Merge early parts of branch 'js/diff'
[git/jnareb-git.git] / diff.c
blob22b643cc2520664416ceac9998d401a6585f07bf
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include <sys/types.h>
5 #include <sys/wait.h>
6 #include <signal.h>
7 #include "cache.h"
8 #include "quote.h"
9 #include "diff.h"
10 #include "diffcore.h"
11 #include "delta.h"
12 #include "xdiff-interface.h"
14 static int use_size_cache;
16 int diff_rename_limit_default = -1;
18 int git_diff_config(const char *var, const char *value)
20 if (!strcmp(var, "diff.renamelimit")) {
21 diff_rename_limit_default = git_config_int(var, value);
22 return 0;
25 return git_default_config(var, value);
28 enum color_diff {
29 DIFF_PLAIN = 0,
30 DIFF_METAINFO = 1,
31 DIFF_FILE_OLD = 2,
32 DIFF_FILE_NEW = 3,
35 static const char *diff_colors[] = {
36 "\033[0;0m",
37 "\033[1;35m",
38 "\033[1;31m",
39 "\033[1;34m",
42 static char *quote_one(const char *str)
44 int needlen;
45 char *xp;
47 if (!str)
48 return NULL;
49 needlen = quote_c_style(str, NULL, NULL, 0);
50 if (!needlen)
51 return strdup(str);
52 xp = xmalloc(needlen + 1);
53 quote_c_style(str, xp, NULL, 0);
54 return xp;
57 static char *quote_two(const char *one, const char *two)
59 int need_one = quote_c_style(one, NULL, NULL, 1);
60 int need_two = quote_c_style(two, NULL, NULL, 1);
61 char *xp;
63 if (need_one + need_two) {
64 if (!need_one) need_one = strlen(one);
65 if (!need_two) need_one = strlen(two);
67 xp = xmalloc(need_one + need_two + 3);
68 xp[0] = '"';
69 quote_c_style(one, xp + 1, NULL, 1);
70 quote_c_style(two, xp + need_one + 1, NULL, 1);
71 strcpy(xp + need_one + need_two + 1, "\"");
72 return xp;
74 need_one = strlen(one);
75 need_two = strlen(two);
76 xp = xmalloc(need_one + need_two + 1);
77 strcpy(xp, one);
78 strcpy(xp + need_one, two);
79 return xp;
82 static const char *external_diff(void)
84 static const char *external_diff_cmd = NULL;
85 static int done_preparing = 0;
87 if (done_preparing)
88 return external_diff_cmd;
89 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
90 done_preparing = 1;
91 return external_diff_cmd;
94 #define TEMPFILE_PATH_LEN 50
96 static struct diff_tempfile {
97 const char *name; /* filename external diff should read from */
98 char hex[41];
99 char mode[10];
100 char tmp_path[TEMPFILE_PATH_LEN];
101 } diff_temp[2];
103 static int count_lines(const char *data, int size)
105 int count, ch, completely_empty = 1, nl_just_seen = 0;
106 count = 0;
107 while (0 < size--) {
108 ch = *data++;
109 if (ch == '\n') {
110 count++;
111 nl_just_seen = 1;
112 completely_empty = 0;
114 else {
115 nl_just_seen = 0;
116 completely_empty = 0;
119 if (completely_empty)
120 return 0;
121 if (!nl_just_seen)
122 count++; /* no trailing newline */
123 return count;
126 static void print_line_count(int count)
128 switch (count) {
129 case 0:
130 printf("0,0");
131 break;
132 case 1:
133 printf("1");
134 break;
135 default:
136 printf("1,%d", count);
137 break;
141 static void copy_file(int prefix, const char *data, int size)
143 int ch, nl_just_seen = 1;
144 while (0 < size--) {
145 ch = *data++;
146 if (nl_just_seen)
147 putchar(prefix);
148 putchar(ch);
149 if (ch == '\n')
150 nl_just_seen = 1;
151 else
152 nl_just_seen = 0;
154 if (!nl_just_seen)
155 printf("\n\\ No newline at end of file\n");
158 static void emit_rewrite_diff(const char *name_a,
159 const char *name_b,
160 struct diff_filespec *one,
161 struct diff_filespec *two)
163 int lc_a, lc_b;
164 diff_populate_filespec(one, 0);
165 diff_populate_filespec(two, 0);
166 lc_a = count_lines(one->data, one->size);
167 lc_b = count_lines(two->data, two->size);
168 printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
169 print_line_count(lc_a);
170 printf(" +");
171 print_line_count(lc_b);
172 printf(" @@\n");
173 if (lc_a)
174 copy_file('-', one->data, one->size);
175 if (lc_b)
176 copy_file('+', two->data, two->size);
179 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
181 if (!DIFF_FILE_VALID(one)) {
182 mf->ptr = ""; /* does not matter */
183 mf->size = 0;
184 return 0;
186 else if (diff_populate_filespec(one, 0))
187 return -1;
188 mf->ptr = one->data;
189 mf->size = one->size;
190 return 0;
193 struct emit_callback {
194 struct xdiff_emit_state xm;
195 int nparents, color_diff;
196 const char **label_path;
199 static inline void color_diff(int diff_use_color, enum color_diff ix)
201 if (diff_use_color)
202 fputs(diff_colors[ix], stdout);
205 static void fn_out_consume(void *priv, char *line, unsigned long len)
207 int i;
208 struct emit_callback *ecbdata = priv;
210 if (ecbdata->label_path[0]) {
211 color_diff(ecbdata->color_diff, DIFF_METAINFO);
212 printf("--- %s\n", ecbdata->label_path[0]);
213 color_diff(ecbdata->color_diff, DIFF_METAINFO);
214 printf("+++ %s\n", ecbdata->label_path[1]);
215 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
218 /* This is not really necessary for now because
219 * this codepath only deals with two-way diffs.
221 for (i = 0; i < len && line[i] == '@'; i++)
223 if (2 <= i && i < len && line[i] == ' ') {
224 ecbdata->nparents = i - 1;
225 color_diff(ecbdata->color_diff, DIFF_METAINFO);
227 else if (len < ecbdata->nparents)
228 color_diff(ecbdata->color_diff, DIFF_PLAIN);
229 else {
230 int nparents = ecbdata->nparents;
231 int color = DIFF_PLAIN;
232 for (i = 0; i < nparents && len; i++) {
233 if (line[i] == '-')
234 color = DIFF_FILE_OLD;
235 else if (line[i] == '+')
236 color = DIFF_FILE_NEW;
238 color_diff(ecbdata->color_diff, color);
240 fwrite(line, len, 1, stdout);
241 color_diff(ecbdata->color_diff, DIFF_PLAIN);
244 static char *pprint_rename(const char *a, const char *b)
246 const char *old = a;
247 const char *new = b;
248 char *name = NULL;
249 int pfx_length, sfx_length;
250 int len_a = strlen(a);
251 int len_b = strlen(b);
253 /* Find common prefix */
254 pfx_length = 0;
255 while (*old && *new && *old == *new) {
256 if (*old == '/')
257 pfx_length = old - a + 1;
258 old++;
259 new++;
262 /* Find common suffix */
263 old = a + len_a;
264 new = b + len_b;
265 sfx_length = 0;
266 while (a <= old && b <= new && *old == *new) {
267 if (*old == '/')
268 sfx_length = len_a - (old - a);
269 old--;
270 new--;
274 * pfx{mid-a => mid-b}sfx
275 * {pfx-a => pfx-b}sfx
276 * pfx{sfx-a => sfx-b}
277 * name-a => name-b
279 if (pfx_length + sfx_length) {
280 int a_midlen = len_a - pfx_length - sfx_length;
281 int b_midlen = len_b - pfx_length - sfx_length;
282 if (a_midlen < 0) a_midlen = 0;
283 if (b_midlen < 0) b_midlen = 0;
285 name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
286 sprintf(name, "%.*s{%.*s => %.*s}%s",
287 pfx_length, a,
288 a_midlen, a + pfx_length,
289 b_midlen, b + pfx_length,
290 a + len_a - sfx_length);
292 else {
293 name = xmalloc(len_a + len_b + 5);
294 sprintf(name, "%s => %s", a, b);
296 return name;
299 struct diffstat_t {
300 struct xdiff_emit_state xm;
302 int nr;
303 int alloc;
304 struct diffstat_file {
305 char *name;
306 unsigned is_unmerged:1;
307 unsigned is_binary:1;
308 unsigned is_renamed:1;
309 unsigned int added, deleted;
310 } **files;
313 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
314 const char *name_a,
315 const char *name_b)
317 struct diffstat_file *x;
318 x = xcalloc(sizeof (*x), 1);
319 if (diffstat->nr == diffstat->alloc) {
320 diffstat->alloc = alloc_nr(diffstat->alloc);
321 diffstat->files = xrealloc(diffstat->files,
322 diffstat->alloc * sizeof(x));
324 diffstat->files[diffstat->nr++] = x;
325 if (name_b) {
326 x->name = pprint_rename(name_a, name_b);
327 x->is_renamed = 1;
329 else
330 x->name = strdup(name_a);
331 return x;
334 static void diffstat_consume(void *priv, char *line, unsigned long len)
336 struct diffstat_t *diffstat = priv;
337 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
339 if (line[0] == '+')
340 x->added++;
341 else if (line[0] == '-')
342 x->deleted++;
345 static const char pluses[] = "++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
346 static const char minuses[]= "----------------------------------------------------------------------";
347 const char mime_boundary_leader[] = "------------";
349 static void show_stats(struct diffstat_t* data)
351 int i, len, add, del, total, adds = 0, dels = 0;
352 int max, max_change = 0, max_len = 0;
353 int total_files = data->nr;
355 if (data->nr == 0)
356 return;
358 for (i = 0; i < data->nr; i++) {
359 struct diffstat_file *file = data->files[i];
361 len = strlen(file->name);
362 if (max_len < len)
363 max_len = len;
365 if (file->is_binary || file->is_unmerged)
366 continue;
367 if (max_change < file->added + file->deleted)
368 max_change = file->added + file->deleted;
371 for (i = 0; i < data->nr; i++) {
372 char *prefix = "";
373 char *name = data->files[i]->name;
374 int added = data->files[i]->added;
375 int deleted = data->files[i]->deleted;
377 if (0 < (len = quote_c_style(name, NULL, NULL, 0))) {
378 char *qname = xmalloc(len + 1);
379 quote_c_style(name, qname, NULL, 0);
380 free(name);
381 data->files[i]->name = name = qname;
385 * "scale" the filename
387 len = strlen(name);
388 max = max_len;
389 if (max > 50)
390 max = 50;
391 if (len > max) {
392 char *slash;
393 prefix = "...";
394 max -= 3;
395 name += len - max;
396 slash = strchr(name, '/');
397 if (slash)
398 name = slash;
400 len = max;
403 * scale the add/delete
405 max = max_change;
406 if (max + len > 70)
407 max = 70 - len;
409 if (data->files[i]->is_binary) {
410 printf(" %s%-*s | Bin\n", prefix, len, name);
411 goto free_diffstat_file;
413 else if (data->files[i]->is_unmerged) {
414 printf(" %s%-*s | Unmerged\n", prefix, len, name);
415 goto free_diffstat_file;
417 else if (!data->files[i]->is_renamed &&
418 (added + deleted == 0)) {
419 total_files--;
420 goto free_diffstat_file;
423 add = added;
424 del = deleted;
425 total = add + del;
426 adds += add;
427 dels += del;
429 if (max_change > 0) {
430 total = (total * max + max_change / 2) / max_change;
431 add = (add * max + max_change / 2) / max_change;
432 del = total - add;
434 printf(" %s%-*s |%5d %.*s%.*s\n", prefix,
435 len, name, added + deleted,
436 add, pluses, del, minuses);
437 free_diffstat_file:
438 free(data->files[i]->name);
439 free(data->files[i]);
441 free(data->files);
442 printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
443 total_files, adds, dels);
446 struct checkdiff_t {
447 struct xdiff_emit_state xm;
448 const char *filename;
449 int lineno;
452 static void checkdiff_consume(void *priv, char *line, unsigned long len)
454 struct checkdiff_t *data = priv;
456 if (line[0] == '+') {
457 int i, spaces = 0;
459 data->lineno++;
461 /* check space before tab */
462 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
463 if (line[i] == ' ')
464 spaces++;
465 if (line[i - 1] == '\t' && spaces)
466 printf("%s:%d: space before tab:%.*s\n",
467 data->filename, data->lineno, (int)len, line);
469 /* check white space at line end */
470 if (line[len - 1] == '\n')
471 len--;
472 if (isspace(line[len - 1]))
473 printf("%s:%d: white space at end: %.*s\n",
474 data->filename, data->lineno, (int)len, line);
475 } else if (line[0] == ' ')
476 data->lineno++;
477 else if (line[0] == '@') {
478 char *plus = strchr(line, '+');
479 if (plus)
480 data->lineno = strtol(plus, NULL, 10);
481 else
482 die("invalid diff");
486 static unsigned char *deflate_it(char *data,
487 unsigned long size,
488 unsigned long *result_size)
490 int bound;
491 unsigned char *deflated;
492 z_stream stream;
494 memset(&stream, 0, sizeof(stream));
495 deflateInit(&stream, Z_BEST_COMPRESSION);
496 bound = deflateBound(&stream, size);
497 deflated = xmalloc(bound);
498 stream.next_out = deflated;
499 stream.avail_out = bound;
501 stream.next_in = (unsigned char *)data;
502 stream.avail_in = size;
503 while (deflate(&stream, Z_FINISH) == Z_OK)
504 ; /* nothing */
505 deflateEnd(&stream);
506 *result_size = stream.total_out;
507 return deflated;
510 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
512 void *cp;
513 void *delta;
514 void *deflated;
515 void *data;
516 unsigned long orig_size;
517 unsigned long delta_size;
518 unsigned long deflate_size;
519 unsigned long data_size;
521 printf("GIT binary patch\n");
522 /* We could do deflated delta, or we could do just deflated two,
523 * whichever is smaller.
525 delta = NULL;
526 deflated = deflate_it(two->ptr, two->size, &deflate_size);
527 if (one->size && two->size) {
528 delta = diff_delta(one->ptr, one->size,
529 two->ptr, two->size,
530 &delta_size, deflate_size);
531 if (delta) {
532 void *to_free = delta;
533 orig_size = delta_size;
534 delta = deflate_it(delta, delta_size, &delta_size);
535 free(to_free);
539 if (delta && delta_size < deflate_size) {
540 printf("delta %lu\n", orig_size);
541 free(deflated);
542 data = delta;
543 data_size = delta_size;
545 else {
546 printf("literal %lu\n", two->size);
547 free(delta);
548 data = deflated;
549 data_size = deflate_size;
552 /* emit data encoded in base85 */
553 cp = data;
554 while (data_size) {
555 int bytes = (52 < data_size) ? 52 : data_size;
556 char line[70];
557 data_size -= bytes;
558 if (bytes <= 26)
559 line[0] = bytes + 'A' - 1;
560 else
561 line[0] = bytes - 26 + 'a' - 1;
562 encode_85(line + 1, cp, bytes);
563 cp = (char *) cp + bytes;
564 puts(line);
566 printf("\n");
567 free(data);
570 #define FIRST_FEW_BYTES 8000
571 static int mmfile_is_binary(mmfile_t *mf)
573 long sz = mf->size;
574 if (FIRST_FEW_BYTES < sz)
575 sz = FIRST_FEW_BYTES;
576 if (memchr(mf->ptr, 0, sz))
577 return 1;
578 return 0;
581 static void builtin_diff(const char *name_a,
582 const char *name_b,
583 struct diff_filespec *one,
584 struct diff_filespec *two,
585 const char *xfrm_msg,
586 struct diff_options *o,
587 int complete_rewrite)
589 mmfile_t mf1, mf2;
590 const char *lbl[2];
591 char *a_one, *b_two;
593 a_one = quote_two("a/", name_a);
594 b_two = quote_two("b/", name_b);
595 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
596 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
597 color_diff(o->color_diff, DIFF_METAINFO);
598 printf("diff --git %s %s\n", a_one, b_two);
599 if (lbl[0][0] == '/') {
600 /* /dev/null */
601 color_diff(o->color_diff, DIFF_METAINFO);
602 printf("new file mode %06o\n", two->mode);
603 if (xfrm_msg && xfrm_msg[0]) {
604 color_diff(o->color_diff, DIFF_METAINFO);
605 puts(xfrm_msg);
608 else if (lbl[1][0] == '/') {
609 printf("deleted file mode %06o\n", one->mode);
610 if (xfrm_msg && xfrm_msg[0]) {
611 color_diff(o->color_diff, DIFF_METAINFO);
612 puts(xfrm_msg);
615 else {
616 if (one->mode != two->mode) {
617 color_diff(o->color_diff, DIFF_METAINFO);
618 printf("old mode %06o\n", one->mode);
619 color_diff(o->color_diff, DIFF_METAINFO);
620 printf("new mode %06o\n", two->mode);
622 if (xfrm_msg && xfrm_msg[0]) {
623 color_diff(o->color_diff, DIFF_METAINFO);
624 puts(xfrm_msg);
627 * we do not run diff between different kind
628 * of objects.
630 if ((one->mode ^ two->mode) & S_IFMT)
631 goto free_ab_and_return;
632 if (complete_rewrite) {
633 color_diff(o->color_diff, DIFF_PLAIN);
634 emit_rewrite_diff(name_a, name_b, one, two);
635 goto free_ab_and_return;
639 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
640 die("unable to read files to diff");
642 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2)) {
643 /* Quite common confusing case */
644 if (mf1.size == mf2.size &&
645 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
646 goto free_ab_and_return;
647 if (o->binary)
648 emit_binary_diff(&mf1, &mf2);
649 else
650 printf("Binary files %s and %s differ\n",
651 lbl[0], lbl[1]);
653 else {
654 /* Crazy xdl interfaces.. */
655 const char *diffopts = getenv("GIT_DIFF_OPTS");
656 xpparam_t xpp;
657 xdemitconf_t xecfg;
658 xdemitcb_t ecb;
659 struct emit_callback ecbdata;
661 memset(&ecbdata, 0, sizeof(ecbdata));
662 ecbdata.label_path = lbl;
663 ecbdata.color_diff = o->color_diff;
664 xpp.flags = XDF_NEED_MINIMAL;
665 xecfg.ctxlen = o->context;
666 xecfg.flags = XDL_EMIT_FUNCNAMES;
667 if (!diffopts)
669 else if (!strncmp(diffopts, "--unified=", 10))
670 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
671 else if (!strncmp(diffopts, "-u", 2))
672 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
673 ecb.outf = xdiff_outf;
674 ecb.priv = &ecbdata;
675 ecbdata.xm.consume = fn_out_consume;
676 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
679 free_ab_and_return:
680 free(a_one);
681 free(b_two);
682 return;
685 static void builtin_diffstat(const char *name_a, const char *name_b,
686 struct diff_filespec *one,
687 struct diff_filespec *two,
688 struct diffstat_t *diffstat,
689 int complete_rewrite)
691 mmfile_t mf1, mf2;
692 struct diffstat_file *data;
694 data = diffstat_add(diffstat, name_a, name_b);
696 if (!one || !two) {
697 data->is_unmerged = 1;
698 return;
700 if (complete_rewrite) {
701 diff_populate_filespec(one, 0);
702 diff_populate_filespec(two, 0);
703 data->deleted = count_lines(one->data, one->size);
704 data->added = count_lines(two->data, two->size);
705 return;
707 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
708 die("unable to read files to diff");
710 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
711 data->is_binary = 1;
712 else {
713 /* Crazy xdl interfaces.. */
714 xpparam_t xpp;
715 xdemitconf_t xecfg;
716 xdemitcb_t ecb;
718 xpp.flags = XDF_NEED_MINIMAL;
719 xecfg.ctxlen = 0;
720 xecfg.flags = 0;
721 ecb.outf = xdiff_outf;
722 ecb.priv = diffstat;
723 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
727 static void builtin_checkdiff(const char *name_a, const char *name_b,
728 struct diff_filespec *one,
729 struct diff_filespec *two)
731 mmfile_t mf1, mf2;
732 struct checkdiff_t data;
734 if (!two)
735 return;
737 memset(&data, 0, sizeof(data));
738 data.xm.consume = checkdiff_consume;
739 data.filename = name_b ? name_b : name_a;
740 data.lineno = 0;
742 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
743 die("unable to read files to diff");
745 if (mmfile_is_binary(&mf2))
746 return;
747 else {
748 /* Crazy xdl interfaces.. */
749 xpparam_t xpp;
750 xdemitconf_t xecfg;
751 xdemitcb_t ecb;
753 xpp.flags = XDF_NEED_MINIMAL;
754 xecfg.ctxlen = 0;
755 xecfg.flags = 0;
756 ecb.outf = xdiff_outf;
757 ecb.priv = &data;
758 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
762 struct diff_filespec *alloc_filespec(const char *path)
764 int namelen = strlen(path);
765 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
767 memset(spec, 0, sizeof(*spec));
768 spec->path = (char *)(spec + 1);
769 memcpy(spec->path, path, namelen+1);
770 return spec;
773 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
774 unsigned short mode)
776 if (mode) {
777 spec->mode = canon_mode(mode);
778 memcpy(spec->sha1, sha1, 20);
779 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
784 * Given a name and sha1 pair, if the dircache tells us the file in
785 * the work tree has that object contents, return true, so that
786 * prepare_temp_file() does not have to inflate and extract.
788 static int work_tree_matches(const char *name, const unsigned char *sha1)
790 struct cache_entry *ce;
791 struct stat st;
792 int pos, len;
794 /* We do not read the cache ourselves here, because the
795 * benchmark with my previous version that always reads cache
796 * shows that it makes things worse for diff-tree comparing
797 * two linux-2.6 kernel trees in an already checked out work
798 * tree. This is because most diff-tree comparisons deal with
799 * only a small number of files, while reading the cache is
800 * expensive for a large project, and its cost outweighs the
801 * savings we get by not inflating the object to a temporary
802 * file. Practically, this code only helps when we are used
803 * by diff-cache --cached, which does read the cache before
804 * calling us.
806 if (!active_cache)
807 return 0;
809 len = strlen(name);
810 pos = cache_name_pos(name, len);
811 if (pos < 0)
812 return 0;
813 ce = active_cache[pos];
814 if ((lstat(name, &st) < 0) ||
815 !S_ISREG(st.st_mode) || /* careful! */
816 ce_match_stat(ce, &st, 0) ||
817 memcmp(sha1, ce->sha1, 20))
818 return 0;
819 /* we return 1 only when we can stat, it is a regular file,
820 * stat information matches, and sha1 recorded in the cache
821 * matches. I.e. we know the file in the work tree really is
822 * the same as the <name, sha1> pair.
824 return 1;
827 static struct sha1_size_cache {
828 unsigned char sha1[20];
829 unsigned long size;
830 } **sha1_size_cache;
831 static int sha1_size_cache_nr, sha1_size_cache_alloc;
833 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
834 int find_only,
835 unsigned long size)
837 int first, last;
838 struct sha1_size_cache *e;
840 first = 0;
841 last = sha1_size_cache_nr;
842 while (last > first) {
843 int cmp, next = (last + first) >> 1;
844 e = sha1_size_cache[next];
845 cmp = memcmp(e->sha1, sha1, 20);
846 if (!cmp)
847 return e;
848 if (cmp < 0) {
849 last = next;
850 continue;
852 first = next+1;
854 /* not found */
855 if (find_only)
856 return NULL;
857 /* insert to make it at "first" */
858 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
859 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
860 sha1_size_cache = xrealloc(sha1_size_cache,
861 sha1_size_cache_alloc *
862 sizeof(*sha1_size_cache));
864 sha1_size_cache_nr++;
865 if (first < sha1_size_cache_nr)
866 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
867 (sha1_size_cache_nr - first - 1) *
868 sizeof(*sha1_size_cache));
869 e = xmalloc(sizeof(struct sha1_size_cache));
870 sha1_size_cache[first] = e;
871 memcpy(e->sha1, sha1, 20);
872 e->size = size;
873 return e;
877 * While doing rename detection and pickaxe operation, we may need to
878 * grab the data for the blob (or file) for our own in-core comparison.
879 * diff_filespec has data and size fields for this purpose.
881 int diff_populate_filespec(struct diff_filespec *s, int size_only)
883 int err = 0;
884 if (!DIFF_FILE_VALID(s))
885 die("internal error: asking to populate invalid file.");
886 if (S_ISDIR(s->mode))
887 return -1;
889 if (!use_size_cache)
890 size_only = 0;
892 if (s->data)
893 return err;
894 if (!s->sha1_valid ||
895 work_tree_matches(s->path, s->sha1)) {
896 struct stat st;
897 int fd;
898 if (lstat(s->path, &st) < 0) {
899 if (errno == ENOENT) {
900 err_empty:
901 err = -1;
902 empty:
903 s->data = "";
904 s->size = 0;
905 return err;
908 s->size = st.st_size;
909 if (!s->size)
910 goto empty;
911 if (size_only)
912 return 0;
913 if (S_ISLNK(st.st_mode)) {
914 int ret;
915 s->data = xmalloc(s->size);
916 s->should_free = 1;
917 ret = readlink(s->path, s->data, s->size);
918 if (ret < 0) {
919 free(s->data);
920 goto err_empty;
922 return 0;
924 fd = open(s->path, O_RDONLY);
925 if (fd < 0)
926 goto err_empty;
927 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
928 close(fd);
929 if (s->data == MAP_FAILED)
930 goto err_empty;
931 s->should_munmap = 1;
933 else {
934 char type[20];
935 struct sha1_size_cache *e;
937 if (size_only) {
938 e = locate_size_cache(s->sha1, 1, 0);
939 if (e) {
940 s->size = e->size;
941 return 0;
943 if (!sha1_object_info(s->sha1, type, &s->size))
944 locate_size_cache(s->sha1, 0, s->size);
946 else {
947 s->data = read_sha1_file(s->sha1, type, &s->size);
948 s->should_free = 1;
951 return 0;
954 void diff_free_filespec_data(struct diff_filespec *s)
956 if (s->should_free)
957 free(s->data);
958 else if (s->should_munmap)
959 munmap(s->data, s->size);
960 s->should_free = s->should_munmap = 0;
961 s->data = NULL;
962 free(s->cnt_data);
963 s->cnt_data = NULL;
966 static void prep_temp_blob(struct diff_tempfile *temp,
967 void *blob,
968 unsigned long size,
969 const unsigned char *sha1,
970 int mode)
972 int fd;
974 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
975 if (fd < 0)
976 die("unable to create temp-file");
977 if (write(fd, blob, size) != size)
978 die("unable to write temp-file");
979 close(fd);
980 temp->name = temp->tmp_path;
981 strcpy(temp->hex, sha1_to_hex(sha1));
982 temp->hex[40] = 0;
983 sprintf(temp->mode, "%06o", mode);
986 static void prepare_temp_file(const char *name,
987 struct diff_tempfile *temp,
988 struct diff_filespec *one)
990 if (!DIFF_FILE_VALID(one)) {
991 not_a_valid_file:
992 /* A '-' entry produces this for file-2, and
993 * a '+' entry produces this for file-1.
995 temp->name = "/dev/null";
996 strcpy(temp->hex, ".");
997 strcpy(temp->mode, ".");
998 return;
1001 if (!one->sha1_valid ||
1002 work_tree_matches(name, one->sha1)) {
1003 struct stat st;
1004 if (lstat(name, &st) < 0) {
1005 if (errno == ENOENT)
1006 goto not_a_valid_file;
1007 die("stat(%s): %s", name, strerror(errno));
1009 if (S_ISLNK(st.st_mode)) {
1010 int ret;
1011 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1012 if (sizeof(buf) <= st.st_size)
1013 die("symlink too long: %s", name);
1014 ret = readlink(name, buf, st.st_size);
1015 if (ret < 0)
1016 die("readlink(%s)", name);
1017 prep_temp_blob(temp, buf, st.st_size,
1018 (one->sha1_valid ?
1019 one->sha1 : null_sha1),
1020 (one->sha1_valid ?
1021 one->mode : S_IFLNK));
1023 else {
1024 /* we can borrow from the file in the work tree */
1025 temp->name = name;
1026 if (!one->sha1_valid)
1027 strcpy(temp->hex, sha1_to_hex(null_sha1));
1028 else
1029 strcpy(temp->hex, sha1_to_hex(one->sha1));
1030 /* Even though we may sometimes borrow the
1031 * contents from the work tree, we always want
1032 * one->mode. mode is trustworthy even when
1033 * !(one->sha1_valid), as long as
1034 * DIFF_FILE_VALID(one).
1036 sprintf(temp->mode, "%06o", one->mode);
1038 return;
1040 else {
1041 if (diff_populate_filespec(one, 0))
1042 die("cannot read data blob for %s", one->path);
1043 prep_temp_blob(temp, one->data, one->size,
1044 one->sha1, one->mode);
1048 static void remove_tempfile(void)
1050 int i;
1052 for (i = 0; i < 2; i++)
1053 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1054 unlink(diff_temp[i].name);
1055 diff_temp[i].name = NULL;
1059 static void remove_tempfile_on_signal(int signo)
1061 remove_tempfile();
1062 signal(SIGINT, SIG_DFL);
1063 raise(signo);
1066 static int spawn_prog(const char *pgm, const char **arg)
1068 pid_t pid;
1069 int status;
1071 fflush(NULL);
1072 pid = fork();
1073 if (pid < 0)
1074 die("unable to fork");
1075 if (!pid) {
1076 execvp(pgm, (char *const*) arg);
1077 exit(255);
1080 while (waitpid(pid, &status, 0) < 0) {
1081 if (errno == EINTR)
1082 continue;
1083 return -1;
1086 /* Earlier we did not check the exit status because
1087 * diff exits non-zero if files are different, and
1088 * we are not interested in knowing that. It was a
1089 * mistake which made it harder to quit a diff-*
1090 * session that uses the git-apply-patch-script as
1091 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
1092 * should also exit non-zero only when it wants to
1093 * abort the entire diff-* session.
1095 if (WIFEXITED(status) && !WEXITSTATUS(status))
1096 return 0;
1097 return -1;
1100 /* An external diff command takes:
1102 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1103 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1106 static void run_external_diff(const char *pgm,
1107 const char *name,
1108 const char *other,
1109 struct diff_filespec *one,
1110 struct diff_filespec *two,
1111 const char *xfrm_msg,
1112 int complete_rewrite)
1114 const char *spawn_arg[10];
1115 struct diff_tempfile *temp = diff_temp;
1116 int retval;
1117 static int atexit_asked = 0;
1118 const char *othername;
1119 const char **arg = &spawn_arg[0];
1121 othername = (other? other : name);
1122 if (one && two) {
1123 prepare_temp_file(name, &temp[0], one);
1124 prepare_temp_file(othername, &temp[1], two);
1125 if (! atexit_asked &&
1126 (temp[0].name == temp[0].tmp_path ||
1127 temp[1].name == temp[1].tmp_path)) {
1128 atexit_asked = 1;
1129 atexit(remove_tempfile);
1131 signal(SIGINT, remove_tempfile_on_signal);
1134 if (one && two) {
1135 *arg++ = pgm;
1136 *arg++ = name;
1137 *arg++ = temp[0].name;
1138 *arg++ = temp[0].hex;
1139 *arg++ = temp[0].mode;
1140 *arg++ = temp[1].name;
1141 *arg++ = temp[1].hex;
1142 *arg++ = temp[1].mode;
1143 if (other) {
1144 *arg++ = other;
1145 *arg++ = xfrm_msg;
1147 } else {
1148 *arg++ = pgm;
1149 *arg++ = name;
1151 *arg = NULL;
1152 retval = spawn_prog(pgm, spawn_arg);
1153 remove_tempfile();
1154 if (retval) {
1155 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1156 exit(1);
1160 static void run_diff_cmd(const char *pgm,
1161 const char *name,
1162 const char *other,
1163 struct diff_filespec *one,
1164 struct diff_filespec *two,
1165 const char *xfrm_msg,
1166 struct diff_options *o,
1167 int complete_rewrite)
1169 if (pgm) {
1170 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1171 complete_rewrite);
1172 return;
1174 if (one && two)
1175 builtin_diff(name, other ? other : name,
1176 one, two, xfrm_msg, o, complete_rewrite);
1177 else
1178 printf("* Unmerged path %s\n", name);
1181 static void diff_fill_sha1_info(struct diff_filespec *one)
1183 if (DIFF_FILE_VALID(one)) {
1184 if (!one->sha1_valid) {
1185 struct stat st;
1186 if (lstat(one->path, &st) < 0)
1187 die("stat %s", one->path);
1188 if (index_path(one->sha1, one->path, &st, 0))
1189 die("cannot hash %s\n", one->path);
1192 else
1193 memset(one->sha1, 0, 20);
1196 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1198 const char *pgm = external_diff();
1199 char msg[PATH_MAX*2+300], *xfrm_msg;
1200 struct diff_filespec *one;
1201 struct diff_filespec *two;
1202 const char *name;
1203 const char *other;
1204 char *name_munged, *other_munged;
1205 int complete_rewrite = 0;
1206 int len;
1208 if (DIFF_PAIR_UNMERGED(p)) {
1209 /* unmerged */
1210 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1211 return;
1214 name = p->one->path;
1215 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1216 name_munged = quote_one(name);
1217 other_munged = quote_one(other);
1218 one = p->one; two = p->two;
1220 diff_fill_sha1_info(one);
1221 diff_fill_sha1_info(two);
1223 len = 0;
1224 switch (p->status) {
1225 case DIFF_STATUS_COPIED:
1226 len += snprintf(msg + len, sizeof(msg) - len,
1227 "similarity index %d%%\n"
1228 "copy from %s\n"
1229 "copy to %s\n",
1230 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1231 name_munged, other_munged);
1232 break;
1233 case DIFF_STATUS_RENAMED:
1234 len += snprintf(msg + len, sizeof(msg) - len,
1235 "similarity index %d%%\n"
1236 "rename from %s\n"
1237 "rename to %s\n",
1238 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1239 name_munged, other_munged);
1240 break;
1241 case DIFF_STATUS_MODIFIED:
1242 if (p->score) {
1243 len += snprintf(msg + len, sizeof(msg) - len,
1244 "dissimilarity index %d%%\n",
1245 (int)(0.5 + p->score *
1246 100.0/MAX_SCORE));
1247 complete_rewrite = 1;
1248 break;
1250 /* fallthru */
1251 default:
1252 /* nothing */
1256 if (memcmp(one->sha1, two->sha1, 20)) {
1257 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1259 len += snprintf(msg + len, sizeof(msg) - len,
1260 "index %.*s..%.*s",
1261 abbrev, sha1_to_hex(one->sha1),
1262 abbrev, sha1_to_hex(two->sha1));
1263 if (one->mode == two->mode)
1264 len += snprintf(msg + len, sizeof(msg) - len,
1265 " %06o", one->mode);
1266 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1269 if (len)
1270 msg[--len] = 0;
1271 xfrm_msg = len ? msg : NULL;
1273 if (!pgm &&
1274 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1275 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1276 /* a filepair that changes between file and symlink
1277 * needs to be split into deletion and creation.
1279 struct diff_filespec *null = alloc_filespec(two->path);
1280 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1281 free(null);
1282 null = alloc_filespec(one->path);
1283 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1284 free(null);
1286 else
1287 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1288 complete_rewrite);
1290 free(name_munged);
1291 free(other_munged);
1294 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1295 struct diffstat_t *diffstat)
1297 const char *name;
1298 const char *other;
1299 int complete_rewrite = 0;
1301 if (DIFF_PAIR_UNMERGED(p)) {
1302 /* unmerged */
1303 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, 0);
1304 return;
1307 name = p->one->path;
1308 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1310 diff_fill_sha1_info(p->one);
1311 diff_fill_sha1_info(p->two);
1313 if (p->status == DIFF_STATUS_MODIFIED && p->score)
1314 complete_rewrite = 1;
1315 builtin_diffstat(name, other, p->one, p->two, diffstat, complete_rewrite);
1318 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
1320 const char *name;
1321 const char *other;
1323 if (DIFF_PAIR_UNMERGED(p)) {
1324 /* unmerged */
1325 return;
1328 name = p->one->path;
1329 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1331 diff_fill_sha1_info(p->one);
1332 diff_fill_sha1_info(p->two);
1334 builtin_checkdiff(name, other, p->one, p->two);
1337 void diff_setup(struct diff_options *options)
1339 memset(options, 0, sizeof(*options));
1340 options->output_format = DIFF_FORMAT_RAW;
1341 options->line_termination = '\n';
1342 options->break_opt = -1;
1343 options->rename_limit = -1;
1344 options->context = 3;
1346 options->change = diff_change;
1347 options->add_remove = diff_addremove;
1350 int diff_setup_done(struct diff_options *options)
1352 if ((options->find_copies_harder &&
1353 options->detect_rename != DIFF_DETECT_COPY) ||
1354 (0 <= options->rename_limit && !options->detect_rename))
1355 return -1;
1358 * These cases always need recursive; we do not drop caller-supplied
1359 * recursive bits for other formats here.
1361 if ((options->output_format == DIFF_FORMAT_PATCH) ||
1362 (options->output_format == DIFF_FORMAT_DIFFSTAT) ||
1363 (options->output_format == DIFF_FORMAT_CHECKDIFF))
1364 options->recursive = 1;
1367 * These combinations do not make sense.
1369 if (options->output_format == DIFF_FORMAT_RAW)
1370 options->with_raw = 0;
1371 if (options->output_format == DIFF_FORMAT_DIFFSTAT)
1372 options->with_stat = 0;
1374 if (options->detect_rename && options->rename_limit < 0)
1375 options->rename_limit = diff_rename_limit_default;
1376 if (options->setup & DIFF_SETUP_USE_CACHE) {
1377 if (!active_cache)
1378 /* read-cache does not die even when it fails
1379 * so it is safe for us to do this here. Also
1380 * it does not smudge active_cache or active_nr
1381 * when it fails, so we do not have to worry about
1382 * cleaning it up ourselves either.
1384 read_cache();
1386 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1387 use_size_cache = 1;
1388 if (options->abbrev <= 0 || 40 < options->abbrev)
1389 options->abbrev = 40; /* full */
1391 return 0;
1394 int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
1396 char c, *eq;
1397 int len;
1399 if (*arg != '-')
1400 return 0;
1401 c = *++arg;
1402 if (!c)
1403 return 0;
1404 if (c == arg_short) {
1405 c = *++arg;
1406 if (!c)
1407 return 1;
1408 if (val && isdigit(c)) {
1409 char *end;
1410 int n = strtoul(arg, &end, 10);
1411 if (*end)
1412 return 0;
1413 *val = n;
1414 return 1;
1416 return 0;
1418 if (c != '-')
1419 return 0;
1420 arg++;
1421 eq = strchr(arg, '=');
1422 if (eq)
1423 len = eq - arg;
1424 else
1425 len = strlen(arg);
1426 if (!len || strncmp(arg, arg_long, len))
1427 return 0;
1428 if (eq) {
1429 int n;
1430 char *end;
1431 if (!isdigit(*++eq))
1432 return 0;
1433 n = strtoul(eq, &end, 10);
1434 if (*end)
1435 return 0;
1436 *val = n;
1438 return 1;
1441 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1443 const char *arg = av[0];
1444 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1445 options->output_format = DIFF_FORMAT_PATCH;
1446 else if (opt_arg(arg, 'U', "unified", &options->context))
1447 options->output_format = DIFF_FORMAT_PATCH;
1448 else if (!strcmp(arg, "--patch-with-raw")) {
1449 options->output_format = DIFF_FORMAT_PATCH;
1450 options->with_raw = 1;
1452 else if (!strcmp(arg, "--stat"))
1453 options->output_format = DIFF_FORMAT_DIFFSTAT;
1454 else if (!strcmp(arg, "--check"))
1455 options->output_format = DIFF_FORMAT_CHECKDIFF;
1456 else if (!strcmp(arg, "--summary"))
1457 options->summary = 1;
1458 else if (!strcmp(arg, "--patch-with-stat")) {
1459 options->output_format = DIFF_FORMAT_PATCH;
1460 options->with_stat = 1;
1462 else if (!strcmp(arg, "-z"))
1463 options->line_termination = 0;
1464 else if (!strncmp(arg, "-l", 2))
1465 options->rename_limit = strtoul(arg+2, NULL, 10);
1466 else if (!strcmp(arg, "--full-index"))
1467 options->full_index = 1;
1468 else if (!strcmp(arg, "--binary")) {
1469 options->output_format = DIFF_FORMAT_PATCH;
1470 options->full_index = options->binary = 1;
1472 else if (!strcmp(arg, "--name-only"))
1473 options->output_format = DIFF_FORMAT_NAME;
1474 else if (!strcmp(arg, "--name-status"))
1475 options->output_format = DIFF_FORMAT_NAME_STATUS;
1476 else if (!strcmp(arg, "-R"))
1477 options->reverse_diff = 1;
1478 else if (!strncmp(arg, "-S", 2))
1479 options->pickaxe = arg + 2;
1480 else if (!strcmp(arg, "-s"))
1481 options->output_format = DIFF_FORMAT_NO_OUTPUT;
1482 else if (!strncmp(arg, "-O", 2))
1483 options->orderfile = arg + 2;
1484 else if (!strncmp(arg, "--diff-filter=", 14))
1485 options->filter = arg + 14;
1486 else if (!strcmp(arg, "--pickaxe-all"))
1487 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1488 else if (!strcmp(arg, "--pickaxe-regex"))
1489 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1490 else if (!strncmp(arg, "-B", 2)) {
1491 if ((options->break_opt =
1492 diff_scoreopt_parse(arg)) == -1)
1493 return -1;
1495 else if (!strncmp(arg, "-M", 2)) {
1496 if ((options->rename_score =
1497 diff_scoreopt_parse(arg)) == -1)
1498 return -1;
1499 options->detect_rename = DIFF_DETECT_RENAME;
1501 else if (!strncmp(arg, "-C", 2)) {
1502 if ((options->rename_score =
1503 diff_scoreopt_parse(arg)) == -1)
1504 return -1;
1505 options->detect_rename = DIFF_DETECT_COPY;
1507 else if (!strcmp(arg, "--find-copies-harder"))
1508 options->find_copies_harder = 1;
1509 else if (!strcmp(arg, "--abbrev"))
1510 options->abbrev = DEFAULT_ABBREV;
1511 else if (!strncmp(arg, "--abbrev=", 9)) {
1512 options->abbrev = strtoul(arg + 9, NULL, 10);
1513 if (options->abbrev < MINIMUM_ABBREV)
1514 options->abbrev = MINIMUM_ABBREV;
1515 else if (40 < options->abbrev)
1516 options->abbrev = 40;
1518 else if (!strcmp(arg, "--color"))
1519 options->color_diff = 1;
1520 else
1521 return 0;
1522 return 1;
1525 static int parse_num(const char **cp_p)
1527 unsigned long num, scale;
1528 int ch, dot;
1529 const char *cp = *cp_p;
1531 num = 0;
1532 scale = 1;
1533 dot = 0;
1534 for(;;) {
1535 ch = *cp;
1536 if ( !dot && ch == '.' ) {
1537 scale = 1;
1538 dot = 1;
1539 } else if ( ch == '%' ) {
1540 scale = dot ? scale*100 : 100;
1541 cp++; /* % is always at the end */
1542 break;
1543 } else if ( ch >= '0' && ch <= '9' ) {
1544 if ( scale < 100000 ) {
1545 scale *= 10;
1546 num = (num*10) + (ch-'0');
1548 } else {
1549 break;
1551 cp++;
1553 *cp_p = cp;
1555 /* user says num divided by scale and we say internally that
1556 * is MAX_SCORE * num / scale.
1558 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
1561 int diff_scoreopt_parse(const char *opt)
1563 int opt1, opt2, cmd;
1565 if (*opt++ != '-')
1566 return -1;
1567 cmd = *opt++;
1568 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
1569 return -1; /* that is not a -M, -C nor -B option */
1571 opt1 = parse_num(&opt);
1572 if (cmd != 'B')
1573 opt2 = 0;
1574 else {
1575 if (*opt == 0)
1576 opt2 = 0;
1577 else if (*opt != '/')
1578 return -1; /* we expect -B80/99 or -B80 */
1579 else {
1580 opt++;
1581 opt2 = parse_num(&opt);
1584 if (*opt != 0)
1585 return -1;
1586 return opt1 | (opt2 << 16);
1589 struct diff_queue_struct diff_queued_diff;
1591 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
1593 if (queue->alloc <= queue->nr) {
1594 queue->alloc = alloc_nr(queue->alloc);
1595 queue->queue = xrealloc(queue->queue,
1596 sizeof(dp) * queue->alloc);
1598 queue->queue[queue->nr++] = dp;
1601 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
1602 struct diff_filespec *one,
1603 struct diff_filespec *two)
1605 struct diff_filepair *dp = xmalloc(sizeof(*dp));
1606 dp->one = one;
1607 dp->two = two;
1608 dp->score = 0;
1609 dp->status = 0;
1610 dp->source_stays = 0;
1611 dp->broken_pair = 0;
1612 if (queue)
1613 diff_q(queue, dp);
1614 return dp;
1617 void diff_free_filepair(struct diff_filepair *p)
1619 diff_free_filespec_data(p->one);
1620 diff_free_filespec_data(p->two);
1621 free(p->one);
1622 free(p->two);
1623 free(p);
1626 /* This is different from find_unique_abbrev() in that
1627 * it stuffs the result with dots for alignment.
1629 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
1631 int abblen;
1632 const char *abbrev;
1633 if (len == 40)
1634 return sha1_to_hex(sha1);
1636 abbrev = find_unique_abbrev(sha1, len);
1637 if (!abbrev)
1638 return sha1_to_hex(sha1);
1639 abblen = strlen(abbrev);
1640 if (abblen < 37) {
1641 static char hex[41];
1642 if (len < abblen && abblen <= len + 2)
1643 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
1644 else
1645 sprintf(hex, "%s...", abbrev);
1646 return hex;
1648 return sha1_to_hex(sha1);
1651 static void diff_flush_raw(struct diff_filepair *p,
1652 int line_termination,
1653 int inter_name_termination,
1654 struct diff_options *options,
1655 int output_format)
1657 int two_paths;
1658 char status[10];
1659 int abbrev = options->abbrev;
1660 const char *path_one, *path_two;
1662 path_one = p->one->path;
1663 path_two = p->two->path;
1664 if (line_termination) {
1665 path_one = quote_one(path_one);
1666 path_two = quote_one(path_two);
1669 if (p->score)
1670 sprintf(status, "%c%03d", p->status,
1671 (int)(0.5 + p->score * 100.0/MAX_SCORE));
1672 else {
1673 status[0] = p->status;
1674 status[1] = 0;
1676 switch (p->status) {
1677 case DIFF_STATUS_COPIED:
1678 case DIFF_STATUS_RENAMED:
1679 two_paths = 1;
1680 break;
1681 case DIFF_STATUS_ADDED:
1682 case DIFF_STATUS_DELETED:
1683 two_paths = 0;
1684 break;
1685 default:
1686 two_paths = 0;
1687 break;
1689 if (output_format != DIFF_FORMAT_NAME_STATUS) {
1690 printf(":%06o %06o %s ",
1691 p->one->mode, p->two->mode,
1692 diff_unique_abbrev(p->one->sha1, abbrev));
1693 printf("%s ",
1694 diff_unique_abbrev(p->two->sha1, abbrev));
1696 printf("%s%c%s", status, inter_name_termination, path_one);
1697 if (two_paths)
1698 printf("%c%s", inter_name_termination, path_two);
1699 putchar(line_termination);
1700 if (path_one != p->one->path)
1701 free((void*)path_one);
1702 if (path_two != p->two->path)
1703 free((void*)path_two);
1706 static void diff_flush_name(struct diff_filepair *p,
1707 int inter_name_termination,
1708 int line_termination)
1710 char *path = p->two->path;
1712 if (line_termination)
1713 path = quote_one(p->two->path);
1714 else
1715 path = p->two->path;
1716 printf("%s%c", path, line_termination);
1717 if (p->two->path != path)
1718 free(path);
1721 int diff_unmodified_pair(struct diff_filepair *p)
1723 /* This function is written stricter than necessary to support
1724 * the currently implemented transformers, but the idea is to
1725 * let transformers to produce diff_filepairs any way they want,
1726 * and filter and clean them up here before producing the output.
1728 struct diff_filespec *one, *two;
1730 if (DIFF_PAIR_UNMERGED(p))
1731 return 0; /* unmerged is interesting */
1733 one = p->one;
1734 two = p->two;
1736 /* deletion, addition, mode or type change
1737 * and rename are all interesting.
1739 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1740 DIFF_PAIR_MODE_CHANGED(p) ||
1741 strcmp(one->path, two->path))
1742 return 0;
1744 /* both are valid and point at the same path. that is, we are
1745 * dealing with a change.
1747 if (one->sha1_valid && two->sha1_valid &&
1748 !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1749 return 1; /* no change */
1750 if (!one->sha1_valid && !two->sha1_valid)
1751 return 1; /* both look at the same file on the filesystem. */
1752 return 0;
1755 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1757 if (diff_unmodified_pair(p))
1758 return;
1760 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1761 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1762 return; /* no tree diffs in patch format */
1764 run_diff(p, o);
1767 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
1768 struct diffstat_t *diffstat)
1770 if (diff_unmodified_pair(p))
1771 return;
1773 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1774 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1775 return; /* no tree diffs in patch format */
1777 run_diffstat(p, o, diffstat);
1780 static void diff_flush_checkdiff(struct diff_filepair *p,
1781 struct diff_options *o)
1783 if (diff_unmodified_pair(p))
1784 return;
1786 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1787 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1788 return; /* no tree diffs in patch format */
1790 run_checkdiff(p, o);
1793 int diff_queue_is_empty(void)
1795 struct diff_queue_struct *q = &diff_queued_diff;
1796 int i;
1797 for (i = 0; i < q->nr; i++)
1798 if (!diff_unmodified_pair(q->queue[i]))
1799 return 0;
1800 return 1;
1803 #if DIFF_DEBUG
1804 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1806 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1807 x, one ? one : "",
1808 s->path,
1809 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1810 s->mode,
1811 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1812 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1813 x, one ? one : "",
1814 s->size, s->xfrm_flags);
1817 void diff_debug_filepair(const struct diff_filepair *p, int i)
1819 diff_debug_filespec(p->one, i, "one");
1820 diff_debug_filespec(p->two, i, "two");
1821 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1822 p->score, p->status ? p->status : '?',
1823 p->source_stays, p->broken_pair);
1826 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1828 int i;
1829 if (msg)
1830 fprintf(stderr, "%s\n", msg);
1831 fprintf(stderr, "q->nr = %d\n", q->nr);
1832 for (i = 0; i < q->nr; i++) {
1833 struct diff_filepair *p = q->queue[i];
1834 diff_debug_filepair(p, i);
1837 #endif
1839 static void diff_resolve_rename_copy(void)
1841 int i, j;
1842 struct diff_filepair *p, *pp;
1843 struct diff_queue_struct *q = &diff_queued_diff;
1845 diff_debug_queue("resolve-rename-copy", q);
1847 for (i = 0; i < q->nr; i++) {
1848 p = q->queue[i];
1849 p->status = 0; /* undecided */
1850 if (DIFF_PAIR_UNMERGED(p))
1851 p->status = DIFF_STATUS_UNMERGED;
1852 else if (!DIFF_FILE_VALID(p->one))
1853 p->status = DIFF_STATUS_ADDED;
1854 else if (!DIFF_FILE_VALID(p->two))
1855 p->status = DIFF_STATUS_DELETED;
1856 else if (DIFF_PAIR_TYPE_CHANGED(p))
1857 p->status = DIFF_STATUS_TYPE_CHANGED;
1859 /* from this point on, we are dealing with a pair
1860 * whose both sides are valid and of the same type, i.e.
1861 * either in-place edit or rename/copy edit.
1863 else if (DIFF_PAIR_RENAME(p)) {
1864 if (p->source_stays) {
1865 p->status = DIFF_STATUS_COPIED;
1866 continue;
1868 /* See if there is some other filepair that
1869 * copies from the same source as us. If so
1870 * we are a copy. Otherwise we are either a
1871 * copy if the path stays, or a rename if it
1872 * does not, but we already handled "stays" case.
1874 for (j = i + 1; j < q->nr; j++) {
1875 pp = q->queue[j];
1876 if (strcmp(pp->one->path, p->one->path))
1877 continue; /* not us */
1878 if (!DIFF_PAIR_RENAME(pp))
1879 continue; /* not a rename/copy */
1880 /* pp is a rename/copy from the same source */
1881 p->status = DIFF_STATUS_COPIED;
1882 break;
1884 if (!p->status)
1885 p->status = DIFF_STATUS_RENAMED;
1887 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1888 p->one->mode != p->two->mode)
1889 p->status = DIFF_STATUS_MODIFIED;
1890 else {
1891 /* This is a "no-change" entry and should not
1892 * happen anymore, but prepare for broken callers.
1894 error("feeding unmodified %s to diffcore",
1895 p->one->path);
1896 p->status = DIFF_STATUS_UNKNOWN;
1899 diff_debug_queue("resolve-rename-copy done", q);
1902 static void flush_one_pair(struct diff_filepair *p,
1903 int diff_output_format,
1904 struct diff_options *options,
1905 struct diffstat_t *diffstat)
1907 int inter_name_termination = '\t';
1908 int line_termination = options->line_termination;
1909 if (!line_termination)
1910 inter_name_termination = 0;
1912 switch (p->status) {
1913 case DIFF_STATUS_UNKNOWN:
1914 break;
1915 case 0:
1916 die("internal error in diff-resolve-rename-copy");
1917 break;
1918 default:
1919 switch (diff_output_format) {
1920 case DIFF_FORMAT_DIFFSTAT:
1921 diff_flush_stat(p, options, diffstat);
1922 break;
1923 case DIFF_FORMAT_CHECKDIFF:
1924 diff_flush_checkdiff(p, options);
1925 break;
1926 case DIFF_FORMAT_PATCH:
1927 diff_flush_patch(p, options);
1928 break;
1929 case DIFF_FORMAT_RAW:
1930 case DIFF_FORMAT_NAME_STATUS:
1931 diff_flush_raw(p, line_termination,
1932 inter_name_termination,
1933 options, diff_output_format);
1934 break;
1935 case DIFF_FORMAT_NAME:
1936 diff_flush_name(p,
1937 inter_name_termination,
1938 line_termination);
1939 break;
1940 case DIFF_FORMAT_NO_OUTPUT:
1941 break;
1946 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
1948 if (fs->mode)
1949 printf(" %s mode %06o %s\n", newdelete, fs->mode, fs->path);
1950 else
1951 printf(" %s %s\n", newdelete, fs->path);
1955 static void show_mode_change(struct diff_filepair *p, int show_name)
1957 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
1958 if (show_name)
1959 printf(" mode change %06o => %06o %s\n",
1960 p->one->mode, p->two->mode, p->two->path);
1961 else
1962 printf(" mode change %06o => %06o\n",
1963 p->one->mode, p->two->mode);
1967 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
1969 const char *old, *new;
1971 /* Find common prefix */
1972 old = p->one->path;
1973 new = p->two->path;
1974 while (1) {
1975 const char *slash_old, *slash_new;
1976 slash_old = strchr(old, '/');
1977 slash_new = strchr(new, '/');
1978 if (!slash_old ||
1979 !slash_new ||
1980 slash_old - old != slash_new - new ||
1981 memcmp(old, new, slash_new - new))
1982 break;
1983 old = slash_old + 1;
1984 new = slash_new + 1;
1986 /* p->one->path thru old is the common prefix, and old and new
1987 * through the end of names are renames
1989 if (old != p->one->path)
1990 printf(" %s %.*s{%s => %s} (%d%%)\n", renamecopy,
1991 (int)(old - p->one->path), p->one->path,
1992 old, new, (int)(0.5 + p->score * 100.0/MAX_SCORE));
1993 else
1994 printf(" %s %s => %s (%d%%)\n", renamecopy,
1995 p->one->path, p->two->path,
1996 (int)(0.5 + p->score * 100.0/MAX_SCORE));
1997 show_mode_change(p, 0);
2000 static void diff_summary(struct diff_filepair *p)
2002 switch(p->status) {
2003 case DIFF_STATUS_DELETED:
2004 show_file_mode_name("delete", p->one);
2005 break;
2006 case DIFF_STATUS_ADDED:
2007 show_file_mode_name("create", p->two);
2008 break;
2009 case DIFF_STATUS_COPIED:
2010 show_rename_copy("copy", p);
2011 break;
2012 case DIFF_STATUS_RENAMED:
2013 show_rename_copy("rename", p);
2014 break;
2015 default:
2016 if (p->score) {
2017 printf(" rewrite %s (%d%%)\n", p->two->path,
2018 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2019 show_mode_change(p, 0);
2020 } else show_mode_change(p, 1);
2021 break;
2025 void diff_flush(struct diff_options *options)
2027 struct diff_queue_struct *q = &diff_queued_diff;
2028 int i;
2029 int diff_output_format = options->output_format;
2030 struct diffstat_t *diffstat = NULL;
2032 if (diff_output_format == DIFF_FORMAT_DIFFSTAT || options->with_stat) {
2033 diffstat = xcalloc(sizeof (struct diffstat_t), 1);
2034 diffstat->xm.consume = diffstat_consume;
2037 if (options->with_raw) {
2038 for (i = 0; i < q->nr; i++) {
2039 struct diff_filepair *p = q->queue[i];
2040 flush_one_pair(p, DIFF_FORMAT_RAW, options, NULL);
2042 putchar(options->line_termination);
2044 if (options->with_stat) {
2045 for (i = 0; i < q->nr; i++) {
2046 struct diff_filepair *p = q->queue[i];
2047 flush_one_pair(p, DIFF_FORMAT_DIFFSTAT, options,
2048 diffstat);
2050 show_stats(diffstat);
2051 free(diffstat);
2052 diffstat = NULL;
2053 if (options->summary)
2054 for (i = 0; i < q->nr; i++)
2055 diff_summary(q->queue[i]);
2056 if (options->stat_sep)
2057 fputs(options->stat_sep, stdout);
2058 else
2059 putchar(options->line_termination);
2061 for (i = 0; i < q->nr; i++) {
2062 struct diff_filepair *p = q->queue[i];
2063 flush_one_pair(p, diff_output_format, options, diffstat);
2066 if (diffstat) {
2067 show_stats(diffstat);
2068 free(diffstat);
2071 for (i = 0; i < q->nr; i++) {
2072 if (diffstat && options->summary)
2073 diff_summary(q->queue[i]);
2074 diff_free_filepair(q->queue[i]);
2077 free(q->queue);
2078 q->queue = NULL;
2079 q->nr = q->alloc = 0;
2082 static void diffcore_apply_filter(const char *filter)
2084 int i;
2085 struct diff_queue_struct *q = &diff_queued_diff;
2086 struct diff_queue_struct outq;
2087 outq.queue = NULL;
2088 outq.nr = outq.alloc = 0;
2090 if (!filter)
2091 return;
2093 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2094 int found;
2095 for (i = found = 0; !found && i < q->nr; i++) {
2096 struct diff_filepair *p = q->queue[i];
2097 if (((p->status == DIFF_STATUS_MODIFIED) &&
2098 ((p->score &&
2099 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2100 (!p->score &&
2101 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2102 ((p->status != DIFF_STATUS_MODIFIED) &&
2103 strchr(filter, p->status)))
2104 found++;
2106 if (found)
2107 return;
2109 /* otherwise we will clear the whole queue
2110 * by copying the empty outq at the end of this
2111 * function, but first clear the current entries
2112 * in the queue.
2114 for (i = 0; i < q->nr; i++)
2115 diff_free_filepair(q->queue[i]);
2117 else {
2118 /* Only the matching ones */
2119 for (i = 0; i < q->nr; i++) {
2120 struct diff_filepair *p = q->queue[i];
2122 if (((p->status == DIFF_STATUS_MODIFIED) &&
2123 ((p->score &&
2124 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2125 (!p->score &&
2126 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2127 ((p->status != DIFF_STATUS_MODIFIED) &&
2128 strchr(filter, p->status)))
2129 diff_q(&outq, p);
2130 else
2131 diff_free_filepair(p);
2134 free(q->queue);
2135 *q = outq;
2138 void diffcore_std(struct diff_options *options)
2140 if (options->break_opt != -1)
2141 diffcore_break(options->break_opt);
2142 if (options->detect_rename)
2143 diffcore_rename(options);
2144 if (options->break_opt != -1)
2145 diffcore_merge_broken();
2146 if (options->pickaxe)
2147 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2148 if (options->orderfile)
2149 diffcore_order(options->orderfile);
2150 diff_resolve_rename_copy();
2151 diffcore_apply_filter(options->filter);
2155 void diffcore_std_no_resolve(struct diff_options *options)
2157 if (options->pickaxe)
2158 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2159 if (options->orderfile)
2160 diffcore_order(options->orderfile);
2161 diffcore_apply_filter(options->filter);
2164 void diff_addremove(struct diff_options *options,
2165 int addremove, unsigned mode,
2166 const unsigned char *sha1,
2167 const char *base, const char *path)
2169 char concatpath[PATH_MAX];
2170 struct diff_filespec *one, *two;
2172 /* This may look odd, but it is a preparation for
2173 * feeding "there are unchanged files which should
2174 * not produce diffs, but when you are doing copy
2175 * detection you would need them, so here they are"
2176 * entries to the diff-core. They will be prefixed
2177 * with something like '=' or '*' (I haven't decided
2178 * which but should not make any difference).
2179 * Feeding the same new and old to diff_change()
2180 * also has the same effect.
2181 * Before the final output happens, they are pruned after
2182 * merged into rename/copy pairs as appropriate.
2184 if (options->reverse_diff)
2185 addremove = (addremove == '+' ? '-' :
2186 addremove == '-' ? '+' : addremove);
2188 if (!path) path = "";
2189 sprintf(concatpath, "%s%s", base, path);
2190 one = alloc_filespec(concatpath);
2191 two = alloc_filespec(concatpath);
2193 if (addremove != '+')
2194 fill_filespec(one, sha1, mode);
2195 if (addremove != '-')
2196 fill_filespec(two, sha1, mode);
2198 diff_queue(&diff_queued_diff, one, two);
2201 void diff_change(struct diff_options *options,
2202 unsigned old_mode, unsigned new_mode,
2203 const unsigned char *old_sha1,
2204 const unsigned char *new_sha1,
2205 const char *base, const char *path)
2207 char concatpath[PATH_MAX];
2208 struct diff_filespec *one, *two;
2210 if (options->reverse_diff) {
2211 unsigned tmp;
2212 const unsigned char *tmp_c;
2213 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
2214 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
2216 if (!path) path = "";
2217 sprintf(concatpath, "%s%s", base, path);
2218 one = alloc_filespec(concatpath);
2219 two = alloc_filespec(concatpath);
2220 fill_filespec(one, old_sha1, old_mode);
2221 fill_filespec(two, new_sha1, new_mode);
2223 diff_queue(&diff_queued_diff, one, two);
2226 void diff_unmerge(struct diff_options *options,
2227 const char *path)
2229 struct diff_filespec *one, *two;
2230 one = alloc_filespec(path);
2231 two = alloc_filespec(path);
2232 diff_queue(&diff_queued_diff, one, two);