pack-objects: Allow missing base objects when creating thin packs
[alt-git.git] / diff.c
blob6954f992c2f268e2733af1e312a48ac3bc27799b
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"
15 #ifdef NO_FAST_WORKING_DIRECTORY
16 #define FAST_WORKING_DIRECTORY 0
17 #else
18 #define FAST_WORKING_DIRECTORY 1
19 #endif
21 static int diff_detect_rename_default;
22 static int diff_rename_limit_default = 200;
23 int diff_use_color_default = -1;
24 static const char *external_diff_cmd_cfg;
25 int diff_auto_refresh_index = 1;
27 static char diff_colors[][COLOR_MAXLEN] = {
28 "\033[m", /* reset */
29 "", /* PLAIN (normal) */
30 "\033[1m", /* METAINFO (bold) */
31 "\033[36m", /* FRAGINFO (cyan) */
32 "\033[31m", /* OLD (red) */
33 "\033[32m", /* NEW (green) */
34 "\033[33m", /* COMMIT (yellow) */
35 "\033[41m", /* WHITESPACE (red background) */
38 static int parse_diff_color_slot(const char *var, int ofs)
40 if (!strcasecmp(var+ofs, "plain"))
41 return DIFF_PLAIN;
42 if (!strcasecmp(var+ofs, "meta"))
43 return DIFF_METAINFO;
44 if (!strcasecmp(var+ofs, "frag"))
45 return DIFF_FRAGINFO;
46 if (!strcasecmp(var+ofs, "old"))
47 return DIFF_FILE_OLD;
48 if (!strcasecmp(var+ofs, "new"))
49 return DIFF_FILE_NEW;
50 if (!strcasecmp(var+ofs, "commit"))
51 return DIFF_COMMIT;
52 if (!strcasecmp(var+ofs, "whitespace"))
53 return DIFF_WHITESPACE;
54 die("bad config variable '%s'", var);
57 static struct ll_diff_driver {
58 const char *name;
59 struct ll_diff_driver *next;
60 const char *cmd;
61 } *user_diff, **user_diff_tail;
64 * Currently there is only "diff.<drivername>.command" variable;
65 * because there are "diff.color.<slot>" variables, we are parsing
66 * this in a bit convoluted way to allow low level diff driver
67 * called "color".
69 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
71 const char *name;
72 int namelen;
73 struct ll_diff_driver *drv;
75 name = var + 5;
76 namelen = ep - name;
77 for (drv = user_diff; drv; drv = drv->next)
78 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
79 break;
80 if (!drv) {
81 drv = xcalloc(1, sizeof(struct ll_diff_driver));
82 drv->name = xmemdupz(name, namelen);
83 if (!user_diff_tail)
84 user_diff_tail = &user_diff;
85 *user_diff_tail = drv;
86 user_diff_tail = &(drv->next);
89 return git_config_string(&(drv->cmd), var, value);
93 * 'diff.<what>.funcname' attribute can be specified in the configuration
94 * to define a customized regexp to find the beginning of a function to
95 * be used for hunk header lines of "diff -p" style output.
97 static struct funcname_pattern {
98 char *name;
99 char *pattern;
100 struct funcname_pattern *next;
101 } *funcname_pattern_list;
103 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
105 const char *name;
106 int namelen;
107 struct funcname_pattern *pp;
109 name = var + 5; /* "diff." */
110 namelen = ep - name;
112 for (pp = funcname_pattern_list; pp; pp = pp->next)
113 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
114 break;
115 if (!pp) {
116 pp = xcalloc(1, sizeof(*pp));
117 pp->name = xmemdupz(name, namelen);
118 pp->next = funcname_pattern_list;
119 funcname_pattern_list = pp;
121 free(pp->pattern);
122 pp->pattern = xstrdup(value);
123 return 0;
127 * These are to give UI layer defaults.
128 * The core-level commands such as git-diff-files should
129 * never be affected by the setting of diff.renames
130 * the user happens to have in the configuration file.
132 int git_diff_ui_config(const char *var, const char *value, void *cb)
134 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
135 diff_use_color_default = git_config_colorbool(var, value, -1);
136 return 0;
138 if (!strcmp(var, "diff.renames")) {
139 if (!value)
140 diff_detect_rename_default = DIFF_DETECT_RENAME;
141 else if (!strcasecmp(value, "copies") ||
142 !strcasecmp(value, "copy"))
143 diff_detect_rename_default = DIFF_DETECT_COPY;
144 else if (git_config_bool(var,value))
145 diff_detect_rename_default = DIFF_DETECT_RENAME;
146 return 0;
148 if (!strcmp(var, "diff.autorefreshindex")) {
149 diff_auto_refresh_index = git_config_bool(var, value);
150 return 0;
152 if (!strcmp(var, "diff.external"))
153 return git_config_string(&external_diff_cmd_cfg, var, value);
154 if (!prefixcmp(var, "diff.")) {
155 const char *ep = strrchr(var, '.');
157 if (ep != var + 4 && !strcmp(ep, ".command"))
158 return parse_lldiff_command(var, ep, value);
161 return git_diff_basic_config(var, value, cb);
164 int git_diff_basic_config(const char *var, const char *value, void *cb)
166 if (!strcmp(var, "diff.renamelimit")) {
167 diff_rename_limit_default = git_config_int(var, value);
168 return 0;
171 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
172 int slot = parse_diff_color_slot(var, 11);
173 if (!value)
174 return config_error_nonbool(var);
175 color_parse(value, var, diff_colors[slot]);
176 return 0;
179 if (!prefixcmp(var, "diff.")) {
180 const char *ep = strrchr(var, '.');
181 if (ep != var + 4) {
182 if (!strcmp(ep, ".funcname")) {
183 if (!value)
184 return config_error_nonbool(var);
185 return parse_funcname_pattern(var, ep, value);
190 return git_color_default_config(var, value, cb);
193 static char *quote_two(const char *one, const char *two)
195 int need_one = quote_c_style(one, NULL, NULL, 1);
196 int need_two = quote_c_style(two, NULL, NULL, 1);
197 struct strbuf res;
199 strbuf_init(&res, 0);
200 if (need_one + need_two) {
201 strbuf_addch(&res, '"');
202 quote_c_style(one, &res, NULL, 1);
203 quote_c_style(two, &res, NULL, 1);
204 strbuf_addch(&res, '"');
205 } else {
206 strbuf_addstr(&res, one);
207 strbuf_addstr(&res, two);
209 return strbuf_detach(&res, NULL);
212 static const char *external_diff(void)
214 static const char *external_diff_cmd = NULL;
215 static int done_preparing = 0;
217 if (done_preparing)
218 return external_diff_cmd;
219 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
220 if (!external_diff_cmd)
221 external_diff_cmd = external_diff_cmd_cfg;
222 done_preparing = 1;
223 return external_diff_cmd;
226 static struct diff_tempfile {
227 const char *name; /* filename external diff should read from */
228 char hex[41];
229 char mode[10];
230 char tmp_path[PATH_MAX];
231 } diff_temp[2];
233 static int count_lines(const char *data, int size)
235 int count, ch, completely_empty = 1, nl_just_seen = 0;
236 count = 0;
237 while (0 < size--) {
238 ch = *data++;
239 if (ch == '\n') {
240 count++;
241 nl_just_seen = 1;
242 completely_empty = 0;
244 else {
245 nl_just_seen = 0;
246 completely_empty = 0;
249 if (completely_empty)
250 return 0;
251 if (!nl_just_seen)
252 count++; /* no trailing newline */
253 return count;
256 static void print_line_count(FILE *file, int count)
258 switch (count) {
259 case 0:
260 fprintf(file, "0,0");
261 break;
262 case 1:
263 fprintf(file, "1");
264 break;
265 default:
266 fprintf(file, "1,%d", count);
267 break;
271 static void copy_file_with_prefix(FILE *file,
272 int prefix, const char *data, int size,
273 const char *set, const char *reset)
275 int ch, nl_just_seen = 1;
276 while (0 < size--) {
277 ch = *data++;
278 if (nl_just_seen) {
279 fputs(set, file);
280 putc(prefix, file);
282 if (ch == '\n') {
283 nl_just_seen = 1;
284 fputs(reset, file);
285 } else
286 nl_just_seen = 0;
287 putc(ch, file);
289 if (!nl_just_seen)
290 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
293 static void emit_rewrite_diff(const char *name_a,
294 const char *name_b,
295 struct diff_filespec *one,
296 struct diff_filespec *two,
297 struct diff_options *o)
299 int lc_a, lc_b;
300 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
301 const char *name_a_tab, *name_b_tab;
302 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
303 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
304 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
305 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
306 const char *reset = diff_get_color(color_diff, DIFF_RESET);
307 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
309 name_a += (*name_a == '/');
310 name_b += (*name_b == '/');
311 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
312 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
314 strbuf_reset(&a_name);
315 strbuf_reset(&b_name);
316 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
317 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
319 diff_populate_filespec(one, 0);
320 diff_populate_filespec(two, 0);
321 lc_a = count_lines(one->data, one->size);
322 lc_b = count_lines(two->data, two->size);
323 fprintf(o->file,
324 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
325 metainfo, a_name.buf, name_a_tab, reset,
326 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
327 print_line_count(o->file, lc_a);
328 fprintf(o->file, " +");
329 print_line_count(o->file, lc_b);
330 fprintf(o->file, " @@%s\n", reset);
331 if (lc_a)
332 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
333 if (lc_b)
334 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
337 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
339 if (!DIFF_FILE_VALID(one)) {
340 mf->ptr = (char *)""; /* does not matter */
341 mf->size = 0;
342 return 0;
344 else if (diff_populate_filespec(one, 0))
345 return -1;
346 mf->ptr = one->data;
347 mf->size = one->size;
348 return 0;
351 struct diff_words_buffer {
352 mmfile_t text;
353 long alloc;
354 long current; /* output pointer */
355 int suppressed_newline;
358 static void diff_words_append(char *line, unsigned long len,
359 struct diff_words_buffer *buffer)
361 if (buffer->text.size + len > buffer->alloc) {
362 buffer->alloc = (buffer->text.size + len) * 3 / 2;
363 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
365 line++;
366 len--;
367 memcpy(buffer->text.ptr + buffer->text.size, line, len);
368 buffer->text.size += len;
371 struct diff_words_data {
372 struct xdiff_emit_state xm;
373 struct diff_words_buffer minus, plus;
374 FILE *file;
377 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
378 int suppress_newline)
380 const char *ptr;
381 int eol = 0;
383 if (len == 0)
384 return;
386 ptr = buffer->text.ptr + buffer->current;
387 buffer->current += len;
389 if (ptr[len - 1] == '\n') {
390 eol = 1;
391 len--;
394 fputs(diff_get_color(1, color), file);
395 fwrite(ptr, len, 1, file);
396 fputs(diff_get_color(1, DIFF_RESET), file);
398 if (eol) {
399 if (suppress_newline)
400 buffer->suppressed_newline = 1;
401 else
402 putc('\n', file);
406 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
408 struct diff_words_data *diff_words = priv;
410 if (diff_words->minus.suppressed_newline) {
411 if (line[0] != '+')
412 putc('\n', diff_words->file);
413 diff_words->minus.suppressed_newline = 0;
416 len--;
417 switch (line[0]) {
418 case '-':
419 print_word(diff_words->file,
420 &diff_words->minus, len, DIFF_FILE_OLD, 1);
421 break;
422 case '+':
423 print_word(diff_words->file,
424 &diff_words->plus, len, DIFF_FILE_NEW, 0);
425 break;
426 case ' ':
427 print_word(diff_words->file,
428 &diff_words->plus, len, DIFF_PLAIN, 0);
429 diff_words->minus.current += len;
430 break;
434 /* this executes the word diff on the accumulated buffers */
435 static void diff_words_show(struct diff_words_data *diff_words)
437 xpparam_t xpp;
438 xdemitconf_t xecfg;
439 xdemitcb_t ecb;
440 mmfile_t minus, plus;
441 int i;
443 memset(&xecfg, 0, sizeof(xecfg));
444 minus.size = diff_words->minus.text.size;
445 minus.ptr = xmalloc(minus.size);
446 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
447 for (i = 0; i < minus.size; i++)
448 if (isspace(minus.ptr[i]))
449 minus.ptr[i] = '\n';
450 diff_words->minus.current = 0;
452 plus.size = diff_words->plus.text.size;
453 plus.ptr = xmalloc(plus.size);
454 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
455 for (i = 0; i < plus.size; i++)
456 if (isspace(plus.ptr[i]))
457 plus.ptr[i] = '\n';
458 diff_words->plus.current = 0;
460 xpp.flags = XDF_NEED_MINIMAL;
461 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
462 ecb.outf = xdiff_outf;
463 ecb.priv = diff_words;
464 diff_words->xm.consume = fn_out_diff_words_aux;
465 xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
467 free(minus.ptr);
468 free(plus.ptr);
469 diff_words->minus.text.size = diff_words->plus.text.size = 0;
471 if (diff_words->minus.suppressed_newline) {
472 putc('\n', diff_words->file);
473 diff_words->minus.suppressed_newline = 0;
477 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
479 struct emit_callback {
480 struct xdiff_emit_state xm;
481 int nparents, color_diff;
482 unsigned ws_rule;
483 sane_truncate_fn truncate;
484 const char **label_path;
485 struct diff_words_data *diff_words;
486 int *found_changesp;
487 FILE *file;
490 static void free_diff_words_data(struct emit_callback *ecbdata)
492 if (ecbdata->diff_words) {
493 /* flush buffers */
494 if (ecbdata->diff_words->minus.text.size ||
495 ecbdata->diff_words->plus.text.size)
496 diff_words_show(ecbdata->diff_words);
498 free (ecbdata->diff_words->minus.text.ptr);
499 free (ecbdata->diff_words->plus.text.ptr);
500 free(ecbdata->diff_words);
501 ecbdata->diff_words = NULL;
505 const char *diff_get_color(int diff_use_color, enum color_diff ix)
507 if (diff_use_color)
508 return diff_colors[ix];
509 return "";
512 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
514 int has_trailing_newline = (len > 0 && line[len-1] == '\n');
515 if (has_trailing_newline)
516 len--;
518 fputs(set, file);
519 fwrite(line, len, 1, file);
520 fputs(reset, file);
521 if (has_trailing_newline)
522 fputc('\n', file);
525 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
527 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
528 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
530 if (!*ws)
531 emit_line(ecbdata->file, set, reset, line, len);
532 else {
533 /* Emit just the prefix, then the rest. */
534 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
535 ws_check_emit(line + ecbdata->nparents,
536 len - ecbdata->nparents, ecbdata->ws_rule,
537 ecbdata->file, set, reset, ws);
541 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
543 const char *cp;
544 unsigned long allot;
545 size_t l = len;
547 if (ecb->truncate)
548 return ecb->truncate(line, len);
549 cp = line;
550 allot = l;
551 while (0 < l) {
552 (void) utf8_width(&cp, &l);
553 if (!cp)
554 break; /* truncated in the middle? */
556 return allot - l;
559 static void fn_out_consume(void *priv, char *line, unsigned long len)
561 int i;
562 int color;
563 struct emit_callback *ecbdata = priv;
564 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
565 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
566 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
568 *(ecbdata->found_changesp) = 1;
570 if (ecbdata->label_path[0]) {
571 const char *name_a_tab, *name_b_tab;
573 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
574 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
576 fprintf(ecbdata->file, "%s--- %s%s%s\n",
577 meta, ecbdata->label_path[0], reset, name_a_tab);
578 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
579 meta, ecbdata->label_path[1], reset, name_b_tab);
580 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
583 /* This is not really necessary for now because
584 * this codepath only deals with two-way diffs.
586 for (i = 0; i < len && line[i] == '@'; i++)
588 if (2 <= i && i < len && line[i] == ' ') {
589 ecbdata->nparents = i - 1;
590 len = sane_truncate_line(ecbdata, line, len);
591 emit_line(ecbdata->file,
592 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
593 reset, line, len);
594 if (line[len-1] != '\n')
595 putc('\n', ecbdata->file);
596 return;
599 if (len < ecbdata->nparents) {
600 emit_line(ecbdata->file, reset, reset, line, len);
601 return;
604 color = DIFF_PLAIN;
605 if (ecbdata->diff_words && ecbdata->nparents != 1)
606 /* fall back to normal diff */
607 free_diff_words_data(ecbdata);
608 if (ecbdata->diff_words) {
609 if (line[0] == '-') {
610 diff_words_append(line, len,
611 &ecbdata->diff_words->minus);
612 return;
613 } else if (line[0] == '+') {
614 diff_words_append(line, len,
615 &ecbdata->diff_words->plus);
616 return;
618 if (ecbdata->diff_words->minus.text.size ||
619 ecbdata->diff_words->plus.text.size)
620 diff_words_show(ecbdata->diff_words);
621 line++;
622 len--;
623 emit_line(ecbdata->file, plain, reset, line, len);
624 return;
626 for (i = 0; i < ecbdata->nparents && len; i++) {
627 if (line[i] == '-')
628 color = DIFF_FILE_OLD;
629 else if (line[i] == '+')
630 color = DIFF_FILE_NEW;
633 if (color != DIFF_FILE_NEW) {
634 emit_line(ecbdata->file,
635 diff_get_color(ecbdata->color_diff, color),
636 reset, line, len);
637 return;
639 emit_add_line(reset, ecbdata, line, len);
642 static char *pprint_rename(const char *a, const char *b)
644 const char *old = a;
645 const char *new = b;
646 struct strbuf name;
647 int pfx_length, sfx_length;
648 int len_a = strlen(a);
649 int len_b = strlen(b);
650 int a_midlen, b_midlen;
651 int qlen_a = quote_c_style(a, NULL, NULL, 0);
652 int qlen_b = quote_c_style(b, NULL, NULL, 0);
654 strbuf_init(&name, 0);
655 if (qlen_a || qlen_b) {
656 quote_c_style(a, &name, NULL, 0);
657 strbuf_addstr(&name, " => ");
658 quote_c_style(b, &name, NULL, 0);
659 return strbuf_detach(&name, NULL);
662 /* Find common prefix */
663 pfx_length = 0;
664 while (*old && *new && *old == *new) {
665 if (*old == '/')
666 pfx_length = old - a + 1;
667 old++;
668 new++;
671 /* Find common suffix */
672 old = a + len_a;
673 new = b + len_b;
674 sfx_length = 0;
675 while (a <= old && b <= new && *old == *new) {
676 if (*old == '/')
677 sfx_length = len_a - (old - a);
678 old--;
679 new--;
683 * pfx{mid-a => mid-b}sfx
684 * {pfx-a => pfx-b}sfx
685 * pfx{sfx-a => sfx-b}
686 * name-a => name-b
688 a_midlen = len_a - pfx_length - sfx_length;
689 b_midlen = len_b - pfx_length - sfx_length;
690 if (a_midlen < 0)
691 a_midlen = 0;
692 if (b_midlen < 0)
693 b_midlen = 0;
695 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
696 if (pfx_length + sfx_length) {
697 strbuf_add(&name, a, pfx_length);
698 strbuf_addch(&name, '{');
700 strbuf_add(&name, a + pfx_length, a_midlen);
701 strbuf_addstr(&name, " => ");
702 strbuf_add(&name, b + pfx_length, b_midlen);
703 if (pfx_length + sfx_length) {
704 strbuf_addch(&name, '}');
705 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
707 return strbuf_detach(&name, NULL);
710 struct diffstat_t {
711 struct xdiff_emit_state xm;
713 int nr;
714 int alloc;
715 struct diffstat_file {
716 char *from_name;
717 char *name;
718 char *print_name;
719 unsigned is_unmerged:1;
720 unsigned is_binary:1;
721 unsigned is_renamed:1;
722 unsigned int added, deleted;
723 } **files;
726 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
727 const char *name_a,
728 const char *name_b)
730 struct diffstat_file *x;
731 x = xcalloc(sizeof (*x), 1);
732 if (diffstat->nr == diffstat->alloc) {
733 diffstat->alloc = alloc_nr(diffstat->alloc);
734 diffstat->files = xrealloc(diffstat->files,
735 diffstat->alloc * sizeof(x));
737 diffstat->files[diffstat->nr++] = x;
738 if (name_b) {
739 x->from_name = xstrdup(name_a);
740 x->name = xstrdup(name_b);
741 x->is_renamed = 1;
743 else {
744 x->from_name = NULL;
745 x->name = xstrdup(name_a);
747 return x;
750 static void diffstat_consume(void *priv, char *line, unsigned long len)
752 struct diffstat_t *diffstat = priv;
753 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
755 if (line[0] == '+')
756 x->added++;
757 else if (line[0] == '-')
758 x->deleted++;
761 const char mime_boundary_leader[] = "------------";
763 static int scale_linear(int it, int width, int max_change)
766 * make sure that at least one '-' is printed if there were deletions,
767 * and likewise for '+'.
769 if (max_change < 2)
770 return it;
771 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
774 static void show_name(FILE *file,
775 const char *prefix, const char *name, int len,
776 const char *reset, const char *set)
778 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
781 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
783 if (cnt <= 0)
784 return;
785 fprintf(file, "%s", set);
786 while (cnt--)
787 putc(ch, file);
788 fprintf(file, "%s", reset);
791 static void fill_print_name(struct diffstat_file *file)
793 char *pname;
795 if (file->print_name)
796 return;
798 if (!file->is_renamed) {
799 struct strbuf buf;
800 strbuf_init(&buf, 0);
801 if (quote_c_style(file->name, &buf, NULL, 0)) {
802 pname = strbuf_detach(&buf, NULL);
803 } else {
804 pname = file->name;
805 strbuf_release(&buf);
807 } else {
808 pname = pprint_rename(file->from_name, file->name);
810 file->print_name = pname;
813 static void show_stats(struct diffstat_t* data, struct diff_options *options)
815 int i, len, add, del, total, adds = 0, dels = 0;
816 int max_change = 0, max_len = 0;
817 int total_files = data->nr;
818 int width, name_width;
819 const char *reset, *set, *add_c, *del_c;
821 if (data->nr == 0)
822 return;
824 width = options->stat_width ? options->stat_width : 80;
825 name_width = options->stat_name_width ? options->stat_name_width : 50;
827 /* Sanity: give at least 5 columns to the graph,
828 * but leave at least 10 columns for the name.
830 if (width < 25)
831 width = 25;
832 if (name_width < 10)
833 name_width = 10;
834 else if (width < name_width + 15)
835 name_width = width - 15;
837 /* Find the longest filename and max number of changes */
838 reset = diff_get_color_opt(options, DIFF_RESET);
839 set = diff_get_color_opt(options, DIFF_PLAIN);
840 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
841 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
843 for (i = 0; i < data->nr; i++) {
844 struct diffstat_file *file = data->files[i];
845 int change = file->added + file->deleted;
846 fill_print_name(file);
847 len = strlen(file->print_name);
848 if (max_len < len)
849 max_len = len;
851 if (file->is_binary || file->is_unmerged)
852 continue;
853 if (max_change < change)
854 max_change = change;
857 /* Compute the width of the graph part;
858 * 10 is for one blank at the beginning of the line plus
859 * " | count " between the name and the graph.
861 * From here on, name_width is the width of the name area,
862 * and width is the width of the graph area.
864 name_width = (name_width < max_len) ? name_width : max_len;
865 if (width < (name_width + 10) + max_change)
866 width = width - (name_width + 10);
867 else
868 width = max_change;
870 for (i = 0; i < data->nr; i++) {
871 const char *prefix = "";
872 char *name = data->files[i]->print_name;
873 int added = data->files[i]->added;
874 int deleted = data->files[i]->deleted;
875 int name_len;
878 * "scale" the filename
880 len = name_width;
881 name_len = strlen(name);
882 if (name_width < name_len) {
883 char *slash;
884 prefix = "...";
885 len -= 3;
886 name += name_len - len;
887 slash = strchr(name, '/');
888 if (slash)
889 name = slash;
892 if (data->files[i]->is_binary) {
893 show_name(options->file, prefix, name, len, reset, set);
894 fprintf(options->file, " Bin ");
895 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
896 fprintf(options->file, " -> ");
897 fprintf(options->file, "%s%d%s", add_c, added, reset);
898 fprintf(options->file, " bytes");
899 fprintf(options->file, "\n");
900 continue;
902 else if (data->files[i]->is_unmerged) {
903 show_name(options->file, prefix, name, len, reset, set);
904 fprintf(options->file, " Unmerged\n");
905 continue;
907 else if (!data->files[i]->is_renamed &&
908 (added + deleted == 0)) {
909 total_files--;
910 continue;
914 * scale the add/delete
916 add = added;
917 del = deleted;
918 total = add + del;
919 adds += add;
920 dels += del;
922 if (width <= max_change) {
923 add = scale_linear(add, width, max_change);
924 del = scale_linear(del, width, max_change);
925 total = add + del;
927 show_name(options->file, prefix, name, len, reset, set);
928 fprintf(options->file, "%5d%s", added + deleted,
929 added + deleted ? " " : "");
930 show_graph(options->file, '+', add, add_c, reset);
931 show_graph(options->file, '-', del, del_c, reset);
932 fprintf(options->file, "\n");
934 fprintf(options->file,
935 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
936 set, total_files, adds, dels, reset);
939 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
941 int i, adds = 0, dels = 0, total_files = data->nr;
943 if (data->nr == 0)
944 return;
946 for (i = 0; i < data->nr; i++) {
947 if (!data->files[i]->is_binary &&
948 !data->files[i]->is_unmerged) {
949 int added = data->files[i]->added;
950 int deleted= data->files[i]->deleted;
951 if (!data->files[i]->is_renamed &&
952 (added + deleted == 0)) {
953 total_files--;
954 } else {
955 adds += added;
956 dels += deleted;
960 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
961 total_files, adds, dels);
964 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
966 int i;
968 if (data->nr == 0)
969 return;
971 for (i = 0; i < data->nr; i++) {
972 struct diffstat_file *file = data->files[i];
974 if (file->is_binary)
975 fprintf(options->file, "-\t-\t");
976 else
977 fprintf(options->file,
978 "%d\t%d\t", file->added, file->deleted);
979 if (options->line_termination) {
980 fill_print_name(file);
981 if (!file->is_renamed)
982 write_name_quoted(file->name, options->file,
983 options->line_termination);
984 else {
985 fputs(file->print_name, options->file);
986 putc(options->line_termination, options->file);
988 } else {
989 if (file->is_renamed) {
990 putc('\0', options->file);
991 write_name_quoted(file->from_name, options->file, '\0');
993 write_name_quoted(file->name, options->file, '\0');
998 struct dirstat_file {
999 const char *name;
1000 unsigned long changed;
1003 struct dirstat_dir {
1004 struct dirstat_file *files;
1005 int alloc, nr, percent, cumulative;
1008 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1010 unsigned long this_dir = 0;
1011 unsigned int sources = 0;
1013 while (dir->nr) {
1014 struct dirstat_file *f = dir->files;
1015 int namelen = strlen(f->name);
1016 unsigned long this;
1017 char *slash;
1019 if (namelen < baselen)
1020 break;
1021 if (memcmp(f->name, base, baselen))
1022 break;
1023 slash = strchr(f->name + baselen, '/');
1024 if (slash) {
1025 int newbaselen = slash + 1 - f->name;
1026 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1027 sources++;
1028 } else {
1029 this = f->changed;
1030 dir->files++;
1031 dir->nr--;
1032 sources += 2;
1034 this_dir += this;
1038 * We don't report dirstat's for
1039 * - the top level
1040 * - or cases where everything came from a single directory
1041 * under this directory (sources == 1).
1043 if (baselen && sources != 1) {
1044 int permille = this_dir * 1000 / changed;
1045 if (permille) {
1046 int percent = permille / 10;
1047 if (percent >= dir->percent) {
1048 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1049 if (!dir->cumulative)
1050 return 0;
1054 return this_dir;
1057 static void show_dirstat(struct diff_options *options)
1059 int i;
1060 unsigned long changed;
1061 struct dirstat_dir dir;
1062 struct diff_queue_struct *q = &diff_queued_diff;
1064 dir.files = NULL;
1065 dir.alloc = 0;
1066 dir.nr = 0;
1067 dir.percent = options->dirstat_percent;
1068 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1070 changed = 0;
1071 for (i = 0; i < q->nr; i++) {
1072 struct diff_filepair *p = q->queue[i];
1073 const char *name;
1074 unsigned long copied, added, damage;
1076 name = p->one->path ? p->one->path : p->two->path;
1078 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1079 diff_populate_filespec(p->one, 0);
1080 diff_populate_filespec(p->two, 0);
1081 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1082 &copied, &added);
1083 diff_free_filespec_data(p->one);
1084 diff_free_filespec_data(p->two);
1085 } else if (DIFF_FILE_VALID(p->one)) {
1086 diff_populate_filespec(p->one, 1);
1087 copied = added = 0;
1088 diff_free_filespec_data(p->one);
1089 } else if (DIFF_FILE_VALID(p->two)) {
1090 diff_populate_filespec(p->two, 1);
1091 copied = 0;
1092 added = p->two->size;
1093 diff_free_filespec_data(p->two);
1094 } else
1095 continue;
1098 * Original minus copied is the removed material,
1099 * added is the new material. They are both damages
1100 * made to the preimage.
1102 damage = (p->one->size - copied) + added;
1104 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1105 dir.files[dir.nr].name = name;
1106 dir.files[dir.nr].changed = damage;
1107 changed += damage;
1108 dir.nr++;
1111 /* This can happen even with many files, if everything was renames */
1112 if (!changed)
1113 return;
1115 /* Show all directories with more than x% of the changes */
1116 gather_dirstat(options->file, &dir, changed, "", 0);
1119 static void free_diffstat_info(struct diffstat_t *diffstat)
1121 int i;
1122 for (i = 0; i < diffstat->nr; i++) {
1123 struct diffstat_file *f = diffstat->files[i];
1124 if (f->name != f->print_name)
1125 free(f->print_name);
1126 free(f->name);
1127 free(f->from_name);
1128 free(f);
1130 free(diffstat->files);
1133 struct checkdiff_t {
1134 struct xdiff_emit_state xm;
1135 const char *filename;
1136 int lineno;
1137 struct diff_options *o;
1138 unsigned ws_rule;
1139 unsigned status;
1140 int trailing_blanks_start;
1143 static int is_conflict_marker(const char *line, unsigned long len)
1145 char firstchar;
1146 int cnt;
1148 if (len < 8)
1149 return 0;
1150 firstchar = line[0];
1151 switch (firstchar) {
1152 case '=': case '>': case '<':
1153 break;
1154 default:
1155 return 0;
1157 for (cnt = 1; cnt < 7; cnt++)
1158 if (line[cnt] != firstchar)
1159 return 0;
1160 /* line[0] thru line[6] are same as firstchar */
1161 if (firstchar == '=') {
1162 /* divider between ours and theirs? */
1163 if (len != 8 || line[7] != '\n')
1164 return 0;
1165 } else if (len < 8 || !isspace(line[7])) {
1166 /* not divider before ours nor after theirs */
1167 return 0;
1169 return 1;
1172 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1174 struct checkdiff_t *data = priv;
1175 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1176 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1177 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1178 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1179 char *err;
1181 if (line[0] == '+') {
1182 unsigned bad;
1183 data->lineno++;
1184 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1185 data->trailing_blanks_start = 0;
1186 else if (!data->trailing_blanks_start)
1187 data->trailing_blanks_start = data->lineno;
1188 if (is_conflict_marker(line + 1, len - 1)) {
1189 data->status |= 1;
1190 fprintf(data->o->file,
1191 "%s:%d: leftover conflict marker\n",
1192 data->filename, data->lineno);
1194 bad = ws_check(line + 1, len - 1, data->ws_rule);
1195 if (!bad)
1196 return;
1197 data->status |= bad;
1198 err = whitespace_error_string(bad);
1199 fprintf(data->o->file, "%s:%d: %s.\n",
1200 data->filename, data->lineno, err);
1201 free(err);
1202 emit_line(data->o->file, set, reset, line, 1);
1203 ws_check_emit(line + 1, len - 1, data->ws_rule,
1204 data->o->file, set, reset, ws);
1205 } else if (line[0] == ' ') {
1206 data->lineno++;
1207 data->trailing_blanks_start = 0;
1208 } else if (line[0] == '@') {
1209 char *plus = strchr(line, '+');
1210 if (plus)
1211 data->lineno = strtol(plus, NULL, 10) - 1;
1212 else
1213 die("invalid diff");
1214 data->trailing_blanks_start = 0;
1218 static unsigned char *deflate_it(char *data,
1219 unsigned long size,
1220 unsigned long *result_size)
1222 int bound;
1223 unsigned char *deflated;
1224 z_stream stream;
1226 memset(&stream, 0, sizeof(stream));
1227 deflateInit(&stream, zlib_compression_level);
1228 bound = deflateBound(&stream, size);
1229 deflated = xmalloc(bound);
1230 stream.next_out = deflated;
1231 stream.avail_out = bound;
1233 stream.next_in = (unsigned char *)data;
1234 stream.avail_in = size;
1235 while (deflate(&stream, Z_FINISH) == Z_OK)
1236 ; /* nothing */
1237 deflateEnd(&stream);
1238 *result_size = stream.total_out;
1239 return deflated;
1242 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1244 void *cp;
1245 void *delta;
1246 void *deflated;
1247 void *data;
1248 unsigned long orig_size;
1249 unsigned long delta_size;
1250 unsigned long deflate_size;
1251 unsigned long data_size;
1253 /* We could do deflated delta, or we could do just deflated two,
1254 * whichever is smaller.
1256 delta = NULL;
1257 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1258 if (one->size && two->size) {
1259 delta = diff_delta(one->ptr, one->size,
1260 two->ptr, two->size,
1261 &delta_size, deflate_size);
1262 if (delta) {
1263 void *to_free = delta;
1264 orig_size = delta_size;
1265 delta = deflate_it(delta, delta_size, &delta_size);
1266 free(to_free);
1270 if (delta && delta_size < deflate_size) {
1271 fprintf(file, "delta %lu\n", orig_size);
1272 free(deflated);
1273 data = delta;
1274 data_size = delta_size;
1276 else {
1277 fprintf(file, "literal %lu\n", two->size);
1278 free(delta);
1279 data = deflated;
1280 data_size = deflate_size;
1283 /* emit data encoded in base85 */
1284 cp = data;
1285 while (data_size) {
1286 int bytes = (52 < data_size) ? 52 : data_size;
1287 char line[70];
1288 data_size -= bytes;
1289 if (bytes <= 26)
1290 line[0] = bytes + 'A' - 1;
1291 else
1292 line[0] = bytes - 26 + 'a' - 1;
1293 encode_85(line + 1, cp, bytes);
1294 cp = (char *) cp + bytes;
1295 fputs(line, file);
1296 fputc('\n', file);
1298 fprintf(file, "\n");
1299 free(data);
1302 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1304 fprintf(file, "GIT binary patch\n");
1305 emit_binary_diff_body(file, one, two);
1306 emit_binary_diff_body(file, two, one);
1309 static void setup_diff_attr_check(struct git_attr_check *check)
1311 static struct git_attr *attr_diff;
1313 if (!attr_diff) {
1314 attr_diff = git_attr("diff", 4);
1316 check[0].attr = attr_diff;
1319 static void diff_filespec_check_attr(struct diff_filespec *one)
1321 struct git_attr_check attr_diff_check;
1322 int check_from_data = 0;
1324 if (one->checked_attr)
1325 return;
1327 setup_diff_attr_check(&attr_diff_check);
1328 one->is_binary = 0;
1329 one->funcname_pattern_ident = NULL;
1331 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1332 const char *value;
1334 /* binaryness */
1335 value = attr_diff_check.value;
1336 if (ATTR_TRUE(value))
1338 else if (ATTR_FALSE(value))
1339 one->is_binary = 1;
1340 else
1341 check_from_data = 1;
1343 /* funcname pattern ident */
1344 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1346 else
1347 one->funcname_pattern_ident = value;
1350 if (check_from_data) {
1351 if (!one->data && DIFF_FILE_VALID(one))
1352 diff_populate_filespec(one, 0);
1354 if (one->data)
1355 one->is_binary = buffer_is_binary(one->data, one->size);
1359 int diff_filespec_is_binary(struct diff_filespec *one)
1361 diff_filespec_check_attr(one);
1362 return one->is_binary;
1365 static const char *funcname_pattern(const char *ident)
1367 struct funcname_pattern *pp;
1369 for (pp = funcname_pattern_list; pp; pp = pp->next)
1370 if (!strcmp(ident, pp->name))
1371 return pp->pattern;
1372 return NULL;
1375 static struct builtin_funcname_pattern {
1376 const char *name;
1377 const char *pattern;
1378 } builtin_funcname_pattern[] = {
1379 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1380 "new\\|return\\|switch\\|throw\\|while\\)\n"
1381 "^[ ]*\\(\\([ ]*"
1382 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1383 "[ ]*([^;]*\\)$" },
1384 { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
1385 "destructor\\|interface\\|implementation\\|"
1386 "initialization\\|finalization\\)[ \t]*.*\\)$"
1387 "\\|"
1388 "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
1390 { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
1391 { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
1394 static const char *diff_funcname_pattern(struct diff_filespec *one)
1396 const char *ident, *pattern;
1397 int i;
1399 diff_filespec_check_attr(one);
1400 ident = one->funcname_pattern_ident;
1402 if (!ident)
1404 * If the config file has "funcname.default" defined, that
1405 * regexp is used; otherwise NULL is returned and xemit uses
1406 * the built-in default.
1408 return funcname_pattern("default");
1410 /* Look up custom "funcname.$ident" regexp from config. */
1411 pattern = funcname_pattern(ident);
1412 if (pattern)
1413 return pattern;
1416 * And define built-in fallback patterns here. Note that
1417 * these can be overridden by the user's config settings.
1419 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1420 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1421 return builtin_funcname_pattern[i].pattern;
1423 return NULL;
1426 static void builtin_diff(const char *name_a,
1427 const char *name_b,
1428 struct diff_filespec *one,
1429 struct diff_filespec *two,
1430 const char *xfrm_msg,
1431 struct diff_options *o,
1432 int complete_rewrite)
1434 mmfile_t mf1, mf2;
1435 const char *lbl[2];
1436 char *a_one, *b_two;
1437 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1438 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1440 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1441 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1442 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1443 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1444 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1445 if (lbl[0][0] == '/') {
1446 /* /dev/null */
1447 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1448 if (xfrm_msg && xfrm_msg[0])
1449 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1451 else if (lbl[1][0] == '/') {
1452 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1453 if (xfrm_msg && xfrm_msg[0])
1454 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1456 else {
1457 if (one->mode != two->mode) {
1458 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1459 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1461 if (xfrm_msg && xfrm_msg[0])
1462 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1464 * we do not run diff between different kind
1465 * of objects.
1467 if ((one->mode ^ two->mode) & S_IFMT)
1468 goto free_ab_and_return;
1469 if (complete_rewrite) {
1470 emit_rewrite_diff(name_a, name_b, one, two, o);
1471 o->found_changes = 1;
1472 goto free_ab_and_return;
1476 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1477 die("unable to read files to diff");
1479 if (!DIFF_OPT_TST(o, TEXT) &&
1480 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1481 /* Quite common confusing case */
1482 if (mf1.size == mf2.size &&
1483 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1484 goto free_ab_and_return;
1485 if (DIFF_OPT_TST(o, BINARY))
1486 emit_binary_diff(o->file, &mf1, &mf2);
1487 else
1488 fprintf(o->file, "Binary files %s and %s differ\n",
1489 lbl[0], lbl[1]);
1490 o->found_changes = 1;
1492 else {
1493 /* Crazy xdl interfaces.. */
1494 const char *diffopts = getenv("GIT_DIFF_OPTS");
1495 xpparam_t xpp;
1496 xdemitconf_t xecfg;
1497 xdemitcb_t ecb;
1498 struct emit_callback ecbdata;
1499 const char *funcname_pattern;
1501 funcname_pattern = diff_funcname_pattern(one);
1502 if (!funcname_pattern)
1503 funcname_pattern = diff_funcname_pattern(two);
1505 memset(&xecfg, 0, sizeof(xecfg));
1506 memset(&ecbdata, 0, sizeof(ecbdata));
1507 ecbdata.label_path = lbl;
1508 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1509 ecbdata.found_changesp = &o->found_changes;
1510 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1511 ecbdata.file = o->file;
1512 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1513 xecfg.ctxlen = o->context;
1514 xecfg.flags = XDL_EMIT_FUNCNAMES;
1515 if (funcname_pattern)
1516 xdiff_set_find_func(&xecfg, funcname_pattern);
1517 if (!diffopts)
1519 else if (!prefixcmp(diffopts, "--unified="))
1520 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1521 else if (!prefixcmp(diffopts, "-u"))
1522 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1523 ecb.outf = xdiff_outf;
1524 ecb.priv = &ecbdata;
1525 ecbdata.xm.consume = fn_out_consume;
1526 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1527 ecbdata.diff_words =
1528 xcalloc(1, sizeof(struct diff_words_data));
1529 ecbdata.diff_words->file = o->file;
1531 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1532 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1533 free_diff_words_data(&ecbdata);
1536 free_ab_and_return:
1537 diff_free_filespec_data(one);
1538 diff_free_filespec_data(two);
1539 free(a_one);
1540 free(b_two);
1541 return;
1544 static void builtin_diffstat(const char *name_a, const char *name_b,
1545 struct diff_filespec *one,
1546 struct diff_filespec *two,
1547 struct diffstat_t *diffstat,
1548 struct diff_options *o,
1549 int complete_rewrite)
1551 mmfile_t mf1, mf2;
1552 struct diffstat_file *data;
1554 data = diffstat_add(diffstat, name_a, name_b);
1556 if (!one || !two) {
1557 data->is_unmerged = 1;
1558 return;
1560 if (complete_rewrite) {
1561 diff_populate_filespec(one, 0);
1562 diff_populate_filespec(two, 0);
1563 data->deleted = count_lines(one->data, one->size);
1564 data->added = count_lines(two->data, two->size);
1565 goto free_and_return;
1567 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1568 die("unable to read files to diff");
1570 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1571 data->is_binary = 1;
1572 data->added = mf2.size;
1573 data->deleted = mf1.size;
1574 } else {
1575 /* Crazy xdl interfaces.. */
1576 xpparam_t xpp;
1577 xdemitconf_t xecfg;
1578 xdemitcb_t ecb;
1580 memset(&xecfg, 0, sizeof(xecfg));
1581 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1582 ecb.outf = xdiff_outf;
1583 ecb.priv = diffstat;
1584 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1587 free_and_return:
1588 diff_free_filespec_data(one);
1589 diff_free_filespec_data(two);
1592 static void builtin_checkdiff(const char *name_a, const char *name_b,
1593 const char *attr_path,
1594 struct diff_filespec *one,
1595 struct diff_filespec *two,
1596 struct diff_options *o)
1598 mmfile_t mf1, mf2;
1599 struct checkdiff_t data;
1601 if (!two)
1602 return;
1604 memset(&data, 0, sizeof(data));
1605 data.xm.consume = checkdiff_consume;
1606 data.filename = name_b ? name_b : name_a;
1607 data.lineno = 0;
1608 data.o = o;
1609 data.ws_rule = whitespace_rule(attr_path);
1611 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1612 die("unable to read files to diff");
1615 * All the other codepaths check both sides, but not checking
1616 * the "old" side here is deliberate. We are checking the newly
1617 * introduced changes, and as long as the "new" side is text, we
1618 * can and should check what it introduces.
1620 if (diff_filespec_is_binary(two))
1621 goto free_and_return;
1622 else {
1623 /* Crazy xdl interfaces.. */
1624 xpparam_t xpp;
1625 xdemitconf_t xecfg;
1626 xdemitcb_t ecb;
1628 memset(&xecfg, 0, sizeof(xecfg));
1629 xpp.flags = XDF_NEED_MINIMAL;
1630 ecb.outf = xdiff_outf;
1631 ecb.priv = &data;
1632 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1634 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1635 data.trailing_blanks_start) {
1636 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1637 data.filename, data.trailing_blanks_start);
1638 data.status = 1; /* report errors */
1641 free_and_return:
1642 diff_free_filespec_data(one);
1643 diff_free_filespec_data(two);
1644 if (data.status)
1645 DIFF_OPT_SET(o, CHECK_FAILED);
1648 struct diff_filespec *alloc_filespec(const char *path)
1650 int namelen = strlen(path);
1651 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1653 memset(spec, 0, sizeof(*spec));
1654 spec->path = (char *)(spec + 1);
1655 memcpy(spec->path, path, namelen+1);
1656 spec->count = 1;
1657 return spec;
1660 void free_filespec(struct diff_filespec *spec)
1662 if (!--spec->count) {
1663 diff_free_filespec_data(spec);
1664 free(spec);
1668 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1669 unsigned short mode)
1671 if (mode) {
1672 spec->mode = canon_mode(mode);
1673 hashcpy(spec->sha1, sha1);
1674 spec->sha1_valid = !is_null_sha1(sha1);
1679 * Given a name and sha1 pair, if the index tells us the file in
1680 * the work tree has that object contents, return true, so that
1681 * prepare_temp_file() does not have to inflate and extract.
1683 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1685 struct cache_entry *ce;
1686 struct stat st;
1687 int pos, len;
1689 /* We do not read the cache ourselves here, because the
1690 * benchmark with my previous version that always reads cache
1691 * shows that it makes things worse for diff-tree comparing
1692 * two linux-2.6 kernel trees in an already checked out work
1693 * tree. This is because most diff-tree comparisons deal with
1694 * only a small number of files, while reading the cache is
1695 * expensive for a large project, and its cost outweighs the
1696 * savings we get by not inflating the object to a temporary
1697 * file. Practically, this code only helps when we are used
1698 * by diff-cache --cached, which does read the cache before
1699 * calling us.
1701 if (!active_cache)
1702 return 0;
1704 /* We want to avoid the working directory if our caller
1705 * doesn't need the data in a normal file, this system
1706 * is rather slow with its stat/open/mmap/close syscalls,
1707 * and the object is contained in a pack file. The pack
1708 * is probably already open and will be faster to obtain
1709 * the data through than the working directory. Loose
1710 * objects however would tend to be slower as they need
1711 * to be individually opened and inflated.
1713 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1714 return 0;
1716 len = strlen(name);
1717 pos = cache_name_pos(name, len);
1718 if (pos < 0)
1719 return 0;
1720 ce = active_cache[pos];
1723 * This is not the sha1 we are looking for, or
1724 * unreusable because it is not a regular file.
1726 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1727 return 0;
1730 * If ce matches the file in the work tree, we can reuse it.
1732 if (ce_uptodate(ce) ||
1733 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1734 return 1;
1736 return 0;
1739 static int populate_from_stdin(struct diff_filespec *s)
1741 struct strbuf buf;
1742 size_t size = 0;
1744 strbuf_init(&buf, 0);
1745 if (strbuf_read(&buf, 0, 0) < 0)
1746 return error("error while reading from stdin %s",
1747 strerror(errno));
1749 s->should_munmap = 0;
1750 s->data = strbuf_detach(&buf, &size);
1751 s->size = size;
1752 s->should_free = 1;
1753 return 0;
1756 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1758 int len;
1759 char *data = xmalloc(100);
1760 len = snprintf(data, 100,
1761 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1762 s->data = data;
1763 s->size = len;
1764 s->should_free = 1;
1765 if (size_only) {
1766 s->data = NULL;
1767 free(data);
1769 return 0;
1773 * While doing rename detection and pickaxe operation, we may need to
1774 * grab the data for the blob (or file) for our own in-core comparison.
1775 * diff_filespec has data and size fields for this purpose.
1777 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1779 int err = 0;
1780 if (!DIFF_FILE_VALID(s))
1781 die("internal error: asking to populate invalid file.");
1782 if (S_ISDIR(s->mode))
1783 return -1;
1785 if (s->data)
1786 return 0;
1788 if (size_only && 0 < s->size)
1789 return 0;
1791 if (S_ISGITLINK(s->mode))
1792 return diff_populate_gitlink(s, size_only);
1794 if (!s->sha1_valid ||
1795 reuse_worktree_file(s->path, s->sha1, 0)) {
1796 struct strbuf buf;
1797 struct stat st;
1798 int fd;
1800 if (!strcmp(s->path, "-"))
1801 return populate_from_stdin(s);
1803 if (lstat(s->path, &st) < 0) {
1804 if (errno == ENOENT) {
1805 err_empty:
1806 err = -1;
1807 empty:
1808 s->data = (char *)"";
1809 s->size = 0;
1810 return err;
1813 s->size = xsize_t(st.st_size);
1814 if (!s->size)
1815 goto empty;
1816 if (size_only)
1817 return 0;
1818 if (S_ISLNK(st.st_mode)) {
1819 int ret;
1820 s->data = xmalloc(s->size);
1821 s->should_free = 1;
1822 ret = readlink(s->path, s->data, s->size);
1823 if (ret < 0) {
1824 free(s->data);
1825 goto err_empty;
1827 return 0;
1829 fd = open(s->path, O_RDONLY);
1830 if (fd < 0)
1831 goto err_empty;
1832 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1833 close(fd);
1834 s->should_munmap = 1;
1837 * Convert from working tree format to canonical git format
1839 strbuf_init(&buf, 0);
1840 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1841 size_t size = 0;
1842 munmap(s->data, s->size);
1843 s->should_munmap = 0;
1844 s->data = strbuf_detach(&buf, &size);
1845 s->size = size;
1846 s->should_free = 1;
1849 else {
1850 enum object_type type;
1851 if (size_only)
1852 type = sha1_object_info(s->sha1, &s->size);
1853 else {
1854 s->data = read_sha1_file(s->sha1, &type, &s->size);
1855 s->should_free = 1;
1858 return 0;
1861 void diff_free_filespec_blob(struct diff_filespec *s)
1863 if (s->should_free)
1864 free(s->data);
1865 else if (s->should_munmap)
1866 munmap(s->data, s->size);
1868 if (s->should_free || s->should_munmap) {
1869 s->should_free = s->should_munmap = 0;
1870 s->data = NULL;
1874 void diff_free_filespec_data(struct diff_filespec *s)
1876 diff_free_filespec_blob(s);
1877 free(s->cnt_data);
1878 s->cnt_data = NULL;
1881 static void prep_temp_blob(struct diff_tempfile *temp,
1882 void *blob,
1883 unsigned long size,
1884 const unsigned char *sha1,
1885 int mode)
1887 int fd;
1889 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1890 if (fd < 0)
1891 die("unable to create temp-file: %s", strerror(errno));
1892 if (write_in_full(fd, blob, size) != size)
1893 die("unable to write temp-file");
1894 close(fd);
1895 temp->name = temp->tmp_path;
1896 strcpy(temp->hex, sha1_to_hex(sha1));
1897 temp->hex[40] = 0;
1898 sprintf(temp->mode, "%06o", mode);
1901 static void prepare_temp_file(const char *name,
1902 struct diff_tempfile *temp,
1903 struct diff_filespec *one)
1905 if (!DIFF_FILE_VALID(one)) {
1906 not_a_valid_file:
1907 /* A '-' entry produces this for file-2, and
1908 * a '+' entry produces this for file-1.
1910 temp->name = "/dev/null";
1911 strcpy(temp->hex, ".");
1912 strcpy(temp->mode, ".");
1913 return;
1916 if (!one->sha1_valid ||
1917 reuse_worktree_file(name, one->sha1, 1)) {
1918 struct stat st;
1919 if (lstat(name, &st) < 0) {
1920 if (errno == ENOENT)
1921 goto not_a_valid_file;
1922 die("stat(%s): %s", name, strerror(errno));
1924 if (S_ISLNK(st.st_mode)) {
1925 int ret;
1926 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1927 size_t sz = xsize_t(st.st_size);
1928 if (sizeof(buf) <= st.st_size)
1929 die("symlink too long: %s", name);
1930 ret = readlink(name, buf, sz);
1931 if (ret < 0)
1932 die("readlink(%s)", name);
1933 prep_temp_blob(temp, buf, sz,
1934 (one->sha1_valid ?
1935 one->sha1 : null_sha1),
1936 (one->sha1_valid ?
1937 one->mode : S_IFLNK));
1939 else {
1940 /* we can borrow from the file in the work tree */
1941 temp->name = name;
1942 if (!one->sha1_valid)
1943 strcpy(temp->hex, sha1_to_hex(null_sha1));
1944 else
1945 strcpy(temp->hex, sha1_to_hex(one->sha1));
1946 /* Even though we may sometimes borrow the
1947 * contents from the work tree, we always want
1948 * one->mode. mode is trustworthy even when
1949 * !(one->sha1_valid), as long as
1950 * DIFF_FILE_VALID(one).
1952 sprintf(temp->mode, "%06o", one->mode);
1954 return;
1956 else {
1957 if (diff_populate_filespec(one, 0))
1958 die("cannot read data blob for %s", one->path);
1959 prep_temp_blob(temp, one->data, one->size,
1960 one->sha1, one->mode);
1964 static void remove_tempfile(void)
1966 int i;
1968 for (i = 0; i < 2; i++)
1969 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1970 unlink(diff_temp[i].name);
1971 diff_temp[i].name = NULL;
1975 static void remove_tempfile_on_signal(int signo)
1977 remove_tempfile();
1978 signal(SIGINT, SIG_DFL);
1979 raise(signo);
1982 /* An external diff command takes:
1984 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1985 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1988 static void run_external_diff(const char *pgm,
1989 const char *name,
1990 const char *other,
1991 struct diff_filespec *one,
1992 struct diff_filespec *two,
1993 const char *xfrm_msg,
1994 int complete_rewrite)
1996 const char *spawn_arg[10];
1997 struct diff_tempfile *temp = diff_temp;
1998 int retval;
1999 static int atexit_asked = 0;
2000 const char *othername;
2001 const char **arg = &spawn_arg[0];
2003 othername = (other? other : name);
2004 if (one && two) {
2005 prepare_temp_file(name, &temp[0], one);
2006 prepare_temp_file(othername, &temp[1], two);
2007 if (! atexit_asked &&
2008 (temp[0].name == temp[0].tmp_path ||
2009 temp[1].name == temp[1].tmp_path)) {
2010 atexit_asked = 1;
2011 atexit(remove_tempfile);
2013 signal(SIGINT, remove_tempfile_on_signal);
2016 if (one && two) {
2017 *arg++ = pgm;
2018 *arg++ = name;
2019 *arg++ = temp[0].name;
2020 *arg++ = temp[0].hex;
2021 *arg++ = temp[0].mode;
2022 *arg++ = temp[1].name;
2023 *arg++ = temp[1].hex;
2024 *arg++ = temp[1].mode;
2025 if (other) {
2026 *arg++ = other;
2027 *arg++ = xfrm_msg;
2029 } else {
2030 *arg++ = pgm;
2031 *arg++ = name;
2033 *arg = NULL;
2034 fflush(NULL);
2035 retval = run_command_v_opt(spawn_arg, 0);
2036 remove_tempfile();
2037 if (retval) {
2038 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2039 exit(1);
2043 static const char *external_diff_attr(const char *name)
2045 struct git_attr_check attr_diff_check;
2047 if (!name)
2048 return NULL;
2050 setup_diff_attr_check(&attr_diff_check);
2051 if (!git_checkattr(name, 1, &attr_diff_check)) {
2052 const char *value = attr_diff_check.value;
2053 if (!ATTR_TRUE(value) &&
2054 !ATTR_FALSE(value) &&
2055 !ATTR_UNSET(value)) {
2056 struct ll_diff_driver *drv;
2058 for (drv = user_diff; drv; drv = drv->next)
2059 if (!strcmp(drv->name, value))
2060 return drv->cmd;
2063 return NULL;
2066 static void run_diff_cmd(const char *pgm,
2067 const char *name,
2068 const char *other,
2069 const char *attr_path,
2070 struct diff_filespec *one,
2071 struct diff_filespec *two,
2072 const char *xfrm_msg,
2073 struct diff_options *o,
2074 int complete_rewrite)
2076 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2077 pgm = NULL;
2078 else {
2079 const char *cmd = external_diff_attr(attr_path);
2080 if (cmd)
2081 pgm = cmd;
2084 if (pgm) {
2085 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2086 complete_rewrite);
2087 return;
2089 if (one && two)
2090 builtin_diff(name, other ? other : name,
2091 one, two, xfrm_msg, o, complete_rewrite);
2092 else
2093 fprintf(o->file, "* Unmerged path %s\n", name);
2096 static void diff_fill_sha1_info(struct diff_filespec *one)
2098 if (DIFF_FILE_VALID(one)) {
2099 if (!one->sha1_valid) {
2100 struct stat st;
2101 if (!strcmp(one->path, "-")) {
2102 hashcpy(one->sha1, null_sha1);
2103 return;
2105 if (lstat(one->path, &st) < 0)
2106 die("stat %s", one->path);
2107 if (index_path(one->sha1, one->path, &st, 0))
2108 die("cannot hash %s\n", one->path);
2111 else
2112 hashclr(one->sha1);
2115 static int similarity_index(struct diff_filepair *p)
2117 return p->score * 100 / MAX_SCORE;
2120 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2122 /* Strip the prefix but do not molest /dev/null and absolute paths */
2123 if (*namep && **namep != '/')
2124 *namep += prefix_length;
2125 if (*otherp && **otherp != '/')
2126 *otherp += prefix_length;
2129 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2131 const char *pgm = external_diff();
2132 struct strbuf msg;
2133 char *xfrm_msg;
2134 struct diff_filespec *one = p->one;
2135 struct diff_filespec *two = p->two;
2136 const char *name;
2137 const char *other;
2138 const char *attr_path;
2139 int complete_rewrite = 0;
2141 name = p->one->path;
2142 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2143 attr_path = name;
2144 if (o->prefix_length)
2145 strip_prefix(o->prefix_length, &name, &other);
2147 if (DIFF_PAIR_UNMERGED(p)) {
2148 run_diff_cmd(pgm, name, NULL, attr_path,
2149 NULL, NULL, NULL, o, 0);
2150 return;
2153 diff_fill_sha1_info(one);
2154 diff_fill_sha1_info(two);
2156 strbuf_init(&msg, PATH_MAX * 2 + 300);
2157 switch (p->status) {
2158 case DIFF_STATUS_COPIED:
2159 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2160 strbuf_addstr(&msg, "\ncopy from ");
2161 quote_c_style(name, &msg, NULL, 0);
2162 strbuf_addstr(&msg, "\ncopy to ");
2163 quote_c_style(other, &msg, NULL, 0);
2164 strbuf_addch(&msg, '\n');
2165 break;
2166 case DIFF_STATUS_RENAMED:
2167 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2168 strbuf_addstr(&msg, "\nrename from ");
2169 quote_c_style(name, &msg, NULL, 0);
2170 strbuf_addstr(&msg, "\nrename to ");
2171 quote_c_style(other, &msg, NULL, 0);
2172 strbuf_addch(&msg, '\n');
2173 break;
2174 case DIFF_STATUS_MODIFIED:
2175 if (p->score) {
2176 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2177 similarity_index(p));
2178 complete_rewrite = 1;
2179 break;
2181 /* fallthru */
2182 default:
2183 /* nothing */
2187 if (hashcmp(one->sha1, two->sha1)) {
2188 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2190 if (DIFF_OPT_TST(o, BINARY)) {
2191 mmfile_t mf;
2192 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2193 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2194 abbrev = 40;
2196 strbuf_addf(&msg, "index %.*s..%.*s",
2197 abbrev, sha1_to_hex(one->sha1),
2198 abbrev, sha1_to_hex(two->sha1));
2199 if (one->mode == two->mode)
2200 strbuf_addf(&msg, " %06o", one->mode);
2201 strbuf_addch(&msg, '\n');
2204 if (msg.len)
2205 strbuf_setlen(&msg, msg.len - 1);
2206 xfrm_msg = msg.len ? msg.buf : NULL;
2208 if (!pgm &&
2209 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2210 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2211 /* a filepair that changes between file and symlink
2212 * needs to be split into deletion and creation.
2214 struct diff_filespec *null = alloc_filespec(two->path);
2215 run_diff_cmd(NULL, name, other, attr_path,
2216 one, null, xfrm_msg, o, 0);
2217 free(null);
2218 null = alloc_filespec(one->path);
2219 run_diff_cmd(NULL, name, other, attr_path,
2220 null, two, xfrm_msg, o, 0);
2221 free(null);
2223 else
2224 run_diff_cmd(pgm, name, other, attr_path,
2225 one, two, xfrm_msg, o, complete_rewrite);
2227 strbuf_release(&msg);
2230 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2231 struct diffstat_t *diffstat)
2233 const char *name;
2234 const char *other;
2235 int complete_rewrite = 0;
2237 if (DIFF_PAIR_UNMERGED(p)) {
2238 /* unmerged */
2239 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2240 return;
2243 name = p->one->path;
2244 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2246 if (o->prefix_length)
2247 strip_prefix(o->prefix_length, &name, &other);
2249 diff_fill_sha1_info(p->one);
2250 diff_fill_sha1_info(p->two);
2252 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2253 complete_rewrite = 1;
2254 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2257 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2259 const char *name;
2260 const char *other;
2261 const char *attr_path;
2263 if (DIFF_PAIR_UNMERGED(p)) {
2264 /* unmerged */
2265 return;
2268 name = p->one->path;
2269 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2270 attr_path = other ? other : name;
2272 if (o->prefix_length)
2273 strip_prefix(o->prefix_length, &name, &other);
2275 diff_fill_sha1_info(p->one);
2276 diff_fill_sha1_info(p->two);
2278 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2281 void diff_setup(struct diff_options *options)
2283 memset(options, 0, sizeof(*options));
2285 options->file = stdout;
2287 options->line_termination = '\n';
2288 options->break_opt = -1;
2289 options->rename_limit = -1;
2290 options->dirstat_percent = 3;
2291 options->context = 3;
2293 options->change = diff_change;
2294 options->add_remove = diff_addremove;
2295 if (diff_use_color_default > 0)
2296 DIFF_OPT_SET(options, COLOR_DIFF);
2297 else
2298 DIFF_OPT_CLR(options, COLOR_DIFF);
2299 options->detect_rename = diff_detect_rename_default;
2301 options->a_prefix = "a/";
2302 options->b_prefix = "b/";
2305 int diff_setup_done(struct diff_options *options)
2307 int count = 0;
2309 if (options->output_format & DIFF_FORMAT_NAME)
2310 count++;
2311 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2312 count++;
2313 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2314 count++;
2315 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2316 count++;
2317 if (count > 1)
2318 die("--name-only, --name-status, --check and -s are mutually exclusive");
2320 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2321 options->detect_rename = DIFF_DETECT_COPY;
2323 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2324 options->prefix = NULL;
2325 if (options->prefix)
2326 options->prefix_length = strlen(options->prefix);
2327 else
2328 options->prefix_length = 0;
2330 if (options->output_format & (DIFF_FORMAT_NAME |
2331 DIFF_FORMAT_NAME_STATUS |
2332 DIFF_FORMAT_CHECKDIFF |
2333 DIFF_FORMAT_NO_OUTPUT))
2334 options->output_format &= ~(DIFF_FORMAT_RAW |
2335 DIFF_FORMAT_NUMSTAT |
2336 DIFF_FORMAT_DIFFSTAT |
2337 DIFF_FORMAT_SHORTSTAT |
2338 DIFF_FORMAT_DIRSTAT |
2339 DIFF_FORMAT_SUMMARY |
2340 DIFF_FORMAT_PATCH);
2343 * These cases always need recursive; we do not drop caller-supplied
2344 * recursive bits for other formats here.
2346 if (options->output_format & (DIFF_FORMAT_PATCH |
2347 DIFF_FORMAT_NUMSTAT |
2348 DIFF_FORMAT_DIFFSTAT |
2349 DIFF_FORMAT_SHORTSTAT |
2350 DIFF_FORMAT_DIRSTAT |
2351 DIFF_FORMAT_SUMMARY |
2352 DIFF_FORMAT_CHECKDIFF))
2353 DIFF_OPT_SET(options, RECURSIVE);
2355 * Also pickaxe would not work very well if you do not say recursive
2357 if (options->pickaxe)
2358 DIFF_OPT_SET(options, RECURSIVE);
2360 if (options->detect_rename && options->rename_limit < 0)
2361 options->rename_limit = diff_rename_limit_default;
2362 if (options->setup & DIFF_SETUP_USE_CACHE) {
2363 if (!active_cache)
2364 /* read-cache does not die even when it fails
2365 * so it is safe for us to do this here. Also
2366 * it does not smudge active_cache or active_nr
2367 * when it fails, so we do not have to worry about
2368 * cleaning it up ourselves either.
2370 read_cache();
2372 if (options->abbrev <= 0 || 40 < options->abbrev)
2373 options->abbrev = 40; /* full */
2376 * It does not make sense to show the first hit we happened
2377 * to have found. It does not make sense not to return with
2378 * exit code in such a case either.
2380 if (DIFF_OPT_TST(options, QUIET)) {
2381 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2382 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2386 * If we postprocess in diffcore, we cannot simply return
2387 * upon the first hit. We need to run diff as usual.
2389 if (options->pickaxe || options->filter)
2390 DIFF_OPT_CLR(options, QUIET);
2392 return 0;
2395 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2397 char c, *eq;
2398 int len;
2400 if (*arg != '-')
2401 return 0;
2402 c = *++arg;
2403 if (!c)
2404 return 0;
2405 if (c == arg_short) {
2406 c = *++arg;
2407 if (!c)
2408 return 1;
2409 if (val && isdigit(c)) {
2410 char *end;
2411 int n = strtoul(arg, &end, 10);
2412 if (*end)
2413 return 0;
2414 *val = n;
2415 return 1;
2417 return 0;
2419 if (c != '-')
2420 return 0;
2421 arg++;
2422 eq = strchr(arg, '=');
2423 if (eq)
2424 len = eq - arg;
2425 else
2426 len = strlen(arg);
2427 if (!len || strncmp(arg, arg_long, len))
2428 return 0;
2429 if (eq) {
2430 int n;
2431 char *end;
2432 if (!isdigit(*++eq))
2433 return 0;
2434 n = strtoul(eq, &end, 10);
2435 if (*end)
2436 return 0;
2437 *val = n;
2439 return 1;
2442 static int diff_scoreopt_parse(const char *opt);
2444 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2446 const char *arg = av[0];
2448 /* Output format options */
2449 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2450 options->output_format |= DIFF_FORMAT_PATCH;
2451 else if (opt_arg(arg, 'U', "unified", &options->context))
2452 options->output_format |= DIFF_FORMAT_PATCH;
2453 else if (!strcmp(arg, "--raw"))
2454 options->output_format |= DIFF_FORMAT_RAW;
2455 else if (!strcmp(arg, "--patch-with-raw"))
2456 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2457 else if (!strcmp(arg, "--numstat"))
2458 options->output_format |= DIFF_FORMAT_NUMSTAT;
2459 else if (!strcmp(arg, "--shortstat"))
2460 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2461 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2462 options->output_format |= DIFF_FORMAT_DIRSTAT;
2463 else if (!strcmp(arg, "--cumulative"))
2464 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2465 else if (!strcmp(arg, "--check"))
2466 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2467 else if (!strcmp(arg, "--summary"))
2468 options->output_format |= DIFF_FORMAT_SUMMARY;
2469 else if (!strcmp(arg, "--patch-with-stat"))
2470 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2471 else if (!strcmp(arg, "--name-only"))
2472 options->output_format |= DIFF_FORMAT_NAME;
2473 else if (!strcmp(arg, "--name-status"))
2474 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2475 else if (!strcmp(arg, "-s"))
2476 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2477 else if (!prefixcmp(arg, "--stat")) {
2478 char *end;
2479 int width = options->stat_width;
2480 int name_width = options->stat_name_width;
2481 arg += 6;
2482 end = (char *)arg;
2484 switch (*arg) {
2485 case '-':
2486 if (!prefixcmp(arg, "-width="))
2487 width = strtoul(arg + 7, &end, 10);
2488 else if (!prefixcmp(arg, "-name-width="))
2489 name_width = strtoul(arg + 12, &end, 10);
2490 break;
2491 case '=':
2492 width = strtoul(arg+1, &end, 10);
2493 if (*end == ',')
2494 name_width = strtoul(end+1, &end, 10);
2497 /* Important! This checks all the error cases! */
2498 if (*end)
2499 return 0;
2500 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2501 options->stat_name_width = name_width;
2502 options->stat_width = width;
2505 /* renames options */
2506 else if (!prefixcmp(arg, "-B")) {
2507 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2508 return -1;
2510 else if (!prefixcmp(arg, "-M")) {
2511 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2512 return -1;
2513 options->detect_rename = DIFF_DETECT_RENAME;
2515 else if (!prefixcmp(arg, "-C")) {
2516 if (options->detect_rename == DIFF_DETECT_COPY)
2517 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2518 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2519 return -1;
2520 options->detect_rename = DIFF_DETECT_COPY;
2522 else if (!strcmp(arg, "--no-renames"))
2523 options->detect_rename = 0;
2524 else if (!strcmp(arg, "--relative"))
2525 DIFF_OPT_SET(options, RELATIVE_NAME);
2526 else if (!prefixcmp(arg, "--relative=")) {
2527 DIFF_OPT_SET(options, RELATIVE_NAME);
2528 options->prefix = arg + 11;
2531 /* xdiff options */
2532 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2533 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2534 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2535 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2536 else if (!strcmp(arg, "--ignore-space-at-eol"))
2537 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2539 /* flags options */
2540 else if (!strcmp(arg, "--binary")) {
2541 options->output_format |= DIFF_FORMAT_PATCH;
2542 DIFF_OPT_SET(options, BINARY);
2544 else if (!strcmp(arg, "--full-index"))
2545 DIFF_OPT_SET(options, FULL_INDEX);
2546 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2547 DIFF_OPT_SET(options, TEXT);
2548 else if (!strcmp(arg, "-R"))
2549 DIFF_OPT_SET(options, REVERSE_DIFF);
2550 else if (!strcmp(arg, "--find-copies-harder"))
2551 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2552 else if (!strcmp(arg, "--follow"))
2553 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2554 else if (!strcmp(arg, "--color"))
2555 DIFF_OPT_SET(options, COLOR_DIFF);
2556 else if (!strcmp(arg, "--no-color"))
2557 DIFF_OPT_CLR(options, COLOR_DIFF);
2558 else if (!strcmp(arg, "--color-words"))
2559 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2560 else if (!strcmp(arg, "--exit-code"))
2561 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2562 else if (!strcmp(arg, "--quiet"))
2563 DIFF_OPT_SET(options, QUIET);
2564 else if (!strcmp(arg, "--ext-diff"))
2565 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2566 else if (!strcmp(arg, "--no-ext-diff"))
2567 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2568 else if (!strcmp(arg, "--ignore-submodules"))
2569 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2571 /* misc options */
2572 else if (!strcmp(arg, "-z"))
2573 options->line_termination = 0;
2574 else if (!prefixcmp(arg, "-l"))
2575 options->rename_limit = strtoul(arg+2, NULL, 10);
2576 else if (!prefixcmp(arg, "-S"))
2577 options->pickaxe = arg + 2;
2578 else if (!strcmp(arg, "--pickaxe-all"))
2579 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2580 else if (!strcmp(arg, "--pickaxe-regex"))
2581 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2582 else if (!prefixcmp(arg, "-O"))
2583 options->orderfile = arg + 2;
2584 else if (!prefixcmp(arg, "--diff-filter="))
2585 options->filter = arg + 14;
2586 else if (!strcmp(arg, "--abbrev"))
2587 options->abbrev = DEFAULT_ABBREV;
2588 else if (!prefixcmp(arg, "--abbrev=")) {
2589 options->abbrev = strtoul(arg + 9, NULL, 10);
2590 if (options->abbrev < MINIMUM_ABBREV)
2591 options->abbrev = MINIMUM_ABBREV;
2592 else if (40 < options->abbrev)
2593 options->abbrev = 40;
2595 else if (!prefixcmp(arg, "--src-prefix="))
2596 options->a_prefix = arg + 13;
2597 else if (!prefixcmp(arg, "--dst-prefix="))
2598 options->b_prefix = arg + 13;
2599 else if (!strcmp(arg, "--no-prefix"))
2600 options->a_prefix = options->b_prefix = "";
2601 else if (!prefixcmp(arg, "--output=")) {
2602 options->file = fopen(arg + strlen("--output="), "w");
2603 options->close_file = 1;
2604 } else
2605 return 0;
2606 return 1;
2609 static int parse_num(const char **cp_p)
2611 unsigned long num, scale;
2612 int ch, dot;
2613 const char *cp = *cp_p;
2615 num = 0;
2616 scale = 1;
2617 dot = 0;
2618 for(;;) {
2619 ch = *cp;
2620 if ( !dot && ch == '.' ) {
2621 scale = 1;
2622 dot = 1;
2623 } else if ( ch == '%' ) {
2624 scale = dot ? scale*100 : 100;
2625 cp++; /* % is always at the end */
2626 break;
2627 } else if ( ch >= '0' && ch <= '9' ) {
2628 if ( scale < 100000 ) {
2629 scale *= 10;
2630 num = (num*10) + (ch-'0');
2632 } else {
2633 break;
2635 cp++;
2637 *cp_p = cp;
2639 /* user says num divided by scale and we say internally that
2640 * is MAX_SCORE * num / scale.
2642 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2645 static int diff_scoreopt_parse(const char *opt)
2647 int opt1, opt2, cmd;
2649 if (*opt++ != '-')
2650 return -1;
2651 cmd = *opt++;
2652 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2653 return -1; /* that is not a -M, -C nor -B option */
2655 opt1 = parse_num(&opt);
2656 if (cmd != 'B')
2657 opt2 = 0;
2658 else {
2659 if (*opt == 0)
2660 opt2 = 0;
2661 else if (*opt != '/')
2662 return -1; /* we expect -B80/99 or -B80 */
2663 else {
2664 opt++;
2665 opt2 = parse_num(&opt);
2668 if (*opt != 0)
2669 return -1;
2670 return opt1 | (opt2 << 16);
2673 struct diff_queue_struct diff_queued_diff;
2675 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2677 if (queue->alloc <= queue->nr) {
2678 queue->alloc = alloc_nr(queue->alloc);
2679 queue->queue = xrealloc(queue->queue,
2680 sizeof(dp) * queue->alloc);
2682 queue->queue[queue->nr++] = dp;
2685 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2686 struct diff_filespec *one,
2687 struct diff_filespec *two)
2689 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2690 dp->one = one;
2691 dp->two = two;
2692 if (queue)
2693 diff_q(queue, dp);
2694 return dp;
2697 void diff_free_filepair(struct diff_filepair *p)
2699 free_filespec(p->one);
2700 free_filespec(p->two);
2701 free(p);
2704 /* This is different from find_unique_abbrev() in that
2705 * it stuffs the result with dots for alignment.
2707 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2709 int abblen;
2710 const char *abbrev;
2711 if (len == 40)
2712 return sha1_to_hex(sha1);
2714 abbrev = find_unique_abbrev(sha1, len);
2715 abblen = strlen(abbrev);
2716 if (abblen < 37) {
2717 static char hex[41];
2718 if (len < abblen && abblen <= len + 2)
2719 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2720 else
2721 sprintf(hex, "%s...", abbrev);
2722 return hex;
2724 return sha1_to_hex(sha1);
2727 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2729 int line_termination = opt->line_termination;
2730 int inter_name_termination = line_termination ? '\t' : '\0';
2732 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2733 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2734 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2735 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2737 if (p->score) {
2738 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2739 inter_name_termination);
2740 } else {
2741 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2744 if (p->status == DIFF_STATUS_COPIED ||
2745 p->status == DIFF_STATUS_RENAMED) {
2746 const char *name_a, *name_b;
2747 name_a = p->one->path;
2748 name_b = p->two->path;
2749 strip_prefix(opt->prefix_length, &name_a, &name_b);
2750 write_name_quoted(name_a, opt->file, inter_name_termination);
2751 write_name_quoted(name_b, opt->file, line_termination);
2752 } else {
2753 const char *name_a, *name_b;
2754 name_a = p->one->mode ? p->one->path : p->two->path;
2755 name_b = NULL;
2756 strip_prefix(opt->prefix_length, &name_a, &name_b);
2757 write_name_quoted(name_a, opt->file, line_termination);
2761 int diff_unmodified_pair(struct diff_filepair *p)
2763 /* This function is written stricter than necessary to support
2764 * the currently implemented transformers, but the idea is to
2765 * let transformers to produce diff_filepairs any way they want,
2766 * and filter and clean them up here before producing the output.
2768 struct diff_filespec *one = p->one, *two = p->two;
2770 if (DIFF_PAIR_UNMERGED(p))
2771 return 0; /* unmerged is interesting */
2773 /* deletion, addition, mode or type change
2774 * and rename are all interesting.
2776 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2777 DIFF_PAIR_MODE_CHANGED(p) ||
2778 strcmp(one->path, two->path))
2779 return 0;
2781 /* both are valid and point at the same path. that is, we are
2782 * dealing with a change.
2784 if (one->sha1_valid && two->sha1_valid &&
2785 !hashcmp(one->sha1, two->sha1))
2786 return 1; /* no change */
2787 if (!one->sha1_valid && !two->sha1_valid)
2788 return 1; /* both look at the same file on the filesystem. */
2789 return 0;
2792 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2794 if (diff_unmodified_pair(p))
2795 return;
2797 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2798 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2799 return; /* no tree diffs in patch format */
2801 run_diff(p, o);
2804 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2805 struct diffstat_t *diffstat)
2807 if (diff_unmodified_pair(p))
2808 return;
2810 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2811 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2812 return; /* no tree diffs in patch format */
2814 run_diffstat(p, o, diffstat);
2817 static void diff_flush_checkdiff(struct diff_filepair *p,
2818 struct diff_options *o)
2820 if (diff_unmodified_pair(p))
2821 return;
2823 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2824 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2825 return; /* no tree diffs in patch format */
2827 run_checkdiff(p, o);
2830 int diff_queue_is_empty(void)
2832 struct diff_queue_struct *q = &diff_queued_diff;
2833 int i;
2834 for (i = 0; i < q->nr; i++)
2835 if (!diff_unmodified_pair(q->queue[i]))
2836 return 0;
2837 return 1;
2840 #if DIFF_DEBUG
2841 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2843 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2844 x, one ? one : "",
2845 s->path,
2846 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2847 s->mode,
2848 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2849 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2850 x, one ? one : "",
2851 s->size, s->xfrm_flags);
2854 void diff_debug_filepair(const struct diff_filepair *p, int i)
2856 diff_debug_filespec(p->one, i, "one");
2857 diff_debug_filespec(p->two, i, "two");
2858 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2859 p->score, p->status ? p->status : '?',
2860 p->one->rename_used, p->broken_pair);
2863 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2865 int i;
2866 if (msg)
2867 fprintf(stderr, "%s\n", msg);
2868 fprintf(stderr, "q->nr = %d\n", q->nr);
2869 for (i = 0; i < q->nr; i++) {
2870 struct diff_filepair *p = q->queue[i];
2871 diff_debug_filepair(p, i);
2874 #endif
2876 static void diff_resolve_rename_copy(void)
2878 int i;
2879 struct diff_filepair *p;
2880 struct diff_queue_struct *q = &diff_queued_diff;
2882 diff_debug_queue("resolve-rename-copy", q);
2884 for (i = 0; i < q->nr; i++) {
2885 p = q->queue[i];
2886 p->status = 0; /* undecided */
2887 if (DIFF_PAIR_UNMERGED(p))
2888 p->status = DIFF_STATUS_UNMERGED;
2889 else if (!DIFF_FILE_VALID(p->one))
2890 p->status = DIFF_STATUS_ADDED;
2891 else if (!DIFF_FILE_VALID(p->two))
2892 p->status = DIFF_STATUS_DELETED;
2893 else if (DIFF_PAIR_TYPE_CHANGED(p))
2894 p->status = DIFF_STATUS_TYPE_CHANGED;
2896 /* from this point on, we are dealing with a pair
2897 * whose both sides are valid and of the same type, i.e.
2898 * either in-place edit or rename/copy edit.
2900 else if (DIFF_PAIR_RENAME(p)) {
2902 * A rename might have re-connected a broken
2903 * pair up, causing the pathnames to be the
2904 * same again. If so, that's not a rename at
2905 * all, just a modification..
2907 * Otherwise, see if this source was used for
2908 * multiple renames, in which case we decrement
2909 * the count, and call it a copy.
2911 if (!strcmp(p->one->path, p->two->path))
2912 p->status = DIFF_STATUS_MODIFIED;
2913 else if (--p->one->rename_used > 0)
2914 p->status = DIFF_STATUS_COPIED;
2915 else
2916 p->status = DIFF_STATUS_RENAMED;
2918 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2919 p->one->mode != p->two->mode ||
2920 is_null_sha1(p->one->sha1))
2921 p->status = DIFF_STATUS_MODIFIED;
2922 else {
2923 /* This is a "no-change" entry and should not
2924 * happen anymore, but prepare for broken callers.
2926 error("feeding unmodified %s to diffcore",
2927 p->one->path);
2928 p->status = DIFF_STATUS_UNKNOWN;
2931 diff_debug_queue("resolve-rename-copy done", q);
2934 static int check_pair_status(struct diff_filepair *p)
2936 switch (p->status) {
2937 case DIFF_STATUS_UNKNOWN:
2938 return 0;
2939 case 0:
2940 die("internal error in diff-resolve-rename-copy");
2941 default:
2942 return 1;
2946 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2948 int fmt = opt->output_format;
2950 if (fmt & DIFF_FORMAT_CHECKDIFF)
2951 diff_flush_checkdiff(p, opt);
2952 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2953 diff_flush_raw(p, opt);
2954 else if (fmt & DIFF_FORMAT_NAME) {
2955 const char *name_a, *name_b;
2956 name_a = p->two->path;
2957 name_b = NULL;
2958 strip_prefix(opt->prefix_length, &name_a, &name_b);
2959 write_name_quoted(name_a, opt->file, opt->line_termination);
2963 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2965 if (fs->mode)
2966 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2967 else
2968 fprintf(file, " %s ", newdelete);
2969 write_name_quoted(fs->path, file, '\n');
2973 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2975 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2976 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2977 show_name ? ' ' : '\n');
2978 if (show_name) {
2979 write_name_quoted(p->two->path, file, '\n');
2984 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2986 char *names = pprint_rename(p->one->path, p->two->path);
2988 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2989 free(names);
2990 show_mode_change(file, p, 0);
2993 static void diff_summary(FILE *file, struct diff_filepair *p)
2995 switch(p->status) {
2996 case DIFF_STATUS_DELETED:
2997 show_file_mode_name(file, "delete", p->one);
2998 break;
2999 case DIFF_STATUS_ADDED:
3000 show_file_mode_name(file, "create", p->two);
3001 break;
3002 case DIFF_STATUS_COPIED:
3003 show_rename_copy(file, "copy", p);
3004 break;
3005 case DIFF_STATUS_RENAMED:
3006 show_rename_copy(file, "rename", p);
3007 break;
3008 default:
3009 if (p->score) {
3010 fputs(" rewrite ", file);
3011 write_name_quoted(p->two->path, file, ' ');
3012 fprintf(file, "(%d%%)\n", similarity_index(p));
3014 show_mode_change(file, p, !p->score);
3015 break;
3019 struct patch_id_t {
3020 struct xdiff_emit_state xm;
3021 SHA_CTX *ctx;
3022 int patchlen;
3025 static int remove_space(char *line, int len)
3027 int i;
3028 char *dst = line;
3029 unsigned char c;
3031 for (i = 0; i < len; i++)
3032 if (!isspace((c = line[i])))
3033 *dst++ = c;
3035 return dst - line;
3038 static void patch_id_consume(void *priv, char *line, unsigned long len)
3040 struct patch_id_t *data = priv;
3041 int new_len;
3043 /* Ignore line numbers when computing the SHA1 of the patch */
3044 if (!prefixcmp(line, "@@ -"))
3045 return;
3047 new_len = remove_space(line, len);
3049 SHA1_Update(data->ctx, line, new_len);
3050 data->patchlen += new_len;
3053 /* returns 0 upon success, and writes result into sha1 */
3054 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3056 struct diff_queue_struct *q = &diff_queued_diff;
3057 int i;
3058 SHA_CTX ctx;
3059 struct patch_id_t data;
3060 char buffer[PATH_MAX * 4 + 20];
3062 SHA1_Init(&ctx);
3063 memset(&data, 0, sizeof(struct patch_id_t));
3064 data.ctx = &ctx;
3065 data.xm.consume = patch_id_consume;
3067 for (i = 0; i < q->nr; i++) {
3068 xpparam_t xpp;
3069 xdemitconf_t xecfg;
3070 xdemitcb_t ecb;
3071 mmfile_t mf1, mf2;
3072 struct diff_filepair *p = q->queue[i];
3073 int len1, len2;
3075 memset(&xecfg, 0, sizeof(xecfg));
3076 if (p->status == 0)
3077 return error("internal diff status error");
3078 if (p->status == DIFF_STATUS_UNKNOWN)
3079 continue;
3080 if (diff_unmodified_pair(p))
3081 continue;
3082 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3083 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3084 continue;
3085 if (DIFF_PAIR_UNMERGED(p))
3086 continue;
3088 diff_fill_sha1_info(p->one);
3089 diff_fill_sha1_info(p->two);
3090 if (fill_mmfile(&mf1, p->one) < 0 ||
3091 fill_mmfile(&mf2, p->two) < 0)
3092 return error("unable to read files to diff");
3094 len1 = remove_space(p->one->path, strlen(p->one->path));
3095 len2 = remove_space(p->two->path, strlen(p->two->path));
3096 if (p->one->mode == 0)
3097 len1 = snprintf(buffer, sizeof(buffer),
3098 "diff--gita/%.*sb/%.*s"
3099 "newfilemode%06o"
3100 "---/dev/null"
3101 "+++b/%.*s",
3102 len1, p->one->path,
3103 len2, p->two->path,
3104 p->two->mode,
3105 len2, p->two->path);
3106 else if (p->two->mode == 0)
3107 len1 = snprintf(buffer, sizeof(buffer),
3108 "diff--gita/%.*sb/%.*s"
3109 "deletedfilemode%06o"
3110 "---a/%.*s"
3111 "+++/dev/null",
3112 len1, p->one->path,
3113 len2, p->two->path,
3114 p->one->mode,
3115 len1, p->one->path);
3116 else
3117 len1 = snprintf(buffer, sizeof(buffer),
3118 "diff--gita/%.*sb/%.*s"
3119 "---a/%.*s"
3120 "+++b/%.*s",
3121 len1, p->one->path,
3122 len2, p->two->path,
3123 len1, p->one->path,
3124 len2, p->two->path);
3125 SHA1_Update(&ctx, buffer, len1);
3127 xpp.flags = XDF_NEED_MINIMAL;
3128 xecfg.ctxlen = 3;
3129 xecfg.flags = XDL_EMIT_FUNCNAMES;
3130 ecb.outf = xdiff_outf;
3131 ecb.priv = &data;
3132 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3135 SHA1_Final(sha1, &ctx);
3136 return 0;
3139 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3141 struct diff_queue_struct *q = &diff_queued_diff;
3142 int i;
3143 int result = diff_get_patch_id(options, sha1);
3145 for (i = 0; i < q->nr; i++)
3146 diff_free_filepair(q->queue[i]);
3148 free(q->queue);
3149 q->queue = NULL;
3150 q->nr = q->alloc = 0;
3152 return result;
3155 static int is_summary_empty(const struct diff_queue_struct *q)
3157 int i;
3159 for (i = 0; i < q->nr; i++) {
3160 const struct diff_filepair *p = q->queue[i];
3162 switch (p->status) {
3163 case DIFF_STATUS_DELETED:
3164 case DIFF_STATUS_ADDED:
3165 case DIFF_STATUS_COPIED:
3166 case DIFF_STATUS_RENAMED:
3167 return 0;
3168 default:
3169 if (p->score)
3170 return 0;
3171 if (p->one->mode && p->two->mode &&
3172 p->one->mode != p->two->mode)
3173 return 0;
3174 break;
3177 return 1;
3180 void diff_flush(struct diff_options *options)
3182 struct diff_queue_struct *q = &diff_queued_diff;
3183 int i, output_format = options->output_format;
3184 int separator = 0;
3187 * Order: raw, stat, summary, patch
3188 * or: name/name-status/checkdiff (other bits clear)
3190 if (!q->nr)
3191 goto free_queue;
3193 if (output_format & (DIFF_FORMAT_RAW |
3194 DIFF_FORMAT_NAME |
3195 DIFF_FORMAT_NAME_STATUS |
3196 DIFF_FORMAT_CHECKDIFF)) {
3197 for (i = 0; i < q->nr; i++) {
3198 struct diff_filepair *p = q->queue[i];
3199 if (check_pair_status(p))
3200 flush_one_pair(p, options);
3202 separator++;
3205 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3206 struct diffstat_t diffstat;
3208 memset(&diffstat, 0, sizeof(struct diffstat_t));
3209 diffstat.xm.consume = diffstat_consume;
3210 for (i = 0; i < q->nr; i++) {
3211 struct diff_filepair *p = q->queue[i];
3212 if (check_pair_status(p))
3213 diff_flush_stat(p, options, &diffstat);
3215 if (output_format & DIFF_FORMAT_NUMSTAT)
3216 show_numstat(&diffstat, options);
3217 if (output_format & DIFF_FORMAT_DIFFSTAT)
3218 show_stats(&diffstat, options);
3219 if (output_format & DIFF_FORMAT_SHORTSTAT)
3220 show_shortstats(&diffstat, options);
3221 free_diffstat_info(&diffstat);
3222 separator++;
3224 if (output_format & DIFF_FORMAT_DIRSTAT)
3225 show_dirstat(options);
3227 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3228 for (i = 0; i < q->nr; i++)
3229 diff_summary(options->file, q->queue[i]);
3230 separator++;
3233 if (output_format & DIFF_FORMAT_PATCH) {
3234 if (separator) {
3235 putc(options->line_termination, options->file);
3236 if (options->stat_sep) {
3237 /* attach patch instead of inline */
3238 fputs(options->stat_sep, options->file);
3242 for (i = 0; i < q->nr; i++) {
3243 struct diff_filepair *p = q->queue[i];
3244 if (check_pair_status(p))
3245 diff_flush_patch(p, options);
3249 if (output_format & DIFF_FORMAT_CALLBACK)
3250 options->format_callback(q, options, options->format_callback_data);
3252 for (i = 0; i < q->nr; i++)
3253 diff_free_filepair(q->queue[i]);
3254 free_queue:
3255 free(q->queue);
3256 q->queue = NULL;
3257 q->nr = q->alloc = 0;
3258 if (options->close_file)
3259 fclose(options->file);
3262 static void diffcore_apply_filter(const char *filter)
3264 int i;
3265 struct diff_queue_struct *q = &diff_queued_diff;
3266 struct diff_queue_struct outq;
3267 outq.queue = NULL;
3268 outq.nr = outq.alloc = 0;
3270 if (!filter)
3271 return;
3273 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3274 int found;
3275 for (i = found = 0; !found && i < q->nr; i++) {
3276 struct diff_filepair *p = q->queue[i];
3277 if (((p->status == DIFF_STATUS_MODIFIED) &&
3278 ((p->score &&
3279 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3280 (!p->score &&
3281 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3282 ((p->status != DIFF_STATUS_MODIFIED) &&
3283 strchr(filter, p->status)))
3284 found++;
3286 if (found)
3287 return;
3289 /* otherwise we will clear the whole queue
3290 * by copying the empty outq at the end of this
3291 * function, but first clear the current entries
3292 * in the queue.
3294 for (i = 0; i < q->nr; i++)
3295 diff_free_filepair(q->queue[i]);
3297 else {
3298 /* Only the matching ones */
3299 for (i = 0; i < q->nr; i++) {
3300 struct diff_filepair *p = q->queue[i];
3302 if (((p->status == DIFF_STATUS_MODIFIED) &&
3303 ((p->score &&
3304 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3305 (!p->score &&
3306 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3307 ((p->status != DIFF_STATUS_MODIFIED) &&
3308 strchr(filter, p->status)))
3309 diff_q(&outq, p);
3310 else
3311 diff_free_filepair(p);
3314 free(q->queue);
3315 *q = outq;
3318 /* Check whether two filespecs with the same mode and size are identical */
3319 static int diff_filespec_is_identical(struct diff_filespec *one,
3320 struct diff_filespec *two)
3322 if (S_ISGITLINK(one->mode))
3323 return 0;
3324 if (diff_populate_filespec(one, 0))
3325 return 0;
3326 if (diff_populate_filespec(two, 0))
3327 return 0;
3328 return !memcmp(one->data, two->data, one->size);
3331 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3333 int i;
3334 struct diff_queue_struct *q = &diff_queued_diff;
3335 struct diff_queue_struct outq;
3336 outq.queue = NULL;
3337 outq.nr = outq.alloc = 0;
3339 for (i = 0; i < q->nr; i++) {
3340 struct diff_filepair *p = q->queue[i];
3343 * 1. Entries that come from stat info dirtyness
3344 * always have both sides (iow, not create/delete),
3345 * one side of the object name is unknown, with
3346 * the same mode and size. Keep the ones that
3347 * do not match these criteria. They have real
3348 * differences.
3350 * 2. At this point, the file is known to be modified,
3351 * with the same mode and size, and the object
3352 * name of one side is unknown. Need to inspect
3353 * the identical contents.
3355 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3356 !DIFF_FILE_VALID(p->two) ||
3357 (p->one->sha1_valid && p->two->sha1_valid) ||
3358 (p->one->mode != p->two->mode) ||
3359 diff_populate_filespec(p->one, 1) ||
3360 diff_populate_filespec(p->two, 1) ||
3361 (p->one->size != p->two->size) ||
3362 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3363 diff_q(&outq, p);
3364 else {
3366 * The caller can subtract 1 from skip_stat_unmatch
3367 * to determine how many paths were dirty only
3368 * due to stat info mismatch.
3370 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3371 diffopt->skip_stat_unmatch++;
3372 diff_free_filepair(p);
3375 free(q->queue);
3376 *q = outq;
3379 void diffcore_std(struct diff_options *options)
3381 if (DIFF_OPT_TST(options, QUIET))
3382 return;
3384 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3385 diffcore_skip_stat_unmatch(options);
3386 if (options->break_opt != -1)
3387 diffcore_break(options->break_opt);
3388 if (options->detect_rename)
3389 diffcore_rename(options);
3390 if (options->break_opt != -1)
3391 diffcore_merge_broken();
3392 if (options->pickaxe)
3393 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3394 if (options->orderfile)
3395 diffcore_order(options->orderfile);
3396 diff_resolve_rename_copy();
3397 diffcore_apply_filter(options->filter);
3399 if (diff_queued_diff.nr)
3400 DIFF_OPT_SET(options, HAS_CHANGES);
3401 else
3402 DIFF_OPT_CLR(options, HAS_CHANGES);
3405 int diff_result_code(struct diff_options *opt, int status)
3407 int result = 0;
3408 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3409 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3410 return status;
3411 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3412 DIFF_OPT_TST(opt, HAS_CHANGES))
3413 result |= 01;
3414 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3415 DIFF_OPT_TST(opt, CHECK_FAILED))
3416 result |= 02;
3417 return result;
3420 void diff_addremove(struct diff_options *options,
3421 int addremove, unsigned mode,
3422 const unsigned char *sha1,
3423 const char *concatpath)
3425 struct diff_filespec *one, *two;
3427 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3428 return;
3430 /* This may look odd, but it is a preparation for
3431 * feeding "there are unchanged files which should
3432 * not produce diffs, but when you are doing copy
3433 * detection you would need them, so here they are"
3434 * entries to the diff-core. They will be prefixed
3435 * with something like '=' or '*' (I haven't decided
3436 * which but should not make any difference).
3437 * Feeding the same new and old to diff_change()
3438 * also has the same effect.
3439 * Before the final output happens, they are pruned after
3440 * merged into rename/copy pairs as appropriate.
3442 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3443 addremove = (addremove == '+' ? '-' :
3444 addremove == '-' ? '+' : addremove);
3446 if (options->prefix &&
3447 strncmp(concatpath, options->prefix, options->prefix_length))
3448 return;
3450 one = alloc_filespec(concatpath);
3451 two = alloc_filespec(concatpath);
3453 if (addremove != '+')
3454 fill_filespec(one, sha1, mode);
3455 if (addremove != '-')
3456 fill_filespec(two, sha1, mode);
3458 diff_queue(&diff_queued_diff, one, two);
3459 DIFF_OPT_SET(options, HAS_CHANGES);
3462 void diff_change(struct diff_options *options,
3463 unsigned old_mode, unsigned new_mode,
3464 const unsigned char *old_sha1,
3465 const unsigned char *new_sha1,
3466 const char *concatpath)
3468 struct diff_filespec *one, *two;
3470 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3471 && S_ISGITLINK(new_mode))
3472 return;
3474 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3475 unsigned tmp;
3476 const unsigned char *tmp_c;
3477 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3478 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3481 if (options->prefix &&
3482 strncmp(concatpath, options->prefix, options->prefix_length))
3483 return;
3485 one = alloc_filespec(concatpath);
3486 two = alloc_filespec(concatpath);
3487 fill_filespec(one, old_sha1, old_mode);
3488 fill_filespec(two, new_sha1, new_mode);
3490 diff_queue(&diff_queued_diff, one, two);
3491 DIFF_OPT_SET(options, HAS_CHANGES);
3494 void diff_unmerge(struct diff_options *options,
3495 const char *path,
3496 unsigned mode, const unsigned char *sha1)
3498 struct diff_filespec *one, *two;
3500 if (options->prefix &&
3501 strncmp(path, options->prefix, options->prefix_length))
3502 return;
3504 one = alloc_filespec(path);
3505 two = alloc_filespec(path);
3506 fill_filespec(one, sha1, mode);
3507 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;