contrib/difftool: add support for Kompare
[git/spearce.git] / diff.c
blob972b3daa6578776ca2f262d8e4d3290bae64e234
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"
16 #ifdef NO_FAST_WORKING_DIRECTORY
17 #define FAST_WORKING_DIRECTORY 0
18 #else
19 #define FAST_WORKING_DIRECTORY 1
20 #endif
22 static int diff_detect_rename_default;
23 static int diff_rename_limit_default = 200;
24 static int diff_suppress_blank_empty;
25 int diff_use_color_default = -1;
26 static const char *diff_word_regex_cfg;
27 static const char *external_diff_cmd_cfg;
28 int diff_auto_refresh_index = 1;
29 static int diff_mnemonic_prefix;
31 static char diff_colors[][COLOR_MAXLEN] = {
32 "\033[m", /* reset */
33 "", /* PLAIN (normal) */
34 "\033[1m", /* METAINFO (bold) */
35 "\033[36m", /* FRAGINFO (cyan) */
36 "\033[31m", /* OLD (red) */
37 "\033[32m", /* NEW (green) */
38 "\033[33m", /* COMMIT (yellow) */
39 "\033[41m", /* WHITESPACE (red background) */
42 static void diff_filespec_load_driver(struct diff_filespec *one);
43 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
45 static int parse_diff_color_slot(const char *var, int ofs)
47 if (!strcasecmp(var+ofs, "plain"))
48 return DIFF_PLAIN;
49 if (!strcasecmp(var+ofs, "meta"))
50 return DIFF_METAINFO;
51 if (!strcasecmp(var+ofs, "frag"))
52 return DIFF_FRAGINFO;
53 if (!strcasecmp(var+ofs, "old"))
54 return DIFF_FILE_OLD;
55 if (!strcasecmp(var+ofs, "new"))
56 return DIFF_FILE_NEW;
57 if (!strcasecmp(var+ofs, "commit"))
58 return DIFF_COMMIT;
59 if (!strcasecmp(var+ofs, "whitespace"))
60 return DIFF_WHITESPACE;
61 die("bad config variable '%s'", var);
65 * These are to give UI layer defaults.
66 * The core-level commands such as git-diff-files should
67 * never be affected by the setting of diff.renames
68 * the user happens to have in the configuration file.
70 int git_diff_ui_config(const char *var, const char *value, void *cb)
72 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
73 diff_use_color_default = git_config_colorbool(var, value, -1);
74 return 0;
76 if (!strcmp(var, "diff.renames")) {
77 if (!value)
78 diff_detect_rename_default = DIFF_DETECT_RENAME;
79 else if (!strcasecmp(value, "copies") ||
80 !strcasecmp(value, "copy"))
81 diff_detect_rename_default = DIFF_DETECT_COPY;
82 else if (git_config_bool(var,value))
83 diff_detect_rename_default = DIFF_DETECT_RENAME;
84 return 0;
86 if (!strcmp(var, "diff.autorefreshindex")) {
87 diff_auto_refresh_index = git_config_bool(var, value);
88 return 0;
90 if (!strcmp(var, "diff.mnemonicprefix")) {
91 diff_mnemonic_prefix = git_config_bool(var, value);
92 return 0;
94 if (!strcmp(var, "diff.external"))
95 return git_config_string(&external_diff_cmd_cfg, var, value);
96 if (!strcmp(var, "diff.wordregex"))
97 return git_config_string(&diff_word_regex_cfg, var, value);
99 return git_diff_basic_config(var, value, cb);
102 int git_diff_basic_config(const char *var, const char *value, void *cb)
104 if (!strcmp(var, "diff.renamelimit")) {
105 diff_rename_limit_default = git_config_int(var, value);
106 return 0;
109 switch (userdiff_config(var, value)) {
110 case 0: break;
111 case -1: return -1;
112 default: return 0;
115 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
116 int slot = parse_diff_color_slot(var, 11);
117 if (!value)
118 return config_error_nonbool(var);
119 color_parse(value, var, diff_colors[slot]);
120 return 0;
123 /* like GNU diff's --suppress-blank-empty option */
124 if (!strcmp(var, "diff.suppressblankempty") ||
125 /* for backwards compatibility */
126 !strcmp(var, "diff.suppress-blank-empty")) {
127 diff_suppress_blank_empty = git_config_bool(var, value);
128 return 0;
131 return git_color_default_config(var, value, cb);
134 static char *quote_two(const char *one, const char *two)
136 int need_one = quote_c_style(one, NULL, NULL, 1);
137 int need_two = quote_c_style(two, NULL, NULL, 1);
138 struct strbuf res = STRBUF_INIT;
140 if (need_one + need_two) {
141 strbuf_addch(&res, '"');
142 quote_c_style(one, &res, NULL, 1);
143 quote_c_style(two, &res, NULL, 1);
144 strbuf_addch(&res, '"');
145 } else {
146 strbuf_addstr(&res, one);
147 strbuf_addstr(&res, two);
149 return strbuf_detach(&res, NULL);
152 static const char *external_diff(void)
154 static const char *external_diff_cmd = NULL;
155 static int done_preparing = 0;
157 if (done_preparing)
158 return external_diff_cmd;
159 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
160 if (!external_diff_cmd)
161 external_diff_cmd = external_diff_cmd_cfg;
162 done_preparing = 1;
163 return external_diff_cmd;
166 static struct diff_tempfile {
167 const char *name; /* filename external diff should read from */
168 char hex[41];
169 char mode[10];
170 char tmp_path[PATH_MAX];
171 } diff_temp[2];
173 static int count_lines(const char *data, int size)
175 int count, ch, completely_empty = 1, nl_just_seen = 0;
176 count = 0;
177 while (0 < size--) {
178 ch = *data++;
179 if (ch == '\n') {
180 count++;
181 nl_just_seen = 1;
182 completely_empty = 0;
184 else {
185 nl_just_seen = 0;
186 completely_empty = 0;
189 if (completely_empty)
190 return 0;
191 if (!nl_just_seen)
192 count++; /* no trailing newline */
193 return count;
196 static void print_line_count(FILE *file, int count)
198 switch (count) {
199 case 0:
200 fprintf(file, "0,0");
201 break;
202 case 1:
203 fprintf(file, "1");
204 break;
205 default:
206 fprintf(file, "1,%d", count);
207 break;
211 static void copy_file_with_prefix(FILE *file,
212 int prefix, const char *data, int size,
213 const char *set, const char *reset)
215 int ch, nl_just_seen = 1;
216 while (0 < size--) {
217 ch = *data++;
218 if (nl_just_seen) {
219 fputs(set, file);
220 putc(prefix, file);
222 if (ch == '\n') {
223 nl_just_seen = 1;
224 fputs(reset, file);
225 } else
226 nl_just_seen = 0;
227 putc(ch, file);
229 if (!nl_just_seen)
230 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
233 static void emit_rewrite_diff(const char *name_a,
234 const char *name_b,
235 struct diff_filespec *one,
236 struct diff_filespec *two,
237 const char *textconv_one,
238 const char *textconv_two,
239 struct diff_options *o)
241 int lc_a, lc_b;
242 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
243 const char *name_a_tab, *name_b_tab;
244 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
245 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
246 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
247 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
248 const char *reset = diff_get_color(color_diff, DIFF_RESET);
249 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
250 const char *a_prefix, *b_prefix;
251 const char *data_one, *data_two;
252 size_t size_one, size_two;
254 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
255 a_prefix = o->b_prefix;
256 b_prefix = o->a_prefix;
257 } else {
258 a_prefix = o->a_prefix;
259 b_prefix = o->b_prefix;
262 name_a += (*name_a == '/');
263 name_b += (*name_b == '/');
264 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
265 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
267 strbuf_reset(&a_name);
268 strbuf_reset(&b_name);
269 quote_two_c_style(&a_name, a_prefix, name_a, 0);
270 quote_two_c_style(&b_name, b_prefix, name_b, 0);
272 diff_populate_filespec(one, 0);
273 diff_populate_filespec(two, 0);
274 if (textconv_one) {
275 data_one = run_textconv(textconv_one, one, &size_one);
276 if (!data_one)
277 die("unable to read files to diff");
279 else {
280 data_one = one->data;
281 size_one = one->size;
283 if (textconv_two) {
284 data_two = run_textconv(textconv_two, two, &size_two);
285 if (!data_two)
286 die("unable to read files to diff");
288 else {
289 data_two = two->data;
290 size_two = two->size;
293 lc_a = count_lines(data_one, size_one);
294 lc_b = count_lines(data_two, size_two);
295 fprintf(o->file,
296 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
297 metainfo, a_name.buf, name_a_tab, reset,
298 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
299 print_line_count(o->file, lc_a);
300 fprintf(o->file, " +");
301 print_line_count(o->file, lc_b);
302 fprintf(o->file, " @@%s\n", reset);
303 if (lc_a)
304 copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
305 if (lc_b)
306 copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
309 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
311 if (!DIFF_FILE_VALID(one)) {
312 mf->ptr = (char *)""; /* does not matter */
313 mf->size = 0;
314 return 0;
316 else if (diff_populate_filespec(one, 0))
317 return -1;
319 mf->ptr = one->data;
320 mf->size = one->size;
321 return 0;
324 struct diff_words_buffer {
325 mmfile_t text;
326 long alloc;
327 struct diff_words_orig {
328 const char *begin, *end;
329 } *orig;
330 int orig_nr, orig_alloc;
333 static void diff_words_append(char *line, unsigned long len,
334 struct diff_words_buffer *buffer)
336 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
337 line++;
338 len--;
339 memcpy(buffer->text.ptr + buffer->text.size, line, len);
340 buffer->text.size += len;
341 buffer->text.ptr[buffer->text.size] = '\0';
344 struct diff_words_data {
345 struct diff_words_buffer minus, plus;
346 const char *current_plus;
347 FILE *file;
348 regex_t *word_regex;
351 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
353 struct diff_words_data *diff_words = priv;
354 int minus_first, minus_len, plus_first, plus_len;
355 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
357 if (line[0] != '@' || parse_hunk_header(line, len,
358 &minus_first, &minus_len, &plus_first, &plus_len))
359 return;
361 /* POSIX requires that first be decremented by one if len == 0... */
362 if (minus_len) {
363 minus_begin = diff_words->minus.orig[minus_first].begin;
364 minus_end =
365 diff_words->minus.orig[minus_first + minus_len - 1].end;
366 } else
367 minus_begin = minus_end =
368 diff_words->minus.orig[minus_first].end;
370 if (plus_len) {
371 plus_begin = diff_words->plus.orig[plus_first].begin;
372 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
373 } else
374 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
376 if (diff_words->current_plus != plus_begin)
377 fwrite(diff_words->current_plus,
378 plus_begin - diff_words->current_plus, 1,
379 diff_words->file);
380 if (minus_begin != minus_end)
381 color_fwrite_lines(diff_words->file,
382 diff_get_color(1, DIFF_FILE_OLD),
383 minus_end - minus_begin, minus_begin);
384 if (plus_begin != plus_end)
385 color_fwrite_lines(diff_words->file,
386 diff_get_color(1, DIFF_FILE_NEW),
387 plus_end - plus_begin, plus_begin);
389 diff_words->current_plus = plus_end;
392 /* This function starts looking at *begin, and returns 0 iff a word was found. */
393 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
394 int *begin, int *end)
396 if (word_regex && *begin < buffer->size) {
397 regmatch_t match[1];
398 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
399 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
400 '\n', match[0].rm_eo - match[0].rm_so);
401 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
402 *begin += match[0].rm_so;
403 return *begin >= *end;
405 return -1;
408 /* find the next word */
409 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
410 (*begin)++;
411 if (*begin >= buffer->size)
412 return -1;
414 /* find the end of the word */
415 *end = *begin + 1;
416 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
417 (*end)++;
419 return 0;
423 * This function splits the words in buffer->text, stores the list with
424 * newline separator into out, and saves the offsets of the original words
425 * in buffer->orig.
427 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
428 regex_t *word_regex)
430 int i, j;
431 long alloc = 0;
433 out->size = 0;
434 out->ptr = NULL;
436 /* fake an empty "0th" word */
437 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
438 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
439 buffer->orig_nr = 1;
441 for (i = 0; i < buffer->text.size; i++) {
442 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
443 return;
445 /* store original boundaries */
446 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
447 buffer->orig_alloc);
448 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
449 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
450 buffer->orig_nr++;
452 /* store one word */
453 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
454 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
455 out->ptr[out->size + j - i] = '\n';
456 out->size += j - i + 1;
458 i = j - 1;
462 /* this executes the word diff on the accumulated buffers */
463 static void diff_words_show(struct diff_words_data *diff_words)
465 xpparam_t xpp;
466 xdemitconf_t xecfg;
467 xdemitcb_t ecb;
468 mmfile_t minus, plus;
470 /* special case: only removal */
471 if (!diff_words->plus.text.size) {
472 color_fwrite_lines(diff_words->file,
473 diff_get_color(1, DIFF_FILE_OLD),
474 diff_words->minus.text.size, diff_words->minus.text.ptr);
475 diff_words->minus.text.size = 0;
476 return;
479 diff_words->current_plus = diff_words->plus.text.ptr;
481 memset(&xpp, 0, sizeof(xpp));
482 memset(&xecfg, 0, sizeof(xecfg));
483 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
484 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
485 xpp.flags = XDF_NEED_MINIMAL;
486 /* as only the hunk header will be parsed, we need a 0-context */
487 xecfg.ctxlen = 0;
488 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
489 &xpp, &xecfg, &ecb);
490 free(minus.ptr);
491 free(plus.ptr);
492 if (diff_words->current_plus != diff_words->plus.text.ptr +
493 diff_words->plus.text.size)
494 fwrite(diff_words->current_plus,
495 diff_words->plus.text.ptr + diff_words->plus.text.size
496 - diff_words->current_plus, 1,
497 diff_words->file);
498 diff_words->minus.text.size = diff_words->plus.text.size = 0;
501 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
503 struct emit_callback {
504 int nparents, color_diff;
505 unsigned ws_rule;
506 sane_truncate_fn truncate;
507 const char **label_path;
508 struct diff_words_data *diff_words;
509 int *found_changesp;
510 FILE *file;
513 static void free_diff_words_data(struct emit_callback *ecbdata)
515 if (ecbdata->diff_words) {
516 /* flush buffers */
517 if (ecbdata->diff_words->minus.text.size ||
518 ecbdata->diff_words->plus.text.size)
519 diff_words_show(ecbdata->diff_words);
521 free (ecbdata->diff_words->minus.text.ptr);
522 free (ecbdata->diff_words->minus.orig);
523 free (ecbdata->diff_words->plus.text.ptr);
524 free (ecbdata->diff_words->plus.orig);
525 free(ecbdata->diff_words->word_regex);
526 free(ecbdata->diff_words);
527 ecbdata->diff_words = NULL;
531 const char *diff_get_color(int diff_use_color, enum color_diff ix)
533 if (diff_use_color)
534 return diff_colors[ix];
535 return "";
538 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
540 int has_trailing_newline, has_trailing_carriage_return;
542 has_trailing_newline = (len > 0 && line[len-1] == '\n');
543 if (has_trailing_newline)
544 len--;
545 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
546 if (has_trailing_carriage_return)
547 len--;
549 fputs(set, file);
550 fwrite(line, len, 1, file);
551 fputs(reset, file);
552 if (has_trailing_carriage_return)
553 fputc('\r', file);
554 if (has_trailing_newline)
555 fputc('\n', file);
558 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
560 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
561 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
563 if (!*ws)
564 emit_line(ecbdata->file, set, reset, line, len);
565 else {
566 /* Emit just the prefix, then the rest. */
567 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
568 ws_check_emit(line + ecbdata->nparents,
569 len - ecbdata->nparents, ecbdata->ws_rule,
570 ecbdata->file, set, reset, ws);
574 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
576 const char *cp;
577 unsigned long allot;
578 size_t l = len;
580 if (ecb->truncate)
581 return ecb->truncate(line, len);
582 cp = line;
583 allot = l;
584 while (0 < l) {
585 (void) utf8_width(&cp, &l);
586 if (!cp)
587 break; /* truncated in the middle? */
589 return allot - l;
592 static void fn_out_consume(void *priv, char *line, unsigned long len)
594 int i;
595 int color;
596 struct emit_callback *ecbdata = priv;
597 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
598 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
599 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
601 *(ecbdata->found_changesp) = 1;
603 if (ecbdata->label_path[0]) {
604 const char *name_a_tab, *name_b_tab;
606 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
607 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
609 fprintf(ecbdata->file, "%s--- %s%s%s\n",
610 meta, ecbdata->label_path[0], reset, name_a_tab);
611 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
612 meta, ecbdata->label_path[1], reset, name_b_tab);
613 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
616 if (diff_suppress_blank_empty
617 && len == 2 && line[0] == ' ' && line[1] == '\n') {
618 line[0] = '\n';
619 len = 1;
622 /* This is not really necessary for now because
623 * this codepath only deals with two-way diffs.
625 for (i = 0; i < len && line[i] == '@'; i++)
627 if (2 <= i && i < len && line[i] == ' ') {
628 ecbdata->nparents = i - 1;
629 len = sane_truncate_line(ecbdata, line, len);
630 emit_line(ecbdata->file,
631 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
632 reset, line, len);
633 if (line[len-1] != '\n')
634 putc('\n', ecbdata->file);
635 return;
638 if (len < ecbdata->nparents) {
639 emit_line(ecbdata->file, reset, reset, line, len);
640 return;
643 color = DIFF_PLAIN;
644 if (ecbdata->diff_words && ecbdata->nparents != 1)
645 /* fall back to normal diff */
646 free_diff_words_data(ecbdata);
647 if (ecbdata->diff_words) {
648 if (line[0] == '-') {
649 diff_words_append(line, len,
650 &ecbdata->diff_words->minus);
651 return;
652 } else if (line[0] == '+') {
653 diff_words_append(line, len,
654 &ecbdata->diff_words->plus);
655 return;
657 if (ecbdata->diff_words->minus.text.size ||
658 ecbdata->diff_words->plus.text.size)
659 diff_words_show(ecbdata->diff_words);
660 line++;
661 len--;
662 emit_line(ecbdata->file, plain, reset, line, len);
663 return;
665 for (i = 0; i < ecbdata->nparents && len; i++) {
666 if (line[i] == '-')
667 color = DIFF_FILE_OLD;
668 else if (line[i] == '+')
669 color = DIFF_FILE_NEW;
672 if (color != DIFF_FILE_NEW) {
673 emit_line(ecbdata->file,
674 diff_get_color(ecbdata->color_diff, color),
675 reset, line, len);
676 return;
678 emit_add_line(reset, ecbdata, line, len);
681 static char *pprint_rename(const char *a, const char *b)
683 const char *old = a;
684 const char *new = b;
685 struct strbuf name = STRBUF_INIT;
686 int pfx_length, sfx_length;
687 int len_a = strlen(a);
688 int len_b = strlen(b);
689 int a_midlen, b_midlen;
690 int qlen_a = quote_c_style(a, NULL, NULL, 0);
691 int qlen_b = quote_c_style(b, NULL, NULL, 0);
693 if (qlen_a || qlen_b) {
694 quote_c_style(a, &name, NULL, 0);
695 strbuf_addstr(&name, " => ");
696 quote_c_style(b, &name, NULL, 0);
697 return strbuf_detach(&name, NULL);
700 /* Find common prefix */
701 pfx_length = 0;
702 while (*old && *new && *old == *new) {
703 if (*old == '/')
704 pfx_length = old - a + 1;
705 old++;
706 new++;
709 /* Find common suffix */
710 old = a + len_a;
711 new = b + len_b;
712 sfx_length = 0;
713 while (a <= old && b <= new && *old == *new) {
714 if (*old == '/')
715 sfx_length = len_a - (old - a);
716 old--;
717 new--;
721 * pfx{mid-a => mid-b}sfx
722 * {pfx-a => pfx-b}sfx
723 * pfx{sfx-a => sfx-b}
724 * name-a => name-b
726 a_midlen = len_a - pfx_length - sfx_length;
727 b_midlen = len_b - pfx_length - sfx_length;
728 if (a_midlen < 0)
729 a_midlen = 0;
730 if (b_midlen < 0)
731 b_midlen = 0;
733 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
734 if (pfx_length + sfx_length) {
735 strbuf_add(&name, a, pfx_length);
736 strbuf_addch(&name, '{');
738 strbuf_add(&name, a + pfx_length, a_midlen);
739 strbuf_addstr(&name, " => ");
740 strbuf_add(&name, b + pfx_length, b_midlen);
741 if (pfx_length + sfx_length) {
742 strbuf_addch(&name, '}');
743 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
745 return strbuf_detach(&name, NULL);
748 struct diffstat_t {
749 int nr;
750 int alloc;
751 struct diffstat_file {
752 char *from_name;
753 char *name;
754 char *print_name;
755 unsigned is_unmerged:1;
756 unsigned is_binary:1;
757 unsigned is_renamed:1;
758 unsigned int added, deleted;
759 } **files;
762 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
763 const char *name_a,
764 const char *name_b)
766 struct diffstat_file *x;
767 x = xcalloc(sizeof (*x), 1);
768 if (diffstat->nr == diffstat->alloc) {
769 diffstat->alloc = alloc_nr(diffstat->alloc);
770 diffstat->files = xrealloc(diffstat->files,
771 diffstat->alloc * sizeof(x));
773 diffstat->files[diffstat->nr++] = x;
774 if (name_b) {
775 x->from_name = xstrdup(name_a);
776 x->name = xstrdup(name_b);
777 x->is_renamed = 1;
779 else {
780 x->from_name = NULL;
781 x->name = xstrdup(name_a);
783 return x;
786 static void diffstat_consume(void *priv, char *line, unsigned long len)
788 struct diffstat_t *diffstat = priv;
789 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
791 if (line[0] == '+')
792 x->added++;
793 else if (line[0] == '-')
794 x->deleted++;
797 const char mime_boundary_leader[] = "------------";
799 static int scale_linear(int it, int width, int max_change)
802 * make sure that at least one '-' is printed if there were deletions,
803 * and likewise for '+'.
805 if (max_change < 2)
806 return it;
807 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
810 static void show_name(FILE *file,
811 const char *prefix, const char *name, int len,
812 const char *reset, const char *set)
814 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
817 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
819 if (cnt <= 0)
820 return;
821 fprintf(file, "%s", set);
822 while (cnt--)
823 putc(ch, file);
824 fprintf(file, "%s", reset);
827 static void fill_print_name(struct diffstat_file *file)
829 char *pname;
831 if (file->print_name)
832 return;
834 if (!file->is_renamed) {
835 struct strbuf buf = STRBUF_INIT;
836 if (quote_c_style(file->name, &buf, NULL, 0)) {
837 pname = strbuf_detach(&buf, NULL);
838 } else {
839 pname = file->name;
840 strbuf_release(&buf);
842 } else {
843 pname = pprint_rename(file->from_name, file->name);
845 file->print_name = pname;
848 static void show_stats(struct diffstat_t* data, struct diff_options *options)
850 int i, len, add, del, total, adds = 0, dels = 0;
851 int max_change = 0, max_len = 0;
852 int total_files = data->nr;
853 int width, name_width;
854 const char *reset, *set, *add_c, *del_c;
856 if (data->nr == 0)
857 return;
859 width = options->stat_width ? options->stat_width : 80;
860 name_width = options->stat_name_width ? options->stat_name_width : 50;
862 /* Sanity: give at least 5 columns to the graph,
863 * but leave at least 10 columns for the name.
865 if (width < 25)
866 width = 25;
867 if (name_width < 10)
868 name_width = 10;
869 else if (width < name_width + 15)
870 name_width = width - 15;
872 /* Find the longest filename and max number of changes */
873 reset = diff_get_color_opt(options, DIFF_RESET);
874 set = diff_get_color_opt(options, DIFF_PLAIN);
875 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
876 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
878 for (i = 0; i < data->nr; i++) {
879 struct diffstat_file *file = data->files[i];
880 int change = file->added + file->deleted;
881 fill_print_name(file);
882 len = strlen(file->print_name);
883 if (max_len < len)
884 max_len = len;
886 if (file->is_binary || file->is_unmerged)
887 continue;
888 if (max_change < change)
889 max_change = change;
892 /* Compute the width of the graph part;
893 * 10 is for one blank at the beginning of the line plus
894 * " | count " between the name and the graph.
896 * From here on, name_width is the width of the name area,
897 * and width is the width of the graph area.
899 name_width = (name_width < max_len) ? name_width : max_len;
900 if (width < (name_width + 10) + max_change)
901 width = width - (name_width + 10);
902 else
903 width = max_change;
905 for (i = 0; i < data->nr; i++) {
906 const char *prefix = "";
907 char *name = data->files[i]->print_name;
908 int added = data->files[i]->added;
909 int deleted = data->files[i]->deleted;
910 int name_len;
913 * "scale" the filename
915 len = name_width;
916 name_len = strlen(name);
917 if (name_width < name_len) {
918 char *slash;
919 prefix = "...";
920 len -= 3;
921 name += name_len - len;
922 slash = strchr(name, '/');
923 if (slash)
924 name = slash;
927 if (data->files[i]->is_binary) {
928 show_name(options->file, prefix, name, len, reset, set);
929 fprintf(options->file, " Bin ");
930 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
931 fprintf(options->file, " -> ");
932 fprintf(options->file, "%s%d%s", add_c, added, reset);
933 fprintf(options->file, " bytes");
934 fprintf(options->file, "\n");
935 continue;
937 else if (data->files[i]->is_unmerged) {
938 show_name(options->file, prefix, name, len, reset, set);
939 fprintf(options->file, " Unmerged\n");
940 continue;
942 else if (!data->files[i]->is_renamed &&
943 (added + deleted == 0)) {
944 total_files--;
945 continue;
949 * scale the add/delete
951 add = added;
952 del = deleted;
953 total = add + del;
954 adds += add;
955 dels += del;
957 if (width <= max_change) {
958 add = scale_linear(add, width, max_change);
959 del = scale_linear(del, width, max_change);
960 total = add + del;
962 show_name(options->file, prefix, name, len, reset, set);
963 fprintf(options->file, "%5d%s", added + deleted,
964 added + deleted ? " " : "");
965 show_graph(options->file, '+', add, add_c, reset);
966 show_graph(options->file, '-', del, del_c, reset);
967 fprintf(options->file, "\n");
969 fprintf(options->file,
970 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
971 set, total_files, adds, dels, reset);
974 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
976 int i, adds = 0, dels = 0, total_files = data->nr;
978 if (data->nr == 0)
979 return;
981 for (i = 0; i < data->nr; i++) {
982 if (!data->files[i]->is_binary &&
983 !data->files[i]->is_unmerged) {
984 int added = data->files[i]->added;
985 int deleted= data->files[i]->deleted;
986 if (!data->files[i]->is_renamed &&
987 (added + deleted == 0)) {
988 total_files--;
989 } else {
990 adds += added;
991 dels += deleted;
995 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
996 total_files, adds, dels);
999 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
1001 int i;
1003 if (data->nr == 0)
1004 return;
1006 for (i = 0; i < data->nr; i++) {
1007 struct diffstat_file *file = data->files[i];
1009 if (file->is_binary)
1010 fprintf(options->file, "-\t-\t");
1011 else
1012 fprintf(options->file,
1013 "%d\t%d\t", file->added, file->deleted);
1014 if (options->line_termination) {
1015 fill_print_name(file);
1016 if (!file->is_renamed)
1017 write_name_quoted(file->name, options->file,
1018 options->line_termination);
1019 else {
1020 fputs(file->print_name, options->file);
1021 putc(options->line_termination, options->file);
1023 } else {
1024 if (file->is_renamed) {
1025 putc('\0', options->file);
1026 write_name_quoted(file->from_name, options->file, '\0');
1028 write_name_quoted(file->name, options->file, '\0');
1033 struct dirstat_file {
1034 const char *name;
1035 unsigned long changed;
1038 struct dirstat_dir {
1039 struct dirstat_file *files;
1040 int alloc, nr, percent, cumulative;
1043 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1045 unsigned long this_dir = 0;
1046 unsigned int sources = 0;
1048 while (dir->nr) {
1049 struct dirstat_file *f = dir->files;
1050 int namelen = strlen(f->name);
1051 unsigned long this;
1052 char *slash;
1054 if (namelen < baselen)
1055 break;
1056 if (memcmp(f->name, base, baselen))
1057 break;
1058 slash = strchr(f->name + baselen, '/');
1059 if (slash) {
1060 int newbaselen = slash + 1 - f->name;
1061 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1062 sources++;
1063 } else {
1064 this = f->changed;
1065 dir->files++;
1066 dir->nr--;
1067 sources += 2;
1069 this_dir += this;
1073 * We don't report dirstat's for
1074 * - the top level
1075 * - or cases where everything came from a single directory
1076 * under this directory (sources == 1).
1078 if (baselen && sources != 1) {
1079 int permille = this_dir * 1000 / changed;
1080 if (permille) {
1081 int percent = permille / 10;
1082 if (percent >= dir->percent) {
1083 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1084 if (!dir->cumulative)
1085 return 0;
1089 return this_dir;
1092 static int dirstat_compare(const void *_a, const void *_b)
1094 const struct dirstat_file *a = _a;
1095 const struct dirstat_file *b = _b;
1096 return strcmp(a->name, b->name);
1099 static void show_dirstat(struct diff_options *options)
1101 int i;
1102 unsigned long changed;
1103 struct dirstat_dir dir;
1104 struct diff_queue_struct *q = &diff_queued_diff;
1106 dir.files = NULL;
1107 dir.alloc = 0;
1108 dir.nr = 0;
1109 dir.percent = options->dirstat_percent;
1110 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1112 changed = 0;
1113 for (i = 0; i < q->nr; i++) {
1114 struct diff_filepair *p = q->queue[i];
1115 const char *name;
1116 unsigned long copied, added, damage;
1118 name = p->one->path ? p->one->path : p->two->path;
1120 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1121 diff_populate_filespec(p->one, 0);
1122 diff_populate_filespec(p->two, 0);
1123 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1124 &copied, &added);
1125 diff_free_filespec_data(p->one);
1126 diff_free_filespec_data(p->two);
1127 } else if (DIFF_FILE_VALID(p->one)) {
1128 diff_populate_filespec(p->one, 1);
1129 copied = added = 0;
1130 diff_free_filespec_data(p->one);
1131 } else if (DIFF_FILE_VALID(p->two)) {
1132 diff_populate_filespec(p->two, 1);
1133 copied = 0;
1134 added = p->two->size;
1135 diff_free_filespec_data(p->two);
1136 } else
1137 continue;
1140 * Original minus copied is the removed material,
1141 * added is the new material. They are both damages
1142 * made to the preimage. In --dirstat-by-file mode, count
1143 * damaged files, not damaged lines. This is done by
1144 * counting only a single damaged line per file.
1146 damage = (p->one->size - copied) + added;
1147 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1148 damage = 1;
1150 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1151 dir.files[dir.nr].name = name;
1152 dir.files[dir.nr].changed = damage;
1153 changed += damage;
1154 dir.nr++;
1157 /* This can happen even with many files, if everything was renames */
1158 if (!changed)
1159 return;
1161 /* Show all directories with more than x% of the changes */
1162 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1163 gather_dirstat(options->file, &dir, changed, "", 0);
1166 static void free_diffstat_info(struct diffstat_t *diffstat)
1168 int i;
1169 for (i = 0; i < diffstat->nr; i++) {
1170 struct diffstat_file *f = diffstat->files[i];
1171 if (f->name != f->print_name)
1172 free(f->print_name);
1173 free(f->name);
1174 free(f->from_name);
1175 free(f);
1177 free(diffstat->files);
1180 struct checkdiff_t {
1181 const char *filename;
1182 int lineno;
1183 struct diff_options *o;
1184 unsigned ws_rule;
1185 unsigned status;
1186 int trailing_blanks_start;
1189 static int is_conflict_marker(const char *line, unsigned long len)
1191 char firstchar;
1192 int cnt;
1194 if (len < 8)
1195 return 0;
1196 firstchar = line[0];
1197 switch (firstchar) {
1198 case '=': case '>': case '<':
1199 break;
1200 default:
1201 return 0;
1203 for (cnt = 1; cnt < 7; cnt++)
1204 if (line[cnt] != firstchar)
1205 return 0;
1206 /* line[0] thru line[6] are same as firstchar */
1207 if (firstchar == '=') {
1208 /* divider between ours and theirs? */
1209 if (len != 8 || line[7] != '\n')
1210 return 0;
1211 } else if (len < 8 || !isspace(line[7])) {
1212 /* not divider before ours nor after theirs */
1213 return 0;
1215 return 1;
1218 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1220 struct checkdiff_t *data = priv;
1221 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1222 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1223 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1224 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1225 char *err;
1227 if (line[0] == '+') {
1228 unsigned bad;
1229 data->lineno++;
1230 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1231 data->trailing_blanks_start = 0;
1232 else if (!data->trailing_blanks_start)
1233 data->trailing_blanks_start = data->lineno;
1234 if (is_conflict_marker(line + 1, len - 1)) {
1235 data->status |= 1;
1236 fprintf(data->o->file,
1237 "%s:%d: leftover conflict marker\n",
1238 data->filename, data->lineno);
1240 bad = ws_check(line + 1, len - 1, data->ws_rule);
1241 if (!bad)
1242 return;
1243 data->status |= bad;
1244 err = whitespace_error_string(bad);
1245 fprintf(data->o->file, "%s:%d: %s.\n",
1246 data->filename, data->lineno, err);
1247 free(err);
1248 emit_line(data->o->file, set, reset, line, 1);
1249 ws_check_emit(line + 1, len - 1, data->ws_rule,
1250 data->o->file, set, reset, ws);
1251 } else if (line[0] == ' ') {
1252 data->lineno++;
1253 data->trailing_blanks_start = 0;
1254 } else if (line[0] == '@') {
1255 char *plus = strchr(line, '+');
1256 if (plus)
1257 data->lineno = strtol(plus, NULL, 10) - 1;
1258 else
1259 die("invalid diff");
1260 data->trailing_blanks_start = 0;
1264 static unsigned char *deflate_it(char *data,
1265 unsigned long size,
1266 unsigned long *result_size)
1268 int bound;
1269 unsigned char *deflated;
1270 z_stream stream;
1272 memset(&stream, 0, sizeof(stream));
1273 deflateInit(&stream, zlib_compression_level);
1274 bound = deflateBound(&stream, size);
1275 deflated = xmalloc(bound);
1276 stream.next_out = deflated;
1277 stream.avail_out = bound;
1279 stream.next_in = (unsigned char *)data;
1280 stream.avail_in = size;
1281 while (deflate(&stream, Z_FINISH) == Z_OK)
1282 ; /* nothing */
1283 deflateEnd(&stream);
1284 *result_size = stream.total_out;
1285 return deflated;
1288 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1290 void *cp;
1291 void *delta;
1292 void *deflated;
1293 void *data;
1294 unsigned long orig_size;
1295 unsigned long delta_size;
1296 unsigned long deflate_size;
1297 unsigned long data_size;
1299 /* We could do deflated delta, or we could do just deflated two,
1300 * whichever is smaller.
1302 delta = NULL;
1303 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1304 if (one->size && two->size) {
1305 delta = diff_delta(one->ptr, one->size,
1306 two->ptr, two->size,
1307 &delta_size, deflate_size);
1308 if (delta) {
1309 void *to_free = delta;
1310 orig_size = delta_size;
1311 delta = deflate_it(delta, delta_size, &delta_size);
1312 free(to_free);
1316 if (delta && delta_size < deflate_size) {
1317 fprintf(file, "delta %lu\n", orig_size);
1318 free(deflated);
1319 data = delta;
1320 data_size = delta_size;
1322 else {
1323 fprintf(file, "literal %lu\n", two->size);
1324 free(delta);
1325 data = deflated;
1326 data_size = deflate_size;
1329 /* emit data encoded in base85 */
1330 cp = data;
1331 while (data_size) {
1332 int bytes = (52 < data_size) ? 52 : data_size;
1333 char line[70];
1334 data_size -= bytes;
1335 if (bytes <= 26)
1336 line[0] = bytes + 'A' - 1;
1337 else
1338 line[0] = bytes - 26 + 'a' - 1;
1339 encode_85(line + 1, cp, bytes);
1340 cp = (char *) cp + bytes;
1341 fputs(line, file);
1342 fputc('\n', file);
1344 fprintf(file, "\n");
1345 free(data);
1348 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1350 fprintf(file, "GIT binary patch\n");
1351 emit_binary_diff_body(file, one, two);
1352 emit_binary_diff_body(file, two, one);
1355 static void diff_filespec_load_driver(struct diff_filespec *one)
1357 if (!one->driver)
1358 one->driver = userdiff_find_by_path(one->path);
1359 if (!one->driver)
1360 one->driver = userdiff_find_by_name("default");
1363 int diff_filespec_is_binary(struct diff_filespec *one)
1365 if (one->is_binary == -1) {
1366 diff_filespec_load_driver(one);
1367 if (one->driver->binary != -1)
1368 one->is_binary = one->driver->binary;
1369 else {
1370 if (!one->data && DIFF_FILE_VALID(one))
1371 diff_populate_filespec(one, 0);
1372 if (one->data)
1373 one->is_binary = buffer_is_binary(one->data,
1374 one->size);
1375 if (one->is_binary == -1)
1376 one->is_binary = 0;
1379 return one->is_binary;
1382 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1384 diff_filespec_load_driver(one);
1385 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1388 static const char *userdiff_word_regex(struct diff_filespec *one)
1390 diff_filespec_load_driver(one);
1391 return one->driver->word_regex;
1394 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1396 if (!options->a_prefix)
1397 options->a_prefix = a;
1398 if (!options->b_prefix)
1399 options->b_prefix = b;
1402 static const char *get_textconv(struct diff_filespec *one)
1404 if (!DIFF_FILE_VALID(one))
1405 return NULL;
1406 if (!S_ISREG(one->mode))
1407 return NULL;
1408 diff_filespec_load_driver(one);
1409 return one->driver->textconv;
1412 static void builtin_diff(const char *name_a,
1413 const char *name_b,
1414 struct diff_filespec *one,
1415 struct diff_filespec *two,
1416 const char *xfrm_msg,
1417 struct diff_options *o,
1418 int complete_rewrite)
1420 mmfile_t mf1, mf2;
1421 const char *lbl[2];
1422 char *a_one, *b_two;
1423 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1424 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1425 const char *a_prefix, *b_prefix;
1426 const char *textconv_one = NULL, *textconv_two = NULL;
1428 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1429 textconv_one = get_textconv(one);
1430 textconv_two = get_textconv(two);
1433 diff_set_mnemonic_prefix(o, "a/", "b/");
1434 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1435 a_prefix = o->b_prefix;
1436 b_prefix = o->a_prefix;
1437 } else {
1438 a_prefix = o->a_prefix;
1439 b_prefix = o->b_prefix;
1442 /* Never use a non-valid filename anywhere if at all possible */
1443 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1444 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1446 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1447 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1448 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1449 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1450 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1451 if (lbl[0][0] == '/') {
1452 /* /dev/null */
1453 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1454 if (xfrm_msg && xfrm_msg[0])
1455 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1457 else if (lbl[1][0] == '/') {
1458 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1459 if (xfrm_msg && xfrm_msg[0])
1460 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1462 else {
1463 if (one->mode != two->mode) {
1464 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1465 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1467 if (xfrm_msg && xfrm_msg[0])
1468 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1470 * we do not run diff between different kind
1471 * of objects.
1473 if ((one->mode ^ two->mode) & S_IFMT)
1474 goto free_ab_and_return;
1475 if (complete_rewrite &&
1476 (textconv_one || !diff_filespec_is_binary(one)) &&
1477 (textconv_two || !diff_filespec_is_binary(two))) {
1478 emit_rewrite_diff(name_a, name_b, one, two,
1479 textconv_one, textconv_two, o);
1480 o->found_changes = 1;
1481 goto free_ab_and_return;
1485 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1486 die("unable to read files to diff");
1488 if (!DIFF_OPT_TST(o, TEXT) &&
1489 ( (diff_filespec_is_binary(one) && !textconv_one) ||
1490 (diff_filespec_is_binary(two) && !textconv_two) )) {
1491 /* Quite common confusing case */
1492 if (mf1.size == mf2.size &&
1493 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1494 goto free_ab_and_return;
1495 if (DIFF_OPT_TST(o, BINARY))
1496 emit_binary_diff(o->file, &mf1, &mf2);
1497 else
1498 fprintf(o->file, "Binary files %s and %s differ\n",
1499 lbl[0], lbl[1]);
1500 o->found_changes = 1;
1502 else {
1503 /* Crazy xdl interfaces.. */
1504 const char *diffopts = getenv("GIT_DIFF_OPTS");
1505 xpparam_t xpp;
1506 xdemitconf_t xecfg;
1507 xdemitcb_t ecb;
1508 struct emit_callback ecbdata;
1509 const struct userdiff_funcname *pe;
1511 if (textconv_one) {
1512 size_t size;
1513 mf1.ptr = run_textconv(textconv_one, one, &size);
1514 if (!mf1.ptr)
1515 die("unable to read files to diff");
1516 mf1.size = size;
1518 if (textconv_two) {
1519 size_t size;
1520 mf2.ptr = run_textconv(textconv_two, two, &size);
1521 if (!mf2.ptr)
1522 die("unable to read files to diff");
1523 mf2.size = size;
1526 pe = diff_funcname_pattern(one);
1527 if (!pe)
1528 pe = diff_funcname_pattern(two);
1530 memset(&xpp, 0, sizeof(xpp));
1531 memset(&xecfg, 0, sizeof(xecfg));
1532 memset(&ecbdata, 0, sizeof(ecbdata));
1533 ecbdata.label_path = lbl;
1534 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1535 ecbdata.found_changesp = &o->found_changes;
1536 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1537 ecbdata.file = o->file;
1538 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1539 xecfg.ctxlen = o->context;
1540 xecfg.interhunkctxlen = o->interhunkcontext;
1541 xecfg.flags = XDL_EMIT_FUNCNAMES;
1542 if (pe)
1543 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1544 if (!diffopts)
1546 else if (!prefixcmp(diffopts, "--unified="))
1547 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1548 else if (!prefixcmp(diffopts, "-u"))
1549 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1550 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1551 ecbdata.diff_words =
1552 xcalloc(1, sizeof(struct diff_words_data));
1553 ecbdata.diff_words->file = o->file;
1554 if (!o->word_regex)
1555 o->word_regex = userdiff_word_regex(one);
1556 if (!o->word_regex)
1557 o->word_regex = userdiff_word_regex(two);
1558 if (!o->word_regex)
1559 o->word_regex = diff_word_regex_cfg;
1560 if (o->word_regex) {
1561 ecbdata.diff_words->word_regex = (regex_t *)
1562 xmalloc(sizeof(regex_t));
1563 if (regcomp(ecbdata.diff_words->word_regex,
1564 o->word_regex,
1565 REG_EXTENDED | REG_NEWLINE))
1566 die ("Invalid regular expression: %s",
1567 o->word_regex);
1570 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1571 &xpp, &xecfg, &ecb);
1572 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1573 free_diff_words_data(&ecbdata);
1574 if (textconv_one)
1575 free(mf1.ptr);
1576 if (textconv_two)
1577 free(mf2.ptr);
1580 free_ab_and_return:
1581 diff_free_filespec_data(one);
1582 diff_free_filespec_data(two);
1583 free(a_one);
1584 free(b_two);
1585 return;
1588 static void builtin_diffstat(const char *name_a, const char *name_b,
1589 struct diff_filespec *one,
1590 struct diff_filespec *two,
1591 struct diffstat_t *diffstat,
1592 struct diff_options *o,
1593 int complete_rewrite)
1595 mmfile_t mf1, mf2;
1596 struct diffstat_file *data;
1598 data = diffstat_add(diffstat, name_a, name_b);
1600 if (!one || !two) {
1601 data->is_unmerged = 1;
1602 return;
1604 if (complete_rewrite) {
1605 diff_populate_filespec(one, 0);
1606 diff_populate_filespec(two, 0);
1607 data->deleted = count_lines(one->data, one->size);
1608 data->added = count_lines(two->data, two->size);
1609 goto free_and_return;
1611 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1612 die("unable to read files to diff");
1614 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1615 data->is_binary = 1;
1616 data->added = mf2.size;
1617 data->deleted = mf1.size;
1618 } else {
1619 /* Crazy xdl interfaces.. */
1620 xpparam_t xpp;
1621 xdemitconf_t xecfg;
1622 xdemitcb_t ecb;
1624 memset(&xpp, 0, sizeof(xpp));
1625 memset(&xecfg, 0, sizeof(xecfg));
1626 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1627 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1628 &xpp, &xecfg, &ecb);
1631 free_and_return:
1632 diff_free_filespec_data(one);
1633 diff_free_filespec_data(two);
1636 static void builtin_checkdiff(const char *name_a, const char *name_b,
1637 const char *attr_path,
1638 struct diff_filespec *one,
1639 struct diff_filespec *two,
1640 struct diff_options *o)
1642 mmfile_t mf1, mf2;
1643 struct checkdiff_t data;
1645 if (!two)
1646 return;
1648 memset(&data, 0, sizeof(data));
1649 data.filename = name_b ? name_b : name_a;
1650 data.lineno = 0;
1651 data.o = o;
1652 data.ws_rule = whitespace_rule(attr_path);
1654 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1655 die("unable to read files to diff");
1658 * All the other codepaths check both sides, but not checking
1659 * the "old" side here is deliberate. We are checking the newly
1660 * introduced changes, and as long as the "new" side is text, we
1661 * can and should check what it introduces.
1663 if (diff_filespec_is_binary(two))
1664 goto free_and_return;
1665 else {
1666 /* Crazy xdl interfaces.. */
1667 xpparam_t xpp;
1668 xdemitconf_t xecfg;
1669 xdemitcb_t ecb;
1671 memset(&xpp, 0, sizeof(xpp));
1672 memset(&xecfg, 0, sizeof(xecfg));
1673 xecfg.ctxlen = 1; /* at least one context line */
1674 xpp.flags = XDF_NEED_MINIMAL;
1675 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1676 &xpp, &xecfg, &ecb);
1678 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1679 data.trailing_blanks_start) {
1680 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1681 data.filename, data.trailing_blanks_start);
1682 data.status = 1; /* report errors */
1685 free_and_return:
1686 diff_free_filespec_data(one);
1687 diff_free_filespec_data(two);
1688 if (data.status)
1689 DIFF_OPT_SET(o, CHECK_FAILED);
1692 struct diff_filespec *alloc_filespec(const char *path)
1694 int namelen = strlen(path);
1695 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1697 memset(spec, 0, sizeof(*spec));
1698 spec->path = (char *)(spec + 1);
1699 memcpy(spec->path, path, namelen+1);
1700 spec->count = 1;
1701 spec->is_binary = -1;
1702 return spec;
1705 void free_filespec(struct diff_filespec *spec)
1707 if (!--spec->count) {
1708 diff_free_filespec_data(spec);
1709 free(spec);
1713 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1714 unsigned short mode)
1716 if (mode) {
1717 spec->mode = canon_mode(mode);
1718 hashcpy(spec->sha1, sha1);
1719 spec->sha1_valid = !is_null_sha1(sha1);
1724 * Given a name and sha1 pair, if the index tells us the file in
1725 * the work tree has that object contents, return true, so that
1726 * prepare_temp_file() does not have to inflate and extract.
1728 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1730 struct cache_entry *ce;
1731 struct stat st;
1732 int pos, len;
1734 /* We do not read the cache ourselves here, because the
1735 * benchmark with my previous version that always reads cache
1736 * shows that it makes things worse for diff-tree comparing
1737 * two linux-2.6 kernel trees in an already checked out work
1738 * tree. This is because most diff-tree comparisons deal with
1739 * only a small number of files, while reading the cache is
1740 * expensive for a large project, and its cost outweighs the
1741 * savings we get by not inflating the object to a temporary
1742 * file. Practically, this code only helps when we are used
1743 * by diff-cache --cached, which does read the cache before
1744 * calling us.
1746 if (!active_cache)
1747 return 0;
1749 /* We want to avoid the working directory if our caller
1750 * doesn't need the data in a normal file, this system
1751 * is rather slow with its stat/open/mmap/close syscalls,
1752 * and the object is contained in a pack file. The pack
1753 * is probably already open and will be faster to obtain
1754 * the data through than the working directory. Loose
1755 * objects however would tend to be slower as they need
1756 * to be individually opened and inflated.
1758 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1759 return 0;
1761 len = strlen(name);
1762 pos = cache_name_pos(name, len);
1763 if (pos < 0)
1764 return 0;
1765 ce = active_cache[pos];
1768 * This is not the sha1 we are looking for, or
1769 * unreusable because it is not a regular file.
1771 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1772 return 0;
1775 * If ce matches the file in the work tree, we can reuse it.
1777 if (ce_uptodate(ce) ||
1778 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1779 return 1;
1781 return 0;
1784 static int populate_from_stdin(struct diff_filespec *s)
1786 struct strbuf buf = STRBUF_INIT;
1787 size_t size = 0;
1789 if (strbuf_read(&buf, 0, 0) < 0)
1790 return error("error while reading from stdin %s",
1791 strerror(errno));
1793 s->should_munmap = 0;
1794 s->data = strbuf_detach(&buf, &size);
1795 s->size = size;
1796 s->should_free = 1;
1797 return 0;
1800 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1802 int len;
1803 char *data = xmalloc(100);
1804 len = snprintf(data, 100,
1805 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1806 s->data = data;
1807 s->size = len;
1808 s->should_free = 1;
1809 if (size_only) {
1810 s->data = NULL;
1811 free(data);
1813 return 0;
1817 * While doing rename detection and pickaxe operation, we may need to
1818 * grab the data for the blob (or file) for our own in-core comparison.
1819 * diff_filespec has data and size fields for this purpose.
1821 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1823 int err = 0;
1824 if (!DIFF_FILE_VALID(s))
1825 die("internal error: asking to populate invalid file.");
1826 if (S_ISDIR(s->mode))
1827 return -1;
1829 if (s->data)
1830 return 0;
1832 if (size_only && 0 < s->size)
1833 return 0;
1835 if (S_ISGITLINK(s->mode))
1836 return diff_populate_gitlink(s, size_only);
1838 if (!s->sha1_valid ||
1839 reuse_worktree_file(s->path, s->sha1, 0)) {
1840 struct strbuf buf = STRBUF_INIT;
1841 struct stat st;
1842 int fd;
1844 if (!strcmp(s->path, "-"))
1845 return populate_from_stdin(s);
1847 if (lstat(s->path, &st) < 0) {
1848 if (errno == ENOENT) {
1849 err_empty:
1850 err = -1;
1851 empty:
1852 s->data = (char *)"";
1853 s->size = 0;
1854 return err;
1857 s->size = xsize_t(st.st_size);
1858 if (!s->size)
1859 goto empty;
1860 if (S_ISLNK(st.st_mode)) {
1861 struct strbuf sb = STRBUF_INIT;
1863 if (strbuf_readlink(&sb, s->path, s->size))
1864 goto err_empty;
1865 s->size = sb.len;
1866 s->data = strbuf_detach(&sb, NULL);
1867 s->should_free = 1;
1868 return 0;
1870 if (size_only)
1871 return 0;
1872 fd = open(s->path, O_RDONLY);
1873 if (fd < 0)
1874 goto err_empty;
1875 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1876 close(fd);
1877 s->should_munmap = 1;
1880 * Convert from working tree format to canonical git format
1882 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1883 size_t size = 0;
1884 munmap(s->data, s->size);
1885 s->should_munmap = 0;
1886 s->data = strbuf_detach(&buf, &size);
1887 s->size = size;
1888 s->should_free = 1;
1891 else {
1892 enum object_type type;
1893 if (size_only)
1894 type = sha1_object_info(s->sha1, &s->size);
1895 else {
1896 s->data = read_sha1_file(s->sha1, &type, &s->size);
1897 s->should_free = 1;
1900 return 0;
1903 void diff_free_filespec_blob(struct diff_filespec *s)
1905 if (s->should_free)
1906 free(s->data);
1907 else if (s->should_munmap)
1908 munmap(s->data, s->size);
1910 if (s->should_free || s->should_munmap) {
1911 s->should_free = s->should_munmap = 0;
1912 s->data = NULL;
1916 void diff_free_filespec_data(struct diff_filespec *s)
1918 diff_free_filespec_blob(s);
1919 free(s->cnt_data);
1920 s->cnt_data = NULL;
1923 static void prep_temp_blob(struct diff_tempfile *temp,
1924 void *blob,
1925 unsigned long size,
1926 const unsigned char *sha1,
1927 int mode)
1929 int fd;
1931 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1932 if (fd < 0)
1933 die("unable to create temp-file: %s", strerror(errno));
1934 if (write_in_full(fd, blob, size) != size)
1935 die("unable to write temp-file");
1936 close(fd);
1937 temp->name = temp->tmp_path;
1938 strcpy(temp->hex, sha1_to_hex(sha1));
1939 temp->hex[40] = 0;
1940 sprintf(temp->mode, "%06o", mode);
1943 static void prepare_temp_file(const char *name,
1944 struct diff_tempfile *temp,
1945 struct diff_filespec *one)
1947 if (!DIFF_FILE_VALID(one)) {
1948 not_a_valid_file:
1949 /* A '-' entry produces this for file-2, and
1950 * a '+' entry produces this for file-1.
1952 temp->name = "/dev/null";
1953 strcpy(temp->hex, ".");
1954 strcpy(temp->mode, ".");
1955 return;
1958 if (!one->sha1_valid ||
1959 reuse_worktree_file(name, one->sha1, 1)) {
1960 struct stat st;
1961 if (lstat(name, &st) < 0) {
1962 if (errno == ENOENT)
1963 goto not_a_valid_file;
1964 die("stat(%s): %s", name, strerror(errno));
1966 if (S_ISLNK(st.st_mode)) {
1967 int ret;
1968 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1969 ret = readlink(name, buf, sizeof(buf));
1970 if (ret < 0)
1971 die("readlink(%s)", name);
1972 if (ret == sizeof(buf))
1973 die("symlink too long: %s", name);
1974 prep_temp_blob(temp, buf, ret,
1975 (one->sha1_valid ?
1976 one->sha1 : null_sha1),
1977 (one->sha1_valid ?
1978 one->mode : S_IFLNK));
1980 else {
1981 /* we can borrow from the file in the work tree */
1982 temp->name = name;
1983 if (!one->sha1_valid)
1984 strcpy(temp->hex, sha1_to_hex(null_sha1));
1985 else
1986 strcpy(temp->hex, sha1_to_hex(one->sha1));
1987 /* Even though we may sometimes borrow the
1988 * contents from the work tree, we always want
1989 * one->mode. mode is trustworthy even when
1990 * !(one->sha1_valid), as long as
1991 * DIFF_FILE_VALID(one).
1993 sprintf(temp->mode, "%06o", one->mode);
1995 return;
1997 else {
1998 if (diff_populate_filespec(one, 0))
1999 die("cannot read data blob for %s", one->path);
2000 prep_temp_blob(temp, one->data, one->size,
2001 one->sha1, one->mode);
2005 static void remove_tempfile(void)
2007 int i;
2009 for (i = 0; i < 2; i++)
2010 if (diff_temp[i].name == diff_temp[i].tmp_path) {
2011 unlink(diff_temp[i].name);
2012 diff_temp[i].name = NULL;
2016 static void remove_tempfile_on_signal(int signo)
2018 remove_tempfile();
2019 signal(SIGINT, SIG_DFL);
2020 raise(signo);
2023 /* An external diff command takes:
2025 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2026 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2029 static void run_external_diff(const char *pgm,
2030 const char *name,
2031 const char *other,
2032 struct diff_filespec *one,
2033 struct diff_filespec *two,
2034 const char *xfrm_msg,
2035 int complete_rewrite)
2037 const char *spawn_arg[10];
2038 struct diff_tempfile *temp = diff_temp;
2039 int retval;
2040 static int atexit_asked = 0;
2041 const char *othername;
2042 const char **arg = &spawn_arg[0];
2044 othername = (other? other : name);
2045 if (one && two) {
2046 prepare_temp_file(name, &temp[0], one);
2047 prepare_temp_file(othername, &temp[1], two);
2048 if (! atexit_asked &&
2049 (temp[0].name == temp[0].tmp_path ||
2050 temp[1].name == temp[1].tmp_path)) {
2051 atexit_asked = 1;
2052 atexit(remove_tempfile);
2054 signal(SIGINT, remove_tempfile_on_signal);
2057 if (one && two) {
2058 *arg++ = pgm;
2059 *arg++ = name;
2060 *arg++ = temp[0].name;
2061 *arg++ = temp[0].hex;
2062 *arg++ = temp[0].mode;
2063 *arg++ = temp[1].name;
2064 *arg++ = temp[1].hex;
2065 *arg++ = temp[1].mode;
2066 if (other) {
2067 *arg++ = other;
2068 *arg++ = xfrm_msg;
2070 } else {
2071 *arg++ = pgm;
2072 *arg++ = name;
2074 *arg = NULL;
2075 fflush(NULL);
2076 retval = run_command_v_opt(spawn_arg, 0);
2077 remove_tempfile();
2078 if (retval) {
2079 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2080 exit(1);
2084 static void run_diff_cmd(const char *pgm,
2085 const char *name,
2086 const char *other,
2087 const char *attr_path,
2088 struct diff_filespec *one,
2089 struct diff_filespec *two,
2090 const char *xfrm_msg,
2091 struct diff_options *o,
2092 int complete_rewrite)
2094 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2095 pgm = NULL;
2096 else {
2097 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2098 if (drv && drv->external)
2099 pgm = drv->external;
2102 if (pgm) {
2103 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2104 complete_rewrite);
2105 return;
2107 if (one && two)
2108 builtin_diff(name, other ? other : name,
2109 one, two, xfrm_msg, o, complete_rewrite);
2110 else
2111 fprintf(o->file, "* Unmerged path %s\n", name);
2114 static void diff_fill_sha1_info(struct diff_filespec *one)
2116 if (DIFF_FILE_VALID(one)) {
2117 if (!one->sha1_valid) {
2118 struct stat st;
2119 if (!strcmp(one->path, "-")) {
2120 hashcpy(one->sha1, null_sha1);
2121 return;
2123 if (lstat(one->path, &st) < 0)
2124 die("stat %s", one->path);
2125 if (index_path(one->sha1, one->path, &st, 0))
2126 die("cannot hash %s", one->path);
2129 else
2130 hashclr(one->sha1);
2133 static int similarity_index(struct diff_filepair *p)
2135 return p->score * 100 / MAX_SCORE;
2138 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2140 /* Strip the prefix but do not molest /dev/null and absolute paths */
2141 if (*namep && **namep != '/')
2142 *namep += prefix_length;
2143 if (*otherp && **otherp != '/')
2144 *otherp += prefix_length;
2147 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2149 const char *pgm = external_diff();
2150 struct strbuf msg;
2151 char *xfrm_msg;
2152 struct diff_filespec *one = p->one;
2153 struct diff_filespec *two = p->two;
2154 const char *name;
2155 const char *other;
2156 const char *attr_path;
2157 int complete_rewrite = 0;
2159 name = p->one->path;
2160 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2161 attr_path = name;
2162 if (o->prefix_length)
2163 strip_prefix(o->prefix_length, &name, &other);
2165 if (DIFF_PAIR_UNMERGED(p)) {
2166 run_diff_cmd(pgm, name, NULL, attr_path,
2167 NULL, NULL, NULL, o, 0);
2168 return;
2171 diff_fill_sha1_info(one);
2172 diff_fill_sha1_info(two);
2174 strbuf_init(&msg, PATH_MAX * 2 + 300);
2175 switch (p->status) {
2176 case DIFF_STATUS_COPIED:
2177 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2178 strbuf_addstr(&msg, "\ncopy from ");
2179 quote_c_style(name, &msg, NULL, 0);
2180 strbuf_addstr(&msg, "\ncopy to ");
2181 quote_c_style(other, &msg, NULL, 0);
2182 strbuf_addch(&msg, '\n');
2183 break;
2184 case DIFF_STATUS_RENAMED:
2185 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2186 strbuf_addstr(&msg, "\nrename from ");
2187 quote_c_style(name, &msg, NULL, 0);
2188 strbuf_addstr(&msg, "\nrename to ");
2189 quote_c_style(other, &msg, NULL, 0);
2190 strbuf_addch(&msg, '\n');
2191 break;
2192 case DIFF_STATUS_MODIFIED:
2193 if (p->score) {
2194 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2195 similarity_index(p));
2196 complete_rewrite = 1;
2197 break;
2199 /* fallthru */
2200 default:
2201 /* nothing */
2205 if (hashcmp(one->sha1, two->sha1)) {
2206 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2208 if (DIFF_OPT_TST(o, BINARY)) {
2209 mmfile_t mf;
2210 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2211 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2212 abbrev = 40;
2214 strbuf_addf(&msg, "index %.*s..%.*s",
2215 abbrev, sha1_to_hex(one->sha1),
2216 abbrev, sha1_to_hex(two->sha1));
2217 if (one->mode == two->mode)
2218 strbuf_addf(&msg, " %06o", one->mode);
2219 strbuf_addch(&msg, '\n');
2222 if (msg.len)
2223 strbuf_setlen(&msg, msg.len - 1);
2224 xfrm_msg = msg.len ? msg.buf : NULL;
2226 if (!pgm &&
2227 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2228 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2229 /* a filepair that changes between file and symlink
2230 * needs to be split into deletion and creation.
2232 struct diff_filespec *null = alloc_filespec(two->path);
2233 run_diff_cmd(NULL, name, other, attr_path,
2234 one, null, xfrm_msg, o, 0);
2235 free(null);
2236 null = alloc_filespec(one->path);
2237 run_diff_cmd(NULL, name, other, attr_path,
2238 null, two, xfrm_msg, o, 0);
2239 free(null);
2241 else
2242 run_diff_cmd(pgm, name, other, attr_path,
2243 one, two, xfrm_msg, o, complete_rewrite);
2245 strbuf_release(&msg);
2248 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2249 struct diffstat_t *diffstat)
2251 const char *name;
2252 const char *other;
2253 int complete_rewrite = 0;
2255 if (DIFF_PAIR_UNMERGED(p)) {
2256 /* unmerged */
2257 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2258 return;
2261 name = p->one->path;
2262 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2264 if (o->prefix_length)
2265 strip_prefix(o->prefix_length, &name, &other);
2267 diff_fill_sha1_info(p->one);
2268 diff_fill_sha1_info(p->two);
2270 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2271 complete_rewrite = 1;
2272 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2275 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2277 const char *name;
2278 const char *other;
2279 const char *attr_path;
2281 if (DIFF_PAIR_UNMERGED(p)) {
2282 /* unmerged */
2283 return;
2286 name = p->one->path;
2287 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2288 attr_path = other ? other : name;
2290 if (o->prefix_length)
2291 strip_prefix(o->prefix_length, &name, &other);
2293 diff_fill_sha1_info(p->one);
2294 diff_fill_sha1_info(p->two);
2296 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2299 void diff_setup(struct diff_options *options)
2301 memset(options, 0, sizeof(*options));
2303 options->file = stdout;
2305 options->line_termination = '\n';
2306 options->break_opt = -1;
2307 options->rename_limit = -1;
2308 options->dirstat_percent = 3;
2309 DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2310 options->context = 3;
2312 options->change = diff_change;
2313 options->add_remove = diff_addremove;
2314 if (diff_use_color_default > 0)
2315 DIFF_OPT_SET(options, COLOR_DIFF);
2316 else
2317 DIFF_OPT_CLR(options, COLOR_DIFF);
2318 options->detect_rename = diff_detect_rename_default;
2320 if (!diff_mnemonic_prefix) {
2321 options->a_prefix = "a/";
2322 options->b_prefix = "b/";
2326 int diff_setup_done(struct diff_options *options)
2328 int count = 0;
2330 if (options->output_format & DIFF_FORMAT_NAME)
2331 count++;
2332 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2333 count++;
2334 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2335 count++;
2336 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2337 count++;
2338 if (count > 1)
2339 die("--name-only, --name-status, --check and -s are mutually exclusive");
2341 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2342 options->detect_rename = DIFF_DETECT_COPY;
2344 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2345 options->prefix = NULL;
2346 if (options->prefix)
2347 options->prefix_length = strlen(options->prefix);
2348 else
2349 options->prefix_length = 0;
2351 if (options->output_format & (DIFF_FORMAT_NAME |
2352 DIFF_FORMAT_NAME_STATUS |
2353 DIFF_FORMAT_CHECKDIFF |
2354 DIFF_FORMAT_NO_OUTPUT))
2355 options->output_format &= ~(DIFF_FORMAT_RAW |
2356 DIFF_FORMAT_NUMSTAT |
2357 DIFF_FORMAT_DIFFSTAT |
2358 DIFF_FORMAT_SHORTSTAT |
2359 DIFF_FORMAT_DIRSTAT |
2360 DIFF_FORMAT_SUMMARY |
2361 DIFF_FORMAT_PATCH);
2364 * These cases always need recursive; we do not drop caller-supplied
2365 * recursive bits for other formats here.
2367 if (options->output_format & (DIFF_FORMAT_PATCH |
2368 DIFF_FORMAT_NUMSTAT |
2369 DIFF_FORMAT_DIFFSTAT |
2370 DIFF_FORMAT_SHORTSTAT |
2371 DIFF_FORMAT_DIRSTAT |
2372 DIFF_FORMAT_SUMMARY |
2373 DIFF_FORMAT_CHECKDIFF))
2374 DIFF_OPT_SET(options, RECURSIVE);
2376 * Also pickaxe would not work very well if you do not say recursive
2378 if (options->pickaxe)
2379 DIFF_OPT_SET(options, RECURSIVE);
2381 if (options->detect_rename && options->rename_limit < 0)
2382 options->rename_limit = diff_rename_limit_default;
2383 if (options->setup & DIFF_SETUP_USE_CACHE) {
2384 if (!active_cache)
2385 /* read-cache does not die even when it fails
2386 * so it is safe for us to do this here. Also
2387 * it does not smudge active_cache or active_nr
2388 * when it fails, so we do not have to worry about
2389 * cleaning it up ourselves either.
2391 read_cache();
2393 if (options->abbrev <= 0 || 40 < options->abbrev)
2394 options->abbrev = 40; /* full */
2397 * It does not make sense to show the first hit we happened
2398 * to have found. It does not make sense not to return with
2399 * exit code in such a case either.
2401 if (DIFF_OPT_TST(options, QUIET)) {
2402 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2403 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2406 return 0;
2409 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2411 char c, *eq;
2412 int len;
2414 if (*arg != '-')
2415 return 0;
2416 c = *++arg;
2417 if (!c)
2418 return 0;
2419 if (c == arg_short) {
2420 c = *++arg;
2421 if (!c)
2422 return 1;
2423 if (val && isdigit(c)) {
2424 char *end;
2425 int n = strtoul(arg, &end, 10);
2426 if (*end)
2427 return 0;
2428 *val = n;
2429 return 1;
2431 return 0;
2433 if (c != '-')
2434 return 0;
2435 arg++;
2436 eq = strchr(arg, '=');
2437 if (eq)
2438 len = eq - arg;
2439 else
2440 len = strlen(arg);
2441 if (!len || strncmp(arg, arg_long, len))
2442 return 0;
2443 if (eq) {
2444 int n;
2445 char *end;
2446 if (!isdigit(*++eq))
2447 return 0;
2448 n = strtoul(eq, &end, 10);
2449 if (*end)
2450 return 0;
2451 *val = n;
2453 return 1;
2456 static int diff_scoreopt_parse(const char *opt);
2458 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2460 const char *arg = av[0];
2462 /* Output format options */
2463 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2464 options->output_format |= DIFF_FORMAT_PATCH;
2465 else if (opt_arg(arg, 'U', "unified", &options->context))
2466 options->output_format |= DIFF_FORMAT_PATCH;
2467 else if (!strcmp(arg, "--raw"))
2468 options->output_format |= DIFF_FORMAT_RAW;
2469 else if (!strcmp(arg, "--patch-with-raw"))
2470 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2471 else if (!strcmp(arg, "--numstat"))
2472 options->output_format |= DIFF_FORMAT_NUMSTAT;
2473 else if (!strcmp(arg, "--shortstat"))
2474 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2475 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2476 options->output_format |= DIFF_FORMAT_DIRSTAT;
2477 else if (!strcmp(arg, "--cumulative")) {
2478 options->output_format |= DIFF_FORMAT_DIRSTAT;
2479 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2480 } else if (opt_arg(arg, 0, "dirstat-by-file",
2481 &options->dirstat_percent)) {
2482 options->output_format |= DIFF_FORMAT_DIRSTAT;
2483 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2485 else if (!strcmp(arg, "--check"))
2486 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2487 else if (!strcmp(arg, "--summary"))
2488 options->output_format |= DIFF_FORMAT_SUMMARY;
2489 else if (!strcmp(arg, "--patch-with-stat"))
2490 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2491 else if (!strcmp(arg, "--name-only"))
2492 options->output_format |= DIFF_FORMAT_NAME;
2493 else if (!strcmp(arg, "--name-status"))
2494 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2495 else if (!strcmp(arg, "-s"))
2496 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2497 else if (!prefixcmp(arg, "--stat")) {
2498 char *end;
2499 int width = options->stat_width;
2500 int name_width = options->stat_name_width;
2501 arg += 6;
2502 end = (char *)arg;
2504 switch (*arg) {
2505 case '-':
2506 if (!prefixcmp(arg, "-width="))
2507 width = strtoul(arg + 7, &end, 10);
2508 else if (!prefixcmp(arg, "-name-width="))
2509 name_width = strtoul(arg + 12, &end, 10);
2510 break;
2511 case '=':
2512 width = strtoul(arg+1, &end, 10);
2513 if (*end == ',')
2514 name_width = strtoul(end+1, &end, 10);
2517 /* Important! This checks all the error cases! */
2518 if (*end)
2519 return 0;
2520 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2521 options->stat_name_width = name_width;
2522 options->stat_width = width;
2525 /* renames options */
2526 else if (!prefixcmp(arg, "-B")) {
2527 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2528 return -1;
2530 else if (!prefixcmp(arg, "-M")) {
2531 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2532 return -1;
2533 options->detect_rename = DIFF_DETECT_RENAME;
2535 else if (!prefixcmp(arg, "-C")) {
2536 if (options->detect_rename == DIFF_DETECT_COPY)
2537 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2538 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2539 return -1;
2540 options->detect_rename = DIFF_DETECT_COPY;
2542 else if (!strcmp(arg, "--no-renames"))
2543 options->detect_rename = 0;
2544 else if (!strcmp(arg, "--relative"))
2545 DIFF_OPT_SET(options, RELATIVE_NAME);
2546 else if (!prefixcmp(arg, "--relative=")) {
2547 DIFF_OPT_SET(options, RELATIVE_NAME);
2548 options->prefix = arg + 11;
2551 /* xdiff options */
2552 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2553 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2554 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2555 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2556 else if (!strcmp(arg, "--ignore-space-at-eol"))
2557 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2558 else if (!strcmp(arg, "--patience"))
2559 options->xdl_opts |= XDF_PATIENCE_DIFF;
2561 /* flags options */
2562 else if (!strcmp(arg, "--binary")) {
2563 options->output_format |= DIFF_FORMAT_PATCH;
2564 DIFF_OPT_SET(options, BINARY);
2566 else if (!strcmp(arg, "--full-index"))
2567 DIFF_OPT_SET(options, FULL_INDEX);
2568 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2569 DIFF_OPT_SET(options, TEXT);
2570 else if (!strcmp(arg, "-R"))
2571 DIFF_OPT_SET(options, REVERSE_DIFF);
2572 else if (!strcmp(arg, "--find-copies-harder"))
2573 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2574 else if (!strcmp(arg, "--follow"))
2575 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2576 else if (!strcmp(arg, "--color"))
2577 DIFF_OPT_SET(options, COLOR_DIFF);
2578 else if (!strcmp(arg, "--no-color"))
2579 DIFF_OPT_CLR(options, COLOR_DIFF);
2580 else if (!strcmp(arg, "--color-words"))
2581 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2582 else if (!prefixcmp(arg, "--color-words=")) {
2583 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2584 options->word_regex = arg + 14;
2586 else if (!strcmp(arg, "--exit-code"))
2587 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2588 else if (!strcmp(arg, "--quiet"))
2589 DIFF_OPT_SET(options, QUIET);
2590 else if (!strcmp(arg, "--ext-diff"))
2591 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2592 else if (!strcmp(arg, "--no-ext-diff"))
2593 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2594 else if (!strcmp(arg, "--textconv"))
2595 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2596 else if (!strcmp(arg, "--no-textconv"))
2597 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2598 else if (!strcmp(arg, "--ignore-submodules"))
2599 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2601 /* misc options */
2602 else if (!strcmp(arg, "-z"))
2603 options->line_termination = 0;
2604 else if (!prefixcmp(arg, "-l"))
2605 options->rename_limit = strtoul(arg+2, NULL, 10);
2606 else if (!prefixcmp(arg, "-S"))
2607 options->pickaxe = arg + 2;
2608 else if (!strcmp(arg, "--pickaxe-all"))
2609 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2610 else if (!strcmp(arg, "--pickaxe-regex"))
2611 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2612 else if (!prefixcmp(arg, "-O"))
2613 options->orderfile = arg + 2;
2614 else if (!prefixcmp(arg, "--diff-filter="))
2615 options->filter = arg + 14;
2616 else if (!strcmp(arg, "--abbrev"))
2617 options->abbrev = DEFAULT_ABBREV;
2618 else if (!prefixcmp(arg, "--abbrev=")) {
2619 options->abbrev = strtoul(arg + 9, NULL, 10);
2620 if (options->abbrev < MINIMUM_ABBREV)
2621 options->abbrev = MINIMUM_ABBREV;
2622 else if (40 < options->abbrev)
2623 options->abbrev = 40;
2625 else if (!prefixcmp(arg, "--src-prefix="))
2626 options->a_prefix = arg + 13;
2627 else if (!prefixcmp(arg, "--dst-prefix="))
2628 options->b_prefix = arg + 13;
2629 else if (!strcmp(arg, "--no-prefix"))
2630 options->a_prefix = options->b_prefix = "";
2631 else if (opt_arg(arg, '\0', "inter-hunk-context",
2632 &options->interhunkcontext))
2634 else if (!prefixcmp(arg, "--output=")) {
2635 options->file = fopen(arg + strlen("--output="), "w");
2636 options->close_file = 1;
2637 } else
2638 return 0;
2639 return 1;
2642 static int parse_num(const char **cp_p)
2644 unsigned long num, scale;
2645 int ch, dot;
2646 const char *cp = *cp_p;
2648 num = 0;
2649 scale = 1;
2650 dot = 0;
2651 for(;;) {
2652 ch = *cp;
2653 if ( !dot && ch == '.' ) {
2654 scale = 1;
2655 dot = 1;
2656 } else if ( ch == '%' ) {
2657 scale = dot ? scale*100 : 100;
2658 cp++; /* % is always at the end */
2659 break;
2660 } else if ( ch >= '0' && ch <= '9' ) {
2661 if ( scale < 100000 ) {
2662 scale *= 10;
2663 num = (num*10) + (ch-'0');
2665 } else {
2666 break;
2668 cp++;
2670 *cp_p = cp;
2672 /* user says num divided by scale and we say internally that
2673 * is MAX_SCORE * num / scale.
2675 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2678 static int diff_scoreopt_parse(const char *opt)
2680 int opt1, opt2, cmd;
2682 if (*opt++ != '-')
2683 return -1;
2684 cmd = *opt++;
2685 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2686 return -1; /* that is not a -M, -C nor -B option */
2688 opt1 = parse_num(&opt);
2689 if (cmd != 'B')
2690 opt2 = 0;
2691 else {
2692 if (*opt == 0)
2693 opt2 = 0;
2694 else if (*opt != '/')
2695 return -1; /* we expect -B80/99 or -B80 */
2696 else {
2697 opt++;
2698 opt2 = parse_num(&opt);
2701 if (*opt != 0)
2702 return -1;
2703 return opt1 | (opt2 << 16);
2706 struct diff_queue_struct diff_queued_diff;
2708 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2710 if (queue->alloc <= queue->nr) {
2711 queue->alloc = alloc_nr(queue->alloc);
2712 queue->queue = xrealloc(queue->queue,
2713 sizeof(dp) * queue->alloc);
2715 queue->queue[queue->nr++] = dp;
2718 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2719 struct diff_filespec *one,
2720 struct diff_filespec *two)
2722 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2723 dp->one = one;
2724 dp->two = two;
2725 if (queue)
2726 diff_q(queue, dp);
2727 return dp;
2730 void diff_free_filepair(struct diff_filepair *p)
2732 free_filespec(p->one);
2733 free_filespec(p->two);
2734 free(p);
2737 /* This is different from find_unique_abbrev() in that
2738 * it stuffs the result with dots for alignment.
2740 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2742 int abblen;
2743 const char *abbrev;
2744 if (len == 40)
2745 return sha1_to_hex(sha1);
2747 abbrev = find_unique_abbrev(sha1, len);
2748 abblen = strlen(abbrev);
2749 if (abblen < 37) {
2750 static char hex[41];
2751 if (len < abblen && abblen <= len + 2)
2752 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2753 else
2754 sprintf(hex, "%s...", abbrev);
2755 return hex;
2757 return sha1_to_hex(sha1);
2760 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2762 int line_termination = opt->line_termination;
2763 int inter_name_termination = line_termination ? '\t' : '\0';
2765 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2766 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2767 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2768 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2770 if (p->score) {
2771 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2772 inter_name_termination);
2773 } else {
2774 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2777 if (p->status == DIFF_STATUS_COPIED ||
2778 p->status == DIFF_STATUS_RENAMED) {
2779 const char *name_a, *name_b;
2780 name_a = p->one->path;
2781 name_b = p->two->path;
2782 strip_prefix(opt->prefix_length, &name_a, &name_b);
2783 write_name_quoted(name_a, opt->file, inter_name_termination);
2784 write_name_quoted(name_b, opt->file, line_termination);
2785 } else {
2786 const char *name_a, *name_b;
2787 name_a = p->one->mode ? p->one->path : p->two->path;
2788 name_b = NULL;
2789 strip_prefix(opt->prefix_length, &name_a, &name_b);
2790 write_name_quoted(name_a, opt->file, line_termination);
2794 int diff_unmodified_pair(struct diff_filepair *p)
2796 /* This function is written stricter than necessary to support
2797 * the currently implemented transformers, but the idea is to
2798 * let transformers to produce diff_filepairs any way they want,
2799 * and filter and clean them up here before producing the output.
2801 struct diff_filespec *one = p->one, *two = p->two;
2803 if (DIFF_PAIR_UNMERGED(p))
2804 return 0; /* unmerged is interesting */
2806 /* deletion, addition, mode or type change
2807 * and rename are all interesting.
2809 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2810 DIFF_PAIR_MODE_CHANGED(p) ||
2811 strcmp(one->path, two->path))
2812 return 0;
2814 /* both are valid and point at the same path. that is, we are
2815 * dealing with a change.
2817 if (one->sha1_valid && two->sha1_valid &&
2818 !hashcmp(one->sha1, two->sha1))
2819 return 1; /* no change */
2820 if (!one->sha1_valid && !two->sha1_valid)
2821 return 1; /* both look at the same file on the filesystem. */
2822 return 0;
2825 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2827 if (diff_unmodified_pair(p))
2828 return;
2830 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2831 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2832 return; /* no tree diffs in patch format */
2834 run_diff(p, o);
2837 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2838 struct diffstat_t *diffstat)
2840 if (diff_unmodified_pair(p))
2841 return;
2843 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2844 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2845 return; /* no tree diffs in patch format */
2847 run_diffstat(p, o, diffstat);
2850 static void diff_flush_checkdiff(struct diff_filepair *p,
2851 struct diff_options *o)
2853 if (diff_unmodified_pair(p))
2854 return;
2856 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2857 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2858 return; /* no tree diffs in patch format */
2860 run_checkdiff(p, o);
2863 int diff_queue_is_empty(void)
2865 struct diff_queue_struct *q = &diff_queued_diff;
2866 int i;
2867 for (i = 0; i < q->nr; i++)
2868 if (!diff_unmodified_pair(q->queue[i]))
2869 return 0;
2870 return 1;
2873 #if DIFF_DEBUG
2874 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2876 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2877 x, one ? one : "",
2878 s->path,
2879 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2880 s->mode,
2881 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2882 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2883 x, one ? one : "",
2884 s->size, s->xfrm_flags);
2887 void diff_debug_filepair(const struct diff_filepair *p, int i)
2889 diff_debug_filespec(p->one, i, "one");
2890 diff_debug_filespec(p->two, i, "two");
2891 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2892 p->score, p->status ? p->status : '?',
2893 p->one->rename_used, p->broken_pair);
2896 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2898 int i;
2899 if (msg)
2900 fprintf(stderr, "%s\n", msg);
2901 fprintf(stderr, "q->nr = %d\n", q->nr);
2902 for (i = 0; i < q->nr; i++) {
2903 struct diff_filepair *p = q->queue[i];
2904 diff_debug_filepair(p, i);
2907 #endif
2909 static void diff_resolve_rename_copy(void)
2911 int i;
2912 struct diff_filepair *p;
2913 struct diff_queue_struct *q = &diff_queued_diff;
2915 diff_debug_queue("resolve-rename-copy", q);
2917 for (i = 0; i < q->nr; i++) {
2918 p = q->queue[i];
2919 p->status = 0; /* undecided */
2920 if (DIFF_PAIR_UNMERGED(p))
2921 p->status = DIFF_STATUS_UNMERGED;
2922 else if (!DIFF_FILE_VALID(p->one))
2923 p->status = DIFF_STATUS_ADDED;
2924 else if (!DIFF_FILE_VALID(p->two))
2925 p->status = DIFF_STATUS_DELETED;
2926 else if (DIFF_PAIR_TYPE_CHANGED(p))
2927 p->status = DIFF_STATUS_TYPE_CHANGED;
2929 /* from this point on, we are dealing with a pair
2930 * whose both sides are valid and of the same type, i.e.
2931 * either in-place edit or rename/copy edit.
2933 else if (DIFF_PAIR_RENAME(p)) {
2935 * A rename might have re-connected a broken
2936 * pair up, causing the pathnames to be the
2937 * same again. If so, that's not a rename at
2938 * all, just a modification..
2940 * Otherwise, see if this source was used for
2941 * multiple renames, in which case we decrement
2942 * the count, and call it a copy.
2944 if (!strcmp(p->one->path, p->two->path))
2945 p->status = DIFF_STATUS_MODIFIED;
2946 else if (--p->one->rename_used > 0)
2947 p->status = DIFF_STATUS_COPIED;
2948 else
2949 p->status = DIFF_STATUS_RENAMED;
2951 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2952 p->one->mode != p->two->mode ||
2953 is_null_sha1(p->one->sha1))
2954 p->status = DIFF_STATUS_MODIFIED;
2955 else {
2956 /* This is a "no-change" entry and should not
2957 * happen anymore, but prepare for broken callers.
2959 error("feeding unmodified %s to diffcore",
2960 p->one->path);
2961 p->status = DIFF_STATUS_UNKNOWN;
2964 diff_debug_queue("resolve-rename-copy done", q);
2967 static int check_pair_status(struct diff_filepair *p)
2969 switch (p->status) {
2970 case DIFF_STATUS_UNKNOWN:
2971 return 0;
2972 case 0:
2973 die("internal error in diff-resolve-rename-copy");
2974 default:
2975 return 1;
2979 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2981 int fmt = opt->output_format;
2983 if (fmt & DIFF_FORMAT_CHECKDIFF)
2984 diff_flush_checkdiff(p, opt);
2985 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2986 diff_flush_raw(p, opt);
2987 else if (fmt & DIFF_FORMAT_NAME) {
2988 const char *name_a, *name_b;
2989 name_a = p->two->path;
2990 name_b = NULL;
2991 strip_prefix(opt->prefix_length, &name_a, &name_b);
2992 write_name_quoted(name_a, opt->file, opt->line_termination);
2996 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2998 if (fs->mode)
2999 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3000 else
3001 fprintf(file, " %s ", newdelete);
3002 write_name_quoted(fs->path, file, '\n');
3006 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3008 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3009 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3010 show_name ? ' ' : '\n');
3011 if (show_name) {
3012 write_name_quoted(p->two->path, file, '\n');
3017 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3019 char *names = pprint_rename(p->one->path, p->two->path);
3021 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3022 free(names);
3023 show_mode_change(file, p, 0);
3026 static void diff_summary(FILE *file, struct diff_filepair *p)
3028 switch(p->status) {
3029 case DIFF_STATUS_DELETED:
3030 show_file_mode_name(file, "delete", p->one);
3031 break;
3032 case DIFF_STATUS_ADDED:
3033 show_file_mode_name(file, "create", p->two);
3034 break;
3035 case DIFF_STATUS_COPIED:
3036 show_rename_copy(file, "copy", p);
3037 break;
3038 case DIFF_STATUS_RENAMED:
3039 show_rename_copy(file, "rename", p);
3040 break;
3041 default:
3042 if (p->score) {
3043 fputs(" rewrite ", file);
3044 write_name_quoted(p->two->path, file, ' ');
3045 fprintf(file, "(%d%%)\n", similarity_index(p));
3047 show_mode_change(file, p, !p->score);
3048 break;
3052 struct patch_id_t {
3053 git_SHA_CTX *ctx;
3054 int patchlen;
3057 static int remove_space(char *line, int len)
3059 int i;
3060 char *dst = line;
3061 unsigned char c;
3063 for (i = 0; i < len; i++)
3064 if (!isspace((c = line[i])))
3065 *dst++ = c;
3067 return dst - line;
3070 static void patch_id_consume(void *priv, char *line, unsigned long len)
3072 struct patch_id_t *data = priv;
3073 int new_len;
3075 /* Ignore line numbers when computing the SHA1 of the patch */
3076 if (!prefixcmp(line, "@@ -"))
3077 return;
3079 new_len = remove_space(line, len);
3081 git_SHA1_Update(data->ctx, line, new_len);
3082 data->patchlen += new_len;
3085 /* returns 0 upon success, and writes result into sha1 */
3086 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3088 struct diff_queue_struct *q = &diff_queued_diff;
3089 int i;
3090 git_SHA_CTX ctx;
3091 struct patch_id_t data;
3092 char buffer[PATH_MAX * 4 + 20];
3094 git_SHA1_Init(&ctx);
3095 memset(&data, 0, sizeof(struct patch_id_t));
3096 data.ctx = &ctx;
3098 for (i = 0; i < q->nr; i++) {
3099 xpparam_t xpp;
3100 xdemitconf_t xecfg;
3101 xdemitcb_t ecb;
3102 mmfile_t mf1, mf2;
3103 struct diff_filepair *p = q->queue[i];
3104 int len1, len2;
3106 memset(&xpp, 0, sizeof(xpp));
3107 memset(&xecfg, 0, sizeof(xecfg));
3108 if (p->status == 0)
3109 return error("internal diff status error");
3110 if (p->status == DIFF_STATUS_UNKNOWN)
3111 continue;
3112 if (diff_unmodified_pair(p))
3113 continue;
3114 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3115 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3116 continue;
3117 if (DIFF_PAIR_UNMERGED(p))
3118 continue;
3120 diff_fill_sha1_info(p->one);
3121 diff_fill_sha1_info(p->two);
3122 if (fill_mmfile(&mf1, p->one) < 0 ||
3123 fill_mmfile(&mf2, p->two) < 0)
3124 return error("unable to read files to diff");
3126 len1 = remove_space(p->one->path, strlen(p->one->path));
3127 len2 = remove_space(p->two->path, strlen(p->two->path));
3128 if (p->one->mode == 0)
3129 len1 = snprintf(buffer, sizeof(buffer),
3130 "diff--gita/%.*sb/%.*s"
3131 "newfilemode%06o"
3132 "---/dev/null"
3133 "+++b/%.*s",
3134 len1, p->one->path,
3135 len2, p->two->path,
3136 p->two->mode,
3137 len2, p->two->path);
3138 else if (p->two->mode == 0)
3139 len1 = snprintf(buffer, sizeof(buffer),
3140 "diff--gita/%.*sb/%.*s"
3141 "deletedfilemode%06o"
3142 "---a/%.*s"
3143 "+++/dev/null",
3144 len1, p->one->path,
3145 len2, p->two->path,
3146 p->one->mode,
3147 len1, p->one->path);
3148 else
3149 len1 = snprintf(buffer, sizeof(buffer),
3150 "diff--gita/%.*sb/%.*s"
3151 "---a/%.*s"
3152 "+++b/%.*s",
3153 len1, p->one->path,
3154 len2, p->two->path,
3155 len1, p->one->path,
3156 len2, p->two->path);
3157 git_SHA1_Update(&ctx, buffer, len1);
3159 xpp.flags = XDF_NEED_MINIMAL;
3160 xecfg.ctxlen = 3;
3161 xecfg.flags = XDL_EMIT_FUNCNAMES;
3162 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3163 &xpp, &xecfg, &ecb);
3166 git_SHA1_Final(sha1, &ctx);
3167 return 0;
3170 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3172 struct diff_queue_struct *q = &diff_queued_diff;
3173 int i;
3174 int result = diff_get_patch_id(options, sha1);
3176 for (i = 0; i < q->nr; i++)
3177 diff_free_filepair(q->queue[i]);
3179 free(q->queue);
3180 q->queue = NULL;
3181 q->nr = q->alloc = 0;
3183 return result;
3186 static int is_summary_empty(const struct diff_queue_struct *q)
3188 int i;
3190 for (i = 0; i < q->nr; i++) {
3191 const struct diff_filepair *p = q->queue[i];
3193 switch (p->status) {
3194 case DIFF_STATUS_DELETED:
3195 case DIFF_STATUS_ADDED:
3196 case DIFF_STATUS_COPIED:
3197 case DIFF_STATUS_RENAMED:
3198 return 0;
3199 default:
3200 if (p->score)
3201 return 0;
3202 if (p->one->mode && p->two->mode &&
3203 p->one->mode != p->two->mode)
3204 return 0;
3205 break;
3208 return 1;
3211 void diff_flush(struct diff_options *options)
3213 struct diff_queue_struct *q = &diff_queued_diff;
3214 int i, output_format = options->output_format;
3215 int separator = 0;
3218 * Order: raw, stat, summary, patch
3219 * or: name/name-status/checkdiff (other bits clear)
3221 if (!q->nr)
3222 goto free_queue;
3224 if (output_format & (DIFF_FORMAT_RAW |
3225 DIFF_FORMAT_NAME |
3226 DIFF_FORMAT_NAME_STATUS |
3227 DIFF_FORMAT_CHECKDIFF)) {
3228 for (i = 0; i < q->nr; i++) {
3229 struct diff_filepair *p = q->queue[i];
3230 if (check_pair_status(p))
3231 flush_one_pair(p, options);
3233 separator++;
3236 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3237 struct diffstat_t diffstat;
3239 memset(&diffstat, 0, sizeof(struct diffstat_t));
3240 for (i = 0; i < q->nr; i++) {
3241 struct diff_filepair *p = q->queue[i];
3242 if (check_pair_status(p))
3243 diff_flush_stat(p, options, &diffstat);
3245 if (output_format & DIFF_FORMAT_NUMSTAT)
3246 show_numstat(&diffstat, options);
3247 if (output_format & DIFF_FORMAT_DIFFSTAT)
3248 show_stats(&diffstat, options);
3249 if (output_format & DIFF_FORMAT_SHORTSTAT)
3250 show_shortstats(&diffstat, options);
3251 free_diffstat_info(&diffstat);
3252 separator++;
3254 if (output_format & DIFF_FORMAT_DIRSTAT)
3255 show_dirstat(options);
3257 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3258 for (i = 0; i < q->nr; i++)
3259 diff_summary(options->file, q->queue[i]);
3260 separator++;
3263 if (output_format & DIFF_FORMAT_PATCH) {
3264 if (separator) {
3265 putc(options->line_termination, options->file);
3266 if (options->stat_sep) {
3267 /* attach patch instead of inline */
3268 fputs(options->stat_sep, options->file);
3272 for (i = 0; i < q->nr; i++) {
3273 struct diff_filepair *p = q->queue[i];
3274 if (check_pair_status(p))
3275 diff_flush_patch(p, options);
3279 if (output_format & DIFF_FORMAT_CALLBACK)
3280 options->format_callback(q, options, options->format_callback_data);
3282 for (i = 0; i < q->nr; i++)
3283 diff_free_filepair(q->queue[i]);
3284 free_queue:
3285 free(q->queue);
3286 q->queue = NULL;
3287 q->nr = q->alloc = 0;
3288 if (options->close_file)
3289 fclose(options->file);
3292 static void diffcore_apply_filter(const char *filter)
3294 int i;
3295 struct diff_queue_struct *q = &diff_queued_diff;
3296 struct diff_queue_struct outq;
3297 outq.queue = NULL;
3298 outq.nr = outq.alloc = 0;
3300 if (!filter)
3301 return;
3303 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3304 int found;
3305 for (i = found = 0; !found && i < q->nr; i++) {
3306 struct diff_filepair *p = q->queue[i];
3307 if (((p->status == DIFF_STATUS_MODIFIED) &&
3308 ((p->score &&
3309 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3310 (!p->score &&
3311 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3312 ((p->status != DIFF_STATUS_MODIFIED) &&
3313 strchr(filter, p->status)))
3314 found++;
3316 if (found)
3317 return;
3319 /* otherwise we will clear the whole queue
3320 * by copying the empty outq at the end of this
3321 * function, but first clear the current entries
3322 * in the queue.
3324 for (i = 0; i < q->nr; i++)
3325 diff_free_filepair(q->queue[i]);
3327 else {
3328 /* Only the matching ones */
3329 for (i = 0; i < q->nr; i++) {
3330 struct diff_filepair *p = q->queue[i];
3332 if (((p->status == DIFF_STATUS_MODIFIED) &&
3333 ((p->score &&
3334 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3335 (!p->score &&
3336 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3337 ((p->status != DIFF_STATUS_MODIFIED) &&
3338 strchr(filter, p->status)))
3339 diff_q(&outq, p);
3340 else
3341 diff_free_filepair(p);
3344 free(q->queue);
3345 *q = outq;
3348 /* Check whether two filespecs with the same mode and size are identical */
3349 static int diff_filespec_is_identical(struct diff_filespec *one,
3350 struct diff_filespec *two)
3352 if (S_ISGITLINK(one->mode))
3353 return 0;
3354 if (diff_populate_filespec(one, 0))
3355 return 0;
3356 if (diff_populate_filespec(two, 0))
3357 return 0;
3358 return !memcmp(one->data, two->data, one->size);
3361 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3363 int i;
3364 struct diff_queue_struct *q = &diff_queued_diff;
3365 struct diff_queue_struct outq;
3366 outq.queue = NULL;
3367 outq.nr = outq.alloc = 0;
3369 for (i = 0; i < q->nr; i++) {
3370 struct diff_filepair *p = q->queue[i];
3373 * 1. Entries that come from stat info dirtyness
3374 * always have both sides (iow, not create/delete),
3375 * one side of the object name is unknown, with
3376 * the same mode and size. Keep the ones that
3377 * do not match these criteria. They have real
3378 * differences.
3380 * 2. At this point, the file is known to be modified,
3381 * with the same mode and size, and the object
3382 * name of one side is unknown. Need to inspect
3383 * the identical contents.
3385 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3386 !DIFF_FILE_VALID(p->two) ||
3387 (p->one->sha1_valid && p->two->sha1_valid) ||
3388 (p->one->mode != p->two->mode) ||
3389 diff_populate_filespec(p->one, 1) ||
3390 diff_populate_filespec(p->two, 1) ||
3391 (p->one->size != p->two->size) ||
3392 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3393 diff_q(&outq, p);
3394 else {
3396 * The caller can subtract 1 from skip_stat_unmatch
3397 * to determine how many paths were dirty only
3398 * due to stat info mismatch.
3400 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3401 diffopt->skip_stat_unmatch++;
3402 diff_free_filepair(p);
3405 free(q->queue);
3406 *q = outq;
3409 void diffcore_std(struct diff_options *options)
3411 if (options->skip_stat_unmatch)
3412 diffcore_skip_stat_unmatch(options);
3413 if (options->break_opt != -1)
3414 diffcore_break(options->break_opt);
3415 if (options->detect_rename)
3416 diffcore_rename(options);
3417 if (options->break_opt != -1)
3418 diffcore_merge_broken();
3419 if (options->pickaxe)
3420 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3421 if (options->orderfile)
3422 diffcore_order(options->orderfile);
3423 diff_resolve_rename_copy();
3424 diffcore_apply_filter(options->filter);
3426 if (diff_queued_diff.nr)
3427 DIFF_OPT_SET(options, HAS_CHANGES);
3428 else
3429 DIFF_OPT_CLR(options, HAS_CHANGES);
3432 int diff_result_code(struct diff_options *opt, int status)
3434 int result = 0;
3435 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3436 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3437 return status;
3438 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3439 DIFF_OPT_TST(opt, HAS_CHANGES))
3440 result |= 01;
3441 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3442 DIFF_OPT_TST(opt, CHECK_FAILED))
3443 result |= 02;
3444 return result;
3447 void diff_addremove(struct diff_options *options,
3448 int addremove, unsigned mode,
3449 const unsigned char *sha1,
3450 const char *concatpath)
3452 struct diff_filespec *one, *two;
3454 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3455 return;
3457 /* This may look odd, but it is a preparation for
3458 * feeding "there are unchanged files which should
3459 * not produce diffs, but when you are doing copy
3460 * detection you would need them, so here they are"
3461 * entries to the diff-core. They will be prefixed
3462 * with something like '=' or '*' (I haven't decided
3463 * which but should not make any difference).
3464 * Feeding the same new and old to diff_change()
3465 * also has the same effect.
3466 * Before the final output happens, they are pruned after
3467 * merged into rename/copy pairs as appropriate.
3469 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3470 addremove = (addremove == '+' ? '-' :
3471 addremove == '-' ? '+' : addremove);
3473 if (options->prefix &&
3474 strncmp(concatpath, options->prefix, options->prefix_length))
3475 return;
3477 one = alloc_filespec(concatpath);
3478 two = alloc_filespec(concatpath);
3480 if (addremove != '+')
3481 fill_filespec(one, sha1, mode);
3482 if (addremove != '-')
3483 fill_filespec(two, sha1, mode);
3485 diff_queue(&diff_queued_diff, one, two);
3486 DIFF_OPT_SET(options, HAS_CHANGES);
3489 void diff_change(struct diff_options *options,
3490 unsigned old_mode, unsigned new_mode,
3491 const unsigned char *old_sha1,
3492 const unsigned char *new_sha1,
3493 const char *concatpath)
3495 struct diff_filespec *one, *two;
3497 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3498 && S_ISGITLINK(new_mode))
3499 return;
3501 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3502 unsigned tmp;
3503 const unsigned char *tmp_c;
3504 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3505 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3508 if (options->prefix &&
3509 strncmp(concatpath, options->prefix, options->prefix_length))
3510 return;
3512 one = alloc_filespec(concatpath);
3513 two = alloc_filespec(concatpath);
3514 fill_filespec(one, old_sha1, old_mode);
3515 fill_filespec(two, new_sha1, new_mode);
3517 diff_queue(&diff_queued_diff, one, two);
3518 DIFF_OPT_SET(options, HAS_CHANGES);
3521 void diff_unmerge(struct diff_options *options,
3522 const char *path,
3523 unsigned mode, const unsigned char *sha1)
3525 struct diff_filespec *one, *two;
3527 if (options->prefix &&
3528 strncmp(path, options->prefix, options->prefix_length))
3529 return;
3531 one = alloc_filespec(path);
3532 two = alloc_filespec(path);
3533 fill_filespec(one, sha1, mode);
3534 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3537 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3538 size_t *outsize)
3540 struct diff_tempfile temp;
3541 const char *argv[3];
3542 const char **arg = argv;
3543 struct child_process child;
3544 struct strbuf buf = STRBUF_INIT;
3546 prepare_temp_file(spec->path, &temp, spec);
3547 *arg++ = pgm;
3548 *arg++ = temp.name;
3549 *arg = NULL;
3551 memset(&child, 0, sizeof(child));
3552 child.argv = argv;
3553 child.out = -1;
3554 if (start_command(&child) != 0 ||
3555 strbuf_read(&buf, child.out, 0) < 0 ||
3556 finish_command(&child) != 0) {
3557 if (temp.name == temp.tmp_path)
3558 unlink(temp.name);
3559 error("error running textconv command '%s'", pgm);
3560 return NULL;
3562 if (temp.name == temp.tmp_path)
3563 unlink(temp.name);
3565 return strbuf_detach(&buf, outsize);