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