diff: do not color --stat output like patch context
[git/jnareb-git/bp-gitweb.git] / diff.c
blobd581d4d9ff13fc5da1c9857073a0ead1ecfe578b
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);
65 static int git_config_rename(const char *var, const char *value)
67 if (!value)
68 return DIFF_DETECT_RENAME;
69 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
70 return DIFF_DETECT_COPY;
71 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
75 * These are to give UI layer defaults.
76 * The core-level commands such as git-diff-files should
77 * never be affected by the setting of diff.renames
78 * the user happens to have in the configuration file.
80 int git_diff_ui_config(const char *var, const char *value, void *cb)
82 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
83 diff_use_color_default = git_config_colorbool(var, value, -1);
84 return 0;
86 if (!strcmp(var, "diff.renames")) {
87 diff_detect_rename_default = git_config_rename(var, value);
88 return 0;
90 if (!strcmp(var, "diff.autorefreshindex")) {
91 diff_auto_refresh_index = git_config_bool(var, value);
92 return 0;
94 if (!strcmp(var, "diff.mnemonicprefix")) {
95 diff_mnemonic_prefix = git_config_bool(var, value);
96 return 0;
98 if (!strcmp(var, "diff.external"))
99 return git_config_string(&external_diff_cmd_cfg, var, value);
100 if (!strcmp(var, "diff.wordregex"))
101 return git_config_string(&diff_word_regex_cfg, var, value);
103 return git_diff_basic_config(var, value, cb);
106 int git_diff_basic_config(const char *var, const char *value, void *cb)
108 if (!strcmp(var, "diff.renamelimit")) {
109 diff_rename_limit_default = git_config_int(var, value);
110 return 0;
113 switch (userdiff_config(var, value)) {
114 case 0: break;
115 case -1: return -1;
116 default: return 0;
119 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
120 int slot = parse_diff_color_slot(var, 11);
121 if (!value)
122 return config_error_nonbool(var);
123 color_parse(value, var, diff_colors[slot]);
124 return 0;
127 /* like GNU diff's --suppress-blank-empty option */
128 if (!strcmp(var, "diff.suppressblankempty") ||
129 /* for backwards compatibility */
130 !strcmp(var, "diff.suppress-blank-empty")) {
131 diff_suppress_blank_empty = git_config_bool(var, value);
132 return 0;
135 return git_color_default_config(var, value, cb);
138 static char *quote_two(const char *one, const char *two)
140 int need_one = quote_c_style(one, NULL, NULL, 1);
141 int need_two = quote_c_style(two, NULL, NULL, 1);
142 struct strbuf res = STRBUF_INIT;
144 if (need_one + need_two) {
145 strbuf_addch(&res, '"');
146 quote_c_style(one, &res, NULL, 1);
147 quote_c_style(two, &res, NULL, 1);
148 strbuf_addch(&res, '"');
149 } else {
150 strbuf_addstr(&res, one);
151 strbuf_addstr(&res, two);
153 return strbuf_detach(&res, NULL);
156 static const char *external_diff(void)
158 static const char *external_diff_cmd = NULL;
159 static int done_preparing = 0;
161 if (done_preparing)
162 return external_diff_cmd;
163 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
164 if (!external_diff_cmd)
165 external_diff_cmd = external_diff_cmd_cfg;
166 done_preparing = 1;
167 return external_diff_cmd;
170 static struct diff_tempfile {
171 const char *name; /* filename external diff should read from */
172 char hex[41];
173 char mode[10];
174 char tmp_path[PATH_MAX];
175 } diff_temp[2];
177 static struct diff_tempfile *claim_diff_tempfile(void) {
178 int i;
179 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
180 if (!diff_temp[i].name)
181 return diff_temp + i;
182 die("BUG: diff is failing to clean up its tempfiles");
185 static int remove_tempfile_installed;
187 static void remove_tempfile(void)
189 int i;
190 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
191 if (diff_temp[i].name == diff_temp[i].tmp_path)
192 unlink(diff_temp[i].name);
193 diff_temp[i].name = NULL;
197 static void remove_tempfile_on_signal(int signo)
199 remove_tempfile();
200 sigchain_pop(signo);
201 raise(signo);
204 static int count_lines(const char *data, int size)
206 int count, ch, completely_empty = 1, nl_just_seen = 0;
207 count = 0;
208 while (0 < size--) {
209 ch = *data++;
210 if (ch == '\n') {
211 count++;
212 nl_just_seen = 1;
213 completely_empty = 0;
215 else {
216 nl_just_seen = 0;
217 completely_empty = 0;
220 if (completely_empty)
221 return 0;
222 if (!nl_just_seen)
223 count++; /* no trailing newline */
224 return count;
227 static void print_line_count(FILE *file, int count)
229 switch (count) {
230 case 0:
231 fprintf(file, "0,0");
232 break;
233 case 1:
234 fprintf(file, "1");
235 break;
236 default:
237 fprintf(file, "1,%d", count);
238 break;
242 static void copy_file_with_prefix(FILE *file,
243 int prefix, const char *data, int size,
244 const char *set, const char *reset)
246 int ch, nl_just_seen = 1;
247 while (0 < size--) {
248 ch = *data++;
249 if (nl_just_seen) {
250 fputs(set, file);
251 putc(prefix, file);
253 if (ch == '\n') {
254 nl_just_seen = 1;
255 fputs(reset, file);
256 } else
257 nl_just_seen = 0;
258 putc(ch, file);
260 if (!nl_just_seen)
261 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
264 static void emit_rewrite_diff(const char *name_a,
265 const char *name_b,
266 struct diff_filespec *one,
267 struct diff_filespec *two,
268 const char *textconv_one,
269 const char *textconv_two,
270 struct diff_options *o)
272 int lc_a, lc_b;
273 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
274 const char *name_a_tab, *name_b_tab;
275 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
276 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
277 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
278 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
279 const char *reset = diff_get_color(color_diff, DIFF_RESET);
280 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
281 const char *a_prefix, *b_prefix;
282 const char *data_one, *data_two;
283 size_t size_one, size_two;
285 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
286 a_prefix = o->b_prefix;
287 b_prefix = o->a_prefix;
288 } else {
289 a_prefix = o->a_prefix;
290 b_prefix = o->b_prefix;
293 name_a += (*name_a == '/');
294 name_b += (*name_b == '/');
295 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
296 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
298 strbuf_reset(&a_name);
299 strbuf_reset(&b_name);
300 quote_two_c_style(&a_name, a_prefix, name_a, 0);
301 quote_two_c_style(&b_name, b_prefix, name_b, 0);
303 diff_populate_filespec(one, 0);
304 diff_populate_filespec(two, 0);
305 if (textconv_one) {
306 data_one = run_textconv(textconv_one, one, &size_one);
307 if (!data_one)
308 die("unable to read files to diff");
310 else {
311 data_one = one->data;
312 size_one = one->size;
314 if (textconv_two) {
315 data_two = run_textconv(textconv_two, two, &size_two);
316 if (!data_two)
317 die("unable to read files to diff");
319 else {
320 data_two = two->data;
321 size_two = two->size;
324 lc_a = count_lines(data_one, size_one);
325 lc_b = count_lines(data_two, size_two);
326 fprintf(o->file,
327 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
328 metainfo, a_name.buf, name_a_tab, reset,
329 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
330 print_line_count(o->file, lc_a);
331 fprintf(o->file, " +");
332 print_line_count(o->file, lc_b);
333 fprintf(o->file, " @@%s\n", reset);
334 if (lc_a)
335 copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
336 if (lc_b)
337 copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
340 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
342 if (!DIFF_FILE_VALID(one)) {
343 mf->ptr = (char *)""; /* does not matter */
344 mf->size = 0;
345 return 0;
347 else if (diff_populate_filespec(one, 0))
348 return -1;
350 mf->ptr = one->data;
351 mf->size = one->size;
352 return 0;
355 struct diff_words_buffer {
356 mmfile_t text;
357 long alloc;
358 struct diff_words_orig {
359 const char *begin, *end;
360 } *orig;
361 int orig_nr, orig_alloc;
364 static void diff_words_append(char *line, unsigned long len,
365 struct diff_words_buffer *buffer)
367 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
368 line++;
369 len--;
370 memcpy(buffer->text.ptr + buffer->text.size, line, len);
371 buffer->text.size += len;
372 buffer->text.ptr[buffer->text.size] = '\0';
375 struct diff_words_data {
376 struct diff_words_buffer minus, plus;
377 const char *current_plus;
378 FILE *file;
379 regex_t *word_regex;
382 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
384 struct diff_words_data *diff_words = priv;
385 int minus_first, minus_len, plus_first, plus_len;
386 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
388 if (line[0] != '@' || parse_hunk_header(line, len,
389 &minus_first, &minus_len, &plus_first, &plus_len))
390 return;
392 /* POSIX requires that first be decremented by one if len == 0... */
393 if (minus_len) {
394 minus_begin = diff_words->minus.orig[minus_first].begin;
395 minus_end =
396 diff_words->minus.orig[minus_first + minus_len - 1].end;
397 } else
398 minus_begin = minus_end =
399 diff_words->minus.orig[minus_first].end;
401 if (plus_len) {
402 plus_begin = diff_words->plus.orig[plus_first].begin;
403 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
404 } else
405 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
407 if (diff_words->current_plus != plus_begin)
408 fwrite(diff_words->current_plus,
409 plus_begin - diff_words->current_plus, 1,
410 diff_words->file);
411 if (minus_begin != minus_end)
412 color_fwrite_lines(diff_words->file,
413 diff_get_color(1, DIFF_FILE_OLD),
414 minus_end - minus_begin, minus_begin);
415 if (plus_begin != plus_end)
416 color_fwrite_lines(diff_words->file,
417 diff_get_color(1, DIFF_FILE_NEW),
418 plus_end - plus_begin, plus_begin);
420 diff_words->current_plus = plus_end;
423 /* This function starts looking at *begin, and returns 0 iff a word was found. */
424 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
425 int *begin, int *end)
427 if (word_regex && *begin < buffer->size) {
428 regmatch_t match[1];
429 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
430 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
431 '\n', match[0].rm_eo - match[0].rm_so);
432 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
433 *begin += match[0].rm_so;
434 return *begin >= *end;
436 return -1;
439 /* find the next word */
440 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
441 (*begin)++;
442 if (*begin >= buffer->size)
443 return -1;
445 /* find the end of the word */
446 *end = *begin + 1;
447 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
448 (*end)++;
450 return 0;
454 * This function splits the words in buffer->text, stores the list with
455 * newline separator into out, and saves the offsets of the original words
456 * in buffer->orig.
458 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
459 regex_t *word_regex)
461 int i, j;
462 long alloc = 0;
464 out->size = 0;
465 out->ptr = NULL;
467 /* fake an empty "0th" word */
468 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
469 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
470 buffer->orig_nr = 1;
472 for (i = 0; i < buffer->text.size; i++) {
473 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
474 return;
476 /* store original boundaries */
477 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
478 buffer->orig_alloc);
479 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
480 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
481 buffer->orig_nr++;
483 /* store one word */
484 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
485 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
486 out->ptr[out->size + j - i] = '\n';
487 out->size += j - i + 1;
489 i = j - 1;
493 /* this executes the word diff on the accumulated buffers */
494 static void diff_words_show(struct diff_words_data *diff_words)
496 xpparam_t xpp;
497 xdemitconf_t xecfg;
498 xdemitcb_t ecb;
499 mmfile_t minus, plus;
501 /* special case: only removal */
502 if (!diff_words->plus.text.size) {
503 color_fwrite_lines(diff_words->file,
504 diff_get_color(1, DIFF_FILE_OLD),
505 diff_words->minus.text.size, diff_words->minus.text.ptr);
506 diff_words->minus.text.size = 0;
507 return;
510 diff_words->current_plus = diff_words->plus.text.ptr;
512 memset(&xpp, 0, sizeof(xpp));
513 memset(&xecfg, 0, sizeof(xecfg));
514 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
515 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
516 xpp.flags = XDF_NEED_MINIMAL;
517 /* as only the hunk header will be parsed, we need a 0-context */
518 xecfg.ctxlen = 0;
519 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
520 &xpp, &xecfg, &ecb);
521 free(minus.ptr);
522 free(plus.ptr);
523 if (diff_words->current_plus != diff_words->plus.text.ptr +
524 diff_words->plus.text.size)
525 fwrite(diff_words->current_plus,
526 diff_words->plus.text.ptr + diff_words->plus.text.size
527 - diff_words->current_plus, 1,
528 diff_words->file);
529 diff_words->minus.text.size = diff_words->plus.text.size = 0;
532 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
534 struct emit_callback {
535 int nparents, color_diff;
536 unsigned ws_rule;
537 sane_truncate_fn truncate;
538 const char **label_path;
539 struct diff_words_data *diff_words;
540 int *found_changesp;
541 FILE *file;
544 static void free_diff_words_data(struct emit_callback *ecbdata)
546 if (ecbdata->diff_words) {
547 /* flush buffers */
548 if (ecbdata->diff_words->minus.text.size ||
549 ecbdata->diff_words->plus.text.size)
550 diff_words_show(ecbdata->diff_words);
552 free (ecbdata->diff_words->minus.text.ptr);
553 free (ecbdata->diff_words->minus.orig);
554 free (ecbdata->diff_words->plus.text.ptr);
555 free (ecbdata->diff_words->plus.orig);
556 free(ecbdata->diff_words->word_regex);
557 free(ecbdata->diff_words);
558 ecbdata->diff_words = NULL;
562 const char *diff_get_color(int diff_use_color, enum color_diff ix)
564 if (diff_use_color)
565 return diff_colors[ix];
566 return "";
569 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
571 int has_trailing_newline, has_trailing_carriage_return;
573 has_trailing_newline = (len > 0 && line[len-1] == '\n');
574 if (has_trailing_newline)
575 len--;
576 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
577 if (has_trailing_carriage_return)
578 len--;
580 fputs(set, file);
581 fwrite(line, len, 1, file);
582 fputs(reset, file);
583 if (has_trailing_carriage_return)
584 fputc('\r', file);
585 if (has_trailing_newline)
586 fputc('\n', file);
589 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
591 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
592 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
594 if (!*ws)
595 emit_line(ecbdata->file, set, reset, line, len);
596 else {
597 /* Emit just the prefix, then the rest. */
598 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
599 ws_check_emit(line + ecbdata->nparents,
600 len - ecbdata->nparents, ecbdata->ws_rule,
601 ecbdata->file, set, reset, ws);
605 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
607 const char *cp;
608 unsigned long allot;
609 size_t l = len;
611 if (ecb->truncate)
612 return ecb->truncate(line, len);
613 cp = line;
614 allot = l;
615 while (0 < l) {
616 (void) utf8_width(&cp, &l);
617 if (!cp)
618 break; /* truncated in the middle? */
620 return allot - l;
623 static void fn_out_consume(void *priv, char *line, unsigned long len)
625 int i;
626 int color;
627 struct emit_callback *ecbdata = priv;
628 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
629 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
630 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
632 *(ecbdata->found_changesp) = 1;
634 if (ecbdata->label_path[0]) {
635 const char *name_a_tab, *name_b_tab;
637 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
638 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
640 fprintf(ecbdata->file, "%s--- %s%s%s\n",
641 meta, ecbdata->label_path[0], reset, name_a_tab);
642 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
643 meta, ecbdata->label_path[1], reset, name_b_tab);
644 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
647 if (diff_suppress_blank_empty
648 && len == 2 && line[0] == ' ' && line[1] == '\n') {
649 line[0] = '\n';
650 len = 1;
653 /* This is not really necessary for now because
654 * this codepath only deals with two-way diffs.
656 for (i = 0; i < len && line[i] == '@'; i++)
658 if (2 <= i && i < len && line[i] == ' ') {
659 ecbdata->nparents = i - 1;
660 len = sane_truncate_line(ecbdata, line, len);
661 emit_line(ecbdata->file,
662 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
663 reset, line, len);
664 if (line[len-1] != '\n')
665 putc('\n', ecbdata->file);
666 return;
669 if (len < ecbdata->nparents) {
670 emit_line(ecbdata->file, reset, reset, line, len);
671 return;
674 color = DIFF_PLAIN;
675 if (ecbdata->diff_words && ecbdata->nparents != 1)
676 /* fall back to normal diff */
677 free_diff_words_data(ecbdata);
678 if (ecbdata->diff_words) {
679 if (line[0] == '-') {
680 diff_words_append(line, len,
681 &ecbdata->diff_words->minus);
682 return;
683 } else if (line[0] == '+') {
684 diff_words_append(line, len,
685 &ecbdata->diff_words->plus);
686 return;
688 if (ecbdata->diff_words->minus.text.size ||
689 ecbdata->diff_words->plus.text.size)
690 diff_words_show(ecbdata->diff_words);
691 line++;
692 len--;
693 emit_line(ecbdata->file, plain, reset, line, len);
694 return;
696 for (i = 0; i < ecbdata->nparents && len; i++) {
697 if (line[i] == '-')
698 color = DIFF_FILE_OLD;
699 else if (line[i] == '+')
700 color = DIFF_FILE_NEW;
703 if (color != DIFF_FILE_NEW) {
704 emit_line(ecbdata->file,
705 diff_get_color(ecbdata->color_diff, color),
706 reset, line, len);
707 return;
709 emit_add_line(reset, ecbdata, line, len);
712 static char *pprint_rename(const char *a, const char *b)
714 const char *old = a;
715 const char *new = b;
716 struct strbuf name = STRBUF_INIT;
717 int pfx_length, sfx_length;
718 int len_a = strlen(a);
719 int len_b = strlen(b);
720 int a_midlen, b_midlen;
721 int qlen_a = quote_c_style(a, NULL, NULL, 0);
722 int qlen_b = quote_c_style(b, NULL, NULL, 0);
724 if (qlen_a || qlen_b) {
725 quote_c_style(a, &name, NULL, 0);
726 strbuf_addstr(&name, " => ");
727 quote_c_style(b, &name, NULL, 0);
728 return strbuf_detach(&name, NULL);
731 /* Find common prefix */
732 pfx_length = 0;
733 while (*old && *new && *old == *new) {
734 if (*old == '/')
735 pfx_length = old - a + 1;
736 old++;
737 new++;
740 /* Find common suffix */
741 old = a + len_a;
742 new = b + len_b;
743 sfx_length = 0;
744 while (a <= old && b <= new && *old == *new) {
745 if (*old == '/')
746 sfx_length = len_a - (old - a);
747 old--;
748 new--;
752 * pfx{mid-a => mid-b}sfx
753 * {pfx-a => pfx-b}sfx
754 * pfx{sfx-a => sfx-b}
755 * name-a => name-b
757 a_midlen = len_a - pfx_length - sfx_length;
758 b_midlen = len_b - pfx_length - sfx_length;
759 if (a_midlen < 0)
760 a_midlen = 0;
761 if (b_midlen < 0)
762 b_midlen = 0;
764 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
765 if (pfx_length + sfx_length) {
766 strbuf_add(&name, a, pfx_length);
767 strbuf_addch(&name, '{');
769 strbuf_add(&name, a + pfx_length, a_midlen);
770 strbuf_addstr(&name, " => ");
771 strbuf_add(&name, b + pfx_length, b_midlen);
772 if (pfx_length + sfx_length) {
773 strbuf_addch(&name, '}');
774 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
776 return strbuf_detach(&name, NULL);
779 struct diffstat_t {
780 int nr;
781 int alloc;
782 struct diffstat_file {
783 char *from_name;
784 char *name;
785 char *print_name;
786 unsigned is_unmerged:1;
787 unsigned is_binary:1;
788 unsigned is_renamed:1;
789 unsigned int added, deleted;
790 } **files;
793 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
794 const char *name_a,
795 const char *name_b)
797 struct diffstat_file *x;
798 x = xcalloc(sizeof (*x), 1);
799 if (diffstat->nr == diffstat->alloc) {
800 diffstat->alloc = alloc_nr(diffstat->alloc);
801 diffstat->files = xrealloc(diffstat->files,
802 diffstat->alloc * sizeof(x));
804 diffstat->files[diffstat->nr++] = x;
805 if (name_b) {
806 x->from_name = xstrdup(name_a);
807 x->name = xstrdup(name_b);
808 x->is_renamed = 1;
810 else {
811 x->from_name = NULL;
812 x->name = xstrdup(name_a);
814 return x;
817 static void diffstat_consume(void *priv, char *line, unsigned long len)
819 struct diffstat_t *diffstat = priv;
820 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
822 if (line[0] == '+')
823 x->added++;
824 else if (line[0] == '-')
825 x->deleted++;
828 const char mime_boundary_leader[] = "------------";
830 static int scale_linear(int it, int width, int max_change)
833 * make sure that at least one '-' is printed if there were deletions,
834 * and likewise for '+'.
836 if (max_change < 2)
837 return it;
838 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
841 static void show_name(FILE *file,
842 const char *prefix, const char *name, int len)
844 fprintf(file, " %s%-*s |", prefix, len, name);
847 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
849 if (cnt <= 0)
850 return;
851 fprintf(file, "%s", set);
852 while (cnt--)
853 putc(ch, file);
854 fprintf(file, "%s", reset);
857 static void fill_print_name(struct diffstat_file *file)
859 char *pname;
861 if (file->print_name)
862 return;
864 if (!file->is_renamed) {
865 struct strbuf buf = STRBUF_INIT;
866 if (quote_c_style(file->name, &buf, NULL, 0)) {
867 pname = strbuf_detach(&buf, NULL);
868 } else {
869 pname = file->name;
870 strbuf_release(&buf);
872 } else {
873 pname = pprint_rename(file->from_name, file->name);
875 file->print_name = pname;
878 static void show_stats(struct diffstat_t* data, struct diff_options *options)
880 int i, len, add, del, adds = 0, dels = 0;
881 int max_change = 0, max_len = 0;
882 int total_files = data->nr;
883 int width, name_width;
884 const char *reset, *set, *add_c, *del_c;
886 if (data->nr == 0)
887 return;
889 width = options->stat_width ? options->stat_width : 80;
890 name_width = options->stat_name_width ? options->stat_name_width : 50;
892 /* Sanity: give at least 5 columns to the graph,
893 * but leave at least 10 columns for the name.
895 if (width < 25)
896 width = 25;
897 if (name_width < 10)
898 name_width = 10;
899 else if (width < name_width + 15)
900 name_width = width - 15;
902 /* Find the longest filename and max number of changes */
903 reset = diff_get_color_opt(options, DIFF_RESET);
904 set = diff_get_color_opt(options, DIFF_PLAIN);
905 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
906 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
908 for (i = 0; i < data->nr; i++) {
909 struct diffstat_file *file = data->files[i];
910 int change = file->added + file->deleted;
911 fill_print_name(file);
912 len = strlen(file->print_name);
913 if (max_len < len)
914 max_len = len;
916 if (file->is_binary || file->is_unmerged)
917 continue;
918 if (max_change < change)
919 max_change = change;
922 /* Compute the width of the graph part;
923 * 10 is for one blank at the beginning of the line plus
924 * " | count " between the name and the graph.
926 * From here on, name_width is the width of the name area,
927 * and width is the width of the graph area.
929 name_width = (name_width < max_len) ? name_width : max_len;
930 if (width < (name_width + 10) + max_change)
931 width = width - (name_width + 10);
932 else
933 width = max_change;
935 for (i = 0; i < data->nr; i++) {
936 const char *prefix = "";
937 char *name = data->files[i]->print_name;
938 int added = data->files[i]->added;
939 int deleted = data->files[i]->deleted;
940 int name_len;
943 * "scale" the filename
945 len = name_width;
946 name_len = strlen(name);
947 if (name_width < name_len) {
948 char *slash;
949 prefix = "...";
950 len -= 3;
951 name += name_len - len;
952 slash = strchr(name, '/');
953 if (slash)
954 name = slash;
957 if (data->files[i]->is_binary) {
958 show_name(options->file, prefix, name, len);
959 fprintf(options->file, " Bin ");
960 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
961 fprintf(options->file, " -> ");
962 fprintf(options->file, "%s%d%s", add_c, added, reset);
963 fprintf(options->file, " bytes");
964 fprintf(options->file, "\n");
965 continue;
967 else if (data->files[i]->is_unmerged) {
968 show_name(options->file, prefix, name, len);
969 fprintf(options->file, " Unmerged\n");
970 continue;
972 else if (!data->files[i]->is_renamed &&
973 (added + deleted == 0)) {
974 total_files--;
975 continue;
979 * scale the add/delete
981 add = added;
982 del = deleted;
983 adds += add;
984 dels += del;
986 if (width <= max_change) {
987 add = scale_linear(add, width, max_change);
988 del = scale_linear(del, width, max_change);
990 show_name(options->file, prefix, name, len);
991 fprintf(options->file, "%5d%s", added + deleted,
992 added + deleted ? " " : "");
993 show_graph(options->file, '+', add, add_c, reset);
994 show_graph(options->file, '-', del, del_c, reset);
995 fprintf(options->file, "\n");
997 fprintf(options->file,
998 " %d files changed, %d insertions(+), %d deletions(-)\n",
999 total_files, adds, dels);
1002 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
1004 int i, adds = 0, dels = 0, total_files = data->nr;
1006 if (data->nr == 0)
1007 return;
1009 for (i = 0; i < data->nr; i++) {
1010 if (!data->files[i]->is_binary &&
1011 !data->files[i]->is_unmerged) {
1012 int added = data->files[i]->added;
1013 int deleted= data->files[i]->deleted;
1014 if (!data->files[i]->is_renamed &&
1015 (added + deleted == 0)) {
1016 total_files--;
1017 } else {
1018 adds += added;
1019 dels += deleted;
1023 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1024 total_files, adds, dels);
1027 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
1029 int i;
1031 if (data->nr == 0)
1032 return;
1034 for (i = 0; i < data->nr; i++) {
1035 struct diffstat_file *file = data->files[i];
1037 if (file->is_binary)
1038 fprintf(options->file, "-\t-\t");
1039 else
1040 fprintf(options->file,
1041 "%d\t%d\t", file->added, file->deleted);
1042 if (options->line_termination) {
1043 fill_print_name(file);
1044 if (!file->is_renamed)
1045 write_name_quoted(file->name, options->file,
1046 options->line_termination);
1047 else {
1048 fputs(file->print_name, options->file);
1049 putc(options->line_termination, options->file);
1051 } else {
1052 if (file->is_renamed) {
1053 putc('\0', options->file);
1054 write_name_quoted(file->from_name, options->file, '\0');
1056 write_name_quoted(file->name, options->file, '\0');
1061 struct dirstat_file {
1062 const char *name;
1063 unsigned long changed;
1066 struct dirstat_dir {
1067 struct dirstat_file *files;
1068 int alloc, nr, percent, cumulative;
1071 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1073 unsigned long this_dir = 0;
1074 unsigned int sources = 0;
1076 while (dir->nr) {
1077 struct dirstat_file *f = dir->files;
1078 int namelen = strlen(f->name);
1079 unsigned long this;
1080 char *slash;
1082 if (namelen < baselen)
1083 break;
1084 if (memcmp(f->name, base, baselen))
1085 break;
1086 slash = strchr(f->name + baselen, '/');
1087 if (slash) {
1088 int newbaselen = slash + 1 - f->name;
1089 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1090 sources++;
1091 } else {
1092 this = f->changed;
1093 dir->files++;
1094 dir->nr--;
1095 sources += 2;
1097 this_dir += this;
1101 * We don't report dirstat's for
1102 * - the top level
1103 * - or cases where everything came from a single directory
1104 * under this directory (sources == 1).
1106 if (baselen && sources != 1) {
1107 int permille = this_dir * 1000 / changed;
1108 if (permille) {
1109 int percent = permille / 10;
1110 if (percent >= dir->percent) {
1111 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1112 if (!dir->cumulative)
1113 return 0;
1117 return this_dir;
1120 static int dirstat_compare(const void *_a, const void *_b)
1122 const struct dirstat_file *a = _a;
1123 const struct dirstat_file *b = _b;
1124 return strcmp(a->name, b->name);
1127 static void show_dirstat(struct diff_options *options)
1129 int i;
1130 unsigned long changed;
1131 struct dirstat_dir dir;
1132 struct diff_queue_struct *q = &diff_queued_diff;
1134 dir.files = NULL;
1135 dir.alloc = 0;
1136 dir.nr = 0;
1137 dir.percent = options->dirstat_percent;
1138 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1140 changed = 0;
1141 for (i = 0; i < q->nr; i++) {
1142 struct diff_filepair *p = q->queue[i];
1143 const char *name;
1144 unsigned long copied, added, damage;
1146 name = p->one->path ? p->one->path : p->two->path;
1148 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1149 diff_populate_filespec(p->one, 0);
1150 diff_populate_filespec(p->two, 0);
1151 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1152 &copied, &added);
1153 diff_free_filespec_data(p->one);
1154 diff_free_filespec_data(p->two);
1155 } else if (DIFF_FILE_VALID(p->one)) {
1156 diff_populate_filespec(p->one, 1);
1157 copied = added = 0;
1158 diff_free_filespec_data(p->one);
1159 } else if (DIFF_FILE_VALID(p->two)) {
1160 diff_populate_filespec(p->two, 1);
1161 copied = 0;
1162 added = p->two->size;
1163 diff_free_filespec_data(p->two);
1164 } else
1165 continue;
1168 * Original minus copied is the removed material,
1169 * added is the new material. They are both damages
1170 * made to the preimage. In --dirstat-by-file mode, count
1171 * damaged files, not damaged lines. This is done by
1172 * counting only a single damaged line per file.
1174 damage = (p->one->size - copied) + added;
1175 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1176 damage = 1;
1178 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1179 dir.files[dir.nr].name = name;
1180 dir.files[dir.nr].changed = damage;
1181 changed += damage;
1182 dir.nr++;
1185 /* This can happen even with many files, if everything was renames */
1186 if (!changed)
1187 return;
1189 /* Show all directories with more than x% of the changes */
1190 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1191 gather_dirstat(options->file, &dir, changed, "", 0);
1194 static void free_diffstat_info(struct diffstat_t *diffstat)
1196 int i;
1197 for (i = 0; i < diffstat->nr; i++) {
1198 struct diffstat_file *f = diffstat->files[i];
1199 if (f->name != f->print_name)
1200 free(f->print_name);
1201 free(f->name);
1202 free(f->from_name);
1203 free(f);
1205 free(diffstat->files);
1208 struct checkdiff_t {
1209 const char *filename;
1210 int lineno;
1211 struct diff_options *o;
1212 unsigned ws_rule;
1213 unsigned status;
1214 int trailing_blanks_start;
1217 static int is_conflict_marker(const char *line, unsigned long len)
1219 char firstchar;
1220 int cnt;
1222 if (len < 8)
1223 return 0;
1224 firstchar = line[0];
1225 switch (firstchar) {
1226 case '=': case '>': case '<':
1227 break;
1228 default:
1229 return 0;
1231 for (cnt = 1; cnt < 7; cnt++)
1232 if (line[cnt] != firstchar)
1233 return 0;
1234 /* line[0] thru line[6] are same as firstchar */
1235 if (firstchar == '=') {
1236 /* divider between ours and theirs? */
1237 if (len != 8 || line[7] != '\n')
1238 return 0;
1239 } else if (len < 8 || !isspace(line[7])) {
1240 /* not divider before ours nor after theirs */
1241 return 0;
1243 return 1;
1246 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1248 struct checkdiff_t *data = priv;
1249 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1250 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1251 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1252 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1253 char *err;
1255 if (line[0] == '+') {
1256 unsigned bad;
1257 data->lineno++;
1258 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1259 data->trailing_blanks_start = 0;
1260 else if (!data->trailing_blanks_start)
1261 data->trailing_blanks_start = data->lineno;
1262 if (is_conflict_marker(line + 1, len - 1)) {
1263 data->status |= 1;
1264 fprintf(data->o->file,
1265 "%s:%d: leftover conflict marker\n",
1266 data->filename, data->lineno);
1268 bad = ws_check(line + 1, len - 1, data->ws_rule);
1269 if (!bad)
1270 return;
1271 data->status |= bad;
1272 err = whitespace_error_string(bad);
1273 fprintf(data->o->file, "%s:%d: %s.\n",
1274 data->filename, data->lineno, err);
1275 free(err);
1276 emit_line(data->o->file, set, reset, line, 1);
1277 ws_check_emit(line + 1, len - 1, data->ws_rule,
1278 data->o->file, set, reset, ws);
1279 } else if (line[0] == ' ') {
1280 data->lineno++;
1281 data->trailing_blanks_start = 0;
1282 } else if (line[0] == '@') {
1283 char *plus = strchr(line, '+');
1284 if (plus)
1285 data->lineno = strtol(plus, NULL, 10) - 1;
1286 else
1287 die("invalid diff");
1288 data->trailing_blanks_start = 0;
1292 static unsigned char *deflate_it(char *data,
1293 unsigned long size,
1294 unsigned long *result_size)
1296 int bound;
1297 unsigned char *deflated;
1298 z_stream stream;
1300 memset(&stream, 0, sizeof(stream));
1301 deflateInit(&stream, zlib_compression_level);
1302 bound = deflateBound(&stream, size);
1303 deflated = xmalloc(bound);
1304 stream.next_out = deflated;
1305 stream.avail_out = bound;
1307 stream.next_in = (unsigned char *)data;
1308 stream.avail_in = size;
1309 while (deflate(&stream, Z_FINISH) == Z_OK)
1310 ; /* nothing */
1311 deflateEnd(&stream);
1312 *result_size = stream.total_out;
1313 return deflated;
1316 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1318 void *cp;
1319 void *delta;
1320 void *deflated;
1321 void *data;
1322 unsigned long orig_size;
1323 unsigned long delta_size;
1324 unsigned long deflate_size;
1325 unsigned long data_size;
1327 /* We could do deflated delta, or we could do just deflated two,
1328 * whichever is smaller.
1330 delta = NULL;
1331 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1332 if (one->size && two->size) {
1333 delta = diff_delta(one->ptr, one->size,
1334 two->ptr, two->size,
1335 &delta_size, deflate_size);
1336 if (delta) {
1337 void *to_free = delta;
1338 orig_size = delta_size;
1339 delta = deflate_it(delta, delta_size, &delta_size);
1340 free(to_free);
1344 if (delta && delta_size < deflate_size) {
1345 fprintf(file, "delta %lu\n", orig_size);
1346 free(deflated);
1347 data = delta;
1348 data_size = delta_size;
1350 else {
1351 fprintf(file, "literal %lu\n", two->size);
1352 free(delta);
1353 data = deflated;
1354 data_size = deflate_size;
1357 /* emit data encoded in base85 */
1358 cp = data;
1359 while (data_size) {
1360 int bytes = (52 < data_size) ? 52 : data_size;
1361 char line[70];
1362 data_size -= bytes;
1363 if (bytes <= 26)
1364 line[0] = bytes + 'A' - 1;
1365 else
1366 line[0] = bytes - 26 + 'a' - 1;
1367 encode_85(line + 1, cp, bytes);
1368 cp = (char *) cp + bytes;
1369 fputs(line, file);
1370 fputc('\n', file);
1372 fprintf(file, "\n");
1373 free(data);
1376 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1378 fprintf(file, "GIT binary patch\n");
1379 emit_binary_diff_body(file, one, two);
1380 emit_binary_diff_body(file, two, one);
1383 static void diff_filespec_load_driver(struct diff_filespec *one)
1385 if (!one->driver)
1386 one->driver = userdiff_find_by_path(one->path);
1387 if (!one->driver)
1388 one->driver = userdiff_find_by_name("default");
1391 int diff_filespec_is_binary(struct diff_filespec *one)
1393 if (one->is_binary == -1) {
1394 diff_filespec_load_driver(one);
1395 if (one->driver->binary != -1)
1396 one->is_binary = one->driver->binary;
1397 else {
1398 if (!one->data && DIFF_FILE_VALID(one))
1399 diff_populate_filespec(one, 0);
1400 if (one->data)
1401 one->is_binary = buffer_is_binary(one->data,
1402 one->size);
1403 if (one->is_binary == -1)
1404 one->is_binary = 0;
1407 return one->is_binary;
1410 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1412 diff_filespec_load_driver(one);
1413 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1416 static const char *userdiff_word_regex(struct diff_filespec *one)
1418 diff_filespec_load_driver(one);
1419 return one->driver->word_regex;
1422 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1424 if (!options->a_prefix)
1425 options->a_prefix = a;
1426 if (!options->b_prefix)
1427 options->b_prefix = b;
1430 static const char *get_textconv(struct diff_filespec *one)
1432 if (!DIFF_FILE_VALID(one))
1433 return NULL;
1434 if (!S_ISREG(one->mode))
1435 return NULL;
1436 diff_filespec_load_driver(one);
1437 return one->driver->textconv;
1440 static void builtin_diff(const char *name_a,
1441 const char *name_b,
1442 struct diff_filespec *one,
1443 struct diff_filespec *two,
1444 const char *xfrm_msg,
1445 struct diff_options *o,
1446 int complete_rewrite)
1448 mmfile_t mf1, mf2;
1449 const char *lbl[2];
1450 char *a_one, *b_two;
1451 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1452 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1453 const char *a_prefix, *b_prefix;
1454 const char *textconv_one = NULL, *textconv_two = NULL;
1456 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1457 textconv_one = get_textconv(one);
1458 textconv_two = get_textconv(two);
1461 diff_set_mnemonic_prefix(o, "a/", "b/");
1462 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1463 a_prefix = o->b_prefix;
1464 b_prefix = o->a_prefix;
1465 } else {
1466 a_prefix = o->a_prefix;
1467 b_prefix = o->b_prefix;
1470 /* Never use a non-valid filename anywhere if at all possible */
1471 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1472 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1474 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1475 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1476 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1477 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1478 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1479 if (lbl[0][0] == '/') {
1480 /* /dev/null */
1481 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1482 if (xfrm_msg && xfrm_msg[0])
1483 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1485 else if (lbl[1][0] == '/') {
1486 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1487 if (xfrm_msg && xfrm_msg[0])
1488 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1490 else {
1491 if (one->mode != two->mode) {
1492 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1493 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1495 if (xfrm_msg && xfrm_msg[0])
1496 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1498 * we do not run diff between different kind
1499 * of objects.
1501 if ((one->mode ^ two->mode) & S_IFMT)
1502 goto free_ab_and_return;
1503 if (complete_rewrite &&
1504 (textconv_one || !diff_filespec_is_binary(one)) &&
1505 (textconv_two || !diff_filespec_is_binary(two))) {
1506 emit_rewrite_diff(name_a, name_b, one, two,
1507 textconv_one, textconv_two, o);
1508 o->found_changes = 1;
1509 goto free_ab_and_return;
1513 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1514 die("unable to read files to diff");
1516 if (!DIFF_OPT_TST(o, TEXT) &&
1517 ( (diff_filespec_is_binary(one) && !textconv_one) ||
1518 (diff_filespec_is_binary(two) && !textconv_two) )) {
1519 /* Quite common confusing case */
1520 if (mf1.size == mf2.size &&
1521 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1522 goto free_ab_and_return;
1523 if (DIFF_OPT_TST(o, BINARY))
1524 emit_binary_diff(o->file, &mf1, &mf2);
1525 else
1526 fprintf(o->file, "Binary files %s and %s differ\n",
1527 lbl[0], lbl[1]);
1528 o->found_changes = 1;
1530 else {
1531 /* Crazy xdl interfaces.. */
1532 const char *diffopts = getenv("GIT_DIFF_OPTS");
1533 xpparam_t xpp;
1534 xdemitconf_t xecfg;
1535 xdemitcb_t ecb;
1536 struct emit_callback ecbdata;
1537 const struct userdiff_funcname *pe;
1539 if (textconv_one) {
1540 size_t size;
1541 mf1.ptr = run_textconv(textconv_one, one, &size);
1542 if (!mf1.ptr)
1543 die("unable to read files to diff");
1544 mf1.size = size;
1546 if (textconv_two) {
1547 size_t size;
1548 mf2.ptr = run_textconv(textconv_two, two, &size);
1549 if (!mf2.ptr)
1550 die("unable to read files to diff");
1551 mf2.size = size;
1554 pe = diff_funcname_pattern(one);
1555 if (!pe)
1556 pe = diff_funcname_pattern(two);
1558 memset(&xpp, 0, sizeof(xpp));
1559 memset(&xecfg, 0, sizeof(xecfg));
1560 memset(&ecbdata, 0, sizeof(ecbdata));
1561 ecbdata.label_path = lbl;
1562 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1563 ecbdata.found_changesp = &o->found_changes;
1564 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1565 ecbdata.file = o->file;
1566 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1567 xecfg.ctxlen = o->context;
1568 xecfg.interhunkctxlen = o->interhunkcontext;
1569 xecfg.flags = XDL_EMIT_FUNCNAMES;
1570 if (pe)
1571 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1572 if (!diffopts)
1574 else if (!prefixcmp(diffopts, "--unified="))
1575 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1576 else if (!prefixcmp(diffopts, "-u"))
1577 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1578 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1579 ecbdata.diff_words =
1580 xcalloc(1, sizeof(struct diff_words_data));
1581 ecbdata.diff_words->file = o->file;
1582 if (!o->word_regex)
1583 o->word_regex = userdiff_word_regex(one);
1584 if (!o->word_regex)
1585 o->word_regex = userdiff_word_regex(two);
1586 if (!o->word_regex)
1587 o->word_regex = diff_word_regex_cfg;
1588 if (o->word_regex) {
1589 ecbdata.diff_words->word_regex = (regex_t *)
1590 xmalloc(sizeof(regex_t));
1591 if (regcomp(ecbdata.diff_words->word_regex,
1592 o->word_regex,
1593 REG_EXTENDED | REG_NEWLINE))
1594 die ("Invalid regular expression: %s",
1595 o->word_regex);
1598 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1599 &xpp, &xecfg, &ecb);
1600 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1601 free_diff_words_data(&ecbdata);
1602 if (textconv_one)
1603 free(mf1.ptr);
1604 if (textconv_two)
1605 free(mf2.ptr);
1608 free_ab_and_return:
1609 diff_free_filespec_data(one);
1610 diff_free_filespec_data(two);
1611 free(a_one);
1612 free(b_two);
1613 return;
1616 static void builtin_diffstat(const char *name_a, const char *name_b,
1617 struct diff_filespec *one,
1618 struct diff_filespec *two,
1619 struct diffstat_t *diffstat,
1620 struct diff_options *o,
1621 int complete_rewrite)
1623 mmfile_t mf1, mf2;
1624 struct diffstat_file *data;
1626 data = diffstat_add(diffstat, name_a, name_b);
1628 if (!one || !two) {
1629 data->is_unmerged = 1;
1630 return;
1632 if (complete_rewrite) {
1633 diff_populate_filespec(one, 0);
1634 diff_populate_filespec(two, 0);
1635 data->deleted = count_lines(one->data, one->size);
1636 data->added = count_lines(two->data, two->size);
1637 goto free_and_return;
1639 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1640 die("unable to read files to diff");
1642 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1643 data->is_binary = 1;
1644 data->added = mf2.size;
1645 data->deleted = mf1.size;
1646 } else {
1647 /* Crazy xdl interfaces.. */
1648 xpparam_t xpp;
1649 xdemitconf_t xecfg;
1650 xdemitcb_t ecb;
1652 memset(&xpp, 0, sizeof(xpp));
1653 memset(&xecfg, 0, sizeof(xecfg));
1654 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1655 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1656 &xpp, &xecfg, &ecb);
1659 free_and_return:
1660 diff_free_filespec_data(one);
1661 diff_free_filespec_data(two);
1664 static void builtin_checkdiff(const char *name_a, const char *name_b,
1665 const char *attr_path,
1666 struct diff_filespec *one,
1667 struct diff_filespec *two,
1668 struct diff_options *o)
1670 mmfile_t mf1, mf2;
1671 struct checkdiff_t data;
1673 if (!two)
1674 return;
1676 memset(&data, 0, sizeof(data));
1677 data.filename = name_b ? name_b : name_a;
1678 data.lineno = 0;
1679 data.o = o;
1680 data.ws_rule = whitespace_rule(attr_path);
1682 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1683 die("unable to read files to diff");
1686 * All the other codepaths check both sides, but not checking
1687 * the "old" side here is deliberate. We are checking the newly
1688 * introduced changes, and as long as the "new" side is text, we
1689 * can and should check what it introduces.
1691 if (diff_filespec_is_binary(two))
1692 goto free_and_return;
1693 else {
1694 /* Crazy xdl interfaces.. */
1695 xpparam_t xpp;
1696 xdemitconf_t xecfg;
1697 xdemitcb_t ecb;
1699 memset(&xpp, 0, sizeof(xpp));
1700 memset(&xecfg, 0, sizeof(xecfg));
1701 xecfg.ctxlen = 1; /* at least one context line */
1702 xpp.flags = XDF_NEED_MINIMAL;
1703 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1704 &xpp, &xecfg, &ecb);
1706 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1707 data.trailing_blanks_start) {
1708 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1709 data.filename, data.trailing_blanks_start);
1710 data.status = 1; /* report errors */
1713 free_and_return:
1714 diff_free_filespec_data(one);
1715 diff_free_filespec_data(two);
1716 if (data.status)
1717 DIFF_OPT_SET(o, CHECK_FAILED);
1720 struct diff_filespec *alloc_filespec(const char *path)
1722 int namelen = strlen(path);
1723 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1725 memset(spec, 0, sizeof(*spec));
1726 spec->path = (char *)(spec + 1);
1727 memcpy(spec->path, path, namelen+1);
1728 spec->count = 1;
1729 spec->is_binary = -1;
1730 return spec;
1733 void free_filespec(struct diff_filespec *spec)
1735 if (!--spec->count) {
1736 diff_free_filespec_data(spec);
1737 free(spec);
1741 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1742 unsigned short mode)
1744 if (mode) {
1745 spec->mode = canon_mode(mode);
1746 hashcpy(spec->sha1, sha1);
1747 spec->sha1_valid = !is_null_sha1(sha1);
1752 * Given a name and sha1 pair, if the index tells us the file in
1753 * the work tree has that object contents, return true, so that
1754 * prepare_temp_file() does not have to inflate and extract.
1756 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1758 struct cache_entry *ce;
1759 struct stat st;
1760 int pos, len;
1763 * We do not read the cache ourselves here, because the
1764 * benchmark with my previous version that always reads cache
1765 * shows that it makes things worse for diff-tree comparing
1766 * two linux-2.6 kernel trees in an already checked out work
1767 * tree. This is because most diff-tree comparisons deal with
1768 * only a small number of files, while reading the cache is
1769 * expensive for a large project, and its cost outweighs the
1770 * savings we get by not inflating the object to a temporary
1771 * file. Practically, this code only helps when we are used
1772 * by diff-cache --cached, which does read the cache before
1773 * calling us.
1775 if (!active_cache)
1776 return 0;
1778 /* We want to avoid the working directory if our caller
1779 * doesn't need the data in a normal file, this system
1780 * is rather slow with its stat/open/mmap/close syscalls,
1781 * and the object is contained in a pack file. The pack
1782 * is probably already open and will be faster to obtain
1783 * the data through than the working directory. Loose
1784 * objects however would tend to be slower as they need
1785 * to be individually opened and inflated.
1787 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1788 return 0;
1790 len = strlen(name);
1791 pos = cache_name_pos(name, len);
1792 if (pos < 0)
1793 return 0;
1794 ce = active_cache[pos];
1797 * This is not the sha1 we are looking for, or
1798 * unreusable because it is not a regular file.
1800 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1801 return 0;
1804 * If ce is marked as "assume unchanged", there is no
1805 * guarantee that work tree matches what we are looking for.
1807 if (ce->ce_flags & CE_VALID)
1808 return 0;
1811 * If ce matches the file in the work tree, we can reuse it.
1813 if (ce_uptodate(ce) ||
1814 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1815 return 1;
1817 return 0;
1820 static int populate_from_stdin(struct diff_filespec *s)
1822 struct strbuf buf = STRBUF_INIT;
1823 size_t size = 0;
1825 if (strbuf_read(&buf, 0, 0) < 0)
1826 return error("error while reading from stdin %s",
1827 strerror(errno));
1829 s->should_munmap = 0;
1830 s->data = strbuf_detach(&buf, &size);
1831 s->size = size;
1832 s->should_free = 1;
1833 return 0;
1836 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1838 int len;
1839 char *data = xmalloc(100);
1840 len = snprintf(data, 100,
1841 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1842 s->data = data;
1843 s->size = len;
1844 s->should_free = 1;
1845 if (size_only) {
1846 s->data = NULL;
1847 free(data);
1849 return 0;
1853 * While doing rename detection and pickaxe operation, we may need to
1854 * grab the data for the blob (or file) for our own in-core comparison.
1855 * diff_filespec has data and size fields for this purpose.
1857 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1859 int err = 0;
1860 if (!DIFF_FILE_VALID(s))
1861 die("internal error: asking to populate invalid file.");
1862 if (S_ISDIR(s->mode))
1863 return -1;
1865 if (s->data)
1866 return 0;
1868 if (size_only && 0 < s->size)
1869 return 0;
1871 if (S_ISGITLINK(s->mode))
1872 return diff_populate_gitlink(s, size_only);
1874 if (!s->sha1_valid ||
1875 reuse_worktree_file(s->path, s->sha1, 0)) {
1876 struct strbuf buf = STRBUF_INIT;
1877 struct stat st;
1878 int fd;
1880 if (!strcmp(s->path, "-"))
1881 return populate_from_stdin(s);
1883 if (lstat(s->path, &st) < 0) {
1884 if (errno == ENOENT) {
1885 err_empty:
1886 err = -1;
1887 empty:
1888 s->data = (char *)"";
1889 s->size = 0;
1890 return err;
1893 s->size = xsize_t(st.st_size);
1894 if (!s->size)
1895 goto empty;
1896 if (S_ISLNK(st.st_mode)) {
1897 struct strbuf sb = STRBUF_INIT;
1899 if (strbuf_readlink(&sb, s->path, s->size))
1900 goto err_empty;
1901 s->size = sb.len;
1902 s->data = strbuf_detach(&sb, NULL);
1903 s->should_free = 1;
1904 return 0;
1906 if (size_only)
1907 return 0;
1908 fd = open(s->path, O_RDONLY);
1909 if (fd < 0)
1910 goto err_empty;
1911 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1912 close(fd);
1913 s->should_munmap = 1;
1916 * Convert from working tree format to canonical git format
1918 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1919 size_t size = 0;
1920 munmap(s->data, s->size);
1921 s->should_munmap = 0;
1922 s->data = strbuf_detach(&buf, &size);
1923 s->size = size;
1924 s->should_free = 1;
1927 else {
1928 enum object_type type;
1929 if (size_only)
1930 type = sha1_object_info(s->sha1, &s->size);
1931 else {
1932 s->data = read_sha1_file(s->sha1, &type, &s->size);
1933 s->should_free = 1;
1936 return 0;
1939 void diff_free_filespec_blob(struct diff_filespec *s)
1941 if (s->should_free)
1942 free(s->data);
1943 else if (s->should_munmap)
1944 munmap(s->data, s->size);
1946 if (s->should_free || s->should_munmap) {
1947 s->should_free = s->should_munmap = 0;
1948 s->data = NULL;
1952 void diff_free_filespec_data(struct diff_filespec *s)
1954 diff_free_filespec_blob(s);
1955 free(s->cnt_data);
1956 s->cnt_data = NULL;
1959 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
1960 void *blob,
1961 unsigned long size,
1962 const unsigned char *sha1,
1963 int mode)
1965 int fd;
1966 struct strbuf buf = STRBUF_INIT;
1968 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1969 if (fd < 0)
1970 die("unable to create temp-file: %s", strerror(errno));
1971 if (convert_to_working_tree(path,
1972 (const char *)blob, (size_t)size, &buf)) {
1973 blob = buf.buf;
1974 size = buf.len;
1976 if (write_in_full(fd, blob, size) != size)
1977 die("unable to write temp-file");
1978 close(fd);
1979 temp->name = temp->tmp_path;
1980 strcpy(temp->hex, sha1_to_hex(sha1));
1981 temp->hex[40] = 0;
1982 sprintf(temp->mode, "%06o", mode);
1983 strbuf_release(&buf);
1986 static struct diff_tempfile *prepare_temp_file(const char *name,
1987 struct diff_filespec *one)
1989 struct diff_tempfile *temp = claim_diff_tempfile();
1991 if (!DIFF_FILE_VALID(one)) {
1992 not_a_valid_file:
1993 /* A '-' entry produces this for file-2, and
1994 * a '+' entry produces this for file-1.
1996 temp->name = "/dev/null";
1997 strcpy(temp->hex, ".");
1998 strcpy(temp->mode, ".");
1999 return temp;
2002 if (!remove_tempfile_installed) {
2003 atexit(remove_tempfile);
2004 sigchain_push_common(remove_tempfile_on_signal);
2005 remove_tempfile_installed = 1;
2008 if (!one->sha1_valid ||
2009 reuse_worktree_file(name, one->sha1, 1)) {
2010 struct stat st;
2011 if (lstat(name, &st) < 0) {
2012 if (errno == ENOENT)
2013 goto not_a_valid_file;
2014 die("stat(%s): %s", name, strerror(errno));
2016 if (S_ISLNK(st.st_mode)) {
2017 int ret;
2018 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
2019 ret = readlink(name, buf, sizeof(buf));
2020 if (ret < 0)
2021 die("readlink(%s)", name);
2022 if (ret == sizeof(buf))
2023 die("symlink too long: %s", name);
2024 prep_temp_blob(name, temp, buf, ret,
2025 (one->sha1_valid ?
2026 one->sha1 : null_sha1),
2027 (one->sha1_valid ?
2028 one->mode : S_IFLNK));
2030 else {
2031 /* we can borrow from the file in the work tree */
2032 temp->name = name;
2033 if (!one->sha1_valid)
2034 strcpy(temp->hex, sha1_to_hex(null_sha1));
2035 else
2036 strcpy(temp->hex, sha1_to_hex(one->sha1));
2037 /* Even though we may sometimes borrow the
2038 * contents from the work tree, we always want
2039 * one->mode. mode is trustworthy even when
2040 * !(one->sha1_valid), as long as
2041 * DIFF_FILE_VALID(one).
2043 sprintf(temp->mode, "%06o", one->mode);
2045 return temp;
2047 else {
2048 if (diff_populate_filespec(one, 0))
2049 die("cannot read data blob for %s", one->path);
2050 prep_temp_blob(name, temp, one->data, one->size,
2051 one->sha1, one->mode);
2053 return temp;
2056 /* An external diff command takes:
2058 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2059 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2062 static void run_external_diff(const char *pgm,
2063 const char *name,
2064 const char *other,
2065 struct diff_filespec *one,
2066 struct diff_filespec *two,
2067 const char *xfrm_msg,
2068 int complete_rewrite)
2070 const char *spawn_arg[10];
2071 int retval;
2072 const char **arg = &spawn_arg[0];
2074 if (one && two) {
2075 struct diff_tempfile *temp_one, *temp_two;
2076 const char *othername = (other ? other : name);
2077 temp_one = prepare_temp_file(name, one);
2078 temp_two = prepare_temp_file(othername, two);
2079 *arg++ = pgm;
2080 *arg++ = name;
2081 *arg++ = temp_one->name;
2082 *arg++ = temp_one->hex;
2083 *arg++ = temp_one->mode;
2084 *arg++ = temp_two->name;
2085 *arg++ = temp_two->hex;
2086 *arg++ = temp_two->mode;
2087 if (other) {
2088 *arg++ = other;
2089 *arg++ = xfrm_msg;
2091 } else {
2092 *arg++ = pgm;
2093 *arg++ = name;
2095 *arg = NULL;
2096 fflush(NULL);
2097 retval = run_command_v_opt(spawn_arg, 0);
2098 remove_tempfile();
2099 if (retval) {
2100 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2101 exit(1);
2105 static int similarity_index(struct diff_filepair *p)
2107 return p->score * 100 / MAX_SCORE;
2110 static void fill_metainfo(struct strbuf *msg,
2111 const char *name,
2112 const char *other,
2113 struct diff_filespec *one,
2114 struct diff_filespec *two,
2115 struct diff_options *o,
2116 struct diff_filepair *p)
2118 strbuf_init(msg, PATH_MAX * 2 + 300);
2119 switch (p->status) {
2120 case DIFF_STATUS_COPIED:
2121 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2122 strbuf_addstr(msg, "\ncopy from ");
2123 quote_c_style(name, msg, NULL, 0);
2124 strbuf_addstr(msg, "\ncopy to ");
2125 quote_c_style(other, msg, NULL, 0);
2126 strbuf_addch(msg, '\n');
2127 break;
2128 case DIFF_STATUS_RENAMED:
2129 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2130 strbuf_addstr(msg, "\nrename from ");
2131 quote_c_style(name, msg, NULL, 0);
2132 strbuf_addstr(msg, "\nrename to ");
2133 quote_c_style(other, msg, NULL, 0);
2134 strbuf_addch(msg, '\n');
2135 break;
2136 case DIFF_STATUS_MODIFIED:
2137 if (p->score) {
2138 strbuf_addf(msg, "dissimilarity index %d%%\n",
2139 similarity_index(p));
2140 break;
2142 /* fallthru */
2143 default:
2144 /* nothing */
2147 if (one && two && hashcmp(one->sha1, two->sha1)) {
2148 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2150 if (DIFF_OPT_TST(o, BINARY)) {
2151 mmfile_t mf;
2152 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2153 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2154 abbrev = 40;
2156 strbuf_addf(msg, "index %.*s..%.*s",
2157 abbrev, sha1_to_hex(one->sha1),
2158 abbrev, sha1_to_hex(two->sha1));
2159 if (one->mode == two->mode)
2160 strbuf_addf(msg, " %06o", one->mode);
2161 strbuf_addch(msg, '\n');
2163 if (msg->len)
2164 strbuf_setlen(msg, msg->len - 1);
2167 static void run_diff_cmd(const char *pgm,
2168 const char *name,
2169 const char *other,
2170 const char *attr_path,
2171 struct diff_filespec *one,
2172 struct diff_filespec *two,
2173 struct strbuf *msg,
2174 struct diff_options *o,
2175 struct diff_filepair *p)
2177 const char *xfrm_msg = NULL;
2178 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2180 if (msg) {
2181 fill_metainfo(msg, name, other, one, two, o, p);
2182 xfrm_msg = msg->len ? msg->buf : NULL;
2185 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2186 pgm = NULL;
2187 else {
2188 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2189 if (drv && drv->external)
2190 pgm = drv->external;
2193 if (pgm) {
2194 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2195 complete_rewrite);
2196 return;
2198 if (one && two)
2199 builtin_diff(name, other ? other : name,
2200 one, two, xfrm_msg, o, complete_rewrite);
2201 else
2202 fprintf(o->file, "* Unmerged path %s\n", name);
2205 static void diff_fill_sha1_info(struct diff_filespec *one)
2207 if (DIFF_FILE_VALID(one)) {
2208 if (!one->sha1_valid) {
2209 struct stat st;
2210 if (!strcmp(one->path, "-")) {
2211 hashcpy(one->sha1, null_sha1);
2212 return;
2214 if (lstat(one->path, &st) < 0)
2215 die("stat %s", one->path);
2216 if (index_path(one->sha1, one->path, &st, 0))
2217 die("cannot hash %s", one->path);
2220 else
2221 hashclr(one->sha1);
2224 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2226 /* Strip the prefix but do not molest /dev/null and absolute paths */
2227 if (*namep && **namep != '/')
2228 *namep += prefix_length;
2229 if (*otherp && **otherp != '/')
2230 *otherp += prefix_length;
2233 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2235 const char *pgm = external_diff();
2236 struct strbuf msg;
2237 struct diff_filespec *one = p->one;
2238 struct diff_filespec *two = p->two;
2239 const char *name;
2240 const char *other;
2241 const char *attr_path;
2243 name = p->one->path;
2244 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2245 attr_path = name;
2246 if (o->prefix_length)
2247 strip_prefix(o->prefix_length, &name, &other);
2249 if (DIFF_PAIR_UNMERGED(p)) {
2250 run_diff_cmd(pgm, name, NULL, attr_path,
2251 NULL, NULL, NULL, o, p);
2252 return;
2255 diff_fill_sha1_info(one);
2256 diff_fill_sha1_info(two);
2258 if (!pgm &&
2259 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2260 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2262 * a filepair that changes between file and symlink
2263 * needs to be split into deletion and creation.
2265 struct diff_filespec *null = alloc_filespec(two->path);
2266 run_diff_cmd(NULL, name, other, attr_path,
2267 one, null, &msg, o, p);
2268 free(null);
2269 strbuf_release(&msg);
2271 null = alloc_filespec(one->path);
2272 run_diff_cmd(NULL, name, other, attr_path,
2273 null, two, &msg, o, p);
2274 free(null);
2276 else
2277 run_diff_cmd(pgm, name, other, attr_path,
2278 one, two, &msg, o, p);
2280 strbuf_release(&msg);
2283 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2284 struct diffstat_t *diffstat)
2286 const char *name;
2287 const char *other;
2288 int complete_rewrite = 0;
2290 if (DIFF_PAIR_UNMERGED(p)) {
2291 /* unmerged */
2292 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2293 return;
2296 name = p->one->path;
2297 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2299 if (o->prefix_length)
2300 strip_prefix(o->prefix_length, &name, &other);
2302 diff_fill_sha1_info(p->one);
2303 diff_fill_sha1_info(p->two);
2305 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2306 complete_rewrite = 1;
2307 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2310 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2312 const char *name;
2313 const char *other;
2314 const char *attr_path;
2316 if (DIFF_PAIR_UNMERGED(p)) {
2317 /* unmerged */
2318 return;
2321 name = p->one->path;
2322 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2323 attr_path = other ? other : name;
2325 if (o->prefix_length)
2326 strip_prefix(o->prefix_length, &name, &other);
2328 diff_fill_sha1_info(p->one);
2329 diff_fill_sha1_info(p->two);
2331 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2334 void diff_setup(struct diff_options *options)
2336 memset(options, 0, sizeof(*options));
2338 options->file = stdout;
2340 options->line_termination = '\n';
2341 options->break_opt = -1;
2342 options->rename_limit = -1;
2343 options->dirstat_percent = 3;
2344 options->context = 3;
2346 options->change = diff_change;
2347 options->add_remove = diff_addremove;
2348 if (diff_use_color_default > 0)
2349 DIFF_OPT_SET(options, COLOR_DIFF);
2350 options->detect_rename = diff_detect_rename_default;
2352 if (!diff_mnemonic_prefix) {
2353 options->a_prefix = "a/";
2354 options->b_prefix = "b/";
2358 int diff_setup_done(struct diff_options *options)
2360 int count = 0;
2362 if (options->output_format & DIFF_FORMAT_NAME)
2363 count++;
2364 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2365 count++;
2366 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2367 count++;
2368 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2369 count++;
2370 if (count > 1)
2371 die("--name-only, --name-status, --check and -s are mutually exclusive");
2373 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2374 options->detect_rename = DIFF_DETECT_COPY;
2376 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2377 options->prefix = NULL;
2378 if (options->prefix)
2379 options->prefix_length = strlen(options->prefix);
2380 else
2381 options->prefix_length = 0;
2383 if (options->output_format & (DIFF_FORMAT_NAME |
2384 DIFF_FORMAT_NAME_STATUS |
2385 DIFF_FORMAT_CHECKDIFF |
2386 DIFF_FORMAT_NO_OUTPUT))
2387 options->output_format &= ~(DIFF_FORMAT_RAW |
2388 DIFF_FORMAT_NUMSTAT |
2389 DIFF_FORMAT_DIFFSTAT |
2390 DIFF_FORMAT_SHORTSTAT |
2391 DIFF_FORMAT_DIRSTAT |
2392 DIFF_FORMAT_SUMMARY |
2393 DIFF_FORMAT_PATCH);
2396 * These cases always need recursive; we do not drop caller-supplied
2397 * recursive bits for other formats here.
2399 if (options->output_format & (DIFF_FORMAT_PATCH |
2400 DIFF_FORMAT_NUMSTAT |
2401 DIFF_FORMAT_DIFFSTAT |
2402 DIFF_FORMAT_SHORTSTAT |
2403 DIFF_FORMAT_DIRSTAT |
2404 DIFF_FORMAT_SUMMARY |
2405 DIFF_FORMAT_CHECKDIFF))
2406 DIFF_OPT_SET(options, RECURSIVE);
2408 * Also pickaxe would not work very well if you do not say recursive
2410 if (options->pickaxe)
2411 DIFF_OPT_SET(options, RECURSIVE);
2413 if (options->detect_rename && options->rename_limit < 0)
2414 options->rename_limit = diff_rename_limit_default;
2415 if (options->setup & DIFF_SETUP_USE_CACHE) {
2416 if (!active_cache)
2417 /* read-cache does not die even when it fails
2418 * so it is safe for us to do this here. Also
2419 * it does not smudge active_cache or active_nr
2420 * when it fails, so we do not have to worry about
2421 * cleaning it up ourselves either.
2423 read_cache();
2425 if (options->abbrev <= 0 || 40 < options->abbrev)
2426 options->abbrev = 40; /* full */
2429 * It does not make sense to show the first hit we happened
2430 * to have found. It does not make sense not to return with
2431 * exit code in such a case either.
2433 if (DIFF_OPT_TST(options, QUIET)) {
2434 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2435 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2438 return 0;
2441 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2443 char c, *eq;
2444 int len;
2446 if (*arg != '-')
2447 return 0;
2448 c = *++arg;
2449 if (!c)
2450 return 0;
2451 if (c == arg_short) {
2452 c = *++arg;
2453 if (!c)
2454 return 1;
2455 if (val && isdigit(c)) {
2456 char *end;
2457 int n = strtoul(arg, &end, 10);
2458 if (*end)
2459 return 0;
2460 *val = n;
2461 return 1;
2463 return 0;
2465 if (c != '-')
2466 return 0;
2467 arg++;
2468 eq = strchr(arg, '=');
2469 if (eq)
2470 len = eq - arg;
2471 else
2472 len = strlen(arg);
2473 if (!len || strncmp(arg, arg_long, len))
2474 return 0;
2475 if (eq) {
2476 int n;
2477 char *end;
2478 if (!isdigit(*++eq))
2479 return 0;
2480 n = strtoul(eq, &end, 10);
2481 if (*end)
2482 return 0;
2483 *val = n;
2485 return 1;
2488 static int diff_scoreopt_parse(const char *opt);
2490 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2492 const char *arg = av[0];
2494 /* Output format options */
2495 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2496 options->output_format |= DIFF_FORMAT_PATCH;
2497 else if (opt_arg(arg, 'U', "unified", &options->context))
2498 options->output_format |= DIFF_FORMAT_PATCH;
2499 else if (!strcmp(arg, "--raw"))
2500 options->output_format |= DIFF_FORMAT_RAW;
2501 else if (!strcmp(arg, "--patch-with-raw"))
2502 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2503 else if (!strcmp(arg, "--numstat"))
2504 options->output_format |= DIFF_FORMAT_NUMSTAT;
2505 else if (!strcmp(arg, "--shortstat"))
2506 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2507 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2508 options->output_format |= DIFF_FORMAT_DIRSTAT;
2509 else if (!strcmp(arg, "--cumulative")) {
2510 options->output_format |= DIFF_FORMAT_DIRSTAT;
2511 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2512 } else if (opt_arg(arg, 0, "dirstat-by-file",
2513 &options->dirstat_percent)) {
2514 options->output_format |= DIFF_FORMAT_DIRSTAT;
2515 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2517 else if (!strcmp(arg, "--check"))
2518 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2519 else if (!strcmp(arg, "--summary"))
2520 options->output_format |= DIFF_FORMAT_SUMMARY;
2521 else if (!strcmp(arg, "--patch-with-stat"))
2522 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2523 else if (!strcmp(arg, "--name-only"))
2524 options->output_format |= DIFF_FORMAT_NAME;
2525 else if (!strcmp(arg, "--name-status"))
2526 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2527 else if (!strcmp(arg, "-s"))
2528 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2529 else if (!prefixcmp(arg, "--stat")) {
2530 char *end;
2531 int width = options->stat_width;
2532 int name_width = options->stat_name_width;
2533 arg += 6;
2534 end = (char *)arg;
2536 switch (*arg) {
2537 case '-':
2538 if (!prefixcmp(arg, "-width="))
2539 width = strtoul(arg + 7, &end, 10);
2540 else if (!prefixcmp(arg, "-name-width="))
2541 name_width = strtoul(arg + 12, &end, 10);
2542 break;
2543 case '=':
2544 width = strtoul(arg+1, &end, 10);
2545 if (*end == ',')
2546 name_width = strtoul(end+1, &end, 10);
2549 /* Important! This checks all the error cases! */
2550 if (*end)
2551 return 0;
2552 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2553 options->stat_name_width = name_width;
2554 options->stat_width = width;
2557 /* renames options */
2558 else if (!prefixcmp(arg, "-B")) {
2559 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2560 return -1;
2562 else if (!prefixcmp(arg, "-M")) {
2563 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2564 return -1;
2565 options->detect_rename = DIFF_DETECT_RENAME;
2567 else if (!prefixcmp(arg, "-C")) {
2568 if (options->detect_rename == DIFF_DETECT_COPY)
2569 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2570 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2571 return -1;
2572 options->detect_rename = DIFF_DETECT_COPY;
2574 else if (!strcmp(arg, "--no-renames"))
2575 options->detect_rename = 0;
2576 else if (!strcmp(arg, "--relative"))
2577 DIFF_OPT_SET(options, RELATIVE_NAME);
2578 else if (!prefixcmp(arg, "--relative=")) {
2579 DIFF_OPT_SET(options, RELATIVE_NAME);
2580 options->prefix = arg + 11;
2583 /* xdiff options */
2584 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2585 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2586 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2587 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2588 else if (!strcmp(arg, "--ignore-space-at-eol"))
2589 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2590 else if (!strcmp(arg, "--patience"))
2591 DIFF_XDL_SET(options, PATIENCE_DIFF);
2593 /* flags options */
2594 else if (!strcmp(arg, "--binary")) {
2595 options->output_format |= DIFF_FORMAT_PATCH;
2596 DIFF_OPT_SET(options, BINARY);
2598 else if (!strcmp(arg, "--full-index"))
2599 DIFF_OPT_SET(options, FULL_INDEX);
2600 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2601 DIFF_OPT_SET(options, TEXT);
2602 else if (!strcmp(arg, "-R"))
2603 DIFF_OPT_SET(options, REVERSE_DIFF);
2604 else if (!strcmp(arg, "--find-copies-harder"))
2605 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2606 else if (!strcmp(arg, "--follow"))
2607 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2608 else if (!strcmp(arg, "--color"))
2609 DIFF_OPT_SET(options, COLOR_DIFF);
2610 else if (!strcmp(arg, "--no-color"))
2611 DIFF_OPT_CLR(options, COLOR_DIFF);
2612 else if (!strcmp(arg, "--color-words")) {
2613 DIFF_OPT_SET(options, COLOR_DIFF);
2614 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2616 else if (!prefixcmp(arg, "--color-words=")) {
2617 DIFF_OPT_SET(options, COLOR_DIFF);
2618 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2619 options->word_regex = arg + 14;
2621 else if (!strcmp(arg, "--exit-code"))
2622 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2623 else if (!strcmp(arg, "--quiet"))
2624 DIFF_OPT_SET(options, QUIET);
2625 else if (!strcmp(arg, "--ext-diff"))
2626 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2627 else if (!strcmp(arg, "--no-ext-diff"))
2628 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2629 else if (!strcmp(arg, "--textconv"))
2630 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2631 else if (!strcmp(arg, "--no-textconv"))
2632 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2633 else if (!strcmp(arg, "--ignore-submodules"))
2634 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2636 /* misc options */
2637 else if (!strcmp(arg, "-z"))
2638 options->line_termination = 0;
2639 else if (!prefixcmp(arg, "-l"))
2640 options->rename_limit = strtoul(arg+2, NULL, 10);
2641 else if (!prefixcmp(arg, "-S"))
2642 options->pickaxe = arg + 2;
2643 else if (!strcmp(arg, "--pickaxe-all"))
2644 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2645 else if (!strcmp(arg, "--pickaxe-regex"))
2646 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2647 else if (!prefixcmp(arg, "-O"))
2648 options->orderfile = arg + 2;
2649 else if (!prefixcmp(arg, "--diff-filter="))
2650 options->filter = arg + 14;
2651 else if (!strcmp(arg, "--abbrev"))
2652 options->abbrev = DEFAULT_ABBREV;
2653 else if (!prefixcmp(arg, "--abbrev=")) {
2654 options->abbrev = strtoul(arg + 9, NULL, 10);
2655 if (options->abbrev < MINIMUM_ABBREV)
2656 options->abbrev = MINIMUM_ABBREV;
2657 else if (40 < options->abbrev)
2658 options->abbrev = 40;
2660 else if (!prefixcmp(arg, "--src-prefix="))
2661 options->a_prefix = arg + 13;
2662 else if (!prefixcmp(arg, "--dst-prefix="))
2663 options->b_prefix = arg + 13;
2664 else if (!strcmp(arg, "--no-prefix"))
2665 options->a_prefix = options->b_prefix = "";
2666 else if (opt_arg(arg, '\0', "inter-hunk-context",
2667 &options->interhunkcontext))
2669 else if (!prefixcmp(arg, "--output=")) {
2670 options->file = fopen(arg + strlen("--output="), "w");
2671 options->close_file = 1;
2672 } else
2673 return 0;
2674 return 1;
2677 static int parse_num(const char **cp_p)
2679 unsigned long num, scale;
2680 int ch, dot;
2681 const char *cp = *cp_p;
2683 num = 0;
2684 scale = 1;
2685 dot = 0;
2686 for(;;) {
2687 ch = *cp;
2688 if ( !dot && ch == '.' ) {
2689 scale = 1;
2690 dot = 1;
2691 } else if ( ch == '%' ) {
2692 scale = dot ? scale*100 : 100;
2693 cp++; /* % is always at the end */
2694 break;
2695 } else if ( ch >= '0' && ch <= '9' ) {
2696 if ( scale < 100000 ) {
2697 scale *= 10;
2698 num = (num*10) + (ch-'0');
2700 } else {
2701 break;
2703 cp++;
2705 *cp_p = cp;
2707 /* user says num divided by scale and we say internally that
2708 * is MAX_SCORE * num / scale.
2710 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2713 static int diff_scoreopt_parse(const char *opt)
2715 int opt1, opt2, cmd;
2717 if (*opt++ != '-')
2718 return -1;
2719 cmd = *opt++;
2720 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2721 return -1; /* that is not a -M, -C nor -B option */
2723 opt1 = parse_num(&opt);
2724 if (cmd != 'B')
2725 opt2 = 0;
2726 else {
2727 if (*opt == 0)
2728 opt2 = 0;
2729 else if (*opt != '/')
2730 return -1; /* we expect -B80/99 or -B80 */
2731 else {
2732 opt++;
2733 opt2 = parse_num(&opt);
2736 if (*opt != 0)
2737 return -1;
2738 return opt1 | (opt2 << 16);
2741 struct diff_queue_struct diff_queued_diff;
2743 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2745 if (queue->alloc <= queue->nr) {
2746 queue->alloc = alloc_nr(queue->alloc);
2747 queue->queue = xrealloc(queue->queue,
2748 sizeof(dp) * queue->alloc);
2750 queue->queue[queue->nr++] = dp;
2753 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2754 struct diff_filespec *one,
2755 struct diff_filespec *two)
2757 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2758 dp->one = one;
2759 dp->two = two;
2760 if (queue)
2761 diff_q(queue, dp);
2762 return dp;
2765 void diff_free_filepair(struct diff_filepair *p)
2767 free_filespec(p->one);
2768 free_filespec(p->two);
2769 free(p);
2772 /* This is different from find_unique_abbrev() in that
2773 * it stuffs the result with dots for alignment.
2775 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2777 int abblen;
2778 const char *abbrev;
2779 if (len == 40)
2780 return sha1_to_hex(sha1);
2782 abbrev = find_unique_abbrev(sha1, len);
2783 abblen = strlen(abbrev);
2784 if (abblen < 37) {
2785 static char hex[41];
2786 if (len < abblen && abblen <= len + 2)
2787 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2788 else
2789 sprintf(hex, "%s...", abbrev);
2790 return hex;
2792 return sha1_to_hex(sha1);
2795 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2797 int line_termination = opt->line_termination;
2798 int inter_name_termination = line_termination ? '\t' : '\0';
2800 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2801 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2802 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2803 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2805 if (p->score) {
2806 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2807 inter_name_termination);
2808 } else {
2809 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2812 if (p->status == DIFF_STATUS_COPIED ||
2813 p->status == DIFF_STATUS_RENAMED) {
2814 const char *name_a, *name_b;
2815 name_a = p->one->path;
2816 name_b = p->two->path;
2817 strip_prefix(opt->prefix_length, &name_a, &name_b);
2818 write_name_quoted(name_a, opt->file, inter_name_termination);
2819 write_name_quoted(name_b, opt->file, line_termination);
2820 } else {
2821 const char *name_a, *name_b;
2822 name_a = p->one->mode ? p->one->path : p->two->path;
2823 name_b = NULL;
2824 strip_prefix(opt->prefix_length, &name_a, &name_b);
2825 write_name_quoted(name_a, opt->file, line_termination);
2829 int diff_unmodified_pair(struct diff_filepair *p)
2831 /* This function is written stricter than necessary to support
2832 * the currently implemented transformers, but the idea is to
2833 * let transformers to produce diff_filepairs any way they want,
2834 * and filter and clean them up here before producing the output.
2836 struct diff_filespec *one = p->one, *two = p->two;
2838 if (DIFF_PAIR_UNMERGED(p))
2839 return 0; /* unmerged is interesting */
2841 /* deletion, addition, mode or type change
2842 * and rename are all interesting.
2844 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2845 DIFF_PAIR_MODE_CHANGED(p) ||
2846 strcmp(one->path, two->path))
2847 return 0;
2849 /* both are valid and point at the same path. that is, we are
2850 * dealing with a change.
2852 if (one->sha1_valid && two->sha1_valid &&
2853 !hashcmp(one->sha1, two->sha1))
2854 return 1; /* no change */
2855 if (!one->sha1_valid && !two->sha1_valid)
2856 return 1; /* both look at the same file on the filesystem. */
2857 return 0;
2860 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2862 if (diff_unmodified_pair(p))
2863 return;
2865 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2866 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2867 return; /* no tree diffs in patch format */
2869 run_diff(p, o);
2872 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2873 struct diffstat_t *diffstat)
2875 if (diff_unmodified_pair(p))
2876 return;
2878 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2879 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2880 return; /* no tree diffs in patch format */
2882 run_diffstat(p, o, diffstat);
2885 static void diff_flush_checkdiff(struct diff_filepair *p,
2886 struct diff_options *o)
2888 if (diff_unmodified_pair(p))
2889 return;
2891 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2892 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2893 return; /* no tree diffs in patch format */
2895 run_checkdiff(p, o);
2898 int diff_queue_is_empty(void)
2900 struct diff_queue_struct *q = &diff_queued_diff;
2901 int i;
2902 for (i = 0; i < q->nr; i++)
2903 if (!diff_unmodified_pair(q->queue[i]))
2904 return 0;
2905 return 1;
2908 #if DIFF_DEBUG
2909 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2911 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2912 x, one ? one : "",
2913 s->path,
2914 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2915 s->mode,
2916 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2917 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2918 x, one ? one : "",
2919 s->size, s->xfrm_flags);
2922 void diff_debug_filepair(const struct diff_filepair *p, int i)
2924 diff_debug_filespec(p->one, i, "one");
2925 diff_debug_filespec(p->two, i, "two");
2926 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2927 p->score, p->status ? p->status : '?',
2928 p->one->rename_used, p->broken_pair);
2931 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2933 int i;
2934 if (msg)
2935 fprintf(stderr, "%s\n", msg);
2936 fprintf(stderr, "q->nr = %d\n", q->nr);
2937 for (i = 0; i < q->nr; i++) {
2938 struct diff_filepair *p = q->queue[i];
2939 diff_debug_filepair(p, i);
2942 #endif
2944 static void diff_resolve_rename_copy(void)
2946 int i;
2947 struct diff_filepair *p;
2948 struct diff_queue_struct *q = &diff_queued_diff;
2950 diff_debug_queue("resolve-rename-copy", q);
2952 for (i = 0; i < q->nr; i++) {
2953 p = q->queue[i];
2954 p->status = 0; /* undecided */
2955 if (DIFF_PAIR_UNMERGED(p))
2956 p->status = DIFF_STATUS_UNMERGED;
2957 else if (!DIFF_FILE_VALID(p->one))
2958 p->status = DIFF_STATUS_ADDED;
2959 else if (!DIFF_FILE_VALID(p->two))
2960 p->status = DIFF_STATUS_DELETED;
2961 else if (DIFF_PAIR_TYPE_CHANGED(p))
2962 p->status = DIFF_STATUS_TYPE_CHANGED;
2964 /* from this point on, we are dealing with a pair
2965 * whose both sides are valid and of the same type, i.e.
2966 * either in-place edit or rename/copy edit.
2968 else if (DIFF_PAIR_RENAME(p)) {
2970 * A rename might have re-connected a broken
2971 * pair up, causing the pathnames to be the
2972 * same again. If so, that's not a rename at
2973 * all, just a modification..
2975 * Otherwise, see if this source was used for
2976 * multiple renames, in which case we decrement
2977 * the count, and call it a copy.
2979 if (!strcmp(p->one->path, p->two->path))
2980 p->status = DIFF_STATUS_MODIFIED;
2981 else if (--p->one->rename_used > 0)
2982 p->status = DIFF_STATUS_COPIED;
2983 else
2984 p->status = DIFF_STATUS_RENAMED;
2986 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2987 p->one->mode != p->two->mode ||
2988 is_null_sha1(p->one->sha1))
2989 p->status = DIFF_STATUS_MODIFIED;
2990 else {
2991 /* This is a "no-change" entry and should not
2992 * happen anymore, but prepare for broken callers.
2994 error("feeding unmodified %s to diffcore",
2995 p->one->path);
2996 p->status = DIFF_STATUS_UNKNOWN;
2999 diff_debug_queue("resolve-rename-copy done", q);
3002 static int check_pair_status(struct diff_filepair *p)
3004 switch (p->status) {
3005 case DIFF_STATUS_UNKNOWN:
3006 return 0;
3007 case 0:
3008 die("internal error in diff-resolve-rename-copy");
3009 default:
3010 return 1;
3014 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3016 int fmt = opt->output_format;
3018 if (fmt & DIFF_FORMAT_CHECKDIFF)
3019 diff_flush_checkdiff(p, opt);
3020 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3021 diff_flush_raw(p, opt);
3022 else if (fmt & DIFF_FORMAT_NAME) {
3023 const char *name_a, *name_b;
3024 name_a = p->two->path;
3025 name_b = NULL;
3026 strip_prefix(opt->prefix_length, &name_a, &name_b);
3027 write_name_quoted(name_a, opt->file, opt->line_termination);
3031 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3033 if (fs->mode)
3034 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3035 else
3036 fprintf(file, " %s ", newdelete);
3037 write_name_quoted(fs->path, file, '\n');
3041 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3043 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3044 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3045 show_name ? ' ' : '\n');
3046 if (show_name) {
3047 write_name_quoted(p->two->path, file, '\n');
3052 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3054 char *names = pprint_rename(p->one->path, p->two->path);
3056 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3057 free(names);
3058 show_mode_change(file, p, 0);
3061 static void diff_summary(FILE *file, struct diff_filepair *p)
3063 switch(p->status) {
3064 case DIFF_STATUS_DELETED:
3065 show_file_mode_name(file, "delete", p->one);
3066 break;
3067 case DIFF_STATUS_ADDED:
3068 show_file_mode_name(file, "create", p->two);
3069 break;
3070 case DIFF_STATUS_COPIED:
3071 show_rename_copy(file, "copy", p);
3072 break;
3073 case DIFF_STATUS_RENAMED:
3074 show_rename_copy(file, "rename", p);
3075 break;
3076 default:
3077 if (p->score) {
3078 fputs(" rewrite ", file);
3079 write_name_quoted(p->two->path, file, ' ');
3080 fprintf(file, "(%d%%)\n", similarity_index(p));
3082 show_mode_change(file, p, !p->score);
3083 break;
3087 struct patch_id_t {
3088 git_SHA_CTX *ctx;
3089 int patchlen;
3092 static int remove_space(char *line, int len)
3094 int i;
3095 char *dst = line;
3096 unsigned char c;
3098 for (i = 0; i < len; i++)
3099 if (!isspace((c = line[i])))
3100 *dst++ = c;
3102 return dst - line;
3105 static void patch_id_consume(void *priv, char *line, unsigned long len)
3107 struct patch_id_t *data = priv;
3108 int new_len;
3110 /* Ignore line numbers when computing the SHA1 of the patch */
3111 if (!prefixcmp(line, "@@ -"))
3112 return;
3114 new_len = remove_space(line, len);
3116 git_SHA1_Update(data->ctx, line, new_len);
3117 data->patchlen += new_len;
3120 /* returns 0 upon success, and writes result into sha1 */
3121 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3123 struct diff_queue_struct *q = &diff_queued_diff;
3124 int i;
3125 git_SHA_CTX ctx;
3126 struct patch_id_t data;
3127 char buffer[PATH_MAX * 4 + 20];
3129 git_SHA1_Init(&ctx);
3130 memset(&data, 0, sizeof(struct patch_id_t));
3131 data.ctx = &ctx;
3133 for (i = 0; i < q->nr; i++) {
3134 xpparam_t xpp;
3135 xdemitconf_t xecfg;
3136 xdemitcb_t ecb;
3137 mmfile_t mf1, mf2;
3138 struct diff_filepair *p = q->queue[i];
3139 int len1, len2;
3141 memset(&xpp, 0, sizeof(xpp));
3142 memset(&xecfg, 0, sizeof(xecfg));
3143 if (p->status == 0)
3144 return error("internal diff status error");
3145 if (p->status == DIFF_STATUS_UNKNOWN)
3146 continue;
3147 if (diff_unmodified_pair(p))
3148 continue;
3149 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3150 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3151 continue;
3152 if (DIFF_PAIR_UNMERGED(p))
3153 continue;
3155 diff_fill_sha1_info(p->one);
3156 diff_fill_sha1_info(p->two);
3157 if (fill_mmfile(&mf1, p->one) < 0 ||
3158 fill_mmfile(&mf2, p->two) < 0)
3159 return error("unable to read files to diff");
3161 len1 = remove_space(p->one->path, strlen(p->one->path));
3162 len2 = remove_space(p->two->path, strlen(p->two->path));
3163 if (p->one->mode == 0)
3164 len1 = snprintf(buffer, sizeof(buffer),
3165 "diff--gita/%.*sb/%.*s"
3166 "newfilemode%06o"
3167 "---/dev/null"
3168 "+++b/%.*s",
3169 len1, p->one->path,
3170 len2, p->two->path,
3171 p->two->mode,
3172 len2, p->two->path);
3173 else if (p->two->mode == 0)
3174 len1 = snprintf(buffer, sizeof(buffer),
3175 "diff--gita/%.*sb/%.*s"
3176 "deletedfilemode%06o"
3177 "---a/%.*s"
3178 "+++/dev/null",
3179 len1, p->one->path,
3180 len2, p->two->path,
3181 p->one->mode,
3182 len1, p->one->path);
3183 else
3184 len1 = snprintf(buffer, sizeof(buffer),
3185 "diff--gita/%.*sb/%.*s"
3186 "---a/%.*s"
3187 "+++b/%.*s",
3188 len1, p->one->path,
3189 len2, p->two->path,
3190 len1, p->one->path,
3191 len2, p->two->path);
3192 git_SHA1_Update(&ctx, buffer, len1);
3194 xpp.flags = XDF_NEED_MINIMAL;
3195 xecfg.ctxlen = 3;
3196 xecfg.flags = XDL_EMIT_FUNCNAMES;
3197 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3198 &xpp, &xecfg, &ecb);
3201 git_SHA1_Final(sha1, &ctx);
3202 return 0;
3205 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3207 struct diff_queue_struct *q = &diff_queued_diff;
3208 int i;
3209 int result = diff_get_patch_id(options, sha1);
3211 for (i = 0; i < q->nr; i++)
3212 diff_free_filepair(q->queue[i]);
3214 free(q->queue);
3215 q->queue = NULL;
3216 q->nr = q->alloc = 0;
3218 return result;
3221 static int is_summary_empty(const struct diff_queue_struct *q)
3223 int i;
3225 for (i = 0; i < q->nr; i++) {
3226 const struct diff_filepair *p = q->queue[i];
3228 switch (p->status) {
3229 case DIFF_STATUS_DELETED:
3230 case DIFF_STATUS_ADDED:
3231 case DIFF_STATUS_COPIED:
3232 case DIFF_STATUS_RENAMED:
3233 return 0;
3234 default:
3235 if (p->score)
3236 return 0;
3237 if (p->one->mode && p->two->mode &&
3238 p->one->mode != p->two->mode)
3239 return 0;
3240 break;
3243 return 1;
3246 void diff_flush(struct diff_options *options)
3248 struct diff_queue_struct *q = &diff_queued_diff;
3249 int i, output_format = options->output_format;
3250 int separator = 0;
3253 * Order: raw, stat, summary, patch
3254 * or: name/name-status/checkdiff (other bits clear)
3256 if (!q->nr)
3257 goto free_queue;
3259 if (output_format & (DIFF_FORMAT_RAW |
3260 DIFF_FORMAT_NAME |
3261 DIFF_FORMAT_NAME_STATUS |
3262 DIFF_FORMAT_CHECKDIFF)) {
3263 for (i = 0; i < q->nr; i++) {
3264 struct diff_filepair *p = q->queue[i];
3265 if (check_pair_status(p))
3266 flush_one_pair(p, options);
3268 separator++;
3271 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3272 struct diffstat_t diffstat;
3274 memset(&diffstat, 0, sizeof(struct diffstat_t));
3275 for (i = 0; i < q->nr; i++) {
3276 struct diff_filepair *p = q->queue[i];
3277 if (check_pair_status(p))
3278 diff_flush_stat(p, options, &diffstat);
3280 if (output_format & DIFF_FORMAT_NUMSTAT)
3281 show_numstat(&diffstat, options);
3282 if (output_format & DIFF_FORMAT_DIFFSTAT)
3283 show_stats(&diffstat, options);
3284 if (output_format & DIFF_FORMAT_SHORTSTAT)
3285 show_shortstats(&diffstat, options);
3286 free_diffstat_info(&diffstat);
3287 separator++;
3289 if (output_format & DIFF_FORMAT_DIRSTAT)
3290 show_dirstat(options);
3292 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3293 for (i = 0; i < q->nr; i++)
3294 diff_summary(options->file, q->queue[i]);
3295 separator++;
3298 if (output_format & DIFF_FORMAT_PATCH) {
3299 if (separator) {
3300 putc(options->line_termination, options->file);
3301 if (options->stat_sep) {
3302 /* attach patch instead of inline */
3303 fputs(options->stat_sep, options->file);
3307 for (i = 0; i < q->nr; i++) {
3308 struct diff_filepair *p = q->queue[i];
3309 if (check_pair_status(p))
3310 diff_flush_patch(p, options);
3314 if (output_format & DIFF_FORMAT_CALLBACK)
3315 options->format_callback(q, options, options->format_callback_data);
3317 for (i = 0; i < q->nr; i++)
3318 diff_free_filepair(q->queue[i]);
3319 free_queue:
3320 free(q->queue);
3321 q->queue = NULL;
3322 q->nr = q->alloc = 0;
3323 if (options->close_file)
3324 fclose(options->file);
3327 static void diffcore_apply_filter(const char *filter)
3329 int i;
3330 struct diff_queue_struct *q = &diff_queued_diff;
3331 struct diff_queue_struct outq;
3332 outq.queue = NULL;
3333 outq.nr = outq.alloc = 0;
3335 if (!filter)
3336 return;
3338 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3339 int found;
3340 for (i = found = 0; !found && i < q->nr; i++) {
3341 struct diff_filepair *p = q->queue[i];
3342 if (((p->status == DIFF_STATUS_MODIFIED) &&
3343 ((p->score &&
3344 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3345 (!p->score &&
3346 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3347 ((p->status != DIFF_STATUS_MODIFIED) &&
3348 strchr(filter, p->status)))
3349 found++;
3351 if (found)
3352 return;
3354 /* otherwise we will clear the whole queue
3355 * by copying the empty outq at the end of this
3356 * function, but first clear the current entries
3357 * in the queue.
3359 for (i = 0; i < q->nr; i++)
3360 diff_free_filepair(q->queue[i]);
3362 else {
3363 /* Only the matching ones */
3364 for (i = 0; i < q->nr; i++) {
3365 struct diff_filepair *p = q->queue[i];
3367 if (((p->status == DIFF_STATUS_MODIFIED) &&
3368 ((p->score &&
3369 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3370 (!p->score &&
3371 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3372 ((p->status != DIFF_STATUS_MODIFIED) &&
3373 strchr(filter, p->status)))
3374 diff_q(&outq, p);
3375 else
3376 diff_free_filepair(p);
3379 free(q->queue);
3380 *q = outq;
3383 /* Check whether two filespecs with the same mode and size are identical */
3384 static int diff_filespec_is_identical(struct diff_filespec *one,
3385 struct diff_filespec *two)
3387 if (S_ISGITLINK(one->mode))
3388 return 0;
3389 if (diff_populate_filespec(one, 0))
3390 return 0;
3391 if (diff_populate_filespec(two, 0))
3392 return 0;
3393 return !memcmp(one->data, two->data, one->size);
3396 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3398 int i;
3399 struct diff_queue_struct *q = &diff_queued_diff;
3400 struct diff_queue_struct outq;
3401 outq.queue = NULL;
3402 outq.nr = outq.alloc = 0;
3404 for (i = 0; i < q->nr; i++) {
3405 struct diff_filepair *p = q->queue[i];
3408 * 1. Entries that come from stat info dirtyness
3409 * always have both sides (iow, not create/delete),
3410 * one side of the object name is unknown, with
3411 * the same mode and size. Keep the ones that
3412 * do not match these criteria. They have real
3413 * differences.
3415 * 2. At this point, the file is known to be modified,
3416 * with the same mode and size, and the object
3417 * name of one side is unknown. Need to inspect
3418 * the identical contents.
3420 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3421 !DIFF_FILE_VALID(p->two) ||
3422 (p->one->sha1_valid && p->two->sha1_valid) ||
3423 (p->one->mode != p->two->mode) ||
3424 diff_populate_filespec(p->one, 1) ||
3425 diff_populate_filespec(p->two, 1) ||
3426 (p->one->size != p->two->size) ||
3427 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3428 diff_q(&outq, p);
3429 else {
3431 * The caller can subtract 1 from skip_stat_unmatch
3432 * to determine how many paths were dirty only
3433 * due to stat info mismatch.
3435 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3436 diffopt->skip_stat_unmatch++;
3437 diff_free_filepair(p);
3440 free(q->queue);
3441 *q = outq;
3444 void diffcore_std(struct diff_options *options)
3446 if (options->skip_stat_unmatch)
3447 diffcore_skip_stat_unmatch(options);
3448 if (options->break_opt != -1)
3449 diffcore_break(options->break_opt);
3450 if (options->detect_rename)
3451 diffcore_rename(options);
3452 if (options->break_opt != -1)
3453 diffcore_merge_broken();
3454 if (options->pickaxe)
3455 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3456 if (options->orderfile)
3457 diffcore_order(options->orderfile);
3458 diff_resolve_rename_copy();
3459 diffcore_apply_filter(options->filter);
3461 if (diff_queued_diff.nr)
3462 DIFF_OPT_SET(options, HAS_CHANGES);
3463 else
3464 DIFF_OPT_CLR(options, HAS_CHANGES);
3467 int diff_result_code(struct diff_options *opt, int status)
3469 int result = 0;
3470 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3471 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3472 return status;
3473 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3474 DIFF_OPT_TST(opt, HAS_CHANGES))
3475 result |= 01;
3476 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3477 DIFF_OPT_TST(opt, CHECK_FAILED))
3478 result |= 02;
3479 return result;
3482 void diff_addremove(struct diff_options *options,
3483 int addremove, unsigned mode,
3484 const unsigned char *sha1,
3485 const char *concatpath)
3487 struct diff_filespec *one, *two;
3489 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3490 return;
3492 /* This may look odd, but it is a preparation for
3493 * feeding "there are unchanged files which should
3494 * not produce diffs, but when you are doing copy
3495 * detection you would need them, so here they are"
3496 * entries to the diff-core. They will be prefixed
3497 * with something like '=' or '*' (I haven't decided
3498 * which but should not make any difference).
3499 * Feeding the same new and old to diff_change()
3500 * also has the same effect.
3501 * Before the final output happens, they are pruned after
3502 * merged into rename/copy pairs as appropriate.
3504 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3505 addremove = (addremove == '+' ? '-' :
3506 addremove == '-' ? '+' : addremove);
3508 if (options->prefix &&
3509 strncmp(concatpath, options->prefix, options->prefix_length))
3510 return;
3512 one = alloc_filespec(concatpath);
3513 two = alloc_filespec(concatpath);
3515 if (addremove != '+')
3516 fill_filespec(one, sha1, mode);
3517 if (addremove != '-')
3518 fill_filespec(two, sha1, mode);
3520 diff_queue(&diff_queued_diff, one, two);
3521 DIFF_OPT_SET(options, HAS_CHANGES);
3524 void diff_change(struct diff_options *options,
3525 unsigned old_mode, unsigned new_mode,
3526 const unsigned char *old_sha1,
3527 const unsigned char *new_sha1,
3528 const char *concatpath)
3530 struct diff_filespec *one, *two;
3532 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3533 && S_ISGITLINK(new_mode))
3534 return;
3536 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3537 unsigned tmp;
3538 const unsigned char *tmp_c;
3539 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3540 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3543 if (options->prefix &&
3544 strncmp(concatpath, options->prefix, options->prefix_length))
3545 return;
3547 one = alloc_filespec(concatpath);
3548 two = alloc_filespec(concatpath);
3549 fill_filespec(one, old_sha1, old_mode);
3550 fill_filespec(two, new_sha1, new_mode);
3552 diff_queue(&diff_queued_diff, one, two);
3553 DIFF_OPT_SET(options, HAS_CHANGES);
3556 void diff_unmerge(struct diff_options *options,
3557 const char *path,
3558 unsigned mode, const unsigned char *sha1)
3560 struct diff_filespec *one, *two;
3562 if (options->prefix &&
3563 strncmp(path, options->prefix, options->prefix_length))
3564 return;
3566 one = alloc_filespec(path);
3567 two = alloc_filespec(path);
3568 fill_filespec(one, sha1, mode);
3569 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3572 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3573 size_t *outsize)
3575 struct diff_tempfile *temp;
3576 const char *argv[3];
3577 const char **arg = argv;
3578 struct child_process child;
3579 struct strbuf buf = STRBUF_INIT;
3581 temp = prepare_temp_file(spec->path, spec);
3582 *arg++ = pgm;
3583 *arg++ = temp->name;
3584 *arg = NULL;
3586 memset(&child, 0, sizeof(child));
3587 child.argv = argv;
3588 child.out = -1;
3589 if (start_command(&child) != 0 ||
3590 strbuf_read(&buf, child.out, 0) < 0 ||
3591 finish_command(&child) != 0) {
3592 remove_tempfile();
3593 error("error running textconv command '%s'", pgm);
3594 return NULL;
3596 remove_tempfile();
3598 return strbuf_detach(&buf, outsize);