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