Merge commit 'git/master'
[git/git-bigfiles.git] / diff.c
blob75d9fab8f87033ba86f687fbc92c5d4ee4c68d7b
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"
17 #ifdef NO_FAST_WORKING_DIRECTORY
18 #define FAST_WORKING_DIRECTORY 0
19 #else
20 #define FAST_WORKING_DIRECTORY 1
21 #endif
23 static int diff_detect_rename_default;
24 static int diff_rename_limit_default = 200;
25 static int diff_suppress_blank_empty;
26 int diff_use_color_default = -1;
27 static const char *diff_word_regex_cfg;
28 static const char *external_diff_cmd_cfg;
29 int diff_auto_refresh_index = 1;
30 static int diff_mnemonic_prefix;
32 static char diff_colors[][COLOR_MAXLEN] = {
33 GIT_COLOR_RESET,
34 GIT_COLOR_NORMAL, /* PLAIN */
35 GIT_COLOR_BOLD, /* METAINFO */
36 GIT_COLOR_CYAN, /* FRAGINFO */
37 GIT_COLOR_RED, /* OLD */
38 GIT_COLOR_GREEN, /* NEW */
39 GIT_COLOR_YELLOW, /* COMMIT */
40 GIT_COLOR_BG_RED, /* WHITESPACE */
43 static void diff_filespec_load_driver(struct diff_filespec *one);
44 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
46 static int parse_diff_color_slot(const char *var, int ofs)
48 if (!strcasecmp(var+ofs, "plain"))
49 return DIFF_PLAIN;
50 if (!strcasecmp(var+ofs, "meta"))
51 return DIFF_METAINFO;
52 if (!strcasecmp(var+ofs, "frag"))
53 return DIFF_FRAGINFO;
54 if (!strcasecmp(var+ofs, "old"))
55 return DIFF_FILE_OLD;
56 if (!strcasecmp(var+ofs, "new"))
57 return DIFF_FILE_NEW;
58 if (!strcasecmp(var+ofs, "commit"))
59 return DIFF_COMMIT;
60 if (!strcasecmp(var+ofs, "whitespace"))
61 return DIFF_WHITESPACE;
62 die("bad config variable '%s'", var);
66 * These are to give UI layer defaults.
67 * The core-level commands such as git-diff-files should
68 * never be affected by the setting of diff.renames
69 * the user happens to have in the configuration file.
71 int git_diff_ui_config(const char *var, const char *value, void *cb)
73 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
74 diff_use_color_default = git_config_colorbool(var, value, -1);
75 return 0;
77 if (!strcmp(var, "diff.renames")) {
78 if (!value)
79 diff_detect_rename_default = DIFF_DETECT_RENAME;
80 else if (!strcasecmp(value, "copies") ||
81 !strcasecmp(value, "copy"))
82 diff_detect_rename_default = DIFF_DETECT_COPY;
83 else if (git_config_bool(var,value))
84 diff_detect_rename_default = DIFF_DETECT_RENAME;
85 return 0;
87 if (!strcmp(var, "diff.autorefreshindex")) {
88 diff_auto_refresh_index = git_config_bool(var, value);
89 return 0;
91 if (!strcmp(var, "diff.mnemonicprefix")) {
92 diff_mnemonic_prefix = git_config_bool(var, value);
93 return 0;
95 if (!strcmp(var, "diff.external"))
96 return git_config_string(&external_diff_cmd_cfg, var, value);
97 if (!strcmp(var, "diff.wordregex"))
98 return git_config_string(&diff_word_regex_cfg, var, value);
100 return git_diff_basic_config(var, value, cb);
103 int git_diff_basic_config(const char *var, const char *value, void *cb)
105 if (!strcmp(var, "diff.renamelimit")) {
106 diff_rename_limit_default = git_config_int(var, value);
107 return 0;
110 switch (userdiff_config(var, value)) {
111 case 0: break;
112 case -1: return -1;
113 default: return 0;
116 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
117 int slot = parse_diff_color_slot(var, 11);
118 if (!value)
119 return config_error_nonbool(var);
120 color_parse(value, var, diff_colors[slot]);
121 return 0;
124 /* like GNU diff's --suppress-blank-empty option */
125 if (!strcmp(var, "diff.suppressblankempty") ||
126 /* for backwards compatibility */
127 !strcmp(var, "diff.suppress-blank-empty")) {
128 diff_suppress_blank_empty = git_config_bool(var, value);
129 return 0;
132 return git_color_default_config(var, value, cb);
135 static char *quote_two(const char *one, const char *two)
137 int need_one = quote_c_style(one, NULL, NULL, 1);
138 int need_two = quote_c_style(two, NULL, NULL, 1);
139 struct strbuf res = STRBUF_INIT;
141 if (need_one + need_two) {
142 strbuf_addch(&res, '"');
143 quote_c_style(one, &res, NULL, 1);
144 quote_c_style(two, &res, NULL, 1);
145 strbuf_addch(&res, '"');
146 } else {
147 strbuf_addstr(&res, one);
148 strbuf_addstr(&res, two);
150 return strbuf_detach(&res, NULL);
153 static const char *external_diff(void)
155 static const char *external_diff_cmd = NULL;
156 static int done_preparing = 0;
158 if (done_preparing)
159 return external_diff_cmd;
160 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
161 if (!external_diff_cmd)
162 external_diff_cmd = external_diff_cmd_cfg;
163 done_preparing = 1;
164 return external_diff_cmd;
167 static struct diff_tempfile {
168 const char *name; /* filename external diff should read from */
169 char hex[41];
170 char mode[10];
171 char tmp_path[PATH_MAX];
172 } diff_temp[2];
174 static struct diff_tempfile *claim_diff_tempfile(void) {
175 int i;
176 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
177 if (!diff_temp[i].name)
178 return diff_temp + i;
179 die("BUG: diff is failing to clean up its tempfiles");
182 static int remove_tempfile_installed;
184 static void remove_tempfile(void)
186 int i;
187 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
188 if (diff_temp[i].name == diff_temp[i].tmp_path)
189 unlink(diff_temp[i].name);
190 diff_temp[i].name = NULL;
194 static void remove_tempfile_on_signal(int signo)
196 remove_tempfile();
197 sigchain_pop(signo);
198 raise(signo);
201 static int count_lines(const char *data, int size)
203 int count, ch, completely_empty = 1, nl_just_seen = 0;
204 count = 0;
205 while (0 < size--) {
206 ch = *data++;
207 if (ch == '\n') {
208 count++;
209 nl_just_seen = 1;
210 completely_empty = 0;
212 else {
213 nl_just_seen = 0;
214 completely_empty = 0;
217 if (completely_empty)
218 return 0;
219 if (!nl_just_seen)
220 count++; /* no trailing newline */
221 return count;
224 static void print_line_count(FILE *file, int count)
226 switch (count) {
227 case 0:
228 fprintf(file, "0,0");
229 break;
230 case 1:
231 fprintf(file, "1");
232 break;
233 default:
234 fprintf(file, "1,%d", count);
235 break;
239 static void copy_file_with_prefix(FILE *file,
240 int prefix, const char *data, int size,
241 const char *set, const char *reset)
243 int ch, nl_just_seen = 1;
244 while (0 < size--) {
245 ch = *data++;
246 if (nl_just_seen) {
247 fputs(set, file);
248 putc(prefix, file);
250 if (ch == '\n') {
251 nl_just_seen = 1;
252 fputs(reset, file);
253 } else
254 nl_just_seen = 0;
255 putc(ch, file);
257 if (!nl_just_seen)
258 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
261 static void emit_rewrite_diff(const char *name_a,
262 const char *name_b,
263 struct diff_filespec *one,
264 struct diff_filespec *two,
265 const char *textconv_one,
266 const char *textconv_two,
267 struct diff_options *o)
269 int lc_a, lc_b;
270 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
271 const char *name_a_tab, *name_b_tab;
272 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
273 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
274 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
275 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
276 const char *reset = diff_get_color(color_diff, DIFF_RESET);
277 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
278 const char *a_prefix, *b_prefix;
279 const char *data_one, *data_two;
280 size_t size_one, size_two;
282 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
283 a_prefix = o->b_prefix;
284 b_prefix = o->a_prefix;
285 } else {
286 a_prefix = o->a_prefix;
287 b_prefix = o->b_prefix;
290 name_a += (*name_a == '/');
291 name_b += (*name_b == '/');
292 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
293 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
295 strbuf_reset(&a_name);
296 strbuf_reset(&b_name);
297 quote_two_c_style(&a_name, a_prefix, name_a, 0);
298 quote_two_c_style(&b_name, b_prefix, name_b, 0);
300 diff_populate_filespec(one, 0);
301 diff_populate_filespec(two, 0);
302 if (textconv_one) {
303 data_one = run_textconv(textconv_one, one, &size_one);
304 if (!data_one)
305 die("unable to read files to diff");
307 else {
308 data_one = one->data;
309 size_one = one->size;
311 if (textconv_two) {
312 data_two = run_textconv(textconv_two, two, &size_two);
313 if (!data_two)
314 die("unable to read files to diff");
316 else {
317 data_two = two->data;
318 size_two = two->size;
321 lc_a = count_lines(data_one, size_one);
322 lc_b = count_lines(data_two, size_two);
323 fprintf(o->file,
324 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
325 metainfo, a_name.buf, name_a_tab, reset,
326 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
327 print_line_count(o->file, lc_a);
328 fprintf(o->file, " +");
329 print_line_count(o->file, lc_b);
330 fprintf(o->file, " @@%s\n", reset);
331 if (lc_a)
332 copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
333 if (lc_b)
334 copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
337 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
339 if (!DIFF_FILE_VALID(one)) {
340 mf->ptr = (char *)""; /* does not matter */
341 mf->size = 0;
342 return 0;
344 else if (diff_populate_filespec(one, 0))
345 return -1;
347 mf->ptr = one->data;
348 mf->size = one->size;
349 return 0;
352 struct diff_words_buffer {
353 mmfile_t text;
354 long alloc;
355 struct diff_words_orig {
356 const char *begin, *end;
357 } *orig;
358 int orig_nr, orig_alloc;
361 static void diff_words_append(char *line, unsigned long len,
362 struct diff_words_buffer *buffer)
364 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
365 line++;
366 len--;
367 memcpy(buffer->text.ptr + buffer->text.size, line, len);
368 buffer->text.size += len;
369 buffer->text.ptr[buffer->text.size] = '\0';
372 struct diff_words_data {
373 struct diff_words_buffer minus, plus;
374 const char *current_plus;
375 FILE *file;
376 regex_t *word_regex;
379 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
381 struct diff_words_data *diff_words = priv;
382 int minus_first, minus_len, plus_first, plus_len;
383 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
385 if (line[0] != '@' || parse_hunk_header(line, len,
386 &minus_first, &minus_len, &plus_first, &plus_len))
387 return;
389 /* POSIX requires that first be decremented by one if len == 0... */
390 if (minus_len) {
391 minus_begin = diff_words->minus.orig[minus_first].begin;
392 minus_end =
393 diff_words->minus.orig[minus_first + minus_len - 1].end;
394 } else
395 minus_begin = minus_end =
396 diff_words->minus.orig[minus_first].end;
398 if (plus_len) {
399 plus_begin = diff_words->plus.orig[plus_first].begin;
400 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
401 } else
402 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
404 if (diff_words->current_plus != plus_begin)
405 fwrite(diff_words->current_plus,
406 plus_begin - diff_words->current_plus, 1,
407 diff_words->file);
408 if (minus_begin != minus_end)
409 color_fwrite_lines(diff_words->file,
410 diff_get_color(1, DIFF_FILE_OLD),
411 minus_end - minus_begin, minus_begin);
412 if (plus_begin != plus_end)
413 color_fwrite_lines(diff_words->file,
414 diff_get_color(1, DIFF_FILE_NEW),
415 plus_end - plus_begin, plus_begin);
417 diff_words->current_plus = plus_end;
420 /* This function starts looking at *begin, and returns 0 iff a word was found. */
421 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
422 int *begin, int *end)
424 if (word_regex && *begin < buffer->size) {
425 regmatch_t match[1];
426 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
427 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
428 '\n', match[0].rm_eo - match[0].rm_so);
429 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
430 *begin += match[0].rm_so;
431 return *begin >= *end;
433 return -1;
436 /* find the next word */
437 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
438 (*begin)++;
439 if (*begin >= buffer->size)
440 return -1;
442 /* find the end of the word */
443 *end = *begin + 1;
444 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
445 (*end)++;
447 return 0;
451 * This function splits the words in buffer->text, stores the list with
452 * newline separator into out, and saves the offsets of the original words
453 * in buffer->orig.
455 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
456 regex_t *word_regex)
458 int i, j;
459 long alloc = 0;
461 out->size = 0;
462 out->ptr = NULL;
464 /* fake an empty "0th" word */
465 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
466 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
467 buffer->orig_nr = 1;
469 for (i = 0; i < buffer->text.size; i++) {
470 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
471 return;
473 /* store original boundaries */
474 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
475 buffer->orig_alloc);
476 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
477 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
478 buffer->orig_nr++;
480 /* store one word */
481 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
482 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
483 out->ptr[out->size + j - i] = '\n';
484 out->size += j - i + 1;
486 i = j - 1;
490 /* this executes the word diff on the accumulated buffers */
491 static void diff_words_show(struct diff_words_data *diff_words)
493 xpparam_t xpp;
494 xdemitconf_t xecfg;
495 xdemitcb_t ecb;
496 mmfile_t minus, plus;
498 /* special case: only removal */
499 if (!diff_words->plus.text.size) {
500 color_fwrite_lines(diff_words->file,
501 diff_get_color(1, DIFF_FILE_OLD),
502 diff_words->minus.text.size, diff_words->minus.text.ptr);
503 diff_words->minus.text.size = 0;
504 return;
507 diff_words->current_plus = diff_words->plus.text.ptr;
509 memset(&xpp, 0, sizeof(xpp));
510 memset(&xecfg, 0, sizeof(xecfg));
511 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
512 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
513 xpp.flags = XDF_NEED_MINIMAL;
514 /* as only the hunk header will be parsed, we need a 0-context */
515 xecfg.ctxlen = 0;
516 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
517 &xpp, &xecfg, &ecb);
518 free(minus.ptr);
519 free(plus.ptr);
520 if (diff_words->current_plus != diff_words->plus.text.ptr +
521 diff_words->plus.text.size)
522 fwrite(diff_words->current_plus,
523 diff_words->plus.text.ptr + diff_words->plus.text.size
524 - diff_words->current_plus, 1,
525 diff_words->file);
526 diff_words->minus.text.size = diff_words->plus.text.size = 0;
529 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
531 struct emit_callback {
532 int nparents, color_diff;
533 unsigned ws_rule;
534 sane_truncate_fn truncate;
535 const char **label_path;
536 struct diff_words_data *diff_words;
537 int *found_changesp;
538 FILE *file;
541 static void free_diff_words_data(struct emit_callback *ecbdata)
543 if (ecbdata->diff_words) {
544 /* flush buffers */
545 if (ecbdata->diff_words->minus.text.size ||
546 ecbdata->diff_words->plus.text.size)
547 diff_words_show(ecbdata->diff_words);
549 free (ecbdata->diff_words->minus.text.ptr);
550 free (ecbdata->diff_words->minus.orig);
551 free (ecbdata->diff_words->plus.text.ptr);
552 free (ecbdata->diff_words->plus.orig);
553 free(ecbdata->diff_words->word_regex);
554 free(ecbdata->diff_words);
555 ecbdata->diff_words = NULL;
559 const char *diff_get_color(int diff_use_color, enum color_diff ix)
561 if (diff_use_color)
562 return diff_colors[ix];
563 return "";
566 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
568 int has_trailing_newline, has_trailing_carriage_return;
570 has_trailing_newline = (len > 0 && line[len-1] == '\n');
571 if (has_trailing_newline)
572 len--;
573 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
574 if (has_trailing_carriage_return)
575 len--;
577 fputs(set, file);
578 fwrite(line, len, 1, file);
579 fputs(reset, file);
580 if (has_trailing_carriage_return)
581 fputc('\r', file);
582 if (has_trailing_newline)
583 fputc('\n', file);
586 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
588 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
589 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
591 if (!*ws)
592 emit_line(ecbdata->file, set, reset, line, len);
593 else {
594 /* Emit just the prefix, then the rest. */
595 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
596 ws_check_emit(line + ecbdata->nparents,
597 len - ecbdata->nparents, ecbdata->ws_rule,
598 ecbdata->file, set, reset, ws);
602 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
604 const char *cp;
605 unsigned long allot;
606 size_t l = len;
608 if (ecb->truncate)
609 return ecb->truncate(line, len);
610 cp = line;
611 allot = l;
612 while (0 < l) {
613 (void) utf8_width(&cp, &l);
614 if (!cp)
615 break; /* truncated in the middle? */
617 return allot - l;
620 static void fn_out_consume(void *priv, char *line, unsigned long len)
622 int i;
623 int color;
624 struct emit_callback *ecbdata = priv;
625 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
626 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
627 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
629 *(ecbdata->found_changesp) = 1;
631 if (ecbdata->label_path[0]) {
632 const char *name_a_tab, *name_b_tab;
634 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
635 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
637 fprintf(ecbdata->file, "%s--- %s%s%s\n",
638 meta, ecbdata->label_path[0], reset, name_a_tab);
639 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
640 meta, ecbdata->label_path[1], reset, name_b_tab);
641 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
644 if (diff_suppress_blank_empty
645 && len == 2 && line[0] == ' ' && line[1] == '\n') {
646 line[0] = '\n';
647 len = 1;
650 /* This is not really necessary for now because
651 * this codepath only deals with two-way diffs.
653 for (i = 0; i < len && line[i] == '@'; i++)
655 if (2 <= i && i < len && line[i] == ' ') {
656 ecbdata->nparents = i - 1;
657 len = sane_truncate_line(ecbdata, line, len);
658 emit_line(ecbdata->file,
659 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
660 reset, line, len);
661 if (line[len-1] != '\n')
662 putc('\n', ecbdata->file);
663 return;
666 if (len < ecbdata->nparents) {
667 emit_line(ecbdata->file, reset, reset, line, len);
668 return;
671 color = DIFF_PLAIN;
672 if (ecbdata->diff_words && ecbdata->nparents != 1)
673 /* fall back to normal diff */
674 free_diff_words_data(ecbdata);
675 if (ecbdata->diff_words) {
676 if (line[0] == '-') {
677 diff_words_append(line, len,
678 &ecbdata->diff_words->minus);
679 return;
680 } else if (line[0] == '+') {
681 diff_words_append(line, len,
682 &ecbdata->diff_words->plus);
683 return;
685 if (ecbdata->diff_words->minus.text.size ||
686 ecbdata->diff_words->plus.text.size)
687 diff_words_show(ecbdata->diff_words);
688 line++;
689 len--;
690 emit_line(ecbdata->file, plain, reset, line, len);
691 return;
693 for (i = 0; i < ecbdata->nparents && len; i++) {
694 if (line[i] == '-')
695 color = DIFF_FILE_OLD;
696 else if (line[i] == '+')
697 color = DIFF_FILE_NEW;
700 if (color != DIFF_FILE_NEW) {
701 emit_line(ecbdata->file,
702 diff_get_color(ecbdata->color_diff, color),
703 reset, line, len);
704 return;
706 emit_add_line(reset, ecbdata, line, len);
709 static char *pprint_rename(const char *a, const char *b)
711 const char *old = a;
712 const char *new = b;
713 struct strbuf name = STRBUF_INIT;
714 int pfx_length, sfx_length;
715 int len_a = strlen(a);
716 int len_b = strlen(b);
717 int a_midlen, b_midlen;
718 int qlen_a = quote_c_style(a, NULL, NULL, 0);
719 int qlen_b = quote_c_style(b, NULL, NULL, 0);
721 if (qlen_a || qlen_b) {
722 quote_c_style(a, &name, NULL, 0);
723 strbuf_addstr(&name, " => ");
724 quote_c_style(b, &name, NULL, 0);
725 return strbuf_detach(&name, NULL);
728 /* Find common prefix */
729 pfx_length = 0;
730 while (*old && *new && *old == *new) {
731 if (*old == '/')
732 pfx_length = old - a + 1;
733 old++;
734 new++;
737 /* Find common suffix */
738 old = a + len_a;
739 new = b + len_b;
740 sfx_length = 0;
741 while (a <= old && b <= new && *old == *new) {
742 if (*old == '/')
743 sfx_length = len_a - (old - a);
744 old--;
745 new--;
749 * pfx{mid-a => mid-b}sfx
750 * {pfx-a => pfx-b}sfx
751 * pfx{sfx-a => sfx-b}
752 * name-a => name-b
754 a_midlen = len_a - pfx_length - sfx_length;
755 b_midlen = len_b - pfx_length - sfx_length;
756 if (a_midlen < 0)
757 a_midlen = 0;
758 if (b_midlen < 0)
759 b_midlen = 0;
761 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
762 if (pfx_length + sfx_length) {
763 strbuf_add(&name, a, pfx_length);
764 strbuf_addch(&name, '{');
766 strbuf_add(&name, a + pfx_length, a_midlen);
767 strbuf_addstr(&name, " => ");
768 strbuf_add(&name, b + pfx_length, b_midlen);
769 if (pfx_length + sfx_length) {
770 strbuf_addch(&name, '}');
771 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
773 return strbuf_detach(&name, NULL);
776 struct diffstat_t {
777 int nr;
778 int alloc;
779 struct diffstat_file {
780 char *from_name;
781 char *name;
782 char *print_name;
783 unsigned is_unmerged:1;
784 unsigned is_binary:1;
785 unsigned is_renamed:1;
786 unsigned int added, deleted;
787 } **files;
790 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
791 const char *name_a,
792 const char *name_b)
794 struct diffstat_file *x;
795 x = xcalloc(sizeof (*x), 1);
796 if (diffstat->nr == diffstat->alloc) {
797 diffstat->alloc = alloc_nr(diffstat->alloc);
798 diffstat->files = xrealloc(diffstat->files,
799 diffstat->alloc * sizeof(x));
801 diffstat->files[diffstat->nr++] = x;
802 if (name_b) {
803 x->from_name = xstrdup(name_a);
804 x->name = xstrdup(name_b);
805 x->is_renamed = 1;
807 else {
808 x->from_name = NULL;
809 x->name = xstrdup(name_a);
811 return x;
814 static void diffstat_consume(void *priv, char *line, unsigned long len)
816 struct diffstat_t *diffstat = priv;
817 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
819 if (line[0] == '+')
820 x->added++;
821 else if (line[0] == '-')
822 x->deleted++;
825 const char mime_boundary_leader[] = "------------";
827 static int scale_linear(int it, int width, int max_change)
830 * make sure that at least one '-' is printed if there were deletions,
831 * and likewise for '+'.
833 if (max_change < 2)
834 return it;
835 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
838 static void show_name(FILE *file,
839 const char *prefix, const char *name, int len,
840 const char *reset, const char *set)
842 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
845 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
847 if (cnt <= 0)
848 return;
849 fprintf(file, "%s", set);
850 while (cnt--)
851 putc(ch, file);
852 fprintf(file, "%s", reset);
855 static void fill_print_name(struct diffstat_file *file)
857 char *pname;
859 if (file->print_name)
860 return;
862 if (!file->is_renamed) {
863 struct strbuf buf = STRBUF_INIT;
864 if (quote_c_style(file->name, &buf, NULL, 0)) {
865 pname = strbuf_detach(&buf, NULL);
866 } else {
867 pname = file->name;
868 strbuf_release(&buf);
870 } else {
871 pname = pprint_rename(file->from_name, file->name);
873 file->print_name = pname;
876 static void show_stats(struct diffstat_t* data, struct diff_options *options)
878 int i, len, add, del, adds = 0, dels = 0;
879 int max_change = 0, max_len = 0;
880 int total_files = data->nr;
881 int width, name_width;
882 const char *reset, *set, *add_c, *del_c;
884 if (data->nr == 0)
885 return;
887 width = options->stat_width ? options->stat_width : 80;
888 name_width = options->stat_name_width ? options->stat_name_width : 50;
890 /* Sanity: give at least 5 columns to the graph,
891 * but leave at least 10 columns for the name.
893 if (width < 25)
894 width = 25;
895 if (name_width < 10)
896 name_width = 10;
897 else if (width < name_width + 15)
898 name_width = width - 15;
900 /* Find the longest filename and max number of changes */
901 reset = diff_get_color_opt(options, DIFF_RESET);
902 set = diff_get_color_opt(options, DIFF_PLAIN);
903 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
904 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
906 for (i = 0; i < data->nr; i++) {
907 struct diffstat_file *file = data->files[i];
908 int change = file->added + file->deleted;
909 fill_print_name(file);
910 len = strlen(file->print_name);
911 if (max_len < len)
912 max_len = len;
914 if (file->is_binary || file->is_unmerged)
915 continue;
916 if (max_change < change)
917 max_change = change;
920 /* Compute the width of the graph part;
921 * 10 is for one blank at the beginning of the line plus
922 * " | count " between the name and the graph.
924 * From here on, name_width is the width of the name area,
925 * and width is the width of the graph area.
927 name_width = (name_width < max_len) ? name_width : max_len;
928 if (width < (name_width + 10) + max_change)
929 width = width - (name_width + 10);
930 else
931 width = max_change;
933 for (i = 0; i < data->nr; i++) {
934 const char *prefix = "";
935 char *name = data->files[i]->print_name;
936 int added = data->files[i]->added;
937 int deleted = data->files[i]->deleted;
938 int name_len;
941 * "scale" the filename
943 len = name_width;
944 name_len = strlen(name);
945 if (name_width < name_len) {
946 char *slash;
947 prefix = "...";
948 len -= 3;
949 name += name_len - len;
950 slash = strchr(name, '/');
951 if (slash)
952 name = slash;
955 if (data->files[i]->is_binary) {
956 show_name(options->file, prefix, name, len, reset, set);
957 fprintf(options->file, " Bin ");
958 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
959 fprintf(options->file, " -> ");
960 fprintf(options->file, "%s%d%s", add_c, added, reset);
961 fprintf(options->file, " bytes");
962 fprintf(options->file, "\n");
963 continue;
965 else if (data->files[i]->is_unmerged) {
966 show_name(options->file, prefix, name, len, reset, set);
967 fprintf(options->file, " Unmerged\n");
968 continue;
970 else if (!data->files[i]->is_renamed &&
971 (added + deleted == 0)) {
972 total_files--;
973 continue;
977 * scale the add/delete
979 add = added;
980 del = deleted;
981 adds += add;
982 dels += del;
984 if (width <= max_change) {
985 add = scale_linear(add, width, max_change);
986 del = scale_linear(del, width, max_change);
988 show_name(options->file, prefix, name, len, reset, set);
989 fprintf(options->file, "%5d%s", added + deleted,
990 added + deleted ? " " : "");
991 show_graph(options->file, '+', add, add_c, reset);
992 show_graph(options->file, '-', del, del_c, reset);
993 fprintf(options->file, "\n");
995 fprintf(options->file,
996 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
997 set, total_files, adds, dels, reset);
1000 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
1002 int i, adds = 0, dels = 0, total_files = data->nr;
1004 if (data->nr == 0)
1005 return;
1007 for (i = 0; i < data->nr; i++) {
1008 if (!data->files[i]->is_binary &&
1009 !data->files[i]->is_unmerged) {
1010 int added = data->files[i]->added;
1011 int deleted= data->files[i]->deleted;
1012 if (!data->files[i]->is_renamed &&
1013 (added + deleted == 0)) {
1014 total_files--;
1015 } else {
1016 adds += added;
1017 dels += deleted;
1021 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1022 total_files, adds, dels);
1025 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
1027 int i;
1029 if (data->nr == 0)
1030 return;
1032 for (i = 0; i < data->nr; i++) {
1033 struct diffstat_file *file = data->files[i];
1035 if (file->is_binary)
1036 fprintf(options->file, "-\t-\t");
1037 else
1038 fprintf(options->file,
1039 "%d\t%d\t", file->added, file->deleted);
1040 if (options->line_termination) {
1041 fill_print_name(file);
1042 if (!file->is_renamed)
1043 write_name_quoted(file->name, options->file,
1044 options->line_termination);
1045 else {
1046 fputs(file->print_name, options->file);
1047 putc(options->line_termination, options->file);
1049 } else {
1050 if (file->is_renamed) {
1051 putc('\0', options->file);
1052 write_name_quoted(file->from_name, options->file, '\0');
1054 write_name_quoted(file->name, options->file, '\0');
1059 struct dirstat_file {
1060 const char *name;
1061 unsigned long changed;
1064 struct dirstat_dir {
1065 struct dirstat_file *files;
1066 int alloc, nr, percent, cumulative;
1069 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1071 unsigned long this_dir = 0;
1072 unsigned int sources = 0;
1074 while (dir->nr) {
1075 struct dirstat_file *f = dir->files;
1076 int namelen = strlen(f->name);
1077 unsigned long this;
1078 char *slash;
1080 if (namelen < baselen)
1081 break;
1082 if (memcmp(f->name, base, baselen))
1083 break;
1084 slash = strchr(f->name + baselen, '/');
1085 if (slash) {
1086 int newbaselen = slash + 1 - f->name;
1087 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1088 sources++;
1089 } else {
1090 this = f->changed;
1091 dir->files++;
1092 dir->nr--;
1093 sources += 2;
1095 this_dir += this;
1099 * We don't report dirstat's for
1100 * - the top level
1101 * - or cases where everything came from a single directory
1102 * under this directory (sources == 1).
1104 if (baselen && sources != 1) {
1105 int permille = this_dir * 1000 / changed;
1106 if (permille) {
1107 int percent = permille / 10;
1108 if (percent >= dir->percent) {
1109 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1110 if (!dir->cumulative)
1111 return 0;
1115 return this_dir;
1118 static int dirstat_compare(const void *_a, const void *_b)
1120 const struct dirstat_file *a = _a;
1121 const struct dirstat_file *b = _b;
1122 return strcmp(a->name, b->name);
1125 static void show_dirstat(struct diff_options *options)
1127 int i;
1128 unsigned long changed;
1129 struct dirstat_dir dir;
1130 struct diff_queue_struct *q = &diff_queued_diff;
1132 dir.files = NULL;
1133 dir.alloc = 0;
1134 dir.nr = 0;
1135 dir.percent = options->dirstat_percent;
1136 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1138 changed = 0;
1139 for (i = 0; i < q->nr; i++) {
1140 struct diff_filepair *p = q->queue[i];
1141 const char *name;
1142 unsigned long copied, added, damage;
1144 name = p->one->path ? p->one->path : p->two->path;
1146 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1147 diff_populate_filespec(p->one, 0);
1148 diff_populate_filespec(p->two, 0);
1149 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1150 &copied, &added);
1151 diff_free_filespec_data(p->one);
1152 diff_free_filespec_data(p->two);
1153 } else if (DIFF_FILE_VALID(p->one)) {
1154 diff_populate_filespec(p->one, 1);
1155 copied = added = 0;
1156 diff_free_filespec_data(p->one);
1157 } else if (DIFF_FILE_VALID(p->two)) {
1158 diff_populate_filespec(p->two, 1);
1159 copied = 0;
1160 added = p->two->size;
1161 diff_free_filespec_data(p->two);
1162 } else
1163 continue;
1166 * Original minus copied is the removed material,
1167 * added is the new material. They are both damages
1168 * made to the preimage. In --dirstat-by-file mode, count
1169 * damaged files, not damaged lines. This is done by
1170 * counting only a single damaged line per file.
1172 damage = (p->one->size - copied) + added;
1173 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1174 damage = 1;
1176 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1177 dir.files[dir.nr].name = name;
1178 dir.files[dir.nr].changed = damage;
1179 changed += damage;
1180 dir.nr++;
1183 /* This can happen even with many files, if everything was renames */
1184 if (!changed)
1185 return;
1187 /* Show all directories with more than x% of the changes */
1188 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1189 gather_dirstat(options->file, &dir, changed, "", 0);
1192 static void free_diffstat_info(struct diffstat_t *diffstat)
1194 int i;
1195 for (i = 0; i < diffstat->nr; i++) {
1196 struct diffstat_file *f = diffstat->files[i];
1197 if (f->name != f->print_name)
1198 free(f->print_name);
1199 free(f->name);
1200 free(f->from_name);
1201 free(f);
1203 free(diffstat->files);
1206 struct checkdiff_t {
1207 const char *filename;
1208 int lineno;
1209 struct diff_options *o;
1210 unsigned ws_rule;
1211 unsigned status;
1212 int trailing_blanks_start;
1215 static int is_conflict_marker(const char *line, unsigned long len)
1217 char firstchar;
1218 int cnt;
1220 if (len < 8)
1221 return 0;
1222 firstchar = line[0];
1223 switch (firstchar) {
1224 case '=': case '>': case '<':
1225 break;
1226 default:
1227 return 0;
1229 for (cnt = 1; cnt < 7; cnt++)
1230 if (line[cnt] != firstchar)
1231 return 0;
1232 /* line[0] thru line[6] are same as firstchar */
1233 if (firstchar == '=') {
1234 /* divider between ours and theirs? */
1235 if (len != 8 || line[7] != '\n')
1236 return 0;
1237 } else if (len < 8 || !isspace(line[7])) {
1238 /* not divider before ours nor after theirs */
1239 return 0;
1241 return 1;
1244 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1246 struct checkdiff_t *data = priv;
1247 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1248 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1249 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1250 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1251 char *err;
1253 if (line[0] == '+') {
1254 unsigned bad;
1255 data->lineno++;
1256 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1257 data->trailing_blanks_start = 0;
1258 else if (!data->trailing_blanks_start)
1259 data->trailing_blanks_start = data->lineno;
1260 if (is_conflict_marker(line + 1, len - 1)) {
1261 data->status |= 1;
1262 fprintf(data->o->file,
1263 "%s:%d: leftover conflict marker\n",
1264 data->filename, data->lineno);
1266 bad = ws_check(line + 1, len - 1, data->ws_rule);
1267 if (!bad)
1268 return;
1269 data->status |= bad;
1270 err = whitespace_error_string(bad);
1271 fprintf(data->o->file, "%s:%d: %s.\n",
1272 data->filename, data->lineno, err);
1273 free(err);
1274 emit_line(data->o->file, set, reset, line, 1);
1275 ws_check_emit(line + 1, len - 1, data->ws_rule,
1276 data->o->file, set, reset, ws);
1277 } else if (line[0] == ' ') {
1278 data->lineno++;
1279 data->trailing_blanks_start = 0;
1280 } else if (line[0] == '@') {
1281 char *plus = strchr(line, '+');
1282 if (plus)
1283 data->lineno = strtol(plus, NULL, 10) - 1;
1284 else
1285 die("invalid diff");
1286 data->trailing_blanks_start = 0;
1290 static unsigned char *deflate_it(char *data,
1291 unsigned long size,
1292 unsigned long *result_size)
1294 int bound;
1295 unsigned char *deflated;
1296 z_stream stream;
1298 memset(&stream, 0, sizeof(stream));
1299 deflateInit(&stream, zlib_compression_level);
1300 bound = deflateBound(&stream, size);
1301 deflated = xmalloc(bound);
1302 stream.next_out = deflated;
1303 stream.avail_out = bound;
1305 stream.next_in = (unsigned char *)data;
1306 stream.avail_in = size;
1307 while (deflate(&stream, Z_FINISH) == Z_OK)
1308 ; /* nothing */
1309 deflateEnd(&stream);
1310 *result_size = stream.total_out;
1311 return deflated;
1314 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1316 void *cp;
1317 void *delta;
1318 void *deflated;
1319 void *data;
1320 unsigned long orig_size;
1321 unsigned long delta_size;
1322 unsigned long deflate_size;
1323 unsigned long data_size;
1325 /* We could do deflated delta, or we could do just deflated two,
1326 * whichever is smaller.
1328 delta = NULL;
1329 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1330 if (one->size && two->size) {
1331 delta = diff_delta(one->ptr, one->size,
1332 two->ptr, two->size,
1333 &delta_size, deflate_size);
1334 if (delta) {
1335 void *to_free = delta;
1336 orig_size = delta_size;
1337 delta = deflate_it(delta, delta_size, &delta_size);
1338 free(to_free);
1342 if (delta && delta_size < deflate_size) {
1343 fprintf(file, "delta %lu\n", orig_size);
1344 free(deflated);
1345 data = delta;
1346 data_size = delta_size;
1348 else {
1349 fprintf(file, "literal %lu\n", two->size);
1350 free(delta);
1351 data = deflated;
1352 data_size = deflate_size;
1355 /* emit data encoded in base85 */
1356 cp = data;
1357 while (data_size) {
1358 int bytes = (52 < data_size) ? 52 : data_size;
1359 char line[70];
1360 data_size -= bytes;
1361 if (bytes <= 26)
1362 line[0] = bytes + 'A' - 1;
1363 else
1364 line[0] = bytes - 26 + 'a' - 1;
1365 encode_85(line + 1, cp, bytes);
1366 cp = (char *) cp + bytes;
1367 fputs(line, file);
1368 fputc('\n', file);
1370 fprintf(file, "\n");
1371 free(data);
1374 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1376 fprintf(file, "GIT binary patch\n");
1377 emit_binary_diff_body(file, one, two);
1378 emit_binary_diff_body(file, two, one);
1381 static void diff_filespec_load_driver(struct diff_filespec *one)
1383 if (!one->driver)
1384 one->driver = userdiff_find_by_path(one->path);
1385 if (!one->driver)
1386 one->driver = userdiff_find_by_name("default");
1389 int diff_filespec_is_binary(struct diff_filespec *one)
1391 if (one->is_binary == -1) {
1392 diff_filespec_load_driver(one);
1393 if (one->driver->binary != -1)
1394 one->is_binary = one->driver->binary;
1395 else {
1396 if (!one->data && DIFF_FILE_VALID(one))
1397 diff_populate_filespec(one, 0);
1398 if (one->data)
1399 one->is_binary = buffer_is_binary(one->data,
1400 one->size);
1401 if (one->is_binary == -1)
1402 one->is_binary = 0;
1405 return one->is_binary;
1408 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1410 diff_filespec_load_driver(one);
1411 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1414 static const char *userdiff_word_regex(struct diff_filespec *one)
1416 diff_filespec_load_driver(one);
1417 return one->driver->word_regex;
1420 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1422 if (!options->a_prefix)
1423 options->a_prefix = a;
1424 if (!options->b_prefix)
1425 options->b_prefix = b;
1428 static const char *get_textconv(struct diff_filespec *one)
1430 if (!DIFF_FILE_VALID(one))
1431 return NULL;
1432 if (!S_ISREG(one->mode))
1433 return NULL;
1434 diff_filespec_load_driver(one);
1435 return one->driver->textconv;
1438 static void builtin_diff(const char *name_a,
1439 const char *name_b,
1440 struct diff_filespec *one,
1441 struct diff_filespec *two,
1442 const char *xfrm_msg,
1443 struct diff_options *o,
1444 int complete_rewrite)
1446 mmfile_t mf1, mf2;
1447 const char *lbl[2];
1448 char *a_one, *b_two;
1449 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1450 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1451 const char *a_prefix, *b_prefix;
1452 const char *textconv_one = NULL, *textconv_two = NULL;
1454 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1455 textconv_one = get_textconv(one);
1456 textconv_two = get_textconv(two);
1459 diff_set_mnemonic_prefix(o, "a/", "b/");
1460 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1461 a_prefix = o->b_prefix;
1462 b_prefix = o->a_prefix;
1463 } else {
1464 a_prefix = o->a_prefix;
1465 b_prefix = o->b_prefix;
1468 /* Never use a non-valid filename anywhere if at all possible */
1469 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1470 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1472 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1473 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1474 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1475 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1476 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1477 if (lbl[0][0] == '/') {
1478 /* /dev/null */
1479 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1480 if (xfrm_msg && xfrm_msg[0])
1481 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1483 else if (lbl[1][0] == '/') {
1484 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1485 if (xfrm_msg && xfrm_msg[0])
1486 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1488 else {
1489 if (one->mode != two->mode) {
1490 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1491 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1493 if (xfrm_msg && xfrm_msg[0])
1494 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1496 * we do not run diff between different kind
1497 * of objects.
1499 if ((one->mode ^ two->mode) & S_IFMT)
1500 goto free_ab_and_return;
1501 if (complete_rewrite &&
1502 (textconv_one || !diff_filespec_is_binary(one)) &&
1503 (textconv_two || !diff_filespec_is_binary(two))) {
1504 emit_rewrite_diff(name_a, name_b, one, two,
1505 textconv_one, textconv_two, o);
1506 o->found_changes = 1;
1507 goto free_ab_and_return;
1511 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1512 die("unable to read files to diff");
1514 if (!DIFF_OPT_TST(o, TEXT) &&
1515 ( (diff_filespec_is_binary(one) && !textconv_one) ||
1516 (diff_filespec_is_binary(two) && !textconv_two) )) {
1517 /* Quite common confusing case */
1518 if (mf1.size == mf2.size &&
1519 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1520 goto free_ab_and_return;
1521 if (DIFF_OPT_TST(o, BINARY))
1522 emit_binary_diff(o->file, &mf1, &mf2);
1523 else
1524 fprintf(o->file, "Binary files %s and %s differ\n",
1525 lbl[0], lbl[1]);
1526 o->found_changes = 1;
1528 else {
1529 /* Crazy xdl interfaces.. */
1530 const char *diffopts = getenv("GIT_DIFF_OPTS");
1531 xpparam_t xpp;
1532 xdemitconf_t xecfg;
1533 xdemitcb_t ecb;
1534 struct emit_callback ecbdata;
1535 const struct userdiff_funcname *pe;
1537 if (textconv_one) {
1538 size_t size;
1539 mf1.ptr = run_textconv(textconv_one, one, &size);
1540 if (!mf1.ptr)
1541 die("unable to read files to diff");
1542 mf1.size = size;
1544 if (textconv_two) {
1545 size_t size;
1546 mf2.ptr = run_textconv(textconv_two, two, &size);
1547 if (!mf2.ptr)
1548 die("unable to read files to diff");
1549 mf2.size = size;
1552 pe = diff_funcname_pattern(one);
1553 if (!pe)
1554 pe = diff_funcname_pattern(two);
1556 memset(&xpp, 0, sizeof(xpp));
1557 memset(&xecfg, 0, sizeof(xecfg));
1558 memset(&ecbdata, 0, sizeof(ecbdata));
1559 ecbdata.label_path = lbl;
1560 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1561 ecbdata.found_changesp = &o->found_changes;
1562 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1563 ecbdata.file = o->file;
1564 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1565 xecfg.ctxlen = o->context;
1566 xecfg.interhunkctxlen = o->interhunkcontext;
1567 xecfg.flags = XDL_EMIT_FUNCNAMES;
1568 if (pe)
1569 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1570 if (!diffopts)
1572 else if (!prefixcmp(diffopts, "--unified="))
1573 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1574 else if (!prefixcmp(diffopts, "-u"))
1575 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1576 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1577 ecbdata.diff_words =
1578 xcalloc(1, sizeof(struct diff_words_data));
1579 ecbdata.diff_words->file = o->file;
1580 if (!o->word_regex)
1581 o->word_regex = userdiff_word_regex(one);
1582 if (!o->word_regex)
1583 o->word_regex = userdiff_word_regex(two);
1584 if (!o->word_regex)
1585 o->word_regex = diff_word_regex_cfg;
1586 if (o->word_regex) {
1587 ecbdata.diff_words->word_regex = (regex_t *)
1588 xmalloc(sizeof(regex_t));
1589 if (regcomp(ecbdata.diff_words->word_regex,
1590 o->word_regex,
1591 REG_EXTENDED | REG_NEWLINE))
1592 die ("Invalid regular expression: %s",
1593 o->word_regex);
1596 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1597 &xpp, &xecfg, &ecb);
1598 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1599 free_diff_words_data(&ecbdata);
1600 if (textconv_one)
1601 free(mf1.ptr);
1602 if (textconv_two)
1603 free(mf2.ptr);
1606 free_ab_and_return:
1607 diff_free_filespec_data(one);
1608 diff_free_filespec_data(two);
1609 free(a_one);
1610 free(b_two);
1611 return;
1614 static void builtin_diffstat(const char *name_a, const char *name_b,
1615 struct diff_filespec *one,
1616 struct diff_filespec *two,
1617 struct diffstat_t *diffstat,
1618 struct diff_options *o,
1619 int complete_rewrite)
1621 mmfile_t mf1, mf2;
1622 struct diffstat_file *data;
1624 data = diffstat_add(diffstat, name_a, name_b);
1626 if (!one || !two) {
1627 data->is_unmerged = 1;
1628 return;
1630 if (complete_rewrite) {
1631 diff_populate_filespec(one, 0);
1632 diff_populate_filespec(two, 0);
1633 data->deleted = count_lines(one->data, one->size);
1634 data->added = count_lines(two->data, two->size);
1635 goto free_and_return;
1637 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1638 die("unable to read files to diff");
1640 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1641 data->is_binary = 1;
1642 data->added = mf2.size;
1643 data->deleted = mf1.size;
1644 } else {
1645 /* Crazy xdl interfaces.. */
1646 xpparam_t xpp;
1647 xdemitconf_t xecfg;
1648 xdemitcb_t ecb;
1650 memset(&xpp, 0, sizeof(xpp));
1651 memset(&xecfg, 0, sizeof(xecfg));
1652 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1653 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1654 &xpp, &xecfg, &ecb);
1657 free_and_return:
1658 diff_free_filespec_data(one);
1659 diff_free_filespec_data(two);
1662 static void builtin_checkdiff(const char *name_a, const char *name_b,
1663 const char *attr_path,
1664 struct diff_filespec *one,
1665 struct diff_filespec *two,
1666 struct diff_options *o)
1668 mmfile_t mf1, mf2;
1669 struct checkdiff_t data;
1671 if (!two)
1672 return;
1674 memset(&data, 0, sizeof(data));
1675 data.filename = name_b ? name_b : name_a;
1676 data.lineno = 0;
1677 data.o = o;
1678 data.ws_rule = whitespace_rule(attr_path);
1680 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1681 die("unable to read files to diff");
1684 * All the other codepaths check both sides, but not checking
1685 * the "old" side here is deliberate. We are checking the newly
1686 * introduced changes, and as long as the "new" side is text, we
1687 * can and should check what it introduces.
1689 if (diff_filespec_is_binary(two))
1690 goto free_and_return;
1691 else {
1692 /* Crazy xdl interfaces.. */
1693 xpparam_t xpp;
1694 xdemitconf_t xecfg;
1695 xdemitcb_t ecb;
1697 memset(&xpp, 0, sizeof(xpp));
1698 memset(&xecfg, 0, sizeof(xecfg));
1699 xecfg.ctxlen = 1; /* at least one context line */
1700 xpp.flags = XDF_NEED_MINIMAL;
1701 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1702 &xpp, &xecfg, &ecb);
1704 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1705 data.trailing_blanks_start) {
1706 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1707 data.filename, data.trailing_blanks_start);
1708 data.status = 1; /* report errors */
1711 free_and_return:
1712 diff_free_filespec_data(one);
1713 diff_free_filespec_data(two);
1714 if (data.status)
1715 DIFF_OPT_SET(o, CHECK_FAILED);
1718 struct diff_filespec *alloc_filespec(const char *path)
1720 int namelen = strlen(path);
1721 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1723 memset(spec, 0, sizeof(*spec));
1724 spec->path = (char *)(spec + 1);
1725 memcpy(spec->path, path, namelen+1);
1726 spec->count = 1;
1727 spec->is_binary = -1;
1728 return spec;
1731 void free_filespec(struct diff_filespec *spec)
1733 if (!--spec->count) {
1734 diff_free_filespec_data(spec);
1735 free(spec);
1739 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1740 unsigned short mode)
1742 if (mode) {
1743 spec->mode = canon_mode(mode);
1744 hashcpy(spec->sha1, sha1);
1745 spec->sha1_valid = !is_null_sha1(sha1);
1750 * Given a name and sha1 pair, if the index tells us the file in
1751 * the work tree has that object contents, return true, so that
1752 * prepare_temp_file() does not have to inflate and extract.
1754 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1756 struct cache_entry *ce;
1757 struct stat st;
1758 int pos, len;
1760 /* We do not read the cache ourselves here, because the
1761 * benchmark with my previous version that always reads cache
1762 * shows that it makes things worse for diff-tree comparing
1763 * two linux-2.6 kernel trees in an already checked out work
1764 * tree. This is because most diff-tree comparisons deal with
1765 * only a small number of files, while reading the cache is
1766 * expensive for a large project, and its cost outweighs the
1767 * savings we get by not inflating the object to a temporary
1768 * file. Practically, this code only helps when we are used
1769 * by diff-cache --cached, which does read the cache before
1770 * calling us.
1772 if (!active_cache)
1773 return 0;
1775 /* We want to avoid the working directory if our caller
1776 * doesn't need the data in a normal file, this system
1777 * is rather slow with its stat/open/mmap/close syscalls,
1778 * and the object is contained in a pack file. The pack
1779 * is probably already open and will be faster to obtain
1780 * the data through than the working directory. Loose
1781 * objects however would tend to be slower as they need
1782 * to be individually opened and inflated.
1784 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1785 return 0;
1787 len = strlen(name);
1788 pos = cache_name_pos(name, len);
1789 if (pos < 0)
1790 return 0;
1791 ce = active_cache[pos];
1794 * This is not the sha1 we are looking for, or
1795 * unreusable because it is not a regular file.
1797 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1798 return 0;
1801 * If ce matches the file in the work tree, we can reuse it.
1803 if (ce_uptodate(ce) ||
1804 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1805 return 1;
1807 return 0;
1810 static int populate_from_stdin(struct diff_filespec *s)
1812 struct strbuf buf = STRBUF_INIT;
1813 size_t size = 0;
1815 if (strbuf_read(&buf, 0, 0) < 0)
1816 return error("error while reading from stdin %s",
1817 strerror(errno));
1819 s->should_munmap = 0;
1820 s->data = strbuf_detach(&buf, &size);
1821 s->size = size;
1822 s->should_free = 1;
1823 return 0;
1826 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1828 int len;
1829 char *data = xmalloc(100);
1830 len = snprintf(data, 100,
1831 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1832 s->data = data;
1833 s->size = len;
1834 s->should_free = 1;
1835 if (size_only) {
1836 s->data = NULL;
1837 free(data);
1839 return 0;
1843 * While doing rename detection and pickaxe operation, we may need to
1844 * grab the data for the blob (or file) for our own in-core comparison.
1845 * diff_filespec has data and size fields for this purpose.
1847 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1849 int err = 0;
1850 if (!DIFF_FILE_VALID(s))
1851 die("internal error: asking to populate invalid file.");
1852 if (S_ISDIR(s->mode))
1853 return -1;
1855 if (s->data)
1856 return 0;
1858 if (size_only && 0 < s->size)
1859 return 0;
1861 if (S_ISGITLINK(s->mode))
1862 return diff_populate_gitlink(s, size_only);
1864 if (!s->sha1_valid ||
1865 reuse_worktree_file(s->path, s->sha1, 0)) {
1866 struct strbuf buf = STRBUF_INIT;
1867 struct stat st;
1868 int fd;
1870 if (!strcmp(s->path, "-"))
1871 return populate_from_stdin(s);
1873 if (lstat(s->path, &st) < 0) {
1874 if (errno == ENOENT) {
1875 err_empty:
1876 err = -1;
1877 empty:
1878 s->data = (char *)"";
1879 s->size = 0;
1880 return err;
1883 s->size = xsize_t(st.st_size);
1884 if (!s->size)
1885 goto empty;
1886 if (S_ISLNK(st.st_mode)) {
1887 struct strbuf sb = STRBUF_INIT;
1889 if (strbuf_readlink(&sb, s->path, s->size))
1890 goto err_empty;
1891 s->size = sb.len;
1892 s->data = strbuf_detach(&sb, NULL);
1893 s->should_free = 1;
1894 return 0;
1896 if (size_only)
1897 return 0;
1898 fd = open(s->path, O_RDONLY);
1899 if (fd < 0)
1900 goto err_empty;
1901 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1902 close(fd);
1903 s->should_munmap = 1;
1906 * Convert from working tree format to canonical git format
1908 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1909 size_t size = 0;
1910 munmap(s->data, s->size);
1911 s->should_munmap = 0;
1912 s->data = strbuf_detach(&buf, &size);
1913 s->size = size;
1914 s->should_free = 1;
1917 else {
1918 enum object_type type;
1919 if (size_only)
1920 type = sha1_object_info(s->sha1, &s->size);
1921 else {
1922 s->data = read_sha1_file(s->sha1, &type, &s->size);
1923 s->should_free = 1;
1926 return 0;
1929 void diff_free_filespec_blob(struct diff_filespec *s)
1931 if (s->should_free)
1932 free(s->data);
1933 else if (s->should_munmap)
1934 munmap(s->data, s->size);
1936 if (s->should_free || s->should_munmap) {
1937 s->should_free = s->should_munmap = 0;
1938 s->data = NULL;
1942 void diff_free_filespec_data(struct diff_filespec *s)
1944 diff_free_filespec_blob(s);
1945 free(s->cnt_data);
1946 s->cnt_data = NULL;
1949 static void prep_temp_blob(struct diff_tempfile *temp,
1950 void *blob,
1951 unsigned long size,
1952 const unsigned char *sha1,
1953 int mode)
1955 int fd;
1957 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1958 if (fd < 0)
1959 die("unable to create temp-file: %s", strerror(errno));
1960 if (write_in_full(fd, blob, size) != size)
1961 die("unable to write temp-file");
1962 close(fd);
1963 temp->name = temp->tmp_path;
1964 strcpy(temp->hex, sha1_to_hex(sha1));
1965 temp->hex[40] = 0;
1966 sprintf(temp->mode, "%06o", mode);
1969 static struct diff_tempfile *prepare_temp_file(const char *name,
1970 struct diff_filespec *one)
1972 struct diff_tempfile *temp = claim_diff_tempfile();
1974 if (!DIFF_FILE_VALID(one)) {
1975 not_a_valid_file:
1976 /* A '-' entry produces this for file-2, and
1977 * a '+' entry produces this for file-1.
1979 temp->name = "/dev/null";
1980 strcpy(temp->hex, ".");
1981 strcpy(temp->mode, ".");
1982 return temp;
1985 if (!remove_tempfile_installed) {
1986 atexit(remove_tempfile);
1987 sigchain_push_common(remove_tempfile_on_signal);
1988 remove_tempfile_installed = 1;
1991 if (!one->sha1_valid ||
1992 reuse_worktree_file(name, one->sha1, 1)) {
1993 struct stat st;
1994 if (lstat(name, &st) < 0) {
1995 if (errno == ENOENT)
1996 goto not_a_valid_file;
1997 die("stat(%s): %s", name, strerror(errno));
1999 if (S_ISLNK(st.st_mode)) {
2000 int ret;
2001 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
2002 ret = readlink(name, buf, sizeof(buf));
2003 if (ret < 0)
2004 die("readlink(%s)", name);
2005 if (ret == sizeof(buf))
2006 die("symlink too long: %s", name);
2007 prep_temp_blob(temp, buf, ret,
2008 (one->sha1_valid ?
2009 one->sha1 : null_sha1),
2010 (one->sha1_valid ?
2011 one->mode : S_IFLNK));
2013 else {
2014 /* we can borrow from the file in the work tree */
2015 temp->name = name;
2016 if (!one->sha1_valid)
2017 strcpy(temp->hex, sha1_to_hex(null_sha1));
2018 else
2019 strcpy(temp->hex, sha1_to_hex(one->sha1));
2020 /* Even though we may sometimes borrow the
2021 * contents from the work tree, we always want
2022 * one->mode. mode is trustworthy even when
2023 * !(one->sha1_valid), as long as
2024 * DIFF_FILE_VALID(one).
2026 sprintf(temp->mode, "%06o", one->mode);
2028 return temp;
2030 else {
2031 if (diff_populate_filespec(one, 0))
2032 die("cannot read data blob for %s", one->path);
2033 prep_temp_blob(temp, one->data, one->size,
2034 one->sha1, one->mode);
2036 return temp;
2039 /* An external diff command takes:
2041 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2042 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2045 static void run_external_diff(const char *pgm,
2046 const char *name,
2047 const char *other,
2048 struct diff_filespec *one,
2049 struct diff_filespec *two,
2050 const char *xfrm_msg,
2051 int complete_rewrite)
2053 const char *spawn_arg[10];
2054 int retval;
2055 const char **arg = &spawn_arg[0];
2057 if (one && two) {
2058 struct diff_tempfile *temp_one, *temp_two;
2059 const char *othername = (other ? other : name);
2060 temp_one = prepare_temp_file(name, one);
2061 temp_two = prepare_temp_file(othername, two);
2062 *arg++ = pgm;
2063 *arg++ = name;
2064 *arg++ = temp_one->name;
2065 *arg++ = temp_one->hex;
2066 *arg++ = temp_one->mode;
2067 *arg++ = temp_two->name;
2068 *arg++ = temp_two->hex;
2069 *arg++ = temp_two->mode;
2070 if (other) {
2071 *arg++ = other;
2072 *arg++ = xfrm_msg;
2074 } else {
2075 *arg++ = pgm;
2076 *arg++ = name;
2078 *arg = NULL;
2079 fflush(NULL);
2080 retval = run_command_v_opt(spawn_arg, 0);
2081 remove_tempfile();
2082 if (retval) {
2083 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2084 exit(1);
2088 static int similarity_index(struct diff_filepair *p)
2090 return p->score * 100 / MAX_SCORE;
2093 static void fill_metainfo(struct strbuf *msg,
2094 const char *name,
2095 const char *other,
2096 struct diff_filespec *one,
2097 struct diff_filespec *two,
2098 struct diff_options *o,
2099 struct diff_filepair *p)
2101 strbuf_init(msg, PATH_MAX * 2 + 300);
2102 switch (p->status) {
2103 case DIFF_STATUS_COPIED:
2104 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2105 strbuf_addstr(msg, "\ncopy from ");
2106 quote_c_style(name, msg, NULL, 0);
2107 strbuf_addstr(msg, "\ncopy to ");
2108 quote_c_style(other, msg, NULL, 0);
2109 strbuf_addch(msg, '\n');
2110 break;
2111 case DIFF_STATUS_RENAMED:
2112 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2113 strbuf_addstr(msg, "\nrename from ");
2114 quote_c_style(name, msg, NULL, 0);
2115 strbuf_addstr(msg, "\nrename to ");
2116 quote_c_style(other, msg, NULL, 0);
2117 strbuf_addch(msg, '\n');
2118 break;
2119 case DIFF_STATUS_MODIFIED:
2120 if (p->score) {
2121 strbuf_addf(msg, "dissimilarity index %d%%\n",
2122 similarity_index(p));
2123 break;
2125 /* fallthru */
2126 default:
2127 /* nothing */
2130 if (one && two && hashcmp(one->sha1, two->sha1)) {
2131 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2133 if (DIFF_OPT_TST(o, BINARY)) {
2134 mmfile_t mf;
2135 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2136 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2137 abbrev = 40;
2139 strbuf_addf(msg, "index %.*s..%.*s",
2140 abbrev, sha1_to_hex(one->sha1),
2141 abbrev, sha1_to_hex(two->sha1));
2142 if (one->mode == two->mode)
2143 strbuf_addf(msg, " %06o", one->mode);
2144 strbuf_addch(msg, '\n');
2146 if (msg->len)
2147 strbuf_setlen(msg, msg->len - 1);
2150 static void run_diff_cmd(const char *pgm,
2151 const char *name,
2152 const char *other,
2153 const char *attr_path,
2154 struct diff_filespec *one,
2155 struct diff_filespec *two,
2156 struct strbuf *msg,
2157 struct diff_options *o,
2158 struct diff_filepair *p)
2160 const char *xfrm_msg = NULL;
2161 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2163 if (msg) {
2164 fill_metainfo(msg, name, other, one, two, o, p);
2165 xfrm_msg = msg->len ? msg->buf : NULL;
2168 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2169 pgm = NULL;
2170 else {
2171 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2172 if (drv && drv->external)
2173 pgm = drv->external;
2176 if (pgm) {
2177 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2178 complete_rewrite);
2179 return;
2181 if (one && two)
2182 builtin_diff(name, other ? other : name,
2183 one, two, xfrm_msg, o, complete_rewrite);
2184 else
2185 fprintf(o->file, "* Unmerged path %s\n", name);
2188 static void diff_fill_sha1_info(struct diff_filespec *one)
2190 if (DIFF_FILE_VALID(one)) {
2191 if (!one->sha1_valid) {
2192 struct stat st;
2193 if (!strcmp(one->path, "-")) {
2194 hashcpy(one->sha1, null_sha1);
2195 return;
2197 if (lstat(one->path, &st) < 0)
2198 die("stat %s", one->path);
2199 if (index_path(one->sha1, one->path, &st, 0))
2200 die("cannot hash %s", one->path);
2203 else
2204 hashclr(one->sha1);
2207 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2209 /* Strip the prefix but do not molest /dev/null and absolute paths */
2210 if (*namep && **namep != '/')
2211 *namep += prefix_length;
2212 if (*otherp && **otherp != '/')
2213 *otherp += prefix_length;
2216 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2218 const char *pgm = external_diff();
2219 struct strbuf msg;
2220 struct diff_filespec *one = p->one;
2221 struct diff_filespec *two = p->two;
2222 const char *name;
2223 const char *other;
2224 const char *attr_path;
2226 name = p->one->path;
2227 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2228 attr_path = name;
2229 if (o->prefix_length)
2230 strip_prefix(o->prefix_length, &name, &other);
2232 if (DIFF_PAIR_UNMERGED(p)) {
2233 run_diff_cmd(pgm, name, NULL, attr_path,
2234 NULL, NULL, NULL, o, p);
2235 return;
2238 diff_fill_sha1_info(one);
2239 diff_fill_sha1_info(two);
2241 if (!pgm &&
2242 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2243 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2245 * a filepair that changes between file and symlink
2246 * needs to be split into deletion and creation.
2248 struct diff_filespec *null = alloc_filespec(two->path);
2249 run_diff_cmd(NULL, name, other, attr_path,
2250 one, null, &msg, o, p);
2251 free(null);
2252 strbuf_release(&msg);
2254 null = alloc_filespec(one->path);
2255 run_diff_cmd(NULL, name, other, attr_path,
2256 null, two, &msg, o, p);
2257 free(null);
2259 else
2260 run_diff_cmd(pgm, name, other, attr_path,
2261 one, two, &msg, o, p);
2263 strbuf_release(&msg);
2266 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2267 struct diffstat_t *diffstat)
2269 const char *name;
2270 const char *other;
2271 int complete_rewrite = 0;
2273 if (DIFF_PAIR_UNMERGED(p)) {
2274 /* unmerged */
2275 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2276 return;
2279 name = p->one->path;
2280 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2282 if (o->prefix_length)
2283 strip_prefix(o->prefix_length, &name, &other);
2285 diff_fill_sha1_info(p->one);
2286 diff_fill_sha1_info(p->two);
2288 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2289 complete_rewrite = 1;
2290 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2293 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2295 const char *name;
2296 const char *other;
2297 const char *attr_path;
2299 if (DIFF_PAIR_UNMERGED(p)) {
2300 /* unmerged */
2301 return;
2304 name = p->one->path;
2305 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2306 attr_path = other ? other : name;
2308 if (o->prefix_length)
2309 strip_prefix(o->prefix_length, &name, &other);
2311 diff_fill_sha1_info(p->one);
2312 diff_fill_sha1_info(p->two);
2314 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2317 void diff_setup(struct diff_options *options)
2319 memset(options, 0, sizeof(*options));
2321 options->file = stdout;
2323 options->line_termination = '\n';
2324 options->break_opt = -1;
2325 options->rename_limit = -1;
2326 options->dirstat_percent = 3;
2327 options->context = 3;
2329 options->change = diff_change;
2330 options->add_remove = diff_addremove;
2331 if (diff_use_color_default > 0)
2332 DIFF_OPT_SET(options, COLOR_DIFF);
2333 options->detect_rename = diff_detect_rename_default;
2335 if (!diff_mnemonic_prefix) {
2336 options->a_prefix = "a/";
2337 options->b_prefix = "b/";
2341 int diff_setup_done(struct diff_options *options)
2343 int count = 0;
2345 if (options->output_format & DIFF_FORMAT_NAME)
2346 count++;
2347 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2348 count++;
2349 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2350 count++;
2351 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2352 count++;
2353 if (count > 1)
2354 die("--name-only, --name-status, --check and -s are mutually exclusive");
2356 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2357 options->detect_rename = DIFF_DETECT_COPY;
2359 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2360 options->prefix = NULL;
2361 if (options->prefix)
2362 options->prefix_length = strlen(options->prefix);
2363 else
2364 options->prefix_length = 0;
2366 if (options->output_format & (DIFF_FORMAT_NAME |
2367 DIFF_FORMAT_NAME_STATUS |
2368 DIFF_FORMAT_CHECKDIFF |
2369 DIFF_FORMAT_NO_OUTPUT))
2370 options->output_format &= ~(DIFF_FORMAT_RAW |
2371 DIFF_FORMAT_NUMSTAT |
2372 DIFF_FORMAT_DIFFSTAT |
2373 DIFF_FORMAT_SHORTSTAT |
2374 DIFF_FORMAT_DIRSTAT |
2375 DIFF_FORMAT_SUMMARY |
2376 DIFF_FORMAT_PATCH);
2379 * These cases always need recursive; we do not drop caller-supplied
2380 * recursive bits for other formats here.
2382 if (options->output_format & (DIFF_FORMAT_PATCH |
2383 DIFF_FORMAT_NUMSTAT |
2384 DIFF_FORMAT_DIFFSTAT |
2385 DIFF_FORMAT_SHORTSTAT |
2386 DIFF_FORMAT_DIRSTAT |
2387 DIFF_FORMAT_SUMMARY |
2388 DIFF_FORMAT_CHECKDIFF))
2389 DIFF_OPT_SET(options, RECURSIVE);
2391 * Also pickaxe would not work very well if you do not say recursive
2393 if (options->pickaxe)
2394 DIFF_OPT_SET(options, RECURSIVE);
2396 if (options->detect_rename && options->rename_limit < 0)
2397 options->rename_limit = diff_rename_limit_default;
2398 if (options->setup & DIFF_SETUP_USE_CACHE) {
2399 if (!active_cache)
2400 /* read-cache does not die even when it fails
2401 * so it is safe for us to do this here. Also
2402 * it does not smudge active_cache or active_nr
2403 * when it fails, so we do not have to worry about
2404 * cleaning it up ourselves either.
2406 read_cache();
2408 if (options->abbrev <= 0 || 40 < options->abbrev)
2409 options->abbrev = 40; /* full */
2412 * It does not make sense to show the first hit we happened
2413 * to have found. It does not make sense not to return with
2414 * exit code in such a case either.
2416 if (DIFF_OPT_TST(options, QUIET)) {
2417 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2418 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2421 return 0;
2424 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2426 char c, *eq;
2427 int len;
2429 if (*arg != '-')
2430 return 0;
2431 c = *++arg;
2432 if (!c)
2433 return 0;
2434 if (c == arg_short) {
2435 c = *++arg;
2436 if (!c)
2437 return 1;
2438 if (val && isdigit(c)) {
2439 char *end;
2440 int n = strtoul(arg, &end, 10);
2441 if (*end)
2442 return 0;
2443 *val = n;
2444 return 1;
2446 return 0;
2448 if (c != '-')
2449 return 0;
2450 arg++;
2451 eq = strchr(arg, '=');
2452 if (eq)
2453 len = eq - arg;
2454 else
2455 len = strlen(arg);
2456 if (!len || strncmp(arg, arg_long, len))
2457 return 0;
2458 if (eq) {
2459 int n;
2460 char *end;
2461 if (!isdigit(*++eq))
2462 return 0;
2463 n = strtoul(eq, &end, 10);
2464 if (*end)
2465 return 0;
2466 *val = n;
2468 return 1;
2471 static int diff_scoreopt_parse(const char *opt);
2473 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2475 const char *arg = av[0];
2477 /* Output format options */
2478 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2479 options->output_format |= DIFF_FORMAT_PATCH;
2480 else if (opt_arg(arg, 'U', "unified", &options->context))
2481 options->output_format |= DIFF_FORMAT_PATCH;
2482 else if (!strcmp(arg, "--raw"))
2483 options->output_format |= DIFF_FORMAT_RAW;
2484 else if (!strcmp(arg, "--patch-with-raw"))
2485 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2486 else if (!strcmp(arg, "--numstat"))
2487 options->output_format |= DIFF_FORMAT_NUMSTAT;
2488 else if (!strcmp(arg, "--shortstat"))
2489 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2490 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2491 options->output_format |= DIFF_FORMAT_DIRSTAT;
2492 else if (!strcmp(arg, "--cumulative")) {
2493 options->output_format |= DIFF_FORMAT_DIRSTAT;
2494 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2495 } else if (opt_arg(arg, 0, "dirstat-by-file",
2496 &options->dirstat_percent)) {
2497 options->output_format |= DIFF_FORMAT_DIRSTAT;
2498 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2500 else if (!strcmp(arg, "--check"))
2501 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2502 else if (!strcmp(arg, "--summary"))
2503 options->output_format |= DIFF_FORMAT_SUMMARY;
2504 else if (!strcmp(arg, "--patch-with-stat"))
2505 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2506 else if (!strcmp(arg, "--name-only"))
2507 options->output_format |= DIFF_FORMAT_NAME;
2508 else if (!strcmp(arg, "--name-status"))
2509 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2510 else if (!strcmp(arg, "-s"))
2511 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2512 else if (!prefixcmp(arg, "--stat")) {
2513 char *end;
2514 int width = options->stat_width;
2515 int name_width = options->stat_name_width;
2516 arg += 6;
2517 end = (char *)arg;
2519 switch (*arg) {
2520 case '-':
2521 if (!prefixcmp(arg, "-width="))
2522 width = strtoul(arg + 7, &end, 10);
2523 else if (!prefixcmp(arg, "-name-width="))
2524 name_width = strtoul(arg + 12, &end, 10);
2525 break;
2526 case '=':
2527 width = strtoul(arg+1, &end, 10);
2528 if (*end == ',')
2529 name_width = strtoul(end+1, &end, 10);
2532 /* Important! This checks all the error cases! */
2533 if (*end)
2534 return 0;
2535 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2536 options->stat_name_width = name_width;
2537 options->stat_width = width;
2540 /* renames options */
2541 else if (!prefixcmp(arg, "-B")) {
2542 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2543 return -1;
2545 else if (!prefixcmp(arg, "-M")) {
2546 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2547 return -1;
2548 options->detect_rename = DIFF_DETECT_RENAME;
2550 else if (!prefixcmp(arg, "-C")) {
2551 if (options->detect_rename == DIFF_DETECT_COPY)
2552 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2553 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2554 return -1;
2555 options->detect_rename = DIFF_DETECT_COPY;
2557 else if (!strcmp(arg, "--no-renames"))
2558 options->detect_rename = 0;
2559 else if (!strcmp(arg, "--relative"))
2560 DIFF_OPT_SET(options, RELATIVE_NAME);
2561 else if (!prefixcmp(arg, "--relative=")) {
2562 DIFF_OPT_SET(options, RELATIVE_NAME);
2563 options->prefix = arg + 11;
2566 /* xdiff options */
2567 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2568 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2569 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2570 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2571 else if (!strcmp(arg, "--ignore-space-at-eol"))
2572 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2573 else if (!strcmp(arg, "--patience"))
2574 DIFF_XDL_SET(options, PATIENCE_DIFF);
2576 /* flags options */
2577 else if (!strcmp(arg, "--binary")) {
2578 options->output_format |= DIFF_FORMAT_PATCH;
2579 DIFF_OPT_SET(options, BINARY);
2581 else if (!strcmp(arg, "--full-index"))
2582 DIFF_OPT_SET(options, FULL_INDEX);
2583 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2584 DIFF_OPT_SET(options, TEXT);
2585 else if (!strcmp(arg, "-R"))
2586 DIFF_OPT_SET(options, REVERSE_DIFF);
2587 else if (!strcmp(arg, "--find-copies-harder"))
2588 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2589 else if (!strcmp(arg, "--follow"))
2590 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2591 else if (!strcmp(arg, "--color"))
2592 DIFF_OPT_SET(options, COLOR_DIFF);
2593 else if (!strcmp(arg, "--no-color"))
2594 DIFF_OPT_CLR(options, COLOR_DIFF);
2595 else if (!strcmp(arg, "--color-words")) {
2596 DIFF_OPT_SET(options, COLOR_DIFF);
2597 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2599 else if (!prefixcmp(arg, "--color-words=")) {
2600 DIFF_OPT_SET(options, COLOR_DIFF);
2601 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2602 options->word_regex = arg + 14;
2604 else if (!strcmp(arg, "--exit-code"))
2605 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2606 else if (!strcmp(arg, "--quiet"))
2607 DIFF_OPT_SET(options, QUIET);
2608 else if (!strcmp(arg, "--ext-diff"))
2609 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2610 else if (!strcmp(arg, "--no-ext-diff"))
2611 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2612 else if (!strcmp(arg, "--textconv"))
2613 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2614 else if (!strcmp(arg, "--no-textconv"))
2615 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2616 else if (!strcmp(arg, "--ignore-submodules"))
2617 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2619 /* misc options */
2620 else if (!strcmp(arg, "-z"))
2621 options->line_termination = 0;
2622 else if (!prefixcmp(arg, "-l"))
2623 options->rename_limit = strtoul(arg+2, NULL, 10);
2624 else if (!prefixcmp(arg, "-S"))
2625 options->pickaxe = arg + 2;
2626 else if (!strcmp(arg, "--pickaxe-all"))
2627 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2628 else if (!strcmp(arg, "--pickaxe-regex"))
2629 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2630 else if (!prefixcmp(arg, "-O"))
2631 options->orderfile = arg + 2;
2632 else if (!prefixcmp(arg, "--diff-filter="))
2633 options->filter = arg + 14;
2634 else if (!strcmp(arg, "--abbrev"))
2635 options->abbrev = DEFAULT_ABBREV;
2636 else if (!prefixcmp(arg, "--abbrev=")) {
2637 options->abbrev = strtoul(arg + 9, NULL, 10);
2638 if (options->abbrev < MINIMUM_ABBREV)
2639 options->abbrev = MINIMUM_ABBREV;
2640 else if (40 < options->abbrev)
2641 options->abbrev = 40;
2643 else if (!prefixcmp(arg, "--src-prefix="))
2644 options->a_prefix = arg + 13;
2645 else if (!prefixcmp(arg, "--dst-prefix="))
2646 options->b_prefix = arg + 13;
2647 else if (!strcmp(arg, "--no-prefix"))
2648 options->a_prefix = options->b_prefix = "";
2649 else if (opt_arg(arg, '\0', "inter-hunk-context",
2650 &options->interhunkcontext))
2652 else if (!prefixcmp(arg, "--output=")) {
2653 options->file = fopen(arg + strlen("--output="), "w");
2654 options->close_file = 1;
2655 } else
2656 return 0;
2657 return 1;
2660 static int parse_num(const char **cp_p)
2662 unsigned long num, scale;
2663 int ch, dot;
2664 const char *cp = *cp_p;
2666 num = 0;
2667 scale = 1;
2668 dot = 0;
2669 for(;;) {
2670 ch = *cp;
2671 if ( !dot && ch == '.' ) {
2672 scale = 1;
2673 dot = 1;
2674 } else if ( ch == '%' ) {
2675 scale = dot ? scale*100 : 100;
2676 cp++; /* % is always at the end */
2677 break;
2678 } else if ( ch >= '0' && ch <= '9' ) {
2679 if ( scale < 100000 ) {
2680 scale *= 10;
2681 num = (num*10) + (ch-'0');
2683 } else {
2684 break;
2686 cp++;
2688 *cp_p = cp;
2690 /* user says num divided by scale and we say internally that
2691 * is MAX_SCORE * num / scale.
2693 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2696 static int diff_scoreopt_parse(const char *opt)
2698 int opt1, opt2, cmd;
2700 if (*opt++ != '-')
2701 return -1;
2702 cmd = *opt++;
2703 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2704 return -1; /* that is not a -M, -C nor -B option */
2706 opt1 = parse_num(&opt);
2707 if (cmd != 'B')
2708 opt2 = 0;
2709 else {
2710 if (*opt == 0)
2711 opt2 = 0;
2712 else if (*opt != '/')
2713 return -1; /* we expect -B80/99 or -B80 */
2714 else {
2715 opt++;
2716 opt2 = parse_num(&opt);
2719 if (*opt != 0)
2720 return -1;
2721 return opt1 | (opt2 << 16);
2724 struct diff_queue_struct diff_queued_diff;
2726 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2728 if (queue->alloc <= queue->nr) {
2729 queue->alloc = alloc_nr(queue->alloc);
2730 queue->queue = xrealloc(queue->queue,
2731 sizeof(dp) * queue->alloc);
2733 queue->queue[queue->nr++] = dp;
2736 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2737 struct diff_filespec *one,
2738 struct diff_filespec *two)
2740 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2741 dp->one = one;
2742 dp->two = two;
2743 if (queue)
2744 diff_q(queue, dp);
2745 return dp;
2748 void diff_free_filepair(struct diff_filepair *p)
2750 free_filespec(p->one);
2751 free_filespec(p->two);
2752 free(p);
2755 /* This is different from find_unique_abbrev() in that
2756 * it stuffs the result with dots for alignment.
2758 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2760 int abblen;
2761 const char *abbrev;
2762 if (len == 40)
2763 return sha1_to_hex(sha1);
2765 abbrev = find_unique_abbrev(sha1, len);
2766 abblen = strlen(abbrev);
2767 if (abblen < 37) {
2768 static char hex[41];
2769 if (len < abblen && abblen <= len + 2)
2770 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2771 else
2772 sprintf(hex, "%s...", abbrev);
2773 return hex;
2775 return sha1_to_hex(sha1);
2778 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2780 int line_termination = opt->line_termination;
2781 int inter_name_termination = line_termination ? '\t' : '\0';
2783 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2784 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2785 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2786 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2788 if (p->score) {
2789 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2790 inter_name_termination);
2791 } else {
2792 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2795 if (p->status == DIFF_STATUS_COPIED ||
2796 p->status == DIFF_STATUS_RENAMED) {
2797 const char *name_a, *name_b;
2798 name_a = p->one->path;
2799 name_b = p->two->path;
2800 strip_prefix(opt->prefix_length, &name_a, &name_b);
2801 write_name_quoted(name_a, opt->file, inter_name_termination);
2802 write_name_quoted(name_b, opt->file, line_termination);
2803 } else {
2804 const char *name_a, *name_b;
2805 name_a = p->one->mode ? p->one->path : p->two->path;
2806 name_b = NULL;
2807 strip_prefix(opt->prefix_length, &name_a, &name_b);
2808 write_name_quoted(name_a, opt->file, line_termination);
2812 int diff_unmodified_pair(struct diff_filepair *p)
2814 /* This function is written stricter than necessary to support
2815 * the currently implemented transformers, but the idea is to
2816 * let transformers to produce diff_filepairs any way they want,
2817 * and filter and clean them up here before producing the output.
2819 struct diff_filespec *one = p->one, *two = p->two;
2821 if (DIFF_PAIR_UNMERGED(p))
2822 return 0; /* unmerged is interesting */
2824 /* deletion, addition, mode or type change
2825 * and rename are all interesting.
2827 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2828 DIFF_PAIR_MODE_CHANGED(p) ||
2829 strcmp(one->path, two->path))
2830 return 0;
2832 /* both are valid and point at the same path. that is, we are
2833 * dealing with a change.
2835 if (one->sha1_valid && two->sha1_valid &&
2836 !hashcmp(one->sha1, two->sha1))
2837 return 1; /* no change */
2838 if (!one->sha1_valid && !two->sha1_valid)
2839 return 1; /* both look at the same file on the filesystem. */
2840 return 0;
2843 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2845 if (diff_unmodified_pair(p))
2846 return;
2848 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2849 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2850 return; /* no tree diffs in patch format */
2852 run_diff(p, o);
2855 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2856 struct diffstat_t *diffstat)
2858 if (diff_unmodified_pair(p))
2859 return;
2861 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2862 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2863 return; /* no tree diffs in patch format */
2865 run_diffstat(p, o, diffstat);
2868 static void diff_flush_checkdiff(struct diff_filepair *p,
2869 struct diff_options *o)
2871 if (diff_unmodified_pair(p))
2872 return;
2874 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2875 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2876 return; /* no tree diffs in patch format */
2878 run_checkdiff(p, o);
2881 int diff_queue_is_empty(void)
2883 struct diff_queue_struct *q = &diff_queued_diff;
2884 int i;
2885 for (i = 0; i < q->nr; i++)
2886 if (!diff_unmodified_pair(q->queue[i]))
2887 return 0;
2888 return 1;
2891 #if DIFF_DEBUG
2892 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2894 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2895 x, one ? one : "",
2896 s->path,
2897 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2898 s->mode,
2899 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2900 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2901 x, one ? one : "",
2902 s->size, s->xfrm_flags);
2905 void diff_debug_filepair(const struct diff_filepair *p, int i)
2907 diff_debug_filespec(p->one, i, "one");
2908 diff_debug_filespec(p->two, i, "two");
2909 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2910 p->score, p->status ? p->status : '?',
2911 p->one->rename_used, p->broken_pair);
2914 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2916 int i;
2917 if (msg)
2918 fprintf(stderr, "%s\n", msg);
2919 fprintf(stderr, "q->nr = %d\n", q->nr);
2920 for (i = 0; i < q->nr; i++) {
2921 struct diff_filepair *p = q->queue[i];
2922 diff_debug_filepair(p, i);
2925 #endif
2927 static void diff_resolve_rename_copy(void)
2929 int i;
2930 struct diff_filepair *p;
2931 struct diff_queue_struct *q = &diff_queued_diff;
2933 diff_debug_queue("resolve-rename-copy", q);
2935 for (i = 0; i < q->nr; i++) {
2936 p = q->queue[i];
2937 p->status = 0; /* undecided */
2938 if (DIFF_PAIR_UNMERGED(p))
2939 p->status = DIFF_STATUS_UNMERGED;
2940 else if (!DIFF_FILE_VALID(p->one))
2941 p->status = DIFF_STATUS_ADDED;
2942 else if (!DIFF_FILE_VALID(p->two))
2943 p->status = DIFF_STATUS_DELETED;
2944 else if (DIFF_PAIR_TYPE_CHANGED(p))
2945 p->status = DIFF_STATUS_TYPE_CHANGED;
2947 /* from this point on, we are dealing with a pair
2948 * whose both sides are valid and of the same type, i.e.
2949 * either in-place edit or rename/copy edit.
2951 else if (DIFF_PAIR_RENAME(p)) {
2953 * A rename might have re-connected a broken
2954 * pair up, causing the pathnames to be the
2955 * same again. If so, that's not a rename at
2956 * all, just a modification..
2958 * Otherwise, see if this source was used for
2959 * multiple renames, in which case we decrement
2960 * the count, and call it a copy.
2962 if (!strcmp(p->one->path, p->two->path))
2963 p->status = DIFF_STATUS_MODIFIED;
2964 else if (--p->one->rename_used > 0)
2965 p->status = DIFF_STATUS_COPIED;
2966 else
2967 p->status = DIFF_STATUS_RENAMED;
2969 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2970 p->one->mode != p->two->mode ||
2971 is_null_sha1(p->one->sha1))
2972 p->status = DIFF_STATUS_MODIFIED;
2973 else {
2974 /* This is a "no-change" entry and should not
2975 * happen anymore, but prepare for broken callers.
2977 error("feeding unmodified %s to diffcore",
2978 p->one->path);
2979 p->status = DIFF_STATUS_UNKNOWN;
2982 diff_debug_queue("resolve-rename-copy done", q);
2985 static int check_pair_status(struct diff_filepair *p)
2987 switch (p->status) {
2988 case DIFF_STATUS_UNKNOWN:
2989 return 0;
2990 case 0:
2991 die("internal error in diff-resolve-rename-copy");
2992 default:
2993 return 1;
2997 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2999 int fmt = opt->output_format;
3001 if (fmt & DIFF_FORMAT_CHECKDIFF)
3002 diff_flush_checkdiff(p, opt);
3003 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3004 diff_flush_raw(p, opt);
3005 else if (fmt & DIFF_FORMAT_NAME) {
3006 const char *name_a, *name_b;
3007 name_a = p->two->path;
3008 name_b = NULL;
3009 strip_prefix(opt->prefix_length, &name_a, &name_b);
3010 write_name_quoted(name_a, opt->file, opt->line_termination);
3014 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3016 if (fs->mode)
3017 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3018 else
3019 fprintf(file, " %s ", newdelete);
3020 write_name_quoted(fs->path, file, '\n');
3024 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3026 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3027 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3028 show_name ? ' ' : '\n');
3029 if (show_name) {
3030 write_name_quoted(p->two->path, file, '\n');
3035 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3037 char *names = pprint_rename(p->one->path, p->two->path);
3039 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3040 free(names);
3041 show_mode_change(file, p, 0);
3044 static void diff_summary(FILE *file, struct diff_filepair *p)
3046 switch(p->status) {
3047 case DIFF_STATUS_DELETED:
3048 show_file_mode_name(file, "delete", p->one);
3049 break;
3050 case DIFF_STATUS_ADDED:
3051 show_file_mode_name(file, "create", p->two);
3052 break;
3053 case DIFF_STATUS_COPIED:
3054 show_rename_copy(file, "copy", p);
3055 break;
3056 case DIFF_STATUS_RENAMED:
3057 show_rename_copy(file, "rename", p);
3058 break;
3059 default:
3060 if (p->score) {
3061 fputs(" rewrite ", file);
3062 write_name_quoted(p->two->path, file, ' ');
3063 fprintf(file, "(%d%%)\n", similarity_index(p));
3065 show_mode_change(file, p, !p->score);
3066 break;
3070 struct patch_id_t {
3071 git_SHA_CTX *ctx;
3072 int patchlen;
3075 static int remove_space(char *line, int len)
3077 int i;
3078 char *dst = line;
3079 unsigned char c;
3081 for (i = 0; i < len; i++)
3082 if (!isspace((c = line[i])))
3083 *dst++ = c;
3085 return dst - line;
3088 static void patch_id_consume(void *priv, char *line, unsigned long len)
3090 struct patch_id_t *data = priv;
3091 int new_len;
3093 /* Ignore line numbers when computing the SHA1 of the patch */
3094 if (!prefixcmp(line, "@@ -"))
3095 return;
3097 new_len = remove_space(line, len);
3099 git_SHA1_Update(data->ctx, line, new_len);
3100 data->patchlen += new_len;
3103 /* returns 0 upon success, and writes result into sha1 */
3104 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3106 struct diff_queue_struct *q = &diff_queued_diff;
3107 int i;
3108 git_SHA_CTX ctx;
3109 struct patch_id_t data;
3110 char buffer[PATH_MAX * 4 + 20];
3112 git_SHA1_Init(&ctx);
3113 memset(&data, 0, sizeof(struct patch_id_t));
3114 data.ctx = &ctx;
3116 for (i = 0; i < q->nr; i++) {
3117 xpparam_t xpp;
3118 xdemitconf_t xecfg;
3119 xdemitcb_t ecb;
3120 mmfile_t mf1, mf2;
3121 struct diff_filepair *p = q->queue[i];
3122 int len1, len2;
3124 memset(&xpp, 0, sizeof(xpp));
3125 memset(&xecfg, 0, sizeof(xecfg));
3126 if (p->status == 0)
3127 return error("internal diff status error");
3128 if (p->status == DIFF_STATUS_UNKNOWN)
3129 continue;
3130 if (diff_unmodified_pair(p))
3131 continue;
3132 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3133 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3134 continue;
3135 if (DIFF_PAIR_UNMERGED(p))
3136 continue;
3138 diff_fill_sha1_info(p->one);
3139 diff_fill_sha1_info(p->two);
3140 if (fill_mmfile(&mf1, p->one) < 0 ||
3141 fill_mmfile(&mf2, p->two) < 0)
3142 return error("unable to read files to diff");
3144 len1 = remove_space(p->one->path, strlen(p->one->path));
3145 len2 = remove_space(p->two->path, strlen(p->two->path));
3146 if (p->one->mode == 0)
3147 len1 = snprintf(buffer, sizeof(buffer),
3148 "diff--gita/%.*sb/%.*s"
3149 "newfilemode%06o"
3150 "---/dev/null"
3151 "+++b/%.*s",
3152 len1, p->one->path,
3153 len2, p->two->path,
3154 p->two->mode,
3155 len2, p->two->path);
3156 else if (p->two->mode == 0)
3157 len1 = snprintf(buffer, sizeof(buffer),
3158 "diff--gita/%.*sb/%.*s"
3159 "deletedfilemode%06o"
3160 "---a/%.*s"
3161 "+++/dev/null",
3162 len1, p->one->path,
3163 len2, p->two->path,
3164 p->one->mode,
3165 len1, p->one->path);
3166 else
3167 len1 = snprintf(buffer, sizeof(buffer),
3168 "diff--gita/%.*sb/%.*s"
3169 "---a/%.*s"
3170 "+++b/%.*s",
3171 len1, p->one->path,
3172 len2, p->two->path,
3173 len1, p->one->path,
3174 len2, p->two->path);
3175 git_SHA1_Update(&ctx, buffer, len1);
3177 xpp.flags = XDF_NEED_MINIMAL;
3178 xecfg.ctxlen = 3;
3179 xecfg.flags = XDL_EMIT_FUNCNAMES;
3180 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3181 &xpp, &xecfg, &ecb);
3184 git_SHA1_Final(sha1, &ctx);
3185 return 0;
3188 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3190 struct diff_queue_struct *q = &diff_queued_diff;
3191 int i;
3192 int result = diff_get_patch_id(options, sha1);
3194 for (i = 0; i < q->nr; i++)
3195 diff_free_filepair(q->queue[i]);
3197 free(q->queue);
3198 q->queue = NULL;
3199 q->nr = q->alloc = 0;
3201 return result;
3204 static int is_summary_empty(const struct diff_queue_struct *q)
3206 int i;
3208 for (i = 0; i < q->nr; i++) {
3209 const struct diff_filepair *p = q->queue[i];
3211 switch (p->status) {
3212 case DIFF_STATUS_DELETED:
3213 case DIFF_STATUS_ADDED:
3214 case DIFF_STATUS_COPIED:
3215 case DIFF_STATUS_RENAMED:
3216 return 0;
3217 default:
3218 if (p->score)
3219 return 0;
3220 if (p->one->mode && p->two->mode &&
3221 p->one->mode != p->two->mode)
3222 return 0;
3223 break;
3226 return 1;
3229 void diff_flush(struct diff_options *options)
3231 struct diff_queue_struct *q = &diff_queued_diff;
3232 int i, output_format = options->output_format;
3233 int separator = 0;
3236 * Order: raw, stat, summary, patch
3237 * or: name/name-status/checkdiff (other bits clear)
3239 if (!q->nr)
3240 goto free_queue;
3242 if (output_format & (DIFF_FORMAT_RAW |
3243 DIFF_FORMAT_NAME |
3244 DIFF_FORMAT_NAME_STATUS |
3245 DIFF_FORMAT_CHECKDIFF)) {
3246 for (i = 0; i < q->nr; i++) {
3247 struct diff_filepair *p = q->queue[i];
3248 if (check_pair_status(p))
3249 flush_one_pair(p, options);
3251 separator++;
3254 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3255 struct diffstat_t diffstat;
3257 memset(&diffstat, 0, sizeof(struct diffstat_t));
3258 for (i = 0; i < q->nr; i++) {
3259 struct diff_filepair *p = q->queue[i];
3260 if (check_pair_status(p))
3261 diff_flush_stat(p, options, &diffstat);
3263 if (output_format & DIFF_FORMAT_NUMSTAT)
3264 show_numstat(&diffstat, options);
3265 if (output_format & DIFF_FORMAT_DIFFSTAT)
3266 show_stats(&diffstat, options);
3267 if (output_format & DIFF_FORMAT_SHORTSTAT)
3268 show_shortstats(&diffstat, options);
3269 free_diffstat_info(&diffstat);
3270 separator++;
3272 if (output_format & DIFF_FORMAT_DIRSTAT)
3273 show_dirstat(options);
3275 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3276 for (i = 0; i < q->nr; i++)
3277 diff_summary(options->file, q->queue[i]);
3278 separator++;
3281 if (output_format & DIFF_FORMAT_PATCH) {
3282 if (separator) {
3283 putc(options->line_termination, options->file);
3284 if (options->stat_sep) {
3285 /* attach patch instead of inline */
3286 fputs(options->stat_sep, options->file);
3290 for (i = 0; i < q->nr; i++) {
3291 struct diff_filepair *p = q->queue[i];
3292 if (check_pair_status(p))
3293 diff_flush_patch(p, options);
3297 if (output_format & DIFF_FORMAT_CALLBACK)
3298 options->format_callback(q, options, options->format_callback_data);
3300 for (i = 0; i < q->nr; i++)
3301 diff_free_filepair(q->queue[i]);
3302 free_queue:
3303 free(q->queue);
3304 q->queue = NULL;
3305 q->nr = q->alloc = 0;
3306 if (options->close_file)
3307 fclose(options->file);
3310 static void diffcore_apply_filter(const char *filter)
3312 int i;
3313 struct diff_queue_struct *q = &diff_queued_diff;
3314 struct diff_queue_struct outq;
3315 outq.queue = NULL;
3316 outq.nr = outq.alloc = 0;
3318 if (!filter)
3319 return;
3321 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3322 int found;
3323 for (i = found = 0; !found && i < q->nr; i++) {
3324 struct diff_filepair *p = q->queue[i];
3325 if (((p->status == DIFF_STATUS_MODIFIED) &&
3326 ((p->score &&
3327 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3328 (!p->score &&
3329 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3330 ((p->status != DIFF_STATUS_MODIFIED) &&
3331 strchr(filter, p->status)))
3332 found++;
3334 if (found)
3335 return;
3337 /* otherwise we will clear the whole queue
3338 * by copying the empty outq at the end of this
3339 * function, but first clear the current entries
3340 * in the queue.
3342 for (i = 0; i < q->nr; i++)
3343 diff_free_filepair(q->queue[i]);
3345 else {
3346 /* Only the matching ones */
3347 for (i = 0; i < q->nr; i++) {
3348 struct diff_filepair *p = q->queue[i];
3350 if (((p->status == DIFF_STATUS_MODIFIED) &&
3351 ((p->score &&
3352 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3353 (!p->score &&
3354 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3355 ((p->status != DIFF_STATUS_MODIFIED) &&
3356 strchr(filter, p->status)))
3357 diff_q(&outq, p);
3358 else
3359 diff_free_filepair(p);
3362 free(q->queue);
3363 *q = outq;
3366 /* Check whether two filespecs with the same mode and size are identical */
3367 static int diff_filespec_is_identical(struct diff_filespec *one,
3368 struct diff_filespec *two)
3370 if (S_ISGITLINK(one->mode))
3371 return 0;
3372 if (diff_populate_filespec(one, 0))
3373 return 0;
3374 if (diff_populate_filespec(two, 0))
3375 return 0;
3376 return !memcmp(one->data, two->data, one->size);
3379 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3381 int i;
3382 struct diff_queue_struct *q = &diff_queued_diff;
3383 struct diff_queue_struct outq;
3384 outq.queue = NULL;
3385 outq.nr = outq.alloc = 0;
3387 for (i = 0; i < q->nr; i++) {
3388 struct diff_filepair *p = q->queue[i];
3391 * 1. Entries that come from stat info dirtyness
3392 * always have both sides (iow, not create/delete),
3393 * one side of the object name is unknown, with
3394 * the same mode and size. Keep the ones that
3395 * do not match these criteria. They have real
3396 * differences.
3398 * 2. At this point, the file is known to be modified,
3399 * with the same mode and size, and the object
3400 * name of one side is unknown. Need to inspect
3401 * the identical contents.
3403 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3404 !DIFF_FILE_VALID(p->two) ||
3405 (p->one->sha1_valid && p->two->sha1_valid) ||
3406 (p->one->mode != p->two->mode) ||
3407 diff_populate_filespec(p->one, 1) ||
3408 diff_populate_filespec(p->two, 1) ||
3409 (p->one->size != p->two->size) ||
3410 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3411 diff_q(&outq, p);
3412 else {
3414 * The caller can subtract 1 from skip_stat_unmatch
3415 * to determine how many paths were dirty only
3416 * due to stat info mismatch.
3418 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3419 diffopt->skip_stat_unmatch++;
3420 diff_free_filepair(p);
3423 free(q->queue);
3424 *q = outq;
3427 void diffcore_std(struct diff_options *options)
3429 if (options->skip_stat_unmatch)
3430 diffcore_skip_stat_unmatch(options);
3431 if (options->break_opt != -1)
3432 diffcore_break(options->break_opt);
3433 if (options->detect_rename)
3434 diffcore_rename(options);
3435 if (options->break_opt != -1)
3436 diffcore_merge_broken();
3437 if (options->pickaxe)
3438 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3439 if (options->orderfile)
3440 diffcore_order(options->orderfile);
3441 diff_resolve_rename_copy();
3442 diffcore_apply_filter(options->filter);
3444 if (diff_queued_diff.nr)
3445 DIFF_OPT_SET(options, HAS_CHANGES);
3446 else
3447 DIFF_OPT_CLR(options, HAS_CHANGES);
3450 int diff_result_code(struct diff_options *opt, int status)
3452 int result = 0;
3453 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3454 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3455 return status;
3456 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3457 DIFF_OPT_TST(opt, HAS_CHANGES))
3458 result |= 01;
3459 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3460 DIFF_OPT_TST(opt, CHECK_FAILED))
3461 result |= 02;
3462 return result;
3465 void diff_addremove(struct diff_options *options,
3466 int addremove, unsigned mode,
3467 const unsigned char *sha1,
3468 const char *concatpath)
3470 struct diff_filespec *one, *two;
3472 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3473 return;
3475 /* This may look odd, but it is a preparation for
3476 * feeding "there are unchanged files which should
3477 * not produce diffs, but when you are doing copy
3478 * detection you would need them, so here they are"
3479 * entries to the diff-core. They will be prefixed
3480 * with something like '=' or '*' (I haven't decided
3481 * which but should not make any difference).
3482 * Feeding the same new and old to diff_change()
3483 * also has the same effect.
3484 * Before the final output happens, they are pruned after
3485 * merged into rename/copy pairs as appropriate.
3487 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3488 addremove = (addremove == '+' ? '-' :
3489 addremove == '-' ? '+' : addremove);
3491 if (options->prefix &&
3492 strncmp(concatpath, options->prefix, options->prefix_length))
3493 return;
3495 one = alloc_filespec(concatpath);
3496 two = alloc_filespec(concatpath);
3498 if (addremove != '+')
3499 fill_filespec(one, sha1, mode);
3500 if (addremove != '-')
3501 fill_filespec(two, sha1, mode);
3503 diff_queue(&diff_queued_diff, one, two);
3504 DIFF_OPT_SET(options, HAS_CHANGES);
3507 void diff_change(struct diff_options *options,
3508 unsigned old_mode, unsigned new_mode,
3509 const unsigned char *old_sha1,
3510 const unsigned char *new_sha1,
3511 const char *concatpath)
3513 struct diff_filespec *one, *two;
3515 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3516 && S_ISGITLINK(new_mode))
3517 return;
3519 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3520 unsigned tmp;
3521 const unsigned char *tmp_c;
3522 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3523 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3526 if (options->prefix &&
3527 strncmp(concatpath, options->prefix, options->prefix_length))
3528 return;
3530 one = alloc_filespec(concatpath);
3531 two = alloc_filespec(concatpath);
3532 fill_filespec(one, old_sha1, old_mode);
3533 fill_filespec(two, new_sha1, new_mode);
3535 diff_queue(&diff_queued_diff, one, two);
3536 DIFF_OPT_SET(options, HAS_CHANGES);
3539 void diff_unmerge(struct diff_options *options,
3540 const char *path,
3541 unsigned mode, const unsigned char *sha1)
3543 struct diff_filespec *one, *two;
3545 if (options->prefix &&
3546 strncmp(path, options->prefix, options->prefix_length))
3547 return;
3549 one = alloc_filespec(path);
3550 two = alloc_filespec(path);
3551 fill_filespec(one, sha1, mode);
3552 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3555 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3556 size_t *outsize)
3558 struct diff_tempfile *temp;
3559 const char *argv[3];
3560 const char **arg = argv;
3561 struct child_process child;
3562 struct strbuf buf = STRBUF_INIT;
3564 temp = prepare_temp_file(spec->path, spec);
3565 *arg++ = pgm;
3566 *arg++ = temp->name;
3567 *arg = NULL;
3569 memset(&child, 0, sizeof(child));
3570 child.argv = argv;
3571 child.out = -1;
3572 if (start_command(&child) != 0 ||
3573 strbuf_read(&buf, child.out, 0) < 0 ||
3574 finish_command(&child) != 0) {
3575 remove_tempfile();
3576 error("error running textconv command '%s'", pgm);
3577 return NULL;
3579 remove_tempfile();
3581 return strbuf_detach(&buf, outsize);