here before producing the output.
4143          */
4144         struct diff_filespec *one = p->one, *two = p->two;
4145
4146         if (DIFF_PAIR_UNMERGED(p))
4147                 return 0; /* unmerged is interesting */
4148
4149         /* deletion, addition, mode or type change
4150          * and rename are all interesting.
4151          */
4152         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4153             DIFF_PAIR_MODE_CHANGED(p) ||
4154             strcmp(one->path, two->path))
4155                 return 0;
4156
4157         /* both are valid and point at the same path.  that is, we are
4158          * dealing with a change.
4159          */
4160         if (one->sha1_valid && two->sha1_valid &&
4161             !hashcmp(one->sha1, two->sha1) &&
4162             !one->dirty_submodule && !two->dirty_submodule)
4163                 return 1; /* no change */
4164         if (!one->sha1_valid && !two->sha1_valid)
4165                 return 1; /* both look at the same file on the filesystem. */
4166         return 0;
4167 }
4168
4169 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4170 {
4171         if (diff_unmodified_pair(p))
4172                 return;
4173
4174         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4175             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4176                 return; /* no tree diffs in patch format */
4177
4178         run_diff(p, o);
4179 }
4180
4181 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4182                             struct diffstat_t *diffstat)
4183 {
4184         if (diff_unmodified_pair(p))
4185                 return;
4186
4187         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4188             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4189                 return; /* no useful stat for tree diffs */
4190
4191         run_diffstat(p, o, diffstat);
4192 }
4193
4194 static void diff_flush_checkdiff(struct diff_filepair *p,
4195                 struct diff_options *o)
4196 {
4197         if (diff_unmodified_pair(p))
4198                 return;
4199
4200         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4201             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4202                 return; /* nothing to check in tree diffs */
4203
4204         run_checkdiff(p, o);
4205 }
4206
4207 int diff_queue_is_empty(void)
4208 {
4209         struct diff_queue_struct *q = &diff_queued_diff;
4210         int i;
4211         for (i = 0; i < q->nr; i++)
4212                 if (!diff_unmodified_pair(q->queue[i]))
4213                         return 0;
4214         return 1;
4215 }
4216
4217 #if DIFF_DEBUG
4218 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4219 {
4220         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4221                 x, one ? one : "",
4222                 s->path,
4223                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4224                 s->mode,
4225                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4226         fprintf(stderr, "queue[%d] %s size %lu\n",
4227                 x, one ? one : "",
4228                 s->size);
4229 }
4230
4231 void diff_debug_filepair(const struct diff_filepair *p, int i)
4232 {
4233         diff_debug_filespec(p->one, i, "one");
4234         diff_debug_filespec(p->two, i, "two");
4235         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4236                 p->score, p->status ? p->status : '?',
4237                 p->one->rename_used, p->broken_pair);
4238 }
4239
4240 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4241 {
4242         int i;
4243         if (msg)
4244                 fprintf(stderr, "%s\n", msg);
4245         fprintf(stderr, "q->nr = %d\n", q->nr);
4246         for (i = 0; i < q->nr; i++) {
4247                 struct diff_filepair *p = q->queue[i];
4248                 diff_debug_filepair(p, i);
4249         }
4250 }
4251 #endif
4252
4253 static void diff_resolve_rename_copy(void)
4254 {
4255         int i;
4256         struct diff_filepair *p;
4257         struct diff_queue_struct *q = &diff_queued_diff;
4258
4259         diff_debug_queue("resolve-rename-copy", q);
4260
4261         for (i = 0; i < q->nr; i++) {
4262                 p = q->queue[i];
4263                 p->status = 0; /* undecided */
4264                 if (DIFF_PAIR_UNMERGED(p))
4265                         p->status = DIFF_STATUS_UNMERGED;
4266                 else if (!DIFF_FILE_VALID(p->one))
4267                         p->status = DIFF_STATUS_ADDED;
4268                 else if (!DIFF_FILE_VALID(p->two))
4269                         p->status = DIFF_STATUS_DELETED;
4270                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4271                         p->status = DIFF_STATUS_TYPE_CHANGED;
4272
4273                 /* from this point on, we are dealing with a pair
4274                  * whose both sides are valid and of the same type, i.e.
4275                  * either in-place edit or rename/copy edit.
4276                  */
4277                 else if (DIFF_PAIR_RENAME(p)) {
4278                         /*
4279                          * A rename might have re-connected a broken
4280                          * pair up, causing the pathnames to be the
4281                          * same again. If so, that's not a rename at
4282                          * all, just a modification..
4283                          *
4284                          * Otherwise, see if this source was used for
4285                          * multiple renames, in which case we decrement
4286                          * the count, and call it a copy.
4287                          */
4288                         if (!strcmp(p->one->path, p->two->path))
4289                                 p->status = DIFF_STATUS_MODIFIED;
4290                         else if (--p->one->rename_used > 0)
4291                                 p->status = DIFF_STATUS_COPIED;
4292                         else
4293                                 p->status = DIFF_STATUS_RENAMED;
4294                 }
4295                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4296                          p->one->mode != p->two->mode ||
4297                          p->one->dirty_submodule ||
4298                          p->two->dirty_submodule ||
4299                          is_null_sha1(p->one->sha1))
4300                         p->status = DIFF_STATUS_MODIFIED;
4301                 else {
4302                         /* This is a "no-change" entry and should not
4303                          * happen anymore, but prepare for broken callers.
4304                          */
4305                         error("feeding unmodified %s to diffcore",
4306                               p->one->path);
4307                         p->status = DIFF_STATUS_UNKNOWN;
4308                 }
4309         }
4310         diff_debug_queue("resolve-rename-copy done", q);
4311 }
4312
4313 static int check_pair_status(struct diff_filepair *p)
4314 {
4315         switch (p->status) {
4316         case DIFF_STATUS_UNKNOWN:
4317                 return 0;
4318         case 0:
4319                 die("internal error in diff-resolve-rename-copy");
4320         default:
4321                 return 1;
4322         }
4323 }
4324
4325 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4326 {
4327         int fmt = opt->output_format;
4328
4329         if (fmt & DIFF_FORMAT_CHECKDIFF)
4330                 diff_flush_checkdiff(p, opt);
4331         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4332                 diff_flush_raw(p, opt);
4333         else if (fmt & DIFF_FORMAT_NAME) {
4334                 const char *name_a, *name_b;
4335                 name_a = p->two->path;
4336                 name_b = NULL;
4337                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4338                 write_name_quoted(name_a, opt->file, opt->line_termination);
4339         }
4340 }
4341
4342 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4343 {
4344         if (fs->mode)
4345                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4346         else
4347                 fprintf(file, " %s ", newdelete);
4348         write_name_quoted(fs->path, file, '\n');
4349 }
4350
4351
4352 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4353                 const char *line_prefix)
4354 {
4355         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4356                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4357                         p->two->mode, show_name ? ' ' : '\n');
4358                 if (show_name) {
4359                         write_name_quoted(p->two->path, file, '\n');
4360                 }
4361         }
4362 }
4363
4364 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4365                         const char *line_prefix)
4366 {
4367         char *names = pprint_rename(p->one->path, p->two->path);
4368
4369         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4370         free(names);
4371         show_mode_change(file, p, 0, line_prefix);
4372 }
4373
4374 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4375 {
4376         FILE *file = opt->file;
4377         const char *line_prefix = diff_line_prefix(opt);
4378
4379         switch(p->status) {
4380         case DIFF_STATUS_DELETED:
4381                 fputs(line_prefix, file);
4382                 show_file_mode_name(file, "delete", p->one);
4383                 break;
4384         case DIFF_STATUS_ADDED:
4385                 fputs(line_prefix, file);
4386                 show_file_mode_name(file, "create", p->two);
4387                 break;
4388         case DIFF_STATUS_COPIED:
4389                 fputs(line_prefix, file);
4390                 show_rename_copy(file, "copy", p, line_prefix);
4391                 break;
4392         case DIFF_STATUS_RENAMED:
4393                 fputs(line_prefix, file);
4394                 show_rename_copy(file, "rename", p, line_prefix);
4395                 break;
4396         default:
4397                 if (p->score) {
4398                         fprintf(file, "%s rewrite ", line_prefix);
4399                         write_name_quoted(p->two->path, file, ' ');
4400                         fprintf(file, "(%d%%)\n", similarity_index(p));
4401                 }
4402                 show_mode_change(file, p, !p->score, line_prefix);
4403                 break;
4404         }
4405 }
4406
4407 struct patch_id_t {
4408         git_SHA_CTX *ctx;
4409         int patchlen;
4410 };
4411
4412 static int remove_space(char *line, int len)
4413 {
4414         int i;
4415         char *dst = line;
4416         unsigned char c;
4417
4418         for (i = 0; i < len; i++)
4419                 if (!isspace((c = line[i])))
4420                         *dst++ = c;
4421
4422         return dst - line;
4423 }
4424
4425 static void patch_id_consume(void *priv, char *line, unsigned long len)
4426 {
4427         struct patch_id_t *data = priv;
4428         int new_len;
4429
4430         /* Ignore line numbers when computing the SHA1 of the patch */
4431         if (starts_with(line, "@@ -"))
4432                 return;
4433
4434         new_len = remove_space(line, len);
4435
4436         git_SHA1_Update(data->ctx, line, new_len);
4437         data->patchlen += new_len;
4438 }
4439
4440 /* returns 0 upon success, and writes result into sha1 */
4441 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4442 {
4443         struct diff_queue_struct *q = &diff_queued_diff;
4444         int i;
4445         git_SHA_CTX ctx;
4446         struct patch_id_t data;
4447         char buffer[PATH_MAX * 4 + 20];
4448
4449         git_SHA1_Init(&ctx);
4450         memset(&data, 0, sizeof(struct patch_id_t));
4451         data.ctx = &ctx;
4452
4453         for (i = 0; i < q->nr; i++) {
4454                 xpparam_t xpp;
4455                 xdemitconf_t xecfg;
4456                 mmfile_t mf1, mf2;
4457                 struct diff_filepair *p = q->queue[i];
4458                 int len1, len2;
4459
4460                 memset(&xpp, 0, sizeof(xpp));
4461                 memset(&xecfg, 0, sizeof(xecfg));
4462                 if (p->status == 0)
4463                         return error("internal diff status error");
4464                 if (p->status == DIFF_STATUS_UNKNOWN)
4465                         continue;
4466                 if (diff_unmodified_pair(p))
4467                         continue;
4468                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4469                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4470                         continue;
4471                 if (DIFF_PAIR_UNMERGED(p))
4472                         continue;
4473
4474                 diff_fill_sha1_info(p->one);
4475                 diff_fill_sha1_info(p->two);
4476                 if (fill_mmfile(&mf1, p->one) < 0 ||
4477                                 fill_mmfile(&mf2, p->two) < 0)
4478                         return error("unable to read files to diff");
4479
4480                 len1 = remove_space(p->one->path, strlen(p->one->path));
4481                 len2 = remove_space(p->two->path, strlen(p->two->path));
4482                 if (p->one->mode == 0)
4483                         len1 = snprintf(buffer, sizeof(buffer),
4484                                         "diff--gita/%.*sb/%.*s"
4485                                         "newfilemode%06o"
4486                                         "---/dev/null"
4487                                         "+++b/%.*s",
4488                                         len1, p->one->path,
4489                                         len2, p->two->path,
4490                                         p->two->mode,
4491                                         len2, p->two->path);
4492                 else if (p->two->mode == 0)
4493                         len1 = snprintf(buffer, sizeof(buffer),
4494                                         "diff--gita/%.*sb/%.*s"
4495                                         "deletedfilemode%06o"
4496                                         "---a/%.*s"
4497                                         "+++/dev/null",
4498                                         len1, p->one->path,
4499                                         len2, p->two->path,
4500                                         p->one->mode,
4501                                         len1, p->one->path);
4502                 else
4503                         len1 = snprintf(buffer, sizeof(buffer),
4504                                         "diff--gita/%.*sb/%.*s"
4505                                         "---a/%.*s"
4506                                         "+++b/%.*s",
4507                                         len1, p->one->path,
4508                                         len2, p->two->path,
4509                                         len1, p->one->path,
4510                                         len2, p->two->path);
4511                 git_SHA1_Update(&ctx, buffer, len1);
4512
4513                 if (diff_filespec_is_binary(p->one) ||
4514                     diff_filespec_is_binary(p->two)) {
4515                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4516                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4517                         continue;
4518                 }
4519
4520                 xpp.flags = 0;
4521                 xecfg.ctxlen = 3;
4522                 xecfg.flags = 0;
4523                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4524                                   &xpp, &xecfg))
4525                         return error("unable to generate patch-id diff for %s",
4526                                      p->one->path);
4527         }
4528
4529         git_SHA1_Final(sha1, &ctx);
4530         return 0;
4531 }
4532
4533 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4534 {
4535         struct diff_queue_struct *q = &diff_queued_diff;
4536         int i;
4537         int result = diff_get_patch_id(options, sha1);
4538
4539         for (i = 0; i < q->nr; i++)
4540                 diff_free_filepair(q->queue[i]);
4541
4542         free(q->queue);
4543         DIFF_QUEUE_CLEAR(q);
4544
4545         return result;
4546 }
4547
4548 static int is_summary_empty(const struct diff_queue_struct *q)
4549 {
4550         int i;
4551
4552         for (i = 0; i < q->nr; i++) {
4553                 const struct diff_filepair *p = q->queue[i];
4554
4555                 switch (p->status) {
4556                 case DIFF_STATUS_DELETED:
4557                 case DIFF_STATUS_ADDED:
4558                 case DIFF_STATUS_COPIED:
4559                 case DIFF_STATUS_RENAMED:
4560                         return 0;
4561                 default:
4562                         if (p->score)
4563                                 return 0;
4564                         if (p->one->mode && p->two->mode &&
4565                             p->one->mode != p->two->mode)
4566                                 return 0;
4567                         break;
4568                 }
4569         }
4570         return 1;
4571 }
4572
4573 static const char rename_limit_warning[] =
4574 "inexact rename detection was skipped due to too many files.";
4575
4576 static const char degrade_cc_to_c_warning[] =
4577 "only found copies from modified paths due to too many files.";
4578
4579 static const char rename_limit_advice[] =
4580 "you may want to set your %s variable to at least "
4581 "%d and retry the command.";
4582
4583 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4584 {
4585         if (degraded_cc)
4586                 warning(degrade_cc_to_c_warning);
4587         else if (needed)
4588                 warning(rename_limit_warning);
4589         else
4590                 return;
4591         if (0 < needed && needed < 32767)
4592                 warning(rename_limit_advice, varname, needed);
4593 }
4594
4595 void diff_flush(struct diff_options *options)
4596 {
4597         struct diff_queue_struct *q = &diff_queued_diff;
4598         int i, output_format = options->output_format;
4599         int separator = 0;
4600         int dirstat_by_line = 0;
4601
4602         /*
4603          * Order: raw, stat, summary, patch
4604          * or:    name/name-status/checkdiff (other bits clear)
4605          */
4606         if (!q->nr)
4607                 goto free_queue;
4608
4609         if (output_format & (DIFF_FORMAT_RAW |
4610                              DIFF_FORMAT_NAME |
4611                              DIFF_FORMAT_NAME_STATUS |
4612                              DIFF_FORMAT_CHECKDIFF)) {
4613                 for (i = 0; i < q->nr; i++) {
4614                         struct diff_filepair *p = q->queue[i];
4615                         if (check_pair_status(p))
4616                                 flush_one_pair(p, options);
4617                 }
4618                 separator++;
4619         }
4620
4621         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4622                 dirstat_by_line = 1;
4623
4624         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4625             dirstat_by_line) {
4626                 struct diffstat_t diffstat;
4627
4628                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4629                 for (i = 0; i < q->nr; i++) {
4630                         struct diff_filepair *p = q->queue[i];
4631                         if (check_pair_status(p))
4632                                 diff_flush_stat(p, options, &diffstat);
4633                 }
4634                 if (output_format & DIFF_FORMAT_NUMSTAT)
4635                         show_numstat(&diffstat, options);
4636                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4637                         show_stats(&diffstat, options);
4638                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4639                         show_shortstats(&diffstat, options);
4640                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4641                         show_dirstat_by_line(&diffstat, options);
4642                 free_diffstat_info(&diffstat);
4643                 separator++;
4644         }
4645         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4646                 show_dirstat(options);
4647
4648         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4649                 for (i = 0; i < q->nr; i++) {
4650                         diff_summary(options, q->queue[i]);
4651                 }
4652                 separator++;
4653         }
4654
4655         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4656             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4657             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4658                 /*
4659                  * run diff_flush_patch for the exit status. setting
4660                  * options->file to /dev/null should be safe, because we
4661                  * aren't supposed to produce any output anyway.
4662                  */
4663                 if (options->close_file)
4664                         fclose(options->file);
4665                 options->file = fopen("/dev/null", "w");
4666                 if (!options->file)
4667                         die_errno("Could not open /dev/null");
4668                 options->close_file = 1;
4669                 for (i = 0; i < q->nr; i++) {
4670                         struct diff_filepair *p = q->queue[i];
4671                         if (check_pair_status(p))
4672                                 diff_flush_patch(p, options);
4673                         if (options->found_changes)
4674                                 break;
4675                 }
4676         }
4677
4678         if (output_format & DIFF_FORMAT_PATCH) {
4679                 if (separator) {
4680                         fprintf(options->file, "%s%c",
4681                                 diff_line_prefix(options),
4682                                 options->line_termination);
4683                         if (options->stat_sep) {
4684                                 /* attach patch instead of inline */
4685                                 fputs(options->stat_sep, options->file);
4686                         }
4687                 }
4688
4689                 for (i = 0; i < q->nr; i++) {
4690                         struct diff_filepair *p = q->queue[i];
4691                         if (check_pair_status(p))
4692                                 diff_flush_patch(p, options);
4693                 }
4694         }
4695
4696         if (output_format & DIFF_FORMAT_CALLBACK)
4697                 options->format_callback(q, options, options->format_callback_data);
4698
4699         for (i = 0; i < q->nr; i++)
4700                 diff_free_filepair(q->queue[i]);
4701 free_queue:
4702         free(q->queue);
4703         DIFF_QUEUE_CLEAR(q);
4704         if (options->close_file)
4705                 fclose(options->file);
4706
4707         /*
4708          * Report the content-level differences with HAS_CHANGES;
4709          * diff_addremove/diff_change does not set the bit when
4710          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4711          */
4712         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4713                 if (options->found_changes)
4714                         DIFF_OPT_SET(options, HAS_CHANGES);
4715                 else
4716                         DIFF_OPT_CLR(options, HAS_CHANGES);
4717         }
4718 }
4719
4720 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4721 {
4722         return (((p->status == DIFF_STATUS_MODIFIED) &&
4723                  ((p->score &&
4724                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4725                   (!p->score &&
4726                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4727                 ((p->status != DIFF_STATUS_MODIFIED) &&
4728                  filter_bit_tst(p->status, options)));
4729 }
4730
4731 static void diffcore_apply_filter(struct diff_options *options)
4732 {
4733         int i;
4734         struct diff_queue_struct *q = &diff_queued_diff;
4735         struct diff_queue_struct outq;
4736
4737         DIFF_QUEUE_CLEAR(&outq);
4738
4739         if (!options->filter)
4740                 return;
4741
4742         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4743                 int found;
4744                 for (i = found = 0; !found && i < q->nr; i++) {
4745                         if (match_filter(options, q->queue[i]))
4746                                 found++;
4747                 }
4748                 if (found)
4749                         return;
4750
4751                 /* otherwise we will clear the whole queue
4752                  * by copying the empty outq at the end of this
4753                  * function, but first clear the current entries
4754                  * in the queue.
4755                  */
4756                 for (i = 0; i < q->nr; i++)
4757                         diff_free_filepair(q->queue[i]);
4758         }
4759         else {
4760                 /* Only the matching ones */
4761                 for (i = 0; i < q->nr; i++) {
4762                         struct diff_filepair *p = q->queue[i];
4763                         if (match_filter(options, p))
4764                                 diff_q(&outq, p);
4765                         else
4766                                 diff_free_filepair(p);
4767                 }
4768         }
4769         free(q->queue);
4770         *q = outq;
4771 }
4772
4773 /* Check whether two filespecs with the same mode and size are identical */
4774 static int diff_filespec_is_identical(struct diff_filespec *one,
4775                                       struct diff_filespec *two)
4776 {
4777         if (S_ISGITLINK(one->mode))
4778                 return 0;
4779         if (diff_populate_filespec(one, 0))
4780                 return 0;
4781         if (diff_populate_filespec(two, 0))
4782                 return 0;
4783         return !memcmp(one->data, two->data, one->size);
4784 }
4785
4786 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4787 {
4788         if (p->done_skip_stat_unmatch)
4789                 return p->skip_stat_unmatch_result;
4790
4791         p->done_skip_stat_unmatch = 1;
4792         p->skip_stat_unmatch_result = 0;
4793         /*
4794          * 1. Entries that come from stat info dirtiness
4795          *    always have both sides (iow, not create/delete),
4796          *    one side of the object name is unknown, with
4797          *    the same mode and size.  Keep the ones that
4798          *    do not match these criteria.  They have real
4799          *    differences.
4800          *
4801          * 2. At this point, the file is known to be modified,
4802          *    with the same mode and size, and the object
4803          *    name of one side is unknown.  Need to inspect
4804          *    the identical contents.
4805          */
4806         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4807             !DIFF_FILE_VALID(p->two) ||
4808             (p->one->sha1_valid && p->two->sha1_valid) ||
4809             (p->one->mode != p->two->mode) ||
4810             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4811             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4812             (p->one->size != p->two->size) ||
4813             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4814                 p->skip_stat_unmatch_result = 1;
4815         return p->skip_stat_unmatch_result;
4816 }
4817
4818 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4819 {
4820         int i;
4821         struct diff_queue_struct *q = &diff_queued_diff;
4822         struct diff_queue_struct outq;
4823         DIFF_QUEUE_CLEAR(&outq);
4824
4825         for (i = 0; i < q->nr; i++) {
4826                 struct diff_filepair *p = q->queue[i];
4827
4828                 if (diff_filespec_check_stat_unmatch(p))
4829                         diff_q(&outq, p);
4830                 else {
4831                         /*
4832                          * The caller can subtract 1 from skip_stat_unmatch
4833                          * to determine how many paths were dirty only
4834                          * due to stat info mismatch.
4835                          */
4836                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4837                                 diffopt->skip_stat_unmatch++;
4838                         diff_free_filepair(p);
4839                 }
4840         }
4841         free(q->queue);
4842         *q = outq;
4843 }
4844
4845 static int diffnamecmp(const void *a_, const void *b_)
4846 {
4847         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4848         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4849         const char *name_a, *name_b;
4850
4851         name_a = a->one ? a->one->path : a->two->path;
4852         name_b = b->one ? b->one->path : b->two->path;
4853         return strcmp(name_a, name_b);
4854 }
4855
4856 void diffcore_fix_diff_index(struct diff_options *options)
4857 {
4858         struct diff_queue_struct *q = &diff_queued_diff;
4859         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4860 }
4861
4862 void diffcore_std(struct diff_options *options)
4863 {
4864         /* NOTE please keep the following in sync with diff_tree_combined() */
4865         if (options->skip_stat_unmatch)
4866                 diffcore_skip_stat_unmatch(options);
4867         if (!options->found_follow) {
4868                 /* See try_to_follow_renames() in tree-diff.c */
4869                 if (options->break_opt != -1)
4870                         diffcore_break(options->break_opt);
4871                 if (options->detect_rename)
4872                         diffcore_rename(options);
4873                 if (options->break_opt != -1)
4874                         diffcore_merge_broken();
4875         }
4876         if (options->pickaxe)
4877                 diffcore_pickaxe(options);
4878         if (options->orderfile)
4879                 diffcore_order(options->orderfile);
4880         if (!options->found_follow)
4881                 /* See try_to_follow_renames() in tree-diff.c */
4882                 diff_resolve_rename_copy();
4883         diffcore_apply_filter(options);
4884
4885         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4886                 DIFF_OPT_SET(options, HAS_CHANGES);
4887         else
4888                 DIFF_OPT_CLR(options, HAS_CHANGES);
4889
4890         options->found_follow = 0;
4891 }
4892
4893 int diff_result_code(struct diff_options *opt, int status)
4894 {
4895         int result = 0;
4896
4897         diff_warn_rename_limit("diff.renameLimit",
4898                                opt->needed_rename_limit,
4899                                opt->degraded_cc_to_c);
4900         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4901             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4902                 return status;
4903         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4904             DIFF_OPT_TST(opt, HAS_CHANGES))
4905                 result |= 01;
4906         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4907             DIFF_OPT_TST(opt, CHECK_FAILED))
4908                 result |= 02;
4909         return result;
4910 }
4911
4912 int diff_can_quit_early(struct diff_options *opt)
4913 {
4914         return (DIFF_OPT_TST(opt, QUICK) &&
4915                 !opt->filter &&
4916                 DIFF_OPT_TST(opt, HAS_CHANGES));
4917 }
4918
4919 /*
4920  * Shall changes to this submodule be ignored?
4921  *
4922  * Submodule changes can be configured to be ignored separately for each path,
4923  * but that configuration can be overridden from the command line.
4924  */
4925 static int is_submodule_ignored(const char *path, struct diff_options *options)
4926 {
4927         int ignored = 0;
4928         unsigned orig_flags = options->flags;
4929         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4930                 set_diffopt_flags_from_submodule_config(options, path);
4931         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4932                 ignored = 1;
4933         options->flags = orig_flags;
4934         return ignored;
4935 }
4936
4937 void diff_addremove(struct diff_options *options,
4938                     int addremove, unsigned mode,
4939                     const unsigned char *sha1,
4940                     int sha1_valid,
4941                     const char *concatpath, unsigned dirty_submodule)
4942 {
4943         struct diff_filespec *one, *two;
4944
4945         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4946                 return;
4947
4948         /* This may look odd, but it is a preparation for
4949          * feeding "there are unchanged files which should
4950          * not produce diffs, but when you are doing copy
4951          * detection you would need them, so here they are"
4952          * entries to the diff-core.  They will be prefixed
4953          * with something like '=' or '*' (I haven't decided
4954          * which but should not make any difference).
4955          * Feeding the same new and old to diff_change()
4956          * also has the same effect.
4957          * Before the final output happens, they are pruned after
4958          * merged into rename/copy pairs as appropriate.
4959          */
4960         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4961                 addremove = (addremove == '+' ? '-' :
4962                              addremove == '-' ? '+' : addremove);
4963
4964         if (options->prefix &&
4965             strncmp(concatpath, options->prefix, options->prefix_length))
4966                 return;
4967
4968         one = alloc_filespec(concatpath);
4969         two = alloc_filespec(concatpath);
4970
4971         if (addremove != '+')
4972                 fill_filespec(one, sha1, sha1_valid, mode);
4973         if (addremove != '-') {
4974                 fill_filespec(two, sha1, sha1_valid, mode);
4975                 two->dirty_submodule = dirty_submodule;
4976         }
4977
4978         diff_queue(&diff_queued_diff, one, two);
4979         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4980                 DIFF_OPT_SET(options, HAS_CHANGES);
4981 }
4982
4983 void diff_change(struct diff_options *options,
4984                  unsigned old_mode, unsigned new_mode,
4985                  const unsigned char *old_sha1,
4986                  const unsigned char *new_sha1,
4987                  int old_sha1_valid, int new_sha1_valid,
4988                  const char *concatpath,
4989                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4990 {
4991         struct diff_filespec *one, *two;
4992         struct diff_filepair *p;
4993
4994         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4995             is_submodule_ignored(concatpath, options))
4996                 return;
4997
4998         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4999                 unsigned tmp;
5000                 const unsigned char *tmp_c;
5001                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
5002                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
5003                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
5004                         new_sha1_valid = tmp;
5005                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
5006                         new_dirty_submodule = tmp;
5007         }
5008
5009         if (options->prefix &&
5010             strncmp(concatpath, options->prefix, options->prefix_length))
5011                 return;
5012
5013         one = alloc_filespec(concatpath);
5014         two = alloc_filespec(concatpath);
5015         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5016         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5017         one->dirty_submodule = old_dirty_submodule;
5018         two->dirty_submodule = new_dirty_submodule;
5019         p = diff_queue(&diff_queued_diff, one, two);
5020
5021         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5022                 return;
5023
5024         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5025             !diff_filespec_check_stat_unmatch(p))
5026                 return;
5027
5028         DIFF_OPT_SET(options, HAS_CHANGES);
5029 }
5030
5031 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5032 {
5033         struct diff_filepair *pair;
5034         struct diff_filespec *one, *two;
5035
5036         if (options->prefix &&
5037             strncmp(path, options->prefix, options->prefix_length))
5038                 return NULL;
5039
5040         one = alloc_filespec(path);
5041         two = alloc_filespec(path);
5042         pair = diff_queue(&diff_queued_diff, one, two);
5043         pair->is_unmerged = 1;
5044         return pair;
5045 }
5046
5047 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5048                 size_t *outsize)
5049 {
5050         struct diff_tempfile *temp;
5051         const char *argv[3];
5052         const char **arg = argv;
5053         struct child_process child = CHILD_PROCESS_INIT;
5054         struct strbuf buf = STRBUF_INIT;
5055         int err = 0;
5056
5057         temp = prepare_temp_file(spec->path, spec);
5058         *arg++ = pgm;
5059         *arg++ = temp->name;
5060         *arg = NULL;
5061
5062         child.use_shell = 1;
5063         child.argv = argv;
5064         child.out = -1;
5065         if (start_command(&child)) {
5066                 remove_tempfile();
5067                 return NULL;
5068         }
5069
5070         if (strbuf_read(&buf, child.out, 0) < 0)
5071                 err = error("error reading from textconv command '%s'", pgm);
5072         close(child.out);
5073
5074         if (finish_command(&child) || err) {
5075                 strbuf_release(&buf);
5076                 remove_tempfile();
5077                 return NULL;
5078         }
5079         remove_tempfile();
5080
5081         return strbuf_detach(&buf, outsize);
5082 }
5083
5084 size_t fill_textconv(struct userdiff_driver *driver,
5085                      struct diff_filespec *df,
5086                      char **outbuf)
5087 {
5088         size_t size;
5089
5090         if (!driver) {
5091                 if (!DIFF_FILE_VALID(df)) {
5092                         *outbuf = "";
5093                         return 0;
5094                 }
5095                 if (diff_populate_filespec(df, 0))
5096                         die("unable to read files to diff");
5097                 *outbuf = df->data;
5098                 return df->size;
5099         }
5100
5101         if (!driver->textconv)
5102                 die("BUG: fill_textconv called with non-textconv driver");
5103
5104         if (driver->textconv_cache && df->sha1_valid) {
5105                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5106                                           &size);
5107                 if (*outbuf)
5108                         return size;
5109         }
5110
5111         *outbuf = run_textconv(driver->textconv, df, &size);
5112         if (!*outbuf)
5113                 die("unable to read files to diff");
5114
5115         if (driver->textconv_cache && df->sha1_valid) {
5116                 /* ignore errors, as we might be in a readonly repository */
5117                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5118                                 size);
5119                 /*
5120                  * we could save up changes and flush them all at the end,
5121                  * but we would need an extra call after all diffing is done.
5122                  * Since generating a cache entry is the slow path anyway,
5123                  * this extra overhead probably isn't a big deal.
5124                  */
5125                 notes_cache_write(driver->textconv_cache);
5126         }
5127
5128         return size;
5129 }
5130
5131 void setup_diff_pager(struct diff_options *opt)
5132 {
5133         /*
5134          * If the user asked for our exit code, then either they want --quiet
5135          * or --exit-code. We should definitely not bother with a pager in the
5136          * former case, as we will generate no output. Since we still properly
5137          * report our exit code even when a pager is run, we _could_ run a
5138          * pager with --exit-code. But since we have not done so historically,
5139          * and because it is easy to find people oneline advising "git diff
5140          * --exit-code" in hooks and other scripts, we do not do so.
5141          */
5142         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5143             check_pager_config("diff") != 0)
5144                 setup_pager();
5145 }