README: tutorial.txt is now called gittutorial.txt
[git/platforms.git] / diff.c
blob342733ba7324b59f8cf1ec29f92a1dae43198913
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 (void)check_and_emit_line(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 < name_width + 15) {
831 if (name_width <= 25)
832 width = name_width + 15;
833 else
834 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 ", added + deleted);
929 show_graph(options->file, '+', add, add_c, reset);
930 show_graph(options->file, '-', del, del_c, reset);
931 fprintf(options->file, "\n");
933 fprintf(options->file,
934 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
935 set, total_files, adds, dels, reset);
938 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
940 int i, adds = 0, dels = 0, total_files = data->nr;
942 if (data->nr == 0)
943 return;
945 for (i = 0; i < data->nr; i++) {
946 if (!data->files[i]->is_binary &&
947 !data->files[i]->is_unmerged) {
948 int added = data->files[i]->added;
949 int deleted= data->files[i]->deleted;
950 if (!data->files[i]->is_renamed &&
951 (added + deleted == 0)) {
952 total_files--;
953 } else {
954 adds += added;
955 dels += deleted;
959 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
960 total_files, adds, dels);
963 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
965 int i;
967 if (data->nr == 0)
968 return;
970 for (i = 0; i < data->nr; i++) {
971 struct diffstat_file *file = data->files[i];
973 if (file->is_binary)
974 fprintf(options->file, "-\t-\t");
975 else
976 fprintf(options->file,
977 "%d\t%d\t", file->added, file->deleted);
978 if (options->line_termination) {
979 fill_print_name(file);
980 if (!file->is_renamed)
981 write_name_quoted(file->name, options->file,
982 options->line_termination);
983 else {
984 fputs(file->print_name, options->file);
985 putc(options->line_termination, options->file);
987 } else {
988 if (file->is_renamed) {
989 putc('\0', options->file);
990 write_name_quoted(file->from_name, options->file, '\0');
992 write_name_quoted(file->name, options->file, '\0');
997 struct dirstat_file {
998 const char *name;
999 unsigned long changed;
1002 struct dirstat_dir {
1003 struct dirstat_file *files;
1004 int alloc, nr, percent, cumulative;
1007 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1009 unsigned long this_dir = 0;
1010 unsigned int sources = 0;
1012 while (dir->nr) {
1013 struct dirstat_file *f = dir->files;
1014 int namelen = strlen(f->name);
1015 unsigned long this;
1016 char *slash;
1018 if (namelen < baselen)
1019 break;
1020 if (memcmp(f->name, base, baselen))
1021 break;
1022 slash = strchr(f->name + baselen, '/');
1023 if (slash) {
1024 int newbaselen = slash + 1 - f->name;
1025 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1026 sources++;
1027 } else {
1028 this = f->changed;
1029 dir->files++;
1030 dir->nr--;
1031 sources += 2;
1033 this_dir += this;
1037 * We don't report dirstat's for
1038 * - the top level
1039 * - or cases where everything came from a single directory
1040 * under this directory (sources == 1).
1042 if (baselen && sources != 1) {
1043 int permille = this_dir * 1000 / changed;
1044 if (permille) {
1045 int percent = permille / 10;
1046 if (percent >= dir->percent) {
1047 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1048 if (!dir->cumulative)
1049 return 0;
1053 return this_dir;
1056 static void show_dirstat(struct diff_options *options)
1058 int i;
1059 unsigned long changed;
1060 struct dirstat_dir dir;
1061 struct diff_queue_struct *q = &diff_queued_diff;
1063 dir.files = NULL;
1064 dir.alloc = 0;
1065 dir.nr = 0;
1066 dir.percent = options->dirstat_percent;
1067 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1069 changed = 0;
1070 for (i = 0; i < q->nr; i++) {
1071 struct diff_filepair *p = q->queue[i];
1072 const char *name;
1073 unsigned long copied, added, damage;
1075 name = p->one->path ? p->one->path : p->two->path;
1077 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1078 diff_populate_filespec(p->one, 0);
1079 diff_populate_filespec(p->two, 0);
1080 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1081 &copied, &added);
1082 diff_free_filespec_data(p->one);
1083 diff_free_filespec_data(p->two);
1084 } else if (DIFF_FILE_VALID(p->one)) {
1085 diff_populate_filespec(p->one, 1);
1086 copied = added = 0;
1087 diff_free_filespec_data(p->one);
1088 } else if (DIFF_FILE_VALID(p->two)) {
1089 diff_populate_filespec(p->two, 1);
1090 copied = 0;
1091 added = p->two->size;
1092 diff_free_filespec_data(p->two);
1093 } else
1094 continue;
1097 * Original minus copied is the removed material,
1098 * added is the new material. They are both damages
1099 * made to the preimage.
1101 damage = (p->one->size - copied) + added;
1103 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1104 dir.files[dir.nr].name = name;
1105 dir.files[dir.nr].changed = damage;
1106 changed += damage;
1107 dir.nr++;
1110 /* This can happen even with many files, if everything was renames */
1111 if (!changed)
1112 return;
1114 /* Show all directories with more than x% of the changes */
1115 gather_dirstat(options->file, &dir, changed, "", 0);
1118 static void free_diffstat_info(struct diffstat_t *diffstat)
1120 int i;
1121 for (i = 0; i < diffstat->nr; i++) {
1122 struct diffstat_file *f = diffstat->files[i];
1123 if (f->name != f->print_name)
1124 free(f->print_name);
1125 free(f->name);
1126 free(f->from_name);
1127 free(f);
1129 free(diffstat->files);
1132 struct checkdiff_t {
1133 struct xdiff_emit_state xm;
1134 const char *filename;
1135 int lineno, color_diff;
1136 unsigned ws_rule;
1137 unsigned status;
1138 FILE *file;
1141 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1143 struct checkdiff_t *data = priv;
1144 const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1145 const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1146 const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1147 char *err;
1149 if (line[0] == '+') {
1150 unsigned bad;
1151 data->lineno++;
1152 bad = check_and_emit_line(line + 1, len - 1,
1153 data->ws_rule, NULL, NULL, NULL, NULL);
1154 if (!bad)
1155 return;
1156 data->status |= bad;
1157 err = whitespace_error_string(bad);
1158 fprintf(data->file, "%s:%d: %s.\n", data->filename, data->lineno, err);
1159 free(err);
1160 emit_line(data->file, set, reset, line, 1);
1161 (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
1162 data->file, set, reset, ws);
1163 } else if (line[0] == ' ')
1164 data->lineno++;
1165 else if (line[0] == '@') {
1166 char *plus = strchr(line, '+');
1167 if (plus)
1168 data->lineno = strtol(plus, NULL, 10) - 1;
1169 else
1170 die("invalid diff");
1174 static unsigned char *deflate_it(char *data,
1175 unsigned long size,
1176 unsigned long *result_size)
1178 int bound;
1179 unsigned char *deflated;
1180 z_stream stream;
1182 memset(&stream, 0, sizeof(stream));
1183 deflateInit(&stream, zlib_compression_level);
1184 bound = deflateBound(&stream, size);
1185 deflated = xmalloc(bound);
1186 stream.next_out = deflated;
1187 stream.avail_out = bound;
1189 stream.next_in = (unsigned char *)data;
1190 stream.avail_in = size;
1191 while (deflate(&stream, Z_FINISH) == Z_OK)
1192 ; /* nothing */
1193 deflateEnd(&stream);
1194 *result_size = stream.total_out;
1195 return deflated;
1198 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1200 void *cp;
1201 void *delta;
1202 void *deflated;
1203 void *data;
1204 unsigned long orig_size;
1205 unsigned long delta_size;
1206 unsigned long deflate_size;
1207 unsigned long data_size;
1209 /* We could do deflated delta, or we could do just deflated two,
1210 * whichever is smaller.
1212 delta = NULL;
1213 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1214 if (one->size && two->size) {
1215 delta = diff_delta(one->ptr, one->size,
1216 two->ptr, two->size,
1217 &delta_size, deflate_size);
1218 if (delta) {
1219 void *to_free = delta;
1220 orig_size = delta_size;
1221 delta = deflate_it(delta, delta_size, &delta_size);
1222 free(to_free);
1226 if (delta && delta_size < deflate_size) {
1227 fprintf(file, "delta %lu\n", orig_size);
1228 free(deflated);
1229 data = delta;
1230 data_size = delta_size;
1232 else {
1233 fprintf(file, "literal %lu\n", two->size);
1234 free(delta);
1235 data = deflated;
1236 data_size = deflate_size;
1239 /* emit data encoded in base85 */
1240 cp = data;
1241 while (data_size) {
1242 int bytes = (52 < data_size) ? 52 : data_size;
1243 char line[70];
1244 data_size -= bytes;
1245 if (bytes <= 26)
1246 line[0] = bytes + 'A' - 1;
1247 else
1248 line[0] = bytes - 26 + 'a' - 1;
1249 encode_85(line + 1, cp, bytes);
1250 cp = (char *) cp + bytes;
1251 fputs(line, file);
1252 fputc('\n', file);
1254 fprintf(file, "\n");
1255 free(data);
1258 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1260 fprintf(file, "GIT binary patch\n");
1261 emit_binary_diff_body(file, one, two);
1262 emit_binary_diff_body(file, two, one);
1265 static void setup_diff_attr_check(struct git_attr_check *check)
1267 static struct git_attr *attr_diff;
1269 if (!attr_diff) {
1270 attr_diff = git_attr("diff", 4);
1272 check[0].attr = attr_diff;
1275 static void diff_filespec_check_attr(struct diff_filespec *one)
1277 struct git_attr_check attr_diff_check;
1278 int check_from_data = 0;
1280 if (one->checked_attr)
1281 return;
1283 setup_diff_attr_check(&attr_diff_check);
1284 one->is_binary = 0;
1285 one->funcname_pattern_ident = NULL;
1287 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1288 const char *value;
1290 /* binaryness */
1291 value = attr_diff_check.value;
1292 if (ATTR_TRUE(value))
1294 else if (ATTR_FALSE(value))
1295 one->is_binary = 1;
1296 else
1297 check_from_data = 1;
1299 /* funcname pattern ident */
1300 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1302 else
1303 one->funcname_pattern_ident = value;
1306 if (check_from_data) {
1307 if (!one->data && DIFF_FILE_VALID(one))
1308 diff_populate_filespec(one, 0);
1310 if (one->data)
1311 one->is_binary = buffer_is_binary(one->data, one->size);
1315 int diff_filespec_is_binary(struct diff_filespec *one)
1317 diff_filespec_check_attr(one);
1318 return one->is_binary;
1321 static const char *funcname_pattern(const char *ident)
1323 struct funcname_pattern *pp;
1325 for (pp = funcname_pattern_list; pp; pp = pp->next)
1326 if (!strcmp(ident, pp->name))
1327 return pp->pattern;
1328 return NULL;
1331 static struct builtin_funcname_pattern {
1332 const char *name;
1333 const char *pattern;
1334 } builtin_funcname_pattern[] = {
1335 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1336 "new\\|return\\|switch\\|throw\\|while\\)\n"
1337 "^[ ]*\\(\\([ ]*"
1338 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1339 "[ ]*([^;]*\\)$" },
1340 { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1343 static const char *diff_funcname_pattern(struct diff_filespec *one)
1345 const char *ident, *pattern;
1346 int i;
1348 diff_filespec_check_attr(one);
1349 ident = one->funcname_pattern_ident;
1351 if (!ident)
1353 * If the config file has "funcname.default" defined, that
1354 * regexp is used; otherwise NULL is returned and xemit uses
1355 * the built-in default.
1357 return funcname_pattern("default");
1359 /* Look up custom "funcname.$ident" regexp from config. */
1360 pattern = funcname_pattern(ident);
1361 if (pattern)
1362 return pattern;
1365 * And define built-in fallback patterns here. Note that
1366 * these can be overridden by the user's config settings.
1368 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1369 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1370 return builtin_funcname_pattern[i].pattern;
1372 return NULL;
1375 static void builtin_diff(const char *name_a,
1376 const char *name_b,
1377 struct diff_filespec *one,
1378 struct diff_filespec *two,
1379 const char *xfrm_msg,
1380 struct diff_options *o,
1381 int complete_rewrite)
1383 mmfile_t mf1, mf2;
1384 const char *lbl[2];
1385 char *a_one, *b_two;
1386 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1387 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1389 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1390 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1391 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1392 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1393 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1394 if (lbl[0][0] == '/') {
1395 /* /dev/null */
1396 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1397 if (xfrm_msg && xfrm_msg[0])
1398 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1400 else if (lbl[1][0] == '/') {
1401 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1402 if (xfrm_msg && xfrm_msg[0])
1403 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1405 else {
1406 if (one->mode != two->mode) {
1407 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1408 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1410 if (xfrm_msg && xfrm_msg[0])
1411 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1413 * we do not run diff between different kind
1414 * of objects.
1416 if ((one->mode ^ two->mode) & S_IFMT)
1417 goto free_ab_and_return;
1418 if (complete_rewrite) {
1419 emit_rewrite_diff(name_a, name_b, one, two, o);
1420 o->found_changes = 1;
1421 goto free_ab_and_return;
1425 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1426 die("unable to read files to diff");
1428 if (!DIFF_OPT_TST(o, TEXT) &&
1429 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1430 /* Quite common confusing case */
1431 if (mf1.size == mf2.size &&
1432 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1433 goto free_ab_and_return;
1434 if (DIFF_OPT_TST(o, BINARY))
1435 emit_binary_diff(o->file, &mf1, &mf2);
1436 else
1437 fprintf(o->file, "Binary files %s and %s differ\n",
1438 lbl[0], lbl[1]);
1439 o->found_changes = 1;
1441 else {
1442 /* Crazy xdl interfaces.. */
1443 const char *diffopts = getenv("GIT_DIFF_OPTS");
1444 xpparam_t xpp;
1445 xdemitconf_t xecfg;
1446 xdemitcb_t ecb;
1447 struct emit_callback ecbdata;
1448 const char *funcname_pattern;
1450 funcname_pattern = diff_funcname_pattern(one);
1451 if (!funcname_pattern)
1452 funcname_pattern = diff_funcname_pattern(two);
1454 memset(&xecfg, 0, sizeof(xecfg));
1455 memset(&ecbdata, 0, sizeof(ecbdata));
1456 ecbdata.label_path = lbl;
1457 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1458 ecbdata.found_changesp = &o->found_changes;
1459 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1460 ecbdata.file = o->file;
1461 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1462 xecfg.ctxlen = o->context;
1463 xecfg.flags = XDL_EMIT_FUNCNAMES;
1464 if (funcname_pattern)
1465 xdiff_set_find_func(&xecfg, funcname_pattern);
1466 if (!diffopts)
1468 else if (!prefixcmp(diffopts, "--unified="))
1469 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1470 else if (!prefixcmp(diffopts, "-u"))
1471 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1472 ecb.outf = xdiff_outf;
1473 ecb.priv = &ecbdata;
1474 ecbdata.xm.consume = fn_out_consume;
1475 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1476 ecbdata.diff_words =
1477 xcalloc(1, sizeof(struct diff_words_data));
1478 ecbdata.diff_words->file = o->file;
1480 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1481 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1482 free_diff_words_data(&ecbdata);
1485 free_ab_and_return:
1486 diff_free_filespec_data(one);
1487 diff_free_filespec_data(two);
1488 free(a_one);
1489 free(b_two);
1490 return;
1493 static void builtin_diffstat(const char *name_a, const char *name_b,
1494 struct diff_filespec *one,
1495 struct diff_filespec *two,
1496 struct diffstat_t *diffstat,
1497 struct diff_options *o,
1498 int complete_rewrite)
1500 mmfile_t mf1, mf2;
1501 struct diffstat_file *data;
1503 data = diffstat_add(diffstat, name_a, name_b);
1505 if (!one || !two) {
1506 data->is_unmerged = 1;
1507 return;
1509 if (complete_rewrite) {
1510 diff_populate_filespec(one, 0);
1511 diff_populate_filespec(two, 0);
1512 data->deleted = count_lines(one->data, one->size);
1513 data->added = count_lines(two->data, two->size);
1514 goto free_and_return;
1516 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1517 die("unable to read files to diff");
1519 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1520 data->is_binary = 1;
1521 data->added = mf2.size;
1522 data->deleted = mf1.size;
1523 } else {
1524 /* Crazy xdl interfaces.. */
1525 xpparam_t xpp;
1526 xdemitconf_t xecfg;
1527 xdemitcb_t ecb;
1529 memset(&xecfg, 0, sizeof(xecfg));
1530 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1531 ecb.outf = xdiff_outf;
1532 ecb.priv = diffstat;
1533 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1536 free_and_return:
1537 diff_free_filespec_data(one);
1538 diff_free_filespec_data(two);
1541 static void builtin_checkdiff(const char *name_a, const char *name_b,
1542 const char *attr_path,
1543 struct diff_filespec *one,
1544 struct diff_filespec *two, struct diff_options *o)
1546 mmfile_t mf1, mf2;
1547 struct checkdiff_t data;
1549 if (!two)
1550 return;
1552 memset(&data, 0, sizeof(data));
1553 data.xm.consume = checkdiff_consume;
1554 data.filename = name_b ? name_b : name_a;
1555 data.lineno = 0;
1556 data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1557 data.ws_rule = whitespace_rule(attr_path);
1558 data.file = o->file;
1560 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1561 die("unable to read files to diff");
1563 if (diff_filespec_is_binary(two))
1564 goto free_and_return;
1565 else {
1566 /* Crazy xdl interfaces.. */
1567 xpparam_t xpp;
1568 xdemitconf_t xecfg;
1569 xdemitcb_t ecb;
1571 memset(&xecfg, 0, sizeof(xecfg));
1572 xpp.flags = XDF_NEED_MINIMAL;
1573 ecb.outf = xdiff_outf;
1574 ecb.priv = &data;
1575 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1577 free_and_return:
1578 diff_free_filespec_data(one);
1579 diff_free_filespec_data(two);
1580 if (data.status)
1581 DIFF_OPT_SET(o, CHECK_FAILED);
1584 struct diff_filespec *alloc_filespec(const char *path)
1586 int namelen = strlen(path);
1587 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1589 memset(spec, 0, sizeof(*spec));
1590 spec->path = (char *)(spec + 1);
1591 memcpy(spec->path, path, namelen+1);
1592 spec->count = 1;
1593 return spec;
1596 void free_filespec(struct diff_filespec *spec)
1598 if (!--spec->count) {
1599 diff_free_filespec_data(spec);
1600 free(spec);
1604 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1605 unsigned short mode)
1607 if (mode) {
1608 spec->mode = canon_mode(mode);
1609 hashcpy(spec->sha1, sha1);
1610 spec->sha1_valid = !is_null_sha1(sha1);
1615 * Given a name and sha1 pair, if the index tells us the file in
1616 * the work tree has that object contents, return true, so that
1617 * prepare_temp_file() does not have to inflate and extract.
1619 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1621 struct cache_entry *ce;
1622 struct stat st;
1623 int pos, len;
1625 /* We do not read the cache ourselves here, because the
1626 * benchmark with my previous version that always reads cache
1627 * shows that it makes things worse for diff-tree comparing
1628 * two linux-2.6 kernel trees in an already checked out work
1629 * tree. This is because most diff-tree comparisons deal with
1630 * only a small number of files, while reading the cache is
1631 * expensive for a large project, and its cost outweighs the
1632 * savings we get by not inflating the object to a temporary
1633 * file. Practically, this code only helps when we are used
1634 * by diff-cache --cached, which does read the cache before
1635 * calling us.
1637 if (!active_cache)
1638 return 0;
1640 /* We want to avoid the working directory if our caller
1641 * doesn't need the data in a normal file, this system
1642 * is rather slow with its stat/open/mmap/close syscalls,
1643 * and the object is contained in a pack file. The pack
1644 * is probably already open and will be faster to obtain
1645 * the data through than the working directory. Loose
1646 * objects however would tend to be slower as they need
1647 * to be individually opened and inflated.
1649 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1650 return 0;
1652 len = strlen(name);
1653 pos = cache_name_pos(name, len);
1654 if (pos < 0)
1655 return 0;
1656 ce = active_cache[pos];
1659 * This is not the sha1 we are looking for, or
1660 * unreusable because it is not a regular file.
1662 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1663 return 0;
1666 * If ce matches the file in the work tree, we can reuse it.
1668 if (ce_uptodate(ce) ||
1669 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1670 return 1;
1672 return 0;
1675 static int populate_from_stdin(struct diff_filespec *s)
1677 struct strbuf buf;
1678 size_t size = 0;
1680 strbuf_init(&buf, 0);
1681 if (strbuf_read(&buf, 0, 0) < 0)
1682 return error("error while reading from stdin %s",
1683 strerror(errno));
1685 s->should_munmap = 0;
1686 s->data = strbuf_detach(&buf, &size);
1687 s->size = size;
1688 s->should_free = 1;
1689 return 0;
1692 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1694 int len;
1695 char *data = xmalloc(100);
1696 len = snprintf(data, 100,
1697 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1698 s->data = data;
1699 s->size = len;
1700 s->should_free = 1;
1701 if (size_only) {
1702 s->data = NULL;
1703 free(data);
1705 return 0;
1709 * While doing rename detection and pickaxe operation, we may need to
1710 * grab the data for the blob (or file) for our own in-core comparison.
1711 * diff_filespec has data and size fields for this purpose.
1713 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1715 int err = 0;
1716 if (!DIFF_FILE_VALID(s))
1717 die("internal error: asking to populate invalid file.");
1718 if (S_ISDIR(s->mode))
1719 return -1;
1721 if (s->data)
1722 return 0;
1724 if (size_only && 0 < s->size)
1725 return 0;
1727 if (S_ISGITLINK(s->mode))
1728 return diff_populate_gitlink(s, size_only);
1730 if (!s->sha1_valid ||
1731 reuse_worktree_file(s->path, s->sha1, 0)) {
1732 struct strbuf buf;
1733 struct stat st;
1734 int fd;
1736 if (!strcmp(s->path, "-"))
1737 return populate_from_stdin(s);
1739 if (lstat(s->path, &st) < 0) {
1740 if (errno == ENOENT) {
1741 err_empty:
1742 err = -1;
1743 empty:
1744 s->data = (char *)"";
1745 s->size = 0;
1746 return err;
1749 s->size = xsize_t(st.st_size);
1750 if (!s->size)
1751 goto empty;
1752 if (size_only)
1753 return 0;
1754 if (S_ISLNK(st.st_mode)) {
1755 int ret;
1756 s->data = xmalloc(s->size);
1757 s->should_free = 1;
1758 ret = readlink(s->path, s->data, s->size);
1759 if (ret < 0) {
1760 free(s->data);
1761 goto err_empty;
1763 return 0;
1765 fd = open(s->path, O_RDONLY);
1766 if (fd < 0)
1767 goto err_empty;
1768 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1769 close(fd);
1770 s->should_munmap = 1;
1773 * Convert from working tree format to canonical git format
1775 strbuf_init(&buf, 0);
1776 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1777 size_t size = 0;
1778 munmap(s->data, s->size);
1779 s->should_munmap = 0;
1780 s->data = strbuf_detach(&buf, &size);
1781 s->size = size;
1782 s->should_free = 1;
1785 else {
1786 enum object_type type;
1787 if (size_only)
1788 type = sha1_object_info(s->sha1, &s->size);
1789 else {
1790 s->data = read_sha1_file(s->sha1, &type, &s->size);
1791 s->should_free = 1;
1794 return 0;
1797 void diff_free_filespec_blob(struct diff_filespec *s)
1799 if (s->should_free)
1800 free(s->data);
1801 else if (s->should_munmap)
1802 munmap(s->data, s->size);
1804 if (s->should_free || s->should_munmap) {
1805 s->should_free = s->should_munmap = 0;
1806 s->data = NULL;
1810 void diff_free_filespec_data(struct diff_filespec *s)
1812 diff_free_filespec_blob(s);
1813 free(s->cnt_data);
1814 s->cnt_data = NULL;
1817 static void prep_temp_blob(struct diff_tempfile *temp,
1818 void *blob,
1819 unsigned long size,
1820 const unsigned char *sha1,
1821 int mode)
1823 int fd;
1825 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1826 if (fd < 0)
1827 die("unable to create temp-file: %s", strerror(errno));
1828 if (write_in_full(fd, blob, size) != size)
1829 die("unable to write temp-file");
1830 close(fd);
1831 temp->name = temp->tmp_path;
1832 strcpy(temp->hex, sha1_to_hex(sha1));
1833 temp->hex[40] = 0;
1834 sprintf(temp->mode, "%06o", mode);
1837 static void prepare_temp_file(const char *name,
1838 struct diff_tempfile *temp,
1839 struct diff_filespec *one)
1841 if (!DIFF_FILE_VALID(one)) {
1842 not_a_valid_file:
1843 /* A '-' entry produces this for file-2, and
1844 * a '+' entry produces this for file-1.
1846 temp->name = "/dev/null";
1847 strcpy(temp->hex, ".");
1848 strcpy(temp->mode, ".");
1849 return;
1852 if (!one->sha1_valid ||
1853 reuse_worktree_file(name, one->sha1, 1)) {
1854 struct stat st;
1855 if (lstat(name, &st) < 0) {
1856 if (errno == ENOENT)
1857 goto not_a_valid_file;
1858 die("stat(%s): %s", name, strerror(errno));
1860 if (S_ISLNK(st.st_mode)) {
1861 int ret;
1862 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1863 size_t sz = xsize_t(st.st_size);
1864 if (sizeof(buf) <= st.st_size)
1865 die("symlink too long: %s", name);
1866 ret = readlink(name, buf, sz);
1867 if (ret < 0)
1868 die("readlink(%s)", name);
1869 prep_temp_blob(temp, buf, sz,
1870 (one->sha1_valid ?
1871 one->sha1 : null_sha1),
1872 (one->sha1_valid ?
1873 one->mode : S_IFLNK));
1875 else {
1876 /* we can borrow from the file in the work tree */
1877 temp->name = name;
1878 if (!one->sha1_valid)
1879 strcpy(temp->hex, sha1_to_hex(null_sha1));
1880 else
1881 strcpy(temp->hex, sha1_to_hex(one->sha1));
1882 /* Even though we may sometimes borrow the
1883 * contents from the work tree, we always want
1884 * one->mode. mode is trustworthy even when
1885 * !(one->sha1_valid), as long as
1886 * DIFF_FILE_VALID(one).
1888 sprintf(temp->mode, "%06o", one->mode);
1890 return;
1892 else {
1893 if (diff_populate_filespec(one, 0))
1894 die("cannot read data blob for %s", one->path);
1895 prep_temp_blob(temp, one->data, one->size,
1896 one->sha1, one->mode);
1900 static void remove_tempfile(void)
1902 int i;
1904 for (i = 0; i < 2; i++)
1905 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1906 unlink(diff_temp[i].name);
1907 diff_temp[i].name = NULL;
1911 static void remove_tempfile_on_signal(int signo)
1913 remove_tempfile();
1914 signal(SIGINT, SIG_DFL);
1915 raise(signo);
1918 /* An external diff command takes:
1920 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1921 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1924 static void run_external_diff(const char *pgm,
1925 const char *name,
1926 const char *other,
1927 struct diff_filespec *one,
1928 struct diff_filespec *two,
1929 const char *xfrm_msg,
1930 int complete_rewrite)
1932 const char *spawn_arg[10];
1933 struct diff_tempfile *temp = diff_temp;
1934 int retval;
1935 static int atexit_asked = 0;
1936 const char *othername;
1937 const char **arg = &spawn_arg[0];
1939 othername = (other? other : name);
1940 if (one && two) {
1941 prepare_temp_file(name, &temp[0], one);
1942 prepare_temp_file(othername, &temp[1], two);
1943 if (! atexit_asked &&
1944 (temp[0].name == temp[0].tmp_path ||
1945 temp[1].name == temp[1].tmp_path)) {
1946 atexit_asked = 1;
1947 atexit(remove_tempfile);
1949 signal(SIGINT, remove_tempfile_on_signal);
1952 if (one && two) {
1953 *arg++ = pgm;
1954 *arg++ = name;
1955 *arg++ = temp[0].name;
1956 *arg++ = temp[0].hex;
1957 *arg++ = temp[0].mode;
1958 *arg++ = temp[1].name;
1959 *arg++ = temp[1].hex;
1960 *arg++ = temp[1].mode;
1961 if (other) {
1962 *arg++ = other;
1963 *arg++ = xfrm_msg;
1965 } else {
1966 *arg++ = pgm;
1967 *arg++ = name;
1969 *arg = NULL;
1970 fflush(NULL);
1971 retval = run_command_v_opt(spawn_arg, 0);
1972 remove_tempfile();
1973 if (retval) {
1974 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1975 exit(1);
1979 static const char *external_diff_attr(const char *name)
1981 struct git_attr_check attr_diff_check;
1983 if (!name)
1984 return NULL;
1986 setup_diff_attr_check(&attr_diff_check);
1987 if (!git_checkattr(name, 1, &attr_diff_check)) {
1988 const char *value = attr_diff_check.value;
1989 if (!ATTR_TRUE(value) &&
1990 !ATTR_FALSE(value) &&
1991 !ATTR_UNSET(value)) {
1992 struct ll_diff_driver *drv;
1994 for (drv = user_diff; drv; drv = drv->next)
1995 if (!strcmp(drv->name, value))
1996 return drv->cmd;
1999 return NULL;
2002 static void run_diff_cmd(const char *pgm,
2003 const char *name,
2004 const char *other,
2005 const char *attr_path,
2006 struct diff_filespec *one,
2007 struct diff_filespec *two,
2008 const char *xfrm_msg,
2009 struct diff_options *o,
2010 int complete_rewrite)
2012 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2013 pgm = NULL;
2014 else {
2015 const char *cmd = external_diff_attr(attr_path);
2016 if (cmd)
2017 pgm = cmd;
2020 if (pgm) {
2021 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2022 complete_rewrite);
2023 return;
2025 if (one && two)
2026 builtin_diff(name, other ? other : name,
2027 one, two, xfrm_msg, o, complete_rewrite);
2028 else
2029 fprintf(o->file, "* Unmerged path %s\n", name);
2032 static void diff_fill_sha1_info(struct diff_filespec *one)
2034 if (DIFF_FILE_VALID(one)) {
2035 if (!one->sha1_valid) {
2036 struct stat st;
2037 if (!strcmp(one->path, "-")) {
2038 hashcpy(one->sha1, null_sha1);
2039 return;
2041 if (lstat(one->path, &st) < 0)
2042 die("stat %s", one->path);
2043 if (index_path(one->sha1, one->path, &st, 0))
2044 die("cannot hash %s\n", one->path);
2047 else
2048 hashclr(one->sha1);
2051 static int similarity_index(struct diff_filepair *p)
2053 return p->score * 100 / MAX_SCORE;
2056 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2058 /* Strip the prefix but do not molest /dev/null and absolute paths */
2059 if (*namep && **namep != '/')
2060 *namep += prefix_length;
2061 if (*otherp && **otherp != '/')
2062 *otherp += prefix_length;
2065 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2067 const char *pgm = external_diff();
2068 struct strbuf msg;
2069 char *xfrm_msg;
2070 struct diff_filespec *one = p->one;
2071 struct diff_filespec *two = p->two;
2072 const char *name;
2073 const char *other;
2074 const char *attr_path;
2075 int complete_rewrite = 0;
2077 name = p->one->path;
2078 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2079 attr_path = name;
2080 if (o->prefix_length)
2081 strip_prefix(o->prefix_length, &name, &other);
2083 if (DIFF_PAIR_UNMERGED(p)) {
2084 run_diff_cmd(pgm, name, NULL, attr_path,
2085 NULL, NULL, NULL, o, 0);
2086 return;
2089 diff_fill_sha1_info(one);
2090 diff_fill_sha1_info(two);
2092 strbuf_init(&msg, PATH_MAX * 2 + 300);
2093 switch (p->status) {
2094 case DIFF_STATUS_COPIED:
2095 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2096 strbuf_addstr(&msg, "\ncopy from ");
2097 quote_c_style(name, &msg, NULL, 0);
2098 strbuf_addstr(&msg, "\ncopy to ");
2099 quote_c_style(other, &msg, NULL, 0);
2100 strbuf_addch(&msg, '\n');
2101 break;
2102 case DIFF_STATUS_RENAMED:
2103 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2104 strbuf_addstr(&msg, "\nrename from ");
2105 quote_c_style(name, &msg, NULL, 0);
2106 strbuf_addstr(&msg, "\nrename to ");
2107 quote_c_style(other, &msg, NULL, 0);
2108 strbuf_addch(&msg, '\n');
2109 break;
2110 case DIFF_STATUS_MODIFIED:
2111 if (p->score) {
2112 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2113 similarity_index(p));
2114 complete_rewrite = 1;
2115 break;
2117 /* fallthru */
2118 default:
2119 /* nothing */
2123 if (hashcmp(one->sha1, two->sha1)) {
2124 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2126 if (DIFF_OPT_TST(o, BINARY)) {
2127 mmfile_t mf;
2128 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2129 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2130 abbrev = 40;
2132 strbuf_addf(&msg, "index %.*s..%.*s",
2133 abbrev, sha1_to_hex(one->sha1),
2134 abbrev, sha1_to_hex(two->sha1));
2135 if (one->mode == two->mode)
2136 strbuf_addf(&msg, " %06o", one->mode);
2137 strbuf_addch(&msg, '\n');
2140 if (msg.len)
2141 strbuf_setlen(&msg, msg.len - 1);
2142 xfrm_msg = msg.len ? msg.buf : NULL;
2144 if (!pgm &&
2145 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2146 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2147 /* a filepair that changes between file and symlink
2148 * needs to be split into deletion and creation.
2150 struct diff_filespec *null = alloc_filespec(two->path);
2151 run_diff_cmd(NULL, name, other, attr_path,
2152 one, null, xfrm_msg, o, 0);
2153 free(null);
2154 null = alloc_filespec(one->path);
2155 run_diff_cmd(NULL, name, other, attr_path,
2156 null, two, xfrm_msg, o, 0);
2157 free(null);
2159 else
2160 run_diff_cmd(pgm, name, other, attr_path,
2161 one, two, xfrm_msg, o, complete_rewrite);
2163 strbuf_release(&msg);
2166 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2167 struct diffstat_t *diffstat)
2169 const char *name;
2170 const char *other;
2171 int complete_rewrite = 0;
2173 if (DIFF_PAIR_UNMERGED(p)) {
2174 /* unmerged */
2175 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2176 return;
2179 name = p->one->path;
2180 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2182 if (o->prefix_length)
2183 strip_prefix(o->prefix_length, &name, &other);
2185 diff_fill_sha1_info(p->one);
2186 diff_fill_sha1_info(p->two);
2188 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2189 complete_rewrite = 1;
2190 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2193 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2195 const char *name;
2196 const char *other;
2197 const char *attr_path;
2199 if (DIFF_PAIR_UNMERGED(p)) {
2200 /* unmerged */
2201 return;
2204 name = p->one->path;
2205 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2206 attr_path = other ? other : name;
2208 if (o->prefix_length)
2209 strip_prefix(o->prefix_length, &name, &other);
2211 diff_fill_sha1_info(p->one);
2212 diff_fill_sha1_info(p->two);
2214 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2217 void diff_setup(struct diff_options *options)
2219 memset(options, 0, sizeof(*options));
2221 options->file = stdout;
2223 options->line_termination = '\n';
2224 options->break_opt = -1;
2225 options->rename_limit = -1;
2226 options->dirstat_percent = 3;
2227 options->context = 3;
2229 options->change = diff_change;
2230 options->add_remove = diff_addremove;
2231 if (diff_use_color_default > 0)
2232 DIFF_OPT_SET(options, COLOR_DIFF);
2233 else
2234 DIFF_OPT_CLR(options, COLOR_DIFF);
2235 options->detect_rename = diff_detect_rename_default;
2237 options->a_prefix = "a/";
2238 options->b_prefix = "b/";
2241 int diff_setup_done(struct diff_options *options)
2243 int count = 0;
2245 if (options->output_format & DIFF_FORMAT_NAME)
2246 count++;
2247 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2248 count++;
2249 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2250 count++;
2251 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2252 count++;
2253 if (count > 1)
2254 die("--name-only, --name-status, --check and -s are mutually exclusive");
2256 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2257 options->detect_rename = DIFF_DETECT_COPY;
2259 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2260 options->prefix = NULL;
2261 if (options->prefix)
2262 options->prefix_length = strlen(options->prefix);
2263 else
2264 options->prefix_length = 0;
2266 if (options->output_format & (DIFF_FORMAT_NAME |
2267 DIFF_FORMAT_NAME_STATUS |
2268 DIFF_FORMAT_CHECKDIFF |
2269 DIFF_FORMAT_NO_OUTPUT))
2270 options->output_format &= ~(DIFF_FORMAT_RAW |
2271 DIFF_FORMAT_NUMSTAT |
2272 DIFF_FORMAT_DIFFSTAT |
2273 DIFF_FORMAT_SHORTSTAT |
2274 DIFF_FORMAT_DIRSTAT |
2275 DIFF_FORMAT_SUMMARY |
2276 DIFF_FORMAT_PATCH);
2279 * These cases always need recursive; we do not drop caller-supplied
2280 * recursive bits for other formats here.
2282 if (options->output_format & (DIFF_FORMAT_PATCH |
2283 DIFF_FORMAT_NUMSTAT |
2284 DIFF_FORMAT_DIFFSTAT |
2285 DIFF_FORMAT_SHORTSTAT |
2286 DIFF_FORMAT_DIRSTAT |
2287 DIFF_FORMAT_SUMMARY |
2288 DIFF_FORMAT_CHECKDIFF))
2289 DIFF_OPT_SET(options, RECURSIVE);
2291 * Also pickaxe would not work very well if you do not say recursive
2293 if (options->pickaxe)
2294 DIFF_OPT_SET(options, RECURSIVE);
2296 if (options->detect_rename && options->rename_limit < 0)
2297 options->rename_limit = diff_rename_limit_default;
2298 if (options->setup & DIFF_SETUP_USE_CACHE) {
2299 if (!active_cache)
2300 /* read-cache does not die even when it fails
2301 * so it is safe for us to do this here. Also
2302 * it does not smudge active_cache or active_nr
2303 * when it fails, so we do not have to worry about
2304 * cleaning it up ourselves either.
2306 read_cache();
2308 if (options->abbrev <= 0 || 40 < options->abbrev)
2309 options->abbrev = 40; /* full */
2312 * It does not make sense to show the first hit we happened
2313 * to have found. It does not make sense not to return with
2314 * exit code in such a case either.
2316 if (DIFF_OPT_TST(options, QUIET)) {
2317 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2318 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2322 * If we postprocess in diffcore, we cannot simply return
2323 * upon the first hit. We need to run diff as usual.
2325 if (options->pickaxe || options->filter)
2326 DIFF_OPT_CLR(options, QUIET);
2328 return 0;
2331 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2333 char c, *eq;
2334 int len;
2336 if (*arg != '-')
2337 return 0;
2338 c = *++arg;
2339 if (!c)
2340 return 0;
2341 if (c == arg_short) {
2342 c = *++arg;
2343 if (!c)
2344 return 1;
2345 if (val && isdigit(c)) {
2346 char *end;
2347 int n = strtoul(arg, &end, 10);
2348 if (*end)
2349 return 0;
2350 *val = n;
2351 return 1;
2353 return 0;
2355 if (c != '-')
2356 return 0;
2357 arg++;
2358 eq = strchr(arg, '=');
2359 if (eq)
2360 len = eq - arg;
2361 else
2362 len = strlen(arg);
2363 if (!len || strncmp(arg, arg_long, len))
2364 return 0;
2365 if (eq) {
2366 int n;
2367 char *end;
2368 if (!isdigit(*++eq))
2369 return 0;
2370 n = strtoul(eq, &end, 10);
2371 if (*end)
2372 return 0;
2373 *val = n;
2375 return 1;
2378 static int diff_scoreopt_parse(const char *opt);
2380 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2382 const char *arg = av[0];
2384 /* Output format options */
2385 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2386 options->output_format |= DIFF_FORMAT_PATCH;
2387 else if (opt_arg(arg, 'U', "unified", &options->context))
2388 options->output_format |= DIFF_FORMAT_PATCH;
2389 else if (!strcmp(arg, "--raw"))
2390 options->output_format |= DIFF_FORMAT_RAW;
2391 else if (!strcmp(arg, "--patch-with-raw"))
2392 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2393 else if (!strcmp(arg, "--numstat"))
2394 options->output_format |= DIFF_FORMAT_NUMSTAT;
2395 else if (!strcmp(arg, "--shortstat"))
2396 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2397 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2398 options->output_format |= DIFF_FORMAT_DIRSTAT;
2399 else if (!strcmp(arg, "--cumulative"))
2400 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2401 else if (!strcmp(arg, "--check"))
2402 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2403 else if (!strcmp(arg, "--summary"))
2404 options->output_format |= DIFF_FORMAT_SUMMARY;
2405 else if (!strcmp(arg, "--patch-with-stat"))
2406 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2407 else if (!strcmp(arg, "--name-only"))
2408 options->output_format |= DIFF_FORMAT_NAME;
2409 else if (!strcmp(arg, "--name-status"))
2410 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2411 else if (!strcmp(arg, "-s"))
2412 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2413 else if (!prefixcmp(arg, "--stat")) {
2414 char *end;
2415 int width = options->stat_width;
2416 int name_width = options->stat_name_width;
2417 arg += 6;
2418 end = (char *)arg;
2420 switch (*arg) {
2421 case '-':
2422 if (!prefixcmp(arg, "-width="))
2423 width = strtoul(arg + 7, &end, 10);
2424 else if (!prefixcmp(arg, "-name-width="))
2425 name_width = strtoul(arg + 12, &end, 10);
2426 break;
2427 case '=':
2428 width = strtoul(arg+1, &end, 10);
2429 if (*end == ',')
2430 name_width = strtoul(end+1, &end, 10);
2433 /* Important! This checks all the error cases! */
2434 if (*end)
2435 return 0;
2436 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2437 options->stat_name_width = name_width;
2438 options->stat_width = width;
2441 /* renames options */
2442 else if (!prefixcmp(arg, "-B")) {
2443 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2444 return -1;
2446 else if (!prefixcmp(arg, "-M")) {
2447 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2448 return -1;
2449 options->detect_rename = DIFF_DETECT_RENAME;
2451 else if (!prefixcmp(arg, "-C")) {
2452 if (options->detect_rename == DIFF_DETECT_COPY)
2453 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2454 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2455 return -1;
2456 options->detect_rename = DIFF_DETECT_COPY;
2458 else if (!strcmp(arg, "--no-renames"))
2459 options->detect_rename = 0;
2460 else if (!strcmp(arg, "--relative"))
2461 DIFF_OPT_SET(options, RELATIVE_NAME);
2462 else if (!prefixcmp(arg, "--relative=")) {
2463 DIFF_OPT_SET(options, RELATIVE_NAME);
2464 options->prefix = arg + 11;
2467 /* xdiff options */
2468 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2469 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2470 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2471 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2472 else if (!strcmp(arg, "--ignore-space-at-eol"))
2473 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2475 /* flags options */
2476 else if (!strcmp(arg, "--binary")) {
2477 options->output_format |= DIFF_FORMAT_PATCH;
2478 DIFF_OPT_SET(options, BINARY);
2480 else if (!strcmp(arg, "--full-index"))
2481 DIFF_OPT_SET(options, FULL_INDEX);
2482 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2483 DIFF_OPT_SET(options, TEXT);
2484 else if (!strcmp(arg, "-R"))
2485 DIFF_OPT_SET(options, REVERSE_DIFF);
2486 else if (!strcmp(arg, "--find-copies-harder"))
2487 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2488 else if (!strcmp(arg, "--follow"))
2489 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2490 else if (!strcmp(arg, "--color"))
2491 DIFF_OPT_SET(options, COLOR_DIFF);
2492 else if (!strcmp(arg, "--no-color"))
2493 DIFF_OPT_CLR(options, COLOR_DIFF);
2494 else if (!strcmp(arg, "--color-words"))
2495 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2496 else if (!strcmp(arg, "--exit-code"))
2497 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2498 else if (!strcmp(arg, "--quiet"))
2499 DIFF_OPT_SET(options, QUIET);
2500 else if (!strcmp(arg, "--ext-diff"))
2501 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2502 else if (!strcmp(arg, "--no-ext-diff"))
2503 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2504 else if (!strcmp(arg, "--ignore-submodules"))
2505 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2507 /* misc options */
2508 else if (!strcmp(arg, "-z"))
2509 options->line_termination = 0;
2510 else if (!prefixcmp(arg, "-l"))
2511 options->rename_limit = strtoul(arg+2, NULL, 10);
2512 else if (!prefixcmp(arg, "-S"))
2513 options->pickaxe = arg + 2;
2514 else if (!strcmp(arg, "--pickaxe-all"))
2515 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2516 else if (!strcmp(arg, "--pickaxe-regex"))
2517 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2518 else if (!prefixcmp(arg, "-O"))
2519 options->orderfile = arg + 2;
2520 else if (!prefixcmp(arg, "--diff-filter="))
2521 options->filter = arg + 14;
2522 else if (!strcmp(arg, "--abbrev"))
2523 options->abbrev = DEFAULT_ABBREV;
2524 else if (!prefixcmp(arg, "--abbrev=")) {
2525 options->abbrev = strtoul(arg + 9, NULL, 10);
2526 if (options->abbrev < MINIMUM_ABBREV)
2527 options->abbrev = MINIMUM_ABBREV;
2528 else if (40 < options->abbrev)
2529 options->abbrev = 40;
2531 else if (!prefixcmp(arg, "--src-prefix="))
2532 options->a_prefix = arg + 13;
2533 else if (!prefixcmp(arg, "--dst-prefix="))
2534 options->b_prefix = arg + 13;
2535 else if (!strcmp(arg, "--no-prefix"))
2536 options->a_prefix = options->b_prefix = "";
2537 else if (!prefixcmp(arg, "--output=")) {
2538 options->file = fopen(arg + strlen("--output="), "w");
2539 options->close_file = 1;
2540 } else
2541 return 0;
2542 return 1;
2545 static int parse_num(const char **cp_p)
2547 unsigned long num, scale;
2548 int ch, dot;
2549 const char *cp = *cp_p;
2551 num = 0;
2552 scale = 1;
2553 dot = 0;
2554 for(;;) {
2555 ch = *cp;
2556 if ( !dot && ch == '.' ) {
2557 scale = 1;
2558 dot = 1;
2559 } else if ( ch == '%' ) {
2560 scale = dot ? scale*100 : 100;
2561 cp++; /* % is always at the end */
2562 break;
2563 } else if ( ch >= '0' && ch <= '9' ) {
2564 if ( scale < 100000 ) {
2565 scale *= 10;
2566 num = (num*10) + (ch-'0');
2568 } else {
2569 break;
2571 cp++;
2573 *cp_p = cp;
2575 /* user says num divided by scale and we say internally that
2576 * is MAX_SCORE * num / scale.
2578 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2581 static int diff_scoreopt_parse(const char *opt)
2583 int opt1, opt2, cmd;
2585 if (*opt++ != '-')
2586 return -1;
2587 cmd = *opt++;
2588 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2589 return -1; /* that is not a -M, -C nor -B option */
2591 opt1 = parse_num(&opt);
2592 if (cmd != 'B')
2593 opt2 = 0;
2594 else {
2595 if (*opt == 0)
2596 opt2 = 0;
2597 else if (*opt != '/')
2598 return -1; /* we expect -B80/99 or -B80 */
2599 else {
2600 opt++;
2601 opt2 = parse_num(&opt);
2604 if (*opt != 0)
2605 return -1;
2606 return opt1 | (opt2 << 16);
2609 struct diff_queue_struct diff_queued_diff;
2611 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2613 if (queue->alloc <= queue->nr) {
2614 queue->alloc = alloc_nr(queue->alloc);
2615 queue->queue = xrealloc(queue->queue,
2616 sizeof(dp) * queue->alloc);
2618 queue->queue[queue->nr++] = dp;
2621 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2622 struct diff_filespec *one,
2623 struct diff_filespec *two)
2625 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2626 dp->one = one;
2627 dp->two = two;
2628 if (queue)
2629 diff_q(queue, dp);
2630 return dp;
2633 void diff_free_filepair(struct diff_filepair *p)
2635 free_filespec(p->one);
2636 free_filespec(p->two);
2637 free(p);
2640 /* This is different from find_unique_abbrev() in that
2641 * it stuffs the result with dots for alignment.
2643 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2645 int abblen;
2646 const char *abbrev;
2647 if (len == 40)
2648 return sha1_to_hex(sha1);
2650 abbrev = find_unique_abbrev(sha1, len);
2651 abblen = strlen(abbrev);
2652 if (abblen < 37) {
2653 static char hex[41];
2654 if (len < abblen && abblen <= len + 2)
2655 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2656 else
2657 sprintf(hex, "%s...", abbrev);
2658 return hex;
2660 return sha1_to_hex(sha1);
2663 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2665 int line_termination = opt->line_termination;
2666 int inter_name_termination = line_termination ? '\t' : '\0';
2668 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2669 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2670 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2671 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2673 if (p->score) {
2674 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2675 inter_name_termination);
2676 } else {
2677 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2680 if (p->status == DIFF_STATUS_COPIED ||
2681 p->status == DIFF_STATUS_RENAMED) {
2682 const char *name_a, *name_b;
2683 name_a = p->one->path;
2684 name_b = p->two->path;
2685 strip_prefix(opt->prefix_length, &name_a, &name_b);
2686 write_name_quoted(name_a, opt->file, inter_name_termination);
2687 write_name_quoted(name_b, opt->file, line_termination);
2688 } else {
2689 const char *name_a, *name_b;
2690 name_a = p->one->mode ? p->one->path : p->two->path;
2691 name_b = NULL;
2692 strip_prefix(opt->prefix_length, &name_a, &name_b);
2693 write_name_quoted(name_a, opt->file, line_termination);
2697 int diff_unmodified_pair(struct diff_filepair *p)
2699 /* This function is written stricter than necessary to support
2700 * the currently implemented transformers, but the idea is to
2701 * let transformers to produce diff_filepairs any way they want,
2702 * and filter and clean them up here before producing the output.
2704 struct diff_filespec *one = p->one, *two = p->two;
2706 if (DIFF_PAIR_UNMERGED(p))
2707 return 0; /* unmerged is interesting */
2709 /* deletion, addition, mode or type change
2710 * and rename are all interesting.
2712 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2713 DIFF_PAIR_MODE_CHANGED(p) ||
2714 strcmp(one->path, two->path))
2715 return 0;
2717 /* both are valid and point at the same path. that is, we are
2718 * dealing with a change.
2720 if (one->sha1_valid && two->sha1_valid &&
2721 !hashcmp(one->sha1, two->sha1))
2722 return 1; /* no change */
2723 if (!one->sha1_valid && !two->sha1_valid)
2724 return 1; /* both look at the same file on the filesystem. */
2725 return 0;
2728 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2730 if (diff_unmodified_pair(p))
2731 return;
2733 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2734 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2735 return; /* no tree diffs in patch format */
2737 run_diff(p, o);
2740 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2741 struct diffstat_t *diffstat)
2743 if (diff_unmodified_pair(p))
2744 return;
2746 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2747 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2748 return; /* no tree diffs in patch format */
2750 run_diffstat(p, o, diffstat);
2753 static void diff_flush_checkdiff(struct diff_filepair *p,
2754 struct diff_options *o)
2756 if (diff_unmodified_pair(p))
2757 return;
2759 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2760 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2761 return; /* no tree diffs in patch format */
2763 run_checkdiff(p, o);
2766 int diff_queue_is_empty(void)
2768 struct diff_queue_struct *q = &diff_queued_diff;
2769 int i;
2770 for (i = 0; i < q->nr; i++)
2771 if (!diff_unmodified_pair(q->queue[i]))
2772 return 0;
2773 return 1;
2776 #if DIFF_DEBUG
2777 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2779 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2780 x, one ? one : "",
2781 s->path,
2782 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2783 s->mode,
2784 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2785 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2786 x, one ? one : "",
2787 s->size, s->xfrm_flags);
2790 void diff_debug_filepair(const struct diff_filepair *p, int i)
2792 diff_debug_filespec(p->one, i, "one");
2793 diff_debug_filespec(p->two, i, "two");
2794 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2795 p->score, p->status ? p->status : '?',
2796 p->one->rename_used, p->broken_pair);
2799 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2801 int i;
2802 if (msg)
2803 fprintf(stderr, "%s\n", msg);
2804 fprintf(stderr, "q->nr = %d\n", q->nr);
2805 for (i = 0; i < q->nr; i++) {
2806 struct diff_filepair *p = q->queue[i];
2807 diff_debug_filepair(p, i);
2810 #endif
2812 static void diff_resolve_rename_copy(void)
2814 int i;
2815 struct diff_filepair *p;
2816 struct diff_queue_struct *q = &diff_queued_diff;
2818 diff_debug_queue("resolve-rename-copy", q);
2820 for (i = 0; i < q->nr; i++) {
2821 p = q->queue[i];
2822 p->status = 0; /* undecided */
2823 if (DIFF_PAIR_UNMERGED(p))
2824 p->status = DIFF_STATUS_UNMERGED;
2825 else if (!DIFF_FILE_VALID(p->one))
2826 p->status = DIFF_STATUS_ADDED;
2827 else if (!DIFF_FILE_VALID(p->two))
2828 p->status = DIFF_STATUS_DELETED;
2829 else if (DIFF_PAIR_TYPE_CHANGED(p))
2830 p->status = DIFF_STATUS_TYPE_CHANGED;
2832 /* from this point on, we are dealing with a pair
2833 * whose both sides are valid and of the same type, i.e.
2834 * either in-place edit or rename/copy edit.
2836 else if (DIFF_PAIR_RENAME(p)) {
2838 * A rename might have re-connected a broken
2839 * pair up, causing the pathnames to be the
2840 * same again. If so, that's not a rename at
2841 * all, just a modification..
2843 * Otherwise, see if this source was used for
2844 * multiple renames, in which case we decrement
2845 * the count, and call it a copy.
2847 if (!strcmp(p->one->path, p->two->path))
2848 p->status = DIFF_STATUS_MODIFIED;
2849 else if (--p->one->rename_used > 0)
2850 p->status = DIFF_STATUS_COPIED;
2851 else
2852 p->status = DIFF_STATUS_RENAMED;
2854 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2855 p->one->mode != p->two->mode ||
2856 is_null_sha1(p->one->sha1))
2857 p->status = DIFF_STATUS_MODIFIED;
2858 else {
2859 /* This is a "no-change" entry and should not
2860 * happen anymore, but prepare for broken callers.
2862 error("feeding unmodified %s to diffcore",
2863 p->one->path);
2864 p->status = DIFF_STATUS_UNKNOWN;
2867 diff_debug_queue("resolve-rename-copy done", q);
2870 static int check_pair_status(struct diff_filepair *p)
2872 switch (p->status) {
2873 case DIFF_STATUS_UNKNOWN:
2874 return 0;
2875 case 0:
2876 die("internal error in diff-resolve-rename-copy");
2877 default:
2878 return 1;
2882 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2884 int fmt = opt->output_format;
2886 if (fmt & DIFF_FORMAT_CHECKDIFF)
2887 diff_flush_checkdiff(p, opt);
2888 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2889 diff_flush_raw(p, opt);
2890 else if (fmt & DIFF_FORMAT_NAME) {
2891 const char *name_a, *name_b;
2892 name_a = p->two->path;
2893 name_b = NULL;
2894 strip_prefix(opt->prefix_length, &name_a, &name_b);
2895 write_name_quoted(name_a, opt->file, opt->line_termination);
2899 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2901 if (fs->mode)
2902 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2903 else
2904 fprintf(file, " %s ", newdelete);
2905 write_name_quoted(fs->path, file, '\n');
2909 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2911 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2912 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2913 show_name ? ' ' : '\n');
2914 if (show_name) {
2915 write_name_quoted(p->two->path, file, '\n');
2920 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2922 char *names = pprint_rename(p->one->path, p->two->path);
2924 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2925 free(names);
2926 show_mode_change(file, p, 0);
2929 static void diff_summary(FILE *file, struct diff_filepair *p)
2931 switch(p->status) {
2932 case DIFF_STATUS_DELETED:
2933 show_file_mode_name(file, "delete", p->one);
2934 break;
2935 case DIFF_STATUS_ADDED:
2936 show_file_mode_name(file, "create", p->two);
2937 break;
2938 case DIFF_STATUS_COPIED:
2939 show_rename_copy(file, "copy", p);
2940 break;
2941 case DIFF_STATUS_RENAMED:
2942 show_rename_copy(file, "rename", p);
2943 break;
2944 default:
2945 if (p->score) {
2946 fputs(" rewrite ", file);
2947 write_name_quoted(p->two->path, file, ' ');
2948 fprintf(file, "(%d%%)\n", similarity_index(p));
2950 show_mode_change(file, p, !p->score);
2951 break;
2955 struct patch_id_t {
2956 struct xdiff_emit_state xm;
2957 SHA_CTX *ctx;
2958 int patchlen;
2961 static int remove_space(char *line, int len)
2963 int i;
2964 char *dst = line;
2965 unsigned char c;
2967 for (i = 0; i < len; i++)
2968 if (!isspace((c = line[i])))
2969 *dst++ = c;
2971 return dst - line;
2974 static void patch_id_consume(void *priv, char *line, unsigned long len)
2976 struct patch_id_t *data = priv;
2977 int new_len;
2979 /* Ignore line numbers when computing the SHA1 of the patch */
2980 if (!prefixcmp(line, "@@ -"))
2981 return;
2983 new_len = remove_space(line, len);
2985 SHA1_Update(data->ctx, line, new_len);
2986 data->patchlen += new_len;
2989 /* returns 0 upon success, and writes result into sha1 */
2990 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2992 struct diff_queue_struct *q = &diff_queued_diff;
2993 int i;
2994 SHA_CTX ctx;
2995 struct patch_id_t data;
2996 char buffer[PATH_MAX * 4 + 20];
2998 SHA1_Init(&ctx);
2999 memset(&data, 0, sizeof(struct patch_id_t));
3000 data.ctx = &ctx;
3001 data.xm.consume = patch_id_consume;
3003 for (i = 0; i < q->nr; i++) {
3004 xpparam_t xpp;
3005 xdemitconf_t xecfg;
3006 xdemitcb_t ecb;
3007 mmfile_t mf1, mf2;
3008 struct diff_filepair *p = q->queue[i];
3009 int len1, len2;
3011 memset(&xecfg, 0, sizeof(xecfg));
3012 if (p->status == 0)
3013 return error("internal diff status error");
3014 if (p->status == DIFF_STATUS_UNKNOWN)
3015 continue;
3016 if (diff_unmodified_pair(p))
3017 continue;
3018 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3019 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3020 continue;
3021 if (DIFF_PAIR_UNMERGED(p))
3022 continue;
3024 diff_fill_sha1_info(p->one);
3025 diff_fill_sha1_info(p->two);
3026 if (fill_mmfile(&mf1, p->one) < 0 ||
3027 fill_mmfile(&mf2, p->two) < 0)
3028 return error("unable to read files to diff");
3030 len1 = remove_space(p->one->path, strlen(p->one->path));
3031 len2 = remove_space(p->two->path, strlen(p->two->path));
3032 if (p->one->mode == 0)
3033 len1 = snprintf(buffer, sizeof(buffer),
3034 "diff--gita/%.*sb/%.*s"
3035 "newfilemode%06o"
3036 "---/dev/null"
3037 "+++b/%.*s",
3038 len1, p->one->path,
3039 len2, p->two->path,
3040 p->two->mode,
3041 len2, p->two->path);
3042 else if (p->two->mode == 0)
3043 len1 = snprintf(buffer, sizeof(buffer),
3044 "diff--gita/%.*sb/%.*s"
3045 "deletedfilemode%06o"
3046 "---a/%.*s"
3047 "+++/dev/null",
3048 len1, p->one->path,
3049 len2, p->two->path,
3050 p->one->mode,
3051 len1, p->one->path);
3052 else
3053 len1 = snprintf(buffer, sizeof(buffer),
3054 "diff--gita/%.*sb/%.*s"
3055 "---a/%.*s"
3056 "+++b/%.*s",
3057 len1, p->one->path,
3058 len2, p->two->path,
3059 len1, p->one->path,
3060 len2, p->two->path);
3061 SHA1_Update(&ctx, buffer, len1);
3063 xpp.flags = XDF_NEED_MINIMAL;
3064 xecfg.ctxlen = 3;
3065 xecfg.flags = XDL_EMIT_FUNCNAMES;
3066 ecb.outf = xdiff_outf;
3067 ecb.priv = &data;
3068 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3071 SHA1_Final(sha1, &ctx);
3072 return 0;
3075 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3077 struct diff_queue_struct *q = &diff_queued_diff;
3078 int i;
3079 int result = diff_get_patch_id(options, sha1);
3081 for (i = 0; i < q->nr; i++)
3082 diff_free_filepair(q->queue[i]);
3084 free(q->queue);
3085 q->queue = NULL;
3086 q->nr = q->alloc = 0;
3088 return result;
3091 static int is_summary_empty(const struct diff_queue_struct *q)
3093 int i;
3095 for (i = 0; i < q->nr; i++) {
3096 const struct diff_filepair *p = q->queue[i];
3098 switch (p->status) {
3099 case DIFF_STATUS_DELETED:
3100 case DIFF_STATUS_ADDED:
3101 case DIFF_STATUS_COPIED:
3102 case DIFF_STATUS_RENAMED:
3103 return 0;
3104 default:
3105 if (p->score)
3106 return 0;
3107 if (p->one->mode && p->two->mode &&
3108 p->one->mode != p->two->mode)
3109 return 0;
3110 break;
3113 return 1;
3116 void diff_flush(struct diff_options *options)
3118 struct diff_queue_struct *q = &diff_queued_diff;
3119 int i, output_format = options->output_format;
3120 int separator = 0;
3123 * Order: raw, stat, summary, patch
3124 * or: name/name-status/checkdiff (other bits clear)
3126 if (!q->nr)
3127 goto free_queue;
3129 if (output_format & (DIFF_FORMAT_RAW |
3130 DIFF_FORMAT_NAME |
3131 DIFF_FORMAT_NAME_STATUS |
3132 DIFF_FORMAT_CHECKDIFF)) {
3133 for (i = 0; i < q->nr; i++) {
3134 struct diff_filepair *p = q->queue[i];
3135 if (check_pair_status(p))
3136 flush_one_pair(p, options);
3138 separator++;
3141 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3142 struct diffstat_t diffstat;
3144 memset(&diffstat, 0, sizeof(struct diffstat_t));
3145 diffstat.xm.consume = diffstat_consume;
3146 for (i = 0; i < q->nr; i++) {
3147 struct diff_filepair *p = q->queue[i];
3148 if (check_pair_status(p))
3149 diff_flush_stat(p, options, &diffstat);
3151 if (output_format & DIFF_FORMAT_NUMSTAT)
3152 show_numstat(&diffstat, options);
3153 if (output_format & DIFF_FORMAT_DIFFSTAT)
3154 show_stats(&diffstat, options);
3155 if (output_format & DIFF_FORMAT_SHORTSTAT)
3156 show_shortstats(&diffstat, options);
3157 free_diffstat_info(&diffstat);
3158 separator++;
3160 if (output_format & DIFF_FORMAT_DIRSTAT)
3161 show_dirstat(options);
3163 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3164 for (i = 0; i < q->nr; i++)
3165 diff_summary(options->file, q->queue[i]);
3166 separator++;
3169 if (output_format & DIFF_FORMAT_PATCH) {
3170 if (separator) {
3171 if (options->stat_sep) {
3172 /* attach patch instead of inline */
3173 fputs(options->stat_sep, options->file);
3174 } else {
3175 putc(options->line_termination, options->file);
3179 for (i = 0; i < q->nr; i++) {
3180 struct diff_filepair *p = q->queue[i];
3181 if (check_pair_status(p))
3182 diff_flush_patch(p, options);
3186 if (output_format & DIFF_FORMAT_CALLBACK)
3187 options->format_callback(q, options, options->format_callback_data);
3189 for (i = 0; i < q->nr; i++)
3190 diff_free_filepair(q->queue[i]);
3191 free_queue:
3192 free(q->queue);
3193 q->queue = NULL;
3194 q->nr = q->alloc = 0;
3195 if (options->close_file)
3196 fclose(options->file);
3199 static void diffcore_apply_filter(const char *filter)
3201 int i;
3202 struct diff_queue_struct *q = &diff_queued_diff;
3203 struct diff_queue_struct outq;
3204 outq.queue = NULL;
3205 outq.nr = outq.alloc = 0;
3207 if (!filter)
3208 return;
3210 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3211 int found;
3212 for (i = found = 0; !found && i < q->nr; i++) {
3213 struct diff_filepair *p = q->queue[i];
3214 if (((p->status == DIFF_STATUS_MODIFIED) &&
3215 ((p->score &&
3216 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3217 (!p->score &&
3218 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3219 ((p->status != DIFF_STATUS_MODIFIED) &&
3220 strchr(filter, p->status)))
3221 found++;
3223 if (found)
3224 return;
3226 /* otherwise we will clear the whole queue
3227 * by copying the empty outq at the end of this
3228 * function, but first clear the current entries
3229 * in the queue.
3231 for (i = 0; i < q->nr; i++)
3232 diff_free_filepair(q->queue[i]);
3234 else {
3235 /* Only the matching ones */
3236 for (i = 0; i < q->nr; i++) {
3237 struct diff_filepair *p = q->queue[i];
3239 if (((p->status == DIFF_STATUS_MODIFIED) &&
3240 ((p->score &&
3241 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3242 (!p->score &&
3243 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3244 ((p->status != DIFF_STATUS_MODIFIED) &&
3245 strchr(filter, p->status)))
3246 diff_q(&outq, p);
3247 else
3248 diff_free_filepair(p);
3251 free(q->queue);
3252 *q = outq;
3255 /* Check whether two filespecs with the same mode and size are identical */
3256 static int diff_filespec_is_identical(struct diff_filespec *one,
3257 struct diff_filespec *two)
3259 if (S_ISGITLINK(one->mode))
3260 return 0;
3261 if (diff_populate_filespec(one, 0))
3262 return 0;
3263 if (diff_populate_filespec(two, 0))
3264 return 0;
3265 return !memcmp(one->data, two->data, one->size);
3268 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3270 int i;
3271 struct diff_queue_struct *q = &diff_queued_diff;
3272 struct diff_queue_struct outq;
3273 outq.queue = NULL;
3274 outq.nr = outq.alloc = 0;
3276 for (i = 0; i < q->nr; i++) {
3277 struct diff_filepair *p = q->queue[i];
3280 * 1. Entries that come from stat info dirtyness
3281 * always have both sides (iow, not create/delete),
3282 * one side of the object name is unknown, with
3283 * the same mode and size. Keep the ones that
3284 * do not match these criteria. They have real
3285 * differences.
3287 * 2. At this point, the file is known to be modified,
3288 * with the same mode and size, and the object
3289 * name of one side is unknown. Need to inspect
3290 * the identical contents.
3292 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3293 !DIFF_FILE_VALID(p->two) ||
3294 (p->one->sha1_valid && p->two->sha1_valid) ||
3295 (p->one->mode != p->two->mode) ||
3296 diff_populate_filespec(p->one, 1) ||
3297 diff_populate_filespec(p->two, 1) ||
3298 (p->one->size != p->two->size) ||
3299 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3300 diff_q(&outq, p);
3301 else {
3303 * The caller can subtract 1 from skip_stat_unmatch
3304 * to determine how many paths were dirty only
3305 * due to stat info mismatch.
3307 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3308 diffopt->skip_stat_unmatch++;
3309 diff_free_filepair(p);
3312 free(q->queue);
3313 *q = outq;
3316 void diffcore_std(struct diff_options *options)
3318 if (DIFF_OPT_TST(options, QUIET))
3319 return;
3321 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3322 diffcore_skip_stat_unmatch(options);
3323 if (options->break_opt != -1)
3324 diffcore_break(options->break_opt);
3325 if (options->detect_rename)
3326 diffcore_rename(options);
3327 if (options->break_opt != -1)
3328 diffcore_merge_broken();
3329 if (options->pickaxe)
3330 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3331 if (options->orderfile)
3332 diffcore_order(options->orderfile);
3333 diff_resolve_rename_copy();
3334 diffcore_apply_filter(options->filter);
3336 if (diff_queued_diff.nr)
3337 DIFF_OPT_SET(options, HAS_CHANGES);
3338 else
3339 DIFF_OPT_CLR(options, HAS_CHANGES);
3342 int diff_result_code(struct diff_options *opt, int status)
3344 int result = 0;
3345 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3346 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3347 return status;
3348 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3349 DIFF_OPT_TST(opt, HAS_CHANGES))
3350 result |= 01;
3351 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3352 DIFF_OPT_TST(opt, CHECK_FAILED))
3353 result |= 02;
3354 return result;
3357 void diff_addremove(struct diff_options *options,
3358 int addremove, unsigned mode,
3359 const unsigned char *sha1,
3360 const char *concatpath)
3362 struct diff_filespec *one, *two;
3364 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3365 return;
3367 /* This may look odd, but it is a preparation for
3368 * feeding "there are unchanged files which should
3369 * not produce diffs, but when you are doing copy
3370 * detection you would need them, so here they are"
3371 * entries to the diff-core. They will be prefixed
3372 * with something like '=' or '*' (I haven't decided
3373 * which but should not make any difference).
3374 * Feeding the same new and old to diff_change()
3375 * also has the same effect.
3376 * Before the final output happens, they are pruned after
3377 * merged into rename/copy pairs as appropriate.
3379 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3380 addremove = (addremove == '+' ? '-' :
3381 addremove == '-' ? '+' : addremove);
3383 if (options->prefix &&
3384 strncmp(concatpath, options->prefix, options->prefix_length))
3385 return;
3387 one = alloc_filespec(concatpath);
3388 two = alloc_filespec(concatpath);
3390 if (addremove != '+')
3391 fill_filespec(one, sha1, mode);
3392 if (addremove != '-')
3393 fill_filespec(two, sha1, mode);
3395 diff_queue(&diff_queued_diff, one, two);
3396 DIFF_OPT_SET(options, HAS_CHANGES);
3399 void diff_change(struct diff_options *options,
3400 unsigned old_mode, unsigned new_mode,
3401 const unsigned char *old_sha1,
3402 const unsigned char *new_sha1,
3403 const char *concatpath)
3405 struct diff_filespec *one, *two;
3407 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3408 && S_ISGITLINK(new_mode))
3409 return;
3411 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3412 unsigned tmp;
3413 const unsigned char *tmp_c;
3414 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3415 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3418 if (options->prefix &&
3419 strncmp(concatpath, options->prefix, options->prefix_length))
3420 return;
3422 one = alloc_filespec(concatpath);
3423 two = alloc_filespec(concatpath);
3424 fill_filespec(one, old_sha1, old_mode);
3425 fill_filespec(two, new_sha1, new_mode);
3427 diff_queue(&diff_queued_diff, one, two);
3428 DIFF_OPT_SET(options, HAS_CHANGES);
3431 void diff_unmerge(struct diff_options *options,
3432 const char *path,
3433 unsigned mode, const unsigned char *sha1)
3435 struct diff_filespec *one, *two;
3437 if (options->prefix &&
3438 strncmp(path, options->prefix, options->prefix_length))
3439 return;
3441 one = alloc_filespec(path);
3442 two = alloc_filespec(path);
3443 fill_filespec(one, sha1, mode);
3444 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;