mailsplit and mailinfo: gracefully handle NUL characters
[git/jnareb-git/bp-gitweb.git] / diff.c
blob439d4746cae343f08b298473d35d9792048abfa0
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)
134 if (!strcmp(var, "diff.renamelimit")) {
135 diff_rename_limit_default = git_config_int(var, value);
136 return 0;
138 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
139 diff_use_color_default = git_config_colorbool(var, value, -1);
140 return 0;
142 if (!strcmp(var, "diff.renames")) {
143 if (!value)
144 diff_detect_rename_default = DIFF_DETECT_RENAME;
145 else if (!strcasecmp(value, "copies") ||
146 !strcasecmp(value, "copy"))
147 diff_detect_rename_default = DIFF_DETECT_COPY;
148 else if (git_config_bool(var,value))
149 diff_detect_rename_default = DIFF_DETECT_RENAME;
150 return 0;
152 if (!strcmp(var, "diff.autorefreshindex")) {
153 diff_auto_refresh_index = git_config_bool(var, value);
154 return 0;
156 if (!strcmp(var, "diff.external")) {
157 if (!value)
158 return config_error_nonbool(var);
159 external_diff_cmd_cfg = xstrdup(value);
160 return 0;
162 if (!prefixcmp(var, "diff.")) {
163 const char *ep = strrchr(var, '.');
165 if (ep != var + 4 && !strcmp(ep, ".command"))
166 return parse_lldiff_command(var, ep, value);
169 return git_diff_basic_config(var, value);
172 int git_diff_basic_config(const char *var, const char *value)
174 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
175 int slot = parse_diff_color_slot(var, 11);
176 if (!value)
177 return config_error_nonbool(var);
178 color_parse(value, var, diff_colors[slot]);
179 return 0;
182 if (!prefixcmp(var, "diff.")) {
183 const char *ep = strrchr(var, '.');
184 if (ep != var + 4) {
185 if (!strcmp(ep, ".funcname")) {
186 if (!value)
187 return config_error_nonbool(var);
188 return parse_funcname_pattern(var, ep, value);
193 return git_color_default_config(var, value);
196 static char *quote_two(const char *one, const char *two)
198 int need_one = quote_c_style(one, NULL, NULL, 1);
199 int need_two = quote_c_style(two, NULL, NULL, 1);
200 struct strbuf res;
202 strbuf_init(&res, 0);
203 if (need_one + need_two) {
204 strbuf_addch(&res, '"');
205 quote_c_style(one, &res, NULL, 1);
206 quote_c_style(two, &res, NULL, 1);
207 strbuf_addch(&res, '"');
208 } else {
209 strbuf_addstr(&res, one);
210 strbuf_addstr(&res, two);
212 return strbuf_detach(&res, NULL);
215 static const char *external_diff(void)
217 static const char *external_diff_cmd = NULL;
218 static int done_preparing = 0;
220 if (done_preparing)
221 return external_diff_cmd;
222 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
223 if (!external_diff_cmd)
224 external_diff_cmd = external_diff_cmd_cfg;
225 done_preparing = 1;
226 return external_diff_cmd;
229 static struct diff_tempfile {
230 const char *name; /* filename external diff should read from */
231 char hex[41];
232 char mode[10];
233 char tmp_path[PATH_MAX];
234 } diff_temp[2];
236 static int count_lines(const char *data, int size)
238 int count, ch, completely_empty = 1, nl_just_seen = 0;
239 count = 0;
240 while (0 < size--) {
241 ch = *data++;
242 if (ch == '\n') {
243 count++;
244 nl_just_seen = 1;
245 completely_empty = 0;
247 else {
248 nl_just_seen = 0;
249 completely_empty = 0;
252 if (completely_empty)
253 return 0;
254 if (!nl_just_seen)
255 count++; /* no trailing newline */
256 return count;
259 static void print_line_count(FILE *file, int count)
261 switch (count) {
262 case 0:
263 fprintf(file, "0,0");
264 break;
265 case 1:
266 fprintf(file, "1");
267 break;
268 default:
269 fprintf(file, "1,%d", count);
270 break;
274 static void copy_file_with_prefix(FILE *file,
275 int prefix, const char *data, int size,
276 const char *set, const char *reset)
278 int ch, nl_just_seen = 1;
279 while (0 < size--) {
280 ch = *data++;
281 if (nl_just_seen) {
282 fputs(set, file);
283 putc(prefix, file);
285 if (ch == '\n') {
286 nl_just_seen = 1;
287 fputs(reset, file);
288 } else
289 nl_just_seen = 0;
290 putc(ch, file);
292 if (!nl_just_seen)
293 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
296 static void emit_rewrite_diff(const char *name_a,
297 const char *name_b,
298 struct diff_filespec *one,
299 struct diff_filespec *two,
300 struct diff_options *o)
302 int lc_a, lc_b;
303 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
304 const char *name_a_tab, *name_b_tab;
305 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
306 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
307 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
308 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
309 const char *reset = diff_get_color(color_diff, DIFF_RESET);
310 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
312 name_a += (*name_a == '/');
313 name_b += (*name_b == '/');
314 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
315 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
317 strbuf_reset(&a_name);
318 strbuf_reset(&b_name);
319 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
320 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
322 diff_populate_filespec(one, 0);
323 diff_populate_filespec(two, 0);
324 lc_a = count_lines(one->data, one->size);
325 lc_b = count_lines(two->data, two->size);
326 fprintf(o->file,
327 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
328 metainfo, a_name.buf, name_a_tab, reset,
329 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
330 print_line_count(o->file, lc_a);
331 fprintf(o->file, " +");
332 print_line_count(o->file, lc_b);
333 fprintf(o->file, " @@%s\n", reset);
334 if (lc_a)
335 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
336 if (lc_b)
337 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
340 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
342 if (!DIFF_FILE_VALID(one)) {
343 mf->ptr = (char *)""; /* does not matter */
344 mf->size = 0;
345 return 0;
347 else if (diff_populate_filespec(one, 0))
348 return -1;
349 mf->ptr = one->data;
350 mf->size = one->size;
351 return 0;
354 struct diff_words_buffer {
355 mmfile_t text;
356 long alloc;
357 long current; /* output pointer */
358 int suppressed_newline;
361 static void diff_words_append(char *line, unsigned long len,
362 struct diff_words_buffer *buffer)
364 if (buffer->text.size + len > buffer->alloc) {
365 buffer->alloc = (buffer->text.size + len) * 3 / 2;
366 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
368 line++;
369 len--;
370 memcpy(buffer->text.ptr + buffer->text.size, line, len);
371 buffer->text.size += len;
374 struct diff_words_data {
375 struct xdiff_emit_state xm;
376 struct diff_words_buffer minus, plus;
377 FILE *file;
380 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
381 int suppress_newline)
383 const char *ptr;
384 int eol = 0;
386 if (len == 0)
387 return;
389 ptr = buffer->text.ptr + buffer->current;
390 buffer->current += len;
392 if (ptr[len - 1] == '\n') {
393 eol = 1;
394 len--;
397 fputs(diff_get_color(1, color), file);
398 fwrite(ptr, len, 1, file);
399 fputs(diff_get_color(1, DIFF_RESET), file);
401 if (eol) {
402 if (suppress_newline)
403 buffer->suppressed_newline = 1;
404 else
405 putc('\n', file);
409 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
411 struct diff_words_data *diff_words = priv;
413 if (diff_words->minus.suppressed_newline) {
414 if (line[0] != '+')
415 putc('\n', diff_words->file);
416 diff_words->minus.suppressed_newline = 0;
419 len--;
420 switch (line[0]) {
421 case '-':
422 print_word(diff_words->file,
423 &diff_words->minus, len, DIFF_FILE_OLD, 1);
424 break;
425 case '+':
426 print_word(diff_words->file,
427 &diff_words->plus, len, DIFF_FILE_NEW, 0);
428 break;
429 case ' ':
430 print_word(diff_words->file,
431 &diff_words->plus, len, DIFF_PLAIN, 0);
432 diff_words->minus.current += len;
433 break;
437 /* this executes the word diff on the accumulated buffers */
438 static void diff_words_show(struct diff_words_data *diff_words)
440 xpparam_t xpp;
441 xdemitconf_t xecfg;
442 xdemitcb_t ecb;
443 mmfile_t minus, plus;
444 int i;
446 memset(&xecfg, 0, sizeof(xecfg));
447 minus.size = diff_words->minus.text.size;
448 minus.ptr = xmalloc(minus.size);
449 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
450 for (i = 0; i < minus.size; i++)
451 if (isspace(minus.ptr[i]))
452 minus.ptr[i] = '\n';
453 diff_words->minus.current = 0;
455 plus.size = diff_words->plus.text.size;
456 plus.ptr = xmalloc(plus.size);
457 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
458 for (i = 0; i < plus.size; i++)
459 if (isspace(plus.ptr[i]))
460 plus.ptr[i] = '\n';
461 diff_words->plus.current = 0;
463 xpp.flags = XDF_NEED_MINIMAL;
464 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
465 ecb.outf = xdiff_outf;
466 ecb.priv = diff_words;
467 diff_words->xm.consume = fn_out_diff_words_aux;
468 xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
470 free(minus.ptr);
471 free(plus.ptr);
472 diff_words->minus.text.size = diff_words->plus.text.size = 0;
474 if (diff_words->minus.suppressed_newline) {
475 putc('\n', diff_words->file);
476 diff_words->minus.suppressed_newline = 0;
480 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
482 struct emit_callback {
483 struct xdiff_emit_state xm;
484 int nparents, color_diff;
485 unsigned ws_rule;
486 sane_truncate_fn truncate;
487 const char **label_path;
488 struct diff_words_data *diff_words;
489 int *found_changesp;
490 FILE *file;
493 static void free_diff_words_data(struct emit_callback *ecbdata)
495 if (ecbdata->diff_words) {
496 /* flush buffers */
497 if (ecbdata->diff_words->minus.text.size ||
498 ecbdata->diff_words->plus.text.size)
499 diff_words_show(ecbdata->diff_words);
501 free (ecbdata->diff_words->minus.text.ptr);
502 free (ecbdata->diff_words->plus.text.ptr);
503 free(ecbdata->diff_words);
504 ecbdata->diff_words = NULL;
508 const char *diff_get_color(int diff_use_color, enum color_diff ix)
510 if (diff_use_color)
511 return diff_colors[ix];
512 return "";
515 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
517 fputs(set, file);
518 fwrite(line, len, 1, file);
519 fputs(reset, file);
522 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
524 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
525 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
527 if (!*ws)
528 emit_line(ecbdata->file, set, reset, line, len);
529 else {
530 /* Emit just the prefix, then the rest. */
531 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
532 (void)check_and_emit_line(line + ecbdata->nparents,
533 len - ecbdata->nparents, ecbdata->ws_rule,
534 ecbdata->file, set, reset, ws);
538 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
540 const char *cp;
541 unsigned long allot;
542 size_t l = len;
544 if (ecb->truncate)
545 return ecb->truncate(line, len);
546 cp = line;
547 allot = l;
548 while (0 < l) {
549 (void) utf8_width(&cp, &l);
550 if (!cp)
551 break; /* truncated in the middle? */
553 return allot - l;
556 static void fn_out_consume(void *priv, char *line, unsigned long len)
558 int i;
559 int color;
560 struct emit_callback *ecbdata = priv;
561 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
562 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
563 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
565 *(ecbdata->found_changesp) = 1;
567 if (ecbdata->label_path[0]) {
568 const char *name_a_tab, *name_b_tab;
570 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
571 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
573 fprintf(ecbdata->file, "%s--- %s%s%s\n",
574 meta, ecbdata->label_path[0], reset, name_a_tab);
575 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
576 meta, ecbdata->label_path[1], reset, name_b_tab);
577 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
580 /* This is not really necessary for now because
581 * this codepath only deals with two-way diffs.
583 for (i = 0; i < len && line[i] == '@'; i++)
585 if (2 <= i && i < len && line[i] == ' ') {
586 ecbdata->nparents = i - 1;
587 len = sane_truncate_line(ecbdata, line, len);
588 emit_line(ecbdata->file,
589 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
590 reset, line, len);
591 if (line[len-1] != '\n')
592 putc('\n', ecbdata->file);
593 return;
596 if (len < ecbdata->nparents) {
597 emit_line(ecbdata->file, reset, reset, line, len);
598 return;
601 color = DIFF_PLAIN;
602 if (ecbdata->diff_words && ecbdata->nparents != 1)
603 /* fall back to normal diff */
604 free_diff_words_data(ecbdata);
605 if (ecbdata->diff_words) {
606 if (line[0] == '-') {
607 diff_words_append(line, len,
608 &ecbdata->diff_words->minus);
609 return;
610 } else if (line[0] == '+') {
611 diff_words_append(line, len,
612 &ecbdata->diff_words->plus);
613 return;
615 if (ecbdata->diff_words->minus.text.size ||
616 ecbdata->diff_words->plus.text.size)
617 diff_words_show(ecbdata->diff_words);
618 line++;
619 len--;
620 emit_line(ecbdata->file, plain, reset, line, len);
621 return;
623 for (i = 0; i < ecbdata->nparents && len; i++) {
624 if (line[i] == '-')
625 color = DIFF_FILE_OLD;
626 else if (line[i] == '+')
627 color = DIFF_FILE_NEW;
630 if (color != DIFF_FILE_NEW) {
631 emit_line(ecbdata->file,
632 diff_get_color(ecbdata->color_diff, color),
633 reset, line, len);
634 return;
636 emit_add_line(reset, ecbdata, line, len);
639 static char *pprint_rename(const char *a, const char *b)
641 const char *old = a;
642 const char *new = b;
643 struct strbuf name;
644 int pfx_length, sfx_length;
645 int len_a = strlen(a);
646 int len_b = strlen(b);
647 int a_midlen, b_midlen;
648 int qlen_a = quote_c_style(a, NULL, NULL, 0);
649 int qlen_b = quote_c_style(b, NULL, NULL, 0);
651 strbuf_init(&name, 0);
652 if (qlen_a || qlen_b) {
653 quote_c_style(a, &name, NULL, 0);
654 strbuf_addstr(&name, " => ");
655 quote_c_style(b, &name, NULL, 0);
656 return strbuf_detach(&name, NULL);
659 /* Find common prefix */
660 pfx_length = 0;
661 while (*old && *new && *old == *new) {
662 if (*old == '/')
663 pfx_length = old - a + 1;
664 old++;
665 new++;
668 /* Find common suffix */
669 old = a + len_a;
670 new = b + len_b;
671 sfx_length = 0;
672 while (a <= old && b <= new && *old == *new) {
673 if (*old == '/')
674 sfx_length = len_a - (old - a);
675 old--;
676 new--;
680 * pfx{mid-a => mid-b}sfx
681 * {pfx-a => pfx-b}sfx
682 * pfx{sfx-a => sfx-b}
683 * name-a => name-b
685 a_midlen = len_a - pfx_length - sfx_length;
686 b_midlen = len_b - pfx_length - sfx_length;
687 if (a_midlen < 0)
688 a_midlen = 0;
689 if (b_midlen < 0)
690 b_midlen = 0;
692 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
693 if (pfx_length + sfx_length) {
694 strbuf_add(&name, a, pfx_length);
695 strbuf_addch(&name, '{');
697 strbuf_add(&name, a + pfx_length, a_midlen);
698 strbuf_addstr(&name, " => ");
699 strbuf_add(&name, b + pfx_length, b_midlen);
700 if (pfx_length + sfx_length) {
701 strbuf_addch(&name, '}');
702 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
704 return strbuf_detach(&name, NULL);
707 struct diffstat_t {
708 struct xdiff_emit_state xm;
710 int nr;
711 int alloc;
712 struct diffstat_file {
713 char *from_name;
714 char *name;
715 char *print_name;
716 unsigned is_unmerged:1;
717 unsigned is_binary:1;
718 unsigned is_renamed:1;
719 unsigned int added, deleted;
720 } **files;
723 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
724 const char *name_a,
725 const char *name_b)
727 struct diffstat_file *x;
728 x = xcalloc(sizeof (*x), 1);
729 if (diffstat->nr == diffstat->alloc) {
730 diffstat->alloc = alloc_nr(diffstat->alloc);
731 diffstat->files = xrealloc(diffstat->files,
732 diffstat->alloc * sizeof(x));
734 diffstat->files[diffstat->nr++] = x;
735 if (name_b) {
736 x->from_name = xstrdup(name_a);
737 x->name = xstrdup(name_b);
738 x->is_renamed = 1;
740 else {
741 x->from_name = NULL;
742 x->name = xstrdup(name_a);
744 return x;
747 static void diffstat_consume(void *priv, char *line, unsigned long len)
749 struct diffstat_t *diffstat = priv;
750 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
752 if (line[0] == '+')
753 x->added++;
754 else if (line[0] == '-')
755 x->deleted++;
758 const char mime_boundary_leader[] = "------------";
760 static int scale_linear(int it, int width, int max_change)
763 * make sure that at least one '-' is printed if there were deletions,
764 * and likewise for '+'.
766 if (max_change < 2)
767 return it;
768 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
771 static void show_name(FILE *file,
772 const char *prefix, const char *name, int len,
773 const char *reset, const char *set)
775 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
778 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
780 if (cnt <= 0)
781 return;
782 fprintf(file, "%s", set);
783 while (cnt--)
784 putc(ch, file);
785 fprintf(file, "%s", reset);
788 static void fill_print_name(struct diffstat_file *file)
790 char *pname;
792 if (file->print_name)
793 return;
795 if (!file->is_renamed) {
796 struct strbuf buf;
797 strbuf_init(&buf, 0);
798 if (quote_c_style(file->name, &buf, NULL, 0)) {
799 pname = strbuf_detach(&buf, NULL);
800 } else {
801 pname = file->name;
802 strbuf_release(&buf);
804 } else {
805 pname = pprint_rename(file->from_name, file->name);
807 file->print_name = pname;
810 static void show_stats(struct diffstat_t* data, struct diff_options *options)
812 int i, len, add, del, total, adds = 0, dels = 0;
813 int max_change = 0, max_len = 0;
814 int total_files = data->nr;
815 int width, name_width;
816 const char *reset, *set, *add_c, *del_c;
818 if (data->nr == 0)
819 return;
821 width = options->stat_width ? options->stat_width : 80;
822 name_width = options->stat_name_width ? options->stat_name_width : 50;
824 /* Sanity: give at least 5 columns to the graph,
825 * but leave at least 10 columns for the name.
827 if (width < name_width + 15) {
828 if (name_width <= 25)
829 width = name_width + 15;
830 else
831 name_width = width - 15;
834 /* Find the longest filename and max number of changes */
835 reset = diff_get_color_opt(options, DIFF_RESET);
836 set = diff_get_color_opt(options, DIFF_PLAIN);
837 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
838 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
840 for (i = 0; i < data->nr; i++) {
841 struct diffstat_file *file = data->files[i];
842 int change = file->added + file->deleted;
843 fill_print_name(file);
844 len = strlen(file->print_name);
845 if (max_len < len)
846 max_len = len;
848 if (file->is_binary || file->is_unmerged)
849 continue;
850 if (max_change < change)
851 max_change = change;
854 /* Compute the width of the graph part;
855 * 10 is for one blank at the beginning of the line plus
856 * " | count " between the name and the graph.
858 * From here on, name_width is the width of the name area,
859 * and width is the width of the graph area.
861 name_width = (name_width < max_len) ? name_width : max_len;
862 if (width < (name_width + 10) + max_change)
863 width = width - (name_width + 10);
864 else
865 width = max_change;
867 for (i = 0; i < data->nr; i++) {
868 const char *prefix = "";
869 char *name = data->files[i]->print_name;
870 int added = data->files[i]->added;
871 int deleted = data->files[i]->deleted;
872 int name_len;
875 * "scale" the filename
877 len = name_width;
878 name_len = strlen(name);
879 if (name_width < name_len) {
880 char *slash;
881 prefix = "...";
882 len -= 3;
883 name += name_len - len;
884 slash = strchr(name, '/');
885 if (slash)
886 name = slash;
889 if (data->files[i]->is_binary) {
890 show_name(options->file, prefix, name, len, reset, set);
891 fprintf(options->file, " Bin ");
892 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
893 fprintf(options->file, " -> ");
894 fprintf(options->file, "%s%d%s", add_c, added, reset);
895 fprintf(options->file, " bytes");
896 fprintf(options->file, "\n");
897 continue;
899 else if (data->files[i]->is_unmerged) {
900 show_name(options->file, prefix, name, len, reset, set);
901 fprintf(options->file, " Unmerged\n");
902 continue;
904 else if (!data->files[i]->is_renamed &&
905 (added + deleted == 0)) {
906 total_files--;
907 continue;
911 * scale the add/delete
913 add = added;
914 del = deleted;
915 total = add + del;
916 adds += add;
917 dels += del;
919 if (width <= max_change) {
920 add = scale_linear(add, width, max_change);
921 del = scale_linear(del, width, max_change);
922 total = add + del;
924 show_name(options->file, prefix, name, len, reset, set);
925 fprintf(options->file, "%5d ", added + deleted);
926 show_graph(options->file, '+', add, add_c, reset);
927 show_graph(options->file, '-', del, del_c, reset);
928 fprintf(options->file, "\n");
930 fprintf(options->file,
931 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
932 set, total_files, adds, dels, reset);
935 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
937 int i, adds = 0, dels = 0, total_files = data->nr;
939 if (data->nr == 0)
940 return;
942 for (i = 0; i < data->nr; i++) {
943 if (!data->files[i]->is_binary &&
944 !data->files[i]->is_unmerged) {
945 int added = data->files[i]->added;
946 int deleted= data->files[i]->deleted;
947 if (!data->files[i]->is_renamed &&
948 (added + deleted == 0)) {
949 total_files--;
950 } else {
951 adds += added;
952 dels += deleted;
956 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
957 total_files, adds, dels);
960 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
962 int i;
964 if (data->nr == 0)
965 return;
967 for (i = 0; i < data->nr; i++) {
968 struct diffstat_file *file = data->files[i];
970 if (file->is_binary)
971 fprintf(options->file, "-\t-\t");
972 else
973 fprintf(options->file,
974 "%d\t%d\t", file->added, file->deleted);
975 if (options->line_termination) {
976 fill_print_name(file);
977 if (!file->is_renamed)
978 write_name_quoted(file->name, options->file,
979 options->line_termination);
980 else {
981 fputs(file->print_name, options->file);
982 putc(options->line_termination, options->file);
984 } else {
985 if (file->is_renamed) {
986 putc('\0', options->file);
987 write_name_quoted(file->from_name, options->file, '\0');
989 write_name_quoted(file->name, options->file, '\0');
994 struct dirstat_file {
995 const char *name;
996 unsigned long changed;
999 struct dirstat_dir {
1000 struct dirstat_file *files;
1001 int alloc, nr, percent, cumulative;
1004 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1006 unsigned long this_dir = 0;
1007 unsigned int sources = 0;
1009 while (dir->nr) {
1010 struct dirstat_file *f = dir->files;
1011 int namelen = strlen(f->name);
1012 unsigned long this;
1013 char *slash;
1015 if (namelen < baselen)
1016 break;
1017 if (memcmp(f->name, base, baselen))
1018 break;
1019 slash = strchr(f->name + baselen, '/');
1020 if (slash) {
1021 int newbaselen = slash + 1 - f->name;
1022 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1023 sources++;
1024 } else {
1025 this = f->changed;
1026 dir->files++;
1027 dir->nr--;
1028 sources += 2;
1030 this_dir += this;
1034 * We don't report dirstat's for
1035 * - the top level
1036 * - or cases where everything came from a single directory
1037 * under this directory (sources == 1).
1039 if (baselen && sources != 1) {
1040 int permille = this_dir * 1000 / changed;
1041 if (permille) {
1042 int percent = permille / 10;
1043 if (percent >= dir->percent) {
1044 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1045 if (!dir->cumulative)
1046 return 0;
1050 return this_dir;
1053 static void show_dirstat(struct diff_options *options)
1055 int i;
1056 unsigned long changed;
1057 struct dirstat_dir dir;
1058 struct diff_queue_struct *q = &diff_queued_diff;
1060 dir.files = NULL;
1061 dir.alloc = 0;
1062 dir.nr = 0;
1063 dir.percent = options->dirstat_percent;
1064 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1066 changed = 0;
1067 for (i = 0; i < q->nr; i++) {
1068 struct diff_filepair *p = q->queue[i];
1069 const char *name;
1070 unsigned long copied, added, damage;
1072 name = p->one->path ? p->one->path : p->two->path;
1074 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1075 diff_populate_filespec(p->one, 0);
1076 diff_populate_filespec(p->two, 0);
1077 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1078 &copied, &added);
1079 diff_free_filespec_data(p->one);
1080 diff_free_filespec_data(p->two);
1081 } else if (DIFF_FILE_VALID(p->one)) {
1082 diff_populate_filespec(p->one, 1);
1083 copied = added = 0;
1084 diff_free_filespec_data(p->one);
1085 } else if (DIFF_FILE_VALID(p->two)) {
1086 diff_populate_filespec(p->two, 1);
1087 copied = 0;
1088 added = p->two->size;
1089 diff_free_filespec_data(p->two);
1090 } else
1091 continue;
1094 * Original minus copied is the removed material,
1095 * added is the new material. They are both damages
1096 * made to the preimage.
1098 damage = (p->one->size - copied) + added;
1100 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1101 dir.files[dir.nr].name = name;
1102 dir.files[dir.nr].changed = damage;
1103 changed += damage;
1104 dir.nr++;
1107 /* This can happen even with many files, if everything was renames */
1108 if (!changed)
1109 return;
1111 /* Show all directories with more than x% of the changes */
1112 gather_dirstat(options->file, &dir, changed, "", 0);
1115 static void free_diffstat_info(struct diffstat_t *diffstat)
1117 int i;
1118 for (i = 0; i < diffstat->nr; i++) {
1119 struct diffstat_file *f = diffstat->files[i];
1120 if (f->name != f->print_name)
1121 free(f->print_name);
1122 free(f->name);
1123 free(f->from_name);
1124 free(f);
1126 free(diffstat->files);
1129 struct checkdiff_t {
1130 struct xdiff_emit_state xm;
1131 const char *filename;
1132 int lineno, color_diff;
1133 unsigned ws_rule;
1134 unsigned status;
1135 FILE *file;
1138 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1140 struct checkdiff_t *data = priv;
1141 const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1142 const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1143 const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1144 char *err;
1146 if (line[0] == '+') {
1147 data->lineno++;
1148 data->status = check_and_emit_line(line + 1, len - 1,
1149 data->ws_rule, NULL, NULL, NULL, NULL);
1150 if (!data->status)
1151 return;
1152 err = whitespace_error_string(data->status);
1153 fprintf(data->file, "%s:%d: %s.\n", data->filename, data->lineno, err);
1154 free(err);
1155 emit_line(data->file, set, reset, line, 1);
1156 (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
1157 data->file, set, reset, ws);
1158 } else if (line[0] == ' ')
1159 data->lineno++;
1160 else if (line[0] == '@') {
1161 char *plus = strchr(line, '+');
1162 if (plus)
1163 data->lineno = strtol(plus, NULL, 10) - 1;
1164 else
1165 die("invalid diff");
1169 static unsigned char *deflate_it(char *data,
1170 unsigned long size,
1171 unsigned long *result_size)
1173 int bound;
1174 unsigned char *deflated;
1175 z_stream stream;
1177 memset(&stream, 0, sizeof(stream));
1178 deflateInit(&stream, zlib_compression_level);
1179 bound = deflateBound(&stream, size);
1180 deflated = xmalloc(bound);
1181 stream.next_out = deflated;
1182 stream.avail_out = bound;
1184 stream.next_in = (unsigned char *)data;
1185 stream.avail_in = size;
1186 while (deflate(&stream, Z_FINISH) == Z_OK)
1187 ; /* nothing */
1188 deflateEnd(&stream);
1189 *result_size = stream.total_out;
1190 return deflated;
1193 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1195 void *cp;
1196 void *delta;
1197 void *deflated;
1198 void *data;
1199 unsigned long orig_size;
1200 unsigned long delta_size;
1201 unsigned long deflate_size;
1202 unsigned long data_size;
1204 /* We could do deflated delta, or we could do just deflated two,
1205 * whichever is smaller.
1207 delta = NULL;
1208 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1209 if (one->size && two->size) {
1210 delta = diff_delta(one->ptr, one->size,
1211 two->ptr, two->size,
1212 &delta_size, deflate_size);
1213 if (delta) {
1214 void *to_free = delta;
1215 orig_size = delta_size;
1216 delta = deflate_it(delta, delta_size, &delta_size);
1217 free(to_free);
1221 if (delta && delta_size < deflate_size) {
1222 fprintf(file, "delta %lu\n", orig_size);
1223 free(deflated);
1224 data = delta;
1225 data_size = delta_size;
1227 else {
1228 fprintf(file, "literal %lu\n", two->size);
1229 free(delta);
1230 data = deflated;
1231 data_size = deflate_size;
1234 /* emit data encoded in base85 */
1235 cp = data;
1236 while (data_size) {
1237 int bytes = (52 < data_size) ? 52 : data_size;
1238 char line[70];
1239 data_size -= bytes;
1240 if (bytes <= 26)
1241 line[0] = bytes + 'A' - 1;
1242 else
1243 line[0] = bytes - 26 + 'a' - 1;
1244 encode_85(line + 1, cp, bytes);
1245 cp = (char *) cp + bytes;
1246 fputs(line, file);
1247 fputc('\n', file);
1249 fprintf(file, "\n");
1250 free(data);
1253 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1255 fprintf(file, "GIT binary patch\n");
1256 emit_binary_diff_body(file, one, two);
1257 emit_binary_diff_body(file, two, one);
1260 static void setup_diff_attr_check(struct git_attr_check *check)
1262 static struct git_attr *attr_diff;
1264 if (!attr_diff) {
1265 attr_diff = git_attr("diff", 4);
1267 check[0].attr = attr_diff;
1270 static void diff_filespec_check_attr(struct diff_filespec *one)
1272 struct git_attr_check attr_diff_check;
1273 int check_from_data = 0;
1275 if (one->checked_attr)
1276 return;
1278 setup_diff_attr_check(&attr_diff_check);
1279 one->is_binary = 0;
1280 one->funcname_pattern_ident = NULL;
1282 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1283 const char *value;
1285 /* binaryness */
1286 value = attr_diff_check.value;
1287 if (ATTR_TRUE(value))
1289 else if (ATTR_FALSE(value))
1290 one->is_binary = 1;
1291 else
1292 check_from_data = 1;
1294 /* funcname pattern ident */
1295 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1297 else
1298 one->funcname_pattern_ident = value;
1301 if (check_from_data) {
1302 if (!one->data && DIFF_FILE_VALID(one))
1303 diff_populate_filespec(one, 0);
1305 if (one->data)
1306 one->is_binary = buffer_is_binary(one->data, one->size);
1310 int diff_filespec_is_binary(struct diff_filespec *one)
1312 diff_filespec_check_attr(one);
1313 return one->is_binary;
1316 static const char *funcname_pattern(const char *ident)
1318 struct funcname_pattern *pp;
1320 for (pp = funcname_pattern_list; pp; pp = pp->next)
1321 if (!strcmp(ident, pp->name))
1322 return pp->pattern;
1323 return NULL;
1326 static struct builtin_funcname_pattern {
1327 const char *name;
1328 const char *pattern;
1329 } builtin_funcname_pattern[] = {
1330 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1331 "new\\|return\\|switch\\|throw\\|while\\)\n"
1332 "^[ ]*\\(\\([ ]*"
1333 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1334 "[ ]*([^;]*\\)$" },
1335 { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1338 static const char *diff_funcname_pattern(struct diff_filespec *one)
1340 const char *ident, *pattern;
1341 int i;
1343 diff_filespec_check_attr(one);
1344 ident = one->funcname_pattern_ident;
1346 if (!ident)
1348 * If the config file has "funcname.default" defined, that
1349 * regexp is used; otherwise NULL is returned and xemit uses
1350 * the built-in default.
1352 return funcname_pattern("default");
1354 /* Look up custom "funcname.$ident" regexp from config. */
1355 pattern = funcname_pattern(ident);
1356 if (pattern)
1357 return pattern;
1360 * And define built-in fallback patterns here. Note that
1361 * these can be overridden by the user's config settings.
1363 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1364 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1365 return builtin_funcname_pattern[i].pattern;
1367 return NULL;
1370 static void builtin_diff(const char *name_a,
1371 const char *name_b,
1372 struct diff_filespec *one,
1373 struct diff_filespec *two,
1374 const char *xfrm_msg,
1375 struct diff_options *o,
1376 int complete_rewrite)
1378 mmfile_t mf1, mf2;
1379 const char *lbl[2];
1380 char *a_one, *b_two;
1381 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1382 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1384 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1385 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1386 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1387 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1388 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1389 if (lbl[0][0] == '/') {
1390 /* /dev/null */
1391 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1392 if (xfrm_msg && xfrm_msg[0])
1393 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1395 else if (lbl[1][0] == '/') {
1396 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1397 if (xfrm_msg && xfrm_msg[0])
1398 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1400 else {
1401 if (one->mode != two->mode) {
1402 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1403 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1405 if (xfrm_msg && xfrm_msg[0])
1406 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1408 * we do not run diff between different kind
1409 * of objects.
1411 if ((one->mode ^ two->mode) & S_IFMT)
1412 goto free_ab_and_return;
1413 if (complete_rewrite) {
1414 emit_rewrite_diff(name_a, name_b, one, two, o);
1415 o->found_changes = 1;
1416 goto free_ab_and_return;
1420 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1421 die("unable to read files to diff");
1423 if (!DIFF_OPT_TST(o, TEXT) &&
1424 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1425 /* Quite common confusing case */
1426 if (mf1.size == mf2.size &&
1427 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1428 goto free_ab_and_return;
1429 if (DIFF_OPT_TST(o, BINARY))
1430 emit_binary_diff(o->file, &mf1, &mf2);
1431 else
1432 fprintf(o->file, "Binary files %s and %s differ\n",
1433 lbl[0], lbl[1]);
1434 o->found_changes = 1;
1436 else {
1437 /* Crazy xdl interfaces.. */
1438 const char *diffopts = getenv("GIT_DIFF_OPTS");
1439 xpparam_t xpp;
1440 xdemitconf_t xecfg;
1441 xdemitcb_t ecb;
1442 struct emit_callback ecbdata;
1443 const char *funcname_pattern;
1445 funcname_pattern = diff_funcname_pattern(one);
1446 if (!funcname_pattern)
1447 funcname_pattern = diff_funcname_pattern(two);
1449 memset(&xecfg, 0, sizeof(xecfg));
1450 memset(&ecbdata, 0, sizeof(ecbdata));
1451 ecbdata.label_path = lbl;
1452 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1453 ecbdata.found_changesp = &o->found_changes;
1454 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1455 ecbdata.file = o->file;
1456 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1457 xecfg.ctxlen = o->context;
1458 xecfg.flags = XDL_EMIT_FUNCNAMES;
1459 if (funcname_pattern)
1460 xdiff_set_find_func(&xecfg, funcname_pattern);
1461 if (!diffopts)
1463 else if (!prefixcmp(diffopts, "--unified="))
1464 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1465 else if (!prefixcmp(diffopts, "-u"))
1466 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1467 ecb.outf = xdiff_outf;
1468 ecb.priv = &ecbdata;
1469 ecbdata.xm.consume = fn_out_consume;
1470 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1471 ecbdata.diff_words =
1472 xcalloc(1, sizeof(struct diff_words_data));
1473 ecbdata.diff_words->file = o->file;
1475 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1476 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1477 free_diff_words_data(&ecbdata);
1480 free_ab_and_return:
1481 diff_free_filespec_data(one);
1482 diff_free_filespec_data(two);
1483 free(a_one);
1484 free(b_two);
1485 return;
1488 static void builtin_diffstat(const char *name_a, const char *name_b,
1489 struct diff_filespec *one,
1490 struct diff_filespec *two,
1491 struct diffstat_t *diffstat,
1492 struct diff_options *o,
1493 int complete_rewrite)
1495 mmfile_t mf1, mf2;
1496 struct diffstat_file *data;
1498 data = diffstat_add(diffstat, name_a, name_b);
1500 if (!one || !two) {
1501 data->is_unmerged = 1;
1502 return;
1504 if (complete_rewrite) {
1505 diff_populate_filespec(one, 0);
1506 diff_populate_filespec(two, 0);
1507 data->deleted = count_lines(one->data, one->size);
1508 data->added = count_lines(two->data, two->size);
1509 goto free_and_return;
1511 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1512 die("unable to read files to diff");
1514 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1515 data->is_binary = 1;
1516 data->added = mf2.size;
1517 data->deleted = mf1.size;
1518 } else {
1519 /* Crazy xdl interfaces.. */
1520 xpparam_t xpp;
1521 xdemitconf_t xecfg;
1522 xdemitcb_t ecb;
1524 memset(&xecfg, 0, sizeof(xecfg));
1525 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1526 ecb.outf = xdiff_outf;
1527 ecb.priv = diffstat;
1528 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1531 free_and_return:
1532 diff_free_filespec_data(one);
1533 diff_free_filespec_data(two);
1536 static void builtin_checkdiff(const char *name_a, const char *name_b,
1537 const char *attr_path,
1538 struct diff_filespec *one,
1539 struct diff_filespec *two, struct diff_options *o)
1541 mmfile_t mf1, mf2;
1542 struct checkdiff_t data;
1544 if (!two)
1545 return;
1547 memset(&data, 0, sizeof(data));
1548 data.xm.consume = checkdiff_consume;
1549 data.filename = name_b ? name_b : name_a;
1550 data.lineno = 0;
1551 data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1552 data.ws_rule = whitespace_rule(attr_path);
1553 data.file = o->file;
1555 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1556 die("unable to read files to diff");
1558 if (diff_filespec_is_binary(two))
1559 goto free_and_return;
1560 else {
1561 /* Crazy xdl interfaces.. */
1562 xpparam_t xpp;
1563 xdemitconf_t xecfg;
1564 xdemitcb_t ecb;
1566 memset(&xecfg, 0, sizeof(xecfg));
1567 xpp.flags = XDF_NEED_MINIMAL;
1568 ecb.outf = xdiff_outf;
1569 ecb.priv = &data;
1570 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1572 free_and_return:
1573 diff_free_filespec_data(one);
1574 diff_free_filespec_data(two);
1575 if (data.status)
1576 DIFF_OPT_SET(o, CHECK_FAILED);
1579 struct diff_filespec *alloc_filespec(const char *path)
1581 int namelen = strlen(path);
1582 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1584 memset(spec, 0, sizeof(*spec));
1585 spec->path = (char *)(spec + 1);
1586 memcpy(spec->path, path, namelen+1);
1587 spec->count = 1;
1588 return spec;
1591 void free_filespec(struct diff_filespec *spec)
1593 if (!--spec->count) {
1594 diff_free_filespec_data(spec);
1595 free(spec);
1599 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1600 unsigned short mode)
1602 if (mode) {
1603 spec->mode = canon_mode(mode);
1604 hashcpy(spec->sha1, sha1);
1605 spec->sha1_valid = !is_null_sha1(sha1);
1610 * Given a name and sha1 pair, if the index tells us the file in
1611 * the work tree has that object contents, return true, so that
1612 * prepare_temp_file() does not have to inflate and extract.
1614 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1616 struct cache_entry *ce;
1617 struct stat st;
1618 int pos, len;
1620 /* We do not read the cache ourselves here, because the
1621 * benchmark with my previous version that always reads cache
1622 * shows that it makes things worse for diff-tree comparing
1623 * two linux-2.6 kernel trees in an already checked out work
1624 * tree. This is because most diff-tree comparisons deal with
1625 * only a small number of files, while reading the cache is
1626 * expensive for a large project, and its cost outweighs the
1627 * savings we get by not inflating the object to a temporary
1628 * file. Practically, this code only helps when we are used
1629 * by diff-cache --cached, which does read the cache before
1630 * calling us.
1632 if (!active_cache)
1633 return 0;
1635 /* We want to avoid the working directory if our caller
1636 * doesn't need the data in a normal file, this system
1637 * is rather slow with its stat/open/mmap/close syscalls,
1638 * and the object is contained in a pack file. The pack
1639 * is probably already open and will be faster to obtain
1640 * the data through than the working directory. Loose
1641 * objects however would tend to be slower as they need
1642 * to be individually opened and inflated.
1644 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1645 return 0;
1647 len = strlen(name);
1648 pos = cache_name_pos(name, len);
1649 if (pos < 0)
1650 return 0;
1651 ce = active_cache[pos];
1654 * This is not the sha1 we are looking for, or
1655 * unreusable because it is not a regular file.
1657 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1658 return 0;
1661 * If ce matches the file in the work tree, we can reuse it.
1663 if (ce_uptodate(ce) ||
1664 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1665 return 1;
1667 return 0;
1670 static int populate_from_stdin(struct diff_filespec *s)
1672 struct strbuf buf;
1673 size_t size = 0;
1675 strbuf_init(&buf, 0);
1676 if (strbuf_read(&buf, 0, 0) < 0)
1677 return error("error while reading from stdin %s",
1678 strerror(errno));
1680 s->should_munmap = 0;
1681 s->data = strbuf_detach(&buf, &size);
1682 s->size = size;
1683 s->should_free = 1;
1684 return 0;
1687 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1689 int len;
1690 char *data = xmalloc(100);
1691 len = snprintf(data, 100,
1692 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1693 s->data = data;
1694 s->size = len;
1695 s->should_free = 1;
1696 if (size_only) {
1697 s->data = NULL;
1698 free(data);
1700 return 0;
1704 * While doing rename detection and pickaxe operation, we may need to
1705 * grab the data for the blob (or file) for our own in-core comparison.
1706 * diff_filespec has data and size fields for this purpose.
1708 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1710 int err = 0;
1711 if (!DIFF_FILE_VALID(s))
1712 die("internal error: asking to populate invalid file.");
1713 if (S_ISDIR(s->mode))
1714 return -1;
1716 if (s->data)
1717 return 0;
1719 if (size_only && 0 < s->size)
1720 return 0;
1722 if (S_ISGITLINK(s->mode))
1723 return diff_populate_gitlink(s, size_only);
1725 if (!s->sha1_valid ||
1726 reuse_worktree_file(s->path, s->sha1, 0)) {
1727 struct strbuf buf;
1728 struct stat st;
1729 int fd;
1731 if (!strcmp(s->path, "-"))
1732 return populate_from_stdin(s);
1734 if (lstat(s->path, &st) < 0) {
1735 if (errno == ENOENT) {
1736 err_empty:
1737 err = -1;
1738 empty:
1739 s->data = (char *)"";
1740 s->size = 0;
1741 return err;
1744 s->size = xsize_t(st.st_size);
1745 if (!s->size)
1746 goto empty;
1747 if (size_only)
1748 return 0;
1749 if (S_ISLNK(st.st_mode)) {
1750 int ret;
1751 s->data = xmalloc(s->size);
1752 s->should_free = 1;
1753 ret = readlink(s->path, s->data, s->size);
1754 if (ret < 0) {
1755 free(s->data);
1756 goto err_empty;
1758 return 0;
1760 fd = open(s->path, O_RDONLY);
1761 if (fd < 0)
1762 goto err_empty;
1763 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1764 close(fd);
1765 s->should_munmap = 1;
1768 * Convert from working tree format to canonical git format
1770 strbuf_init(&buf, 0);
1771 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1772 size_t size = 0;
1773 munmap(s->data, s->size);
1774 s->should_munmap = 0;
1775 s->data = strbuf_detach(&buf, &size);
1776 s->size = size;
1777 s->should_free = 1;
1780 else {
1781 enum object_type type;
1782 if (size_only)
1783 type = sha1_object_info(s->sha1, &s->size);
1784 else {
1785 s->data = read_sha1_file(s->sha1, &type, &s->size);
1786 s->should_free = 1;
1789 return 0;
1792 void diff_free_filespec_blob(struct diff_filespec *s)
1794 if (s->should_free)
1795 free(s->data);
1796 else if (s->should_munmap)
1797 munmap(s->data, s->size);
1799 if (s->should_free || s->should_munmap) {
1800 s->should_free = s->should_munmap = 0;
1801 s->data = NULL;
1805 void diff_free_filespec_data(struct diff_filespec *s)
1807 diff_free_filespec_blob(s);
1808 free(s->cnt_data);
1809 s->cnt_data = NULL;
1812 static void prep_temp_blob(struct diff_tempfile *temp,
1813 void *blob,
1814 unsigned long size,
1815 const unsigned char *sha1,
1816 int mode)
1818 int fd;
1820 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1821 if (fd < 0)
1822 die("unable to create temp-file: %s", strerror(errno));
1823 if (write_in_full(fd, blob, size) != size)
1824 die("unable to write temp-file");
1825 close(fd);
1826 temp->name = temp->tmp_path;
1827 strcpy(temp->hex, sha1_to_hex(sha1));
1828 temp->hex[40] = 0;
1829 sprintf(temp->mode, "%06o", mode);
1832 static void prepare_temp_file(const char *name,
1833 struct diff_tempfile *temp,
1834 struct diff_filespec *one)
1836 if (!DIFF_FILE_VALID(one)) {
1837 not_a_valid_file:
1838 /* A '-' entry produces this for file-2, and
1839 * a '+' entry produces this for file-1.
1841 temp->name = "/dev/null";
1842 strcpy(temp->hex, ".");
1843 strcpy(temp->mode, ".");
1844 return;
1847 if (!one->sha1_valid ||
1848 reuse_worktree_file(name, one->sha1, 1)) {
1849 struct stat st;
1850 if (lstat(name, &st) < 0) {
1851 if (errno == ENOENT)
1852 goto not_a_valid_file;
1853 die("stat(%s): %s", name, strerror(errno));
1855 if (S_ISLNK(st.st_mode)) {
1856 int ret;
1857 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1858 size_t sz = xsize_t(st.st_size);
1859 if (sizeof(buf) <= st.st_size)
1860 die("symlink too long: %s", name);
1861 ret = readlink(name, buf, sz);
1862 if (ret < 0)
1863 die("readlink(%s)", name);
1864 prep_temp_blob(temp, buf, sz,
1865 (one->sha1_valid ?
1866 one->sha1 : null_sha1),
1867 (one->sha1_valid ?
1868 one->mode : S_IFLNK));
1870 else {
1871 /* we can borrow from the file in the work tree */
1872 temp->name = name;
1873 if (!one->sha1_valid)
1874 strcpy(temp->hex, sha1_to_hex(null_sha1));
1875 else
1876 strcpy(temp->hex, sha1_to_hex(one->sha1));
1877 /* Even though we may sometimes borrow the
1878 * contents from the work tree, we always want
1879 * one->mode. mode is trustworthy even when
1880 * !(one->sha1_valid), as long as
1881 * DIFF_FILE_VALID(one).
1883 sprintf(temp->mode, "%06o", one->mode);
1885 return;
1887 else {
1888 if (diff_populate_filespec(one, 0))
1889 die("cannot read data blob for %s", one->path);
1890 prep_temp_blob(temp, one->data, one->size,
1891 one->sha1, one->mode);
1895 static void remove_tempfile(void)
1897 int i;
1899 for (i = 0; i < 2; i++)
1900 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1901 unlink(diff_temp[i].name);
1902 diff_temp[i].name = NULL;
1906 static void remove_tempfile_on_signal(int signo)
1908 remove_tempfile();
1909 signal(SIGINT, SIG_DFL);
1910 raise(signo);
1913 /* An external diff command takes:
1915 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1916 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1919 static void run_external_diff(const char *pgm,
1920 const char *name,
1921 const char *other,
1922 struct diff_filespec *one,
1923 struct diff_filespec *two,
1924 const char *xfrm_msg,
1925 int complete_rewrite)
1927 const char *spawn_arg[10];
1928 struct diff_tempfile *temp = diff_temp;
1929 int retval;
1930 static int atexit_asked = 0;
1931 const char *othername;
1932 const char **arg = &spawn_arg[0];
1934 othername = (other? other : name);
1935 if (one && two) {
1936 prepare_temp_file(name, &temp[0], one);
1937 prepare_temp_file(othername, &temp[1], two);
1938 if (! atexit_asked &&
1939 (temp[0].name == temp[0].tmp_path ||
1940 temp[1].name == temp[1].tmp_path)) {
1941 atexit_asked = 1;
1942 atexit(remove_tempfile);
1944 signal(SIGINT, remove_tempfile_on_signal);
1947 if (one && two) {
1948 *arg++ = pgm;
1949 *arg++ = name;
1950 *arg++ = temp[0].name;
1951 *arg++ = temp[0].hex;
1952 *arg++ = temp[0].mode;
1953 *arg++ = temp[1].name;
1954 *arg++ = temp[1].hex;
1955 *arg++ = temp[1].mode;
1956 if (other) {
1957 *arg++ = other;
1958 *arg++ = xfrm_msg;
1960 } else {
1961 *arg++ = pgm;
1962 *arg++ = name;
1964 *arg = NULL;
1965 fflush(NULL);
1966 retval = run_command_v_opt(spawn_arg, 0);
1967 remove_tempfile();
1968 if (retval) {
1969 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1970 exit(1);
1974 static const char *external_diff_attr(const char *name)
1976 struct git_attr_check attr_diff_check;
1978 if (!name)
1979 return NULL;
1981 setup_diff_attr_check(&attr_diff_check);
1982 if (!git_checkattr(name, 1, &attr_diff_check)) {
1983 const char *value = attr_diff_check.value;
1984 if (!ATTR_TRUE(value) &&
1985 !ATTR_FALSE(value) &&
1986 !ATTR_UNSET(value)) {
1987 struct ll_diff_driver *drv;
1989 for (drv = user_diff; drv; drv = drv->next)
1990 if (!strcmp(drv->name, value))
1991 return drv->cmd;
1994 return NULL;
1997 static void run_diff_cmd(const char *pgm,
1998 const char *name,
1999 const char *other,
2000 const char *attr_path,
2001 struct diff_filespec *one,
2002 struct diff_filespec *two,
2003 const char *xfrm_msg,
2004 struct diff_options *o,
2005 int complete_rewrite)
2007 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2008 pgm = NULL;
2009 else {
2010 const char *cmd = external_diff_attr(attr_path);
2011 if (cmd)
2012 pgm = cmd;
2015 if (pgm) {
2016 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2017 complete_rewrite);
2018 return;
2020 if (one && two)
2021 builtin_diff(name, other ? other : name,
2022 one, two, xfrm_msg, o, complete_rewrite);
2023 else
2024 fprintf(o->file, "* Unmerged path %s\n", name);
2027 static void diff_fill_sha1_info(struct diff_filespec *one)
2029 if (DIFF_FILE_VALID(one)) {
2030 if (!one->sha1_valid) {
2031 struct stat st;
2032 if (!strcmp(one->path, "-")) {
2033 hashcpy(one->sha1, null_sha1);
2034 return;
2036 if (lstat(one->path, &st) < 0)
2037 die("stat %s", one->path);
2038 if (index_path(one->sha1, one->path, &st, 0))
2039 die("cannot hash %s\n", one->path);
2042 else
2043 hashclr(one->sha1);
2046 static int similarity_index(struct diff_filepair *p)
2048 return p->score * 100 / MAX_SCORE;
2051 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2053 /* Strip the prefix but do not molest /dev/null and absolute paths */
2054 if (*namep && **namep != '/')
2055 *namep += prefix_length;
2056 if (*otherp && **otherp != '/')
2057 *otherp += prefix_length;
2060 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2062 const char *pgm = external_diff();
2063 struct strbuf msg;
2064 char *xfrm_msg;
2065 struct diff_filespec *one = p->one;
2066 struct diff_filespec *two = p->two;
2067 const char *name;
2068 const char *other;
2069 const char *attr_path;
2070 int complete_rewrite = 0;
2072 name = p->one->path;
2073 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2074 attr_path = name;
2075 if (o->prefix_length)
2076 strip_prefix(o->prefix_length, &name, &other);
2078 if (DIFF_PAIR_UNMERGED(p)) {
2079 run_diff_cmd(pgm, name, NULL, attr_path,
2080 NULL, NULL, NULL, o, 0);
2081 return;
2084 diff_fill_sha1_info(one);
2085 diff_fill_sha1_info(two);
2087 strbuf_init(&msg, PATH_MAX * 2 + 300);
2088 switch (p->status) {
2089 case DIFF_STATUS_COPIED:
2090 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2091 strbuf_addstr(&msg, "\ncopy from ");
2092 quote_c_style(name, &msg, NULL, 0);
2093 strbuf_addstr(&msg, "\ncopy to ");
2094 quote_c_style(other, &msg, NULL, 0);
2095 strbuf_addch(&msg, '\n');
2096 break;
2097 case DIFF_STATUS_RENAMED:
2098 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2099 strbuf_addstr(&msg, "\nrename from ");
2100 quote_c_style(name, &msg, NULL, 0);
2101 strbuf_addstr(&msg, "\nrename to ");
2102 quote_c_style(other, &msg, NULL, 0);
2103 strbuf_addch(&msg, '\n');
2104 break;
2105 case DIFF_STATUS_MODIFIED:
2106 if (p->score) {
2107 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2108 similarity_index(p));
2109 complete_rewrite = 1;
2110 break;
2112 /* fallthru */
2113 default:
2114 /* nothing */
2118 if (hashcmp(one->sha1, two->sha1)) {
2119 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2121 if (DIFF_OPT_TST(o, BINARY)) {
2122 mmfile_t mf;
2123 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2124 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2125 abbrev = 40;
2127 strbuf_addf(&msg, "index %.*s..%.*s",
2128 abbrev, sha1_to_hex(one->sha1),
2129 abbrev, sha1_to_hex(two->sha1));
2130 if (one->mode == two->mode)
2131 strbuf_addf(&msg, " %06o", one->mode);
2132 strbuf_addch(&msg, '\n');
2135 if (msg.len)
2136 strbuf_setlen(&msg, msg.len - 1);
2137 xfrm_msg = msg.len ? msg.buf : NULL;
2139 if (!pgm &&
2140 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2141 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2142 /* a filepair that changes between file and symlink
2143 * needs to be split into deletion and creation.
2145 struct diff_filespec *null = alloc_filespec(two->path);
2146 run_diff_cmd(NULL, name, other, attr_path,
2147 one, null, xfrm_msg, o, 0);
2148 free(null);
2149 null = alloc_filespec(one->path);
2150 run_diff_cmd(NULL, name, other, attr_path,
2151 null, two, xfrm_msg, o, 0);
2152 free(null);
2154 else
2155 run_diff_cmd(pgm, name, other, attr_path,
2156 one, two, xfrm_msg, o, complete_rewrite);
2158 strbuf_release(&msg);
2161 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2162 struct diffstat_t *diffstat)
2164 const char *name;
2165 const char *other;
2166 int complete_rewrite = 0;
2168 if (DIFF_PAIR_UNMERGED(p)) {
2169 /* unmerged */
2170 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2171 return;
2174 name = p->one->path;
2175 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2177 if (o->prefix_length)
2178 strip_prefix(o->prefix_length, &name, &other);
2180 diff_fill_sha1_info(p->one);
2181 diff_fill_sha1_info(p->two);
2183 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2184 complete_rewrite = 1;
2185 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2188 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2190 const char *name;
2191 const char *other;
2192 const char *attr_path;
2194 if (DIFF_PAIR_UNMERGED(p)) {
2195 /* unmerged */
2196 return;
2199 name = p->one->path;
2200 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2201 attr_path = other ? other : name;
2203 if (o->prefix_length)
2204 strip_prefix(o->prefix_length, &name, &other);
2206 diff_fill_sha1_info(p->one);
2207 diff_fill_sha1_info(p->two);
2209 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2212 void diff_setup(struct diff_options *options)
2214 memset(options, 0, sizeof(*options));
2216 options->file = stdout;
2218 options->line_termination = '\n';
2219 options->break_opt = -1;
2220 options->rename_limit = -1;
2221 options->dirstat_percent = 3;
2222 options->context = 3;
2224 options->change = diff_change;
2225 options->add_remove = diff_addremove;
2226 if (diff_use_color_default > 0)
2227 DIFF_OPT_SET(options, COLOR_DIFF);
2228 else
2229 DIFF_OPT_CLR(options, COLOR_DIFF);
2230 options->detect_rename = diff_detect_rename_default;
2232 options->a_prefix = "a/";
2233 options->b_prefix = "b/";
2236 int diff_setup_done(struct diff_options *options)
2238 int count = 0;
2240 if (options->output_format & DIFF_FORMAT_NAME)
2241 count++;
2242 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2243 count++;
2244 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2245 count++;
2246 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2247 count++;
2248 if (count > 1)
2249 die("--name-only, --name-status, --check and -s are mutually exclusive");
2251 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2252 options->detect_rename = DIFF_DETECT_COPY;
2254 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2255 options->prefix = NULL;
2256 if (options->prefix)
2257 options->prefix_length = strlen(options->prefix);
2258 else
2259 options->prefix_length = 0;
2261 if (options->output_format & (DIFF_FORMAT_NAME |
2262 DIFF_FORMAT_NAME_STATUS |
2263 DIFF_FORMAT_CHECKDIFF |
2264 DIFF_FORMAT_NO_OUTPUT))
2265 options->output_format &= ~(DIFF_FORMAT_RAW |
2266 DIFF_FORMAT_NUMSTAT |
2267 DIFF_FORMAT_DIFFSTAT |
2268 DIFF_FORMAT_SHORTSTAT |
2269 DIFF_FORMAT_DIRSTAT |
2270 DIFF_FORMAT_SUMMARY |
2271 DIFF_FORMAT_PATCH);
2274 * These cases always need recursive; we do not drop caller-supplied
2275 * recursive bits for other formats here.
2277 if (options->output_format & (DIFF_FORMAT_PATCH |
2278 DIFF_FORMAT_NUMSTAT |
2279 DIFF_FORMAT_DIFFSTAT |
2280 DIFF_FORMAT_SHORTSTAT |
2281 DIFF_FORMAT_DIRSTAT |
2282 DIFF_FORMAT_SUMMARY |
2283 DIFF_FORMAT_CHECKDIFF))
2284 DIFF_OPT_SET(options, RECURSIVE);
2286 * Also pickaxe would not work very well if you do not say recursive
2288 if (options->pickaxe)
2289 DIFF_OPT_SET(options, RECURSIVE);
2291 if (options->detect_rename && options->rename_limit < 0)
2292 options->rename_limit = diff_rename_limit_default;
2293 if (options->setup & DIFF_SETUP_USE_CACHE) {
2294 if (!active_cache)
2295 /* read-cache does not die even when it fails
2296 * so it is safe for us to do this here. Also
2297 * it does not smudge active_cache or active_nr
2298 * when it fails, so we do not have to worry about
2299 * cleaning it up ourselves either.
2301 read_cache();
2303 if (options->abbrev <= 0 || 40 < options->abbrev)
2304 options->abbrev = 40; /* full */
2307 * It does not make sense to show the first hit we happened
2308 * to have found. It does not make sense not to return with
2309 * exit code in such a case either.
2311 if (DIFF_OPT_TST(options, QUIET)) {
2312 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2313 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2317 * If we postprocess in diffcore, we cannot simply return
2318 * upon the first hit. We need to run diff as usual.
2320 if (options->pickaxe || options->filter)
2321 DIFF_OPT_CLR(options, QUIET);
2323 return 0;
2326 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2328 char c, *eq;
2329 int len;
2331 if (*arg != '-')
2332 return 0;
2333 c = *++arg;
2334 if (!c)
2335 return 0;
2336 if (c == arg_short) {
2337 c = *++arg;
2338 if (!c)
2339 return 1;
2340 if (val && isdigit(c)) {
2341 char *end;
2342 int n = strtoul(arg, &end, 10);
2343 if (*end)
2344 return 0;
2345 *val = n;
2346 return 1;
2348 return 0;
2350 if (c != '-')
2351 return 0;
2352 arg++;
2353 eq = strchr(arg, '=');
2354 if (eq)
2355 len = eq - arg;
2356 else
2357 len = strlen(arg);
2358 if (!len || strncmp(arg, arg_long, len))
2359 return 0;
2360 if (eq) {
2361 int n;
2362 char *end;
2363 if (!isdigit(*++eq))
2364 return 0;
2365 n = strtoul(eq, &end, 10);
2366 if (*end)
2367 return 0;
2368 *val = n;
2370 return 1;
2373 static int diff_scoreopt_parse(const char *opt);
2375 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2377 const char *arg = av[0];
2379 /* Output format options */
2380 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2381 options->output_format |= DIFF_FORMAT_PATCH;
2382 else if (opt_arg(arg, 'U', "unified", &options->context))
2383 options->output_format |= DIFF_FORMAT_PATCH;
2384 else if (!strcmp(arg, "--raw"))
2385 options->output_format |= DIFF_FORMAT_RAW;
2386 else if (!strcmp(arg, "--patch-with-raw"))
2387 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2388 else if (!strcmp(arg, "--numstat"))
2389 options->output_format |= DIFF_FORMAT_NUMSTAT;
2390 else if (!strcmp(arg, "--shortstat"))
2391 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2392 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2393 options->output_format |= DIFF_FORMAT_DIRSTAT;
2394 else if (!strcmp(arg, "--cumulative"))
2395 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2396 else if (!strcmp(arg, "--check"))
2397 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2398 else if (!strcmp(arg, "--summary"))
2399 options->output_format |= DIFF_FORMAT_SUMMARY;
2400 else if (!strcmp(arg, "--patch-with-stat"))
2401 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2402 else if (!strcmp(arg, "--name-only"))
2403 options->output_format |= DIFF_FORMAT_NAME;
2404 else if (!strcmp(arg, "--name-status"))
2405 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2406 else if (!strcmp(arg, "-s"))
2407 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2408 else if (!prefixcmp(arg, "--stat")) {
2409 char *end;
2410 int width = options->stat_width;
2411 int name_width = options->stat_name_width;
2412 arg += 6;
2413 end = (char *)arg;
2415 switch (*arg) {
2416 case '-':
2417 if (!prefixcmp(arg, "-width="))
2418 width = strtoul(arg + 7, &end, 10);
2419 else if (!prefixcmp(arg, "-name-width="))
2420 name_width = strtoul(arg + 12, &end, 10);
2421 break;
2422 case '=':
2423 width = strtoul(arg+1, &end, 10);
2424 if (*end == ',')
2425 name_width = strtoul(end+1, &end, 10);
2428 /* Important! This checks all the error cases! */
2429 if (*end)
2430 return 0;
2431 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2432 options->stat_name_width = name_width;
2433 options->stat_width = width;
2436 /* renames options */
2437 else if (!prefixcmp(arg, "-B")) {
2438 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2439 return -1;
2441 else if (!prefixcmp(arg, "-M")) {
2442 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2443 return -1;
2444 options->detect_rename = DIFF_DETECT_RENAME;
2446 else if (!prefixcmp(arg, "-C")) {
2447 if (options->detect_rename == DIFF_DETECT_COPY)
2448 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2449 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2450 return -1;
2451 options->detect_rename = DIFF_DETECT_COPY;
2453 else if (!strcmp(arg, "--no-renames"))
2454 options->detect_rename = 0;
2455 else if (!strcmp(arg, "--relative"))
2456 DIFF_OPT_SET(options, RELATIVE_NAME);
2457 else if (!prefixcmp(arg, "--relative=")) {
2458 DIFF_OPT_SET(options, RELATIVE_NAME);
2459 options->prefix = arg + 11;
2462 /* xdiff options */
2463 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2464 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2465 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2466 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2467 else if (!strcmp(arg, "--ignore-space-at-eol"))
2468 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2470 /* flags options */
2471 else if (!strcmp(arg, "--binary")) {
2472 options->output_format |= DIFF_FORMAT_PATCH;
2473 DIFF_OPT_SET(options, BINARY);
2475 else if (!strcmp(arg, "--full-index"))
2476 DIFF_OPT_SET(options, FULL_INDEX);
2477 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2478 DIFF_OPT_SET(options, TEXT);
2479 else if (!strcmp(arg, "-R"))
2480 DIFF_OPT_SET(options, REVERSE_DIFF);
2481 else if (!strcmp(arg, "--find-copies-harder"))
2482 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2483 else if (!strcmp(arg, "--follow"))
2484 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2485 else if (!strcmp(arg, "--color"))
2486 DIFF_OPT_SET(options, COLOR_DIFF);
2487 else if (!strcmp(arg, "--no-color"))
2488 DIFF_OPT_CLR(options, COLOR_DIFF);
2489 else if (!strcmp(arg, "--color-words"))
2490 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2491 else if (!strcmp(arg, "--exit-code"))
2492 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2493 else if (!strcmp(arg, "--quiet"))
2494 DIFF_OPT_SET(options, QUIET);
2495 else if (!strcmp(arg, "--ext-diff"))
2496 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2497 else if (!strcmp(arg, "--no-ext-diff"))
2498 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2500 /* misc options */
2501 else if (!strcmp(arg, "-z"))
2502 options->line_termination = 0;
2503 else if (!prefixcmp(arg, "-l"))
2504 options->rename_limit = strtoul(arg+2, NULL, 10);
2505 else if (!prefixcmp(arg, "-S"))
2506 options->pickaxe = arg + 2;
2507 else if (!strcmp(arg, "--pickaxe-all"))
2508 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2509 else if (!strcmp(arg, "--pickaxe-regex"))
2510 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2511 else if (!prefixcmp(arg, "-O"))
2512 options->orderfile = arg + 2;
2513 else if (!prefixcmp(arg, "--diff-filter="))
2514 options->filter = arg + 14;
2515 else if (!strcmp(arg, "--abbrev"))
2516 options->abbrev = DEFAULT_ABBREV;
2517 else if (!prefixcmp(arg, "--abbrev=")) {
2518 options->abbrev = strtoul(arg + 9, NULL, 10);
2519 if (options->abbrev < MINIMUM_ABBREV)
2520 options->abbrev = MINIMUM_ABBREV;
2521 else if (40 < options->abbrev)
2522 options->abbrev = 40;
2524 else if (!prefixcmp(arg, "--src-prefix="))
2525 options->a_prefix = arg + 13;
2526 else if (!prefixcmp(arg, "--dst-prefix="))
2527 options->b_prefix = arg + 13;
2528 else if (!strcmp(arg, "--no-prefix"))
2529 options->a_prefix = options->b_prefix = "";
2530 else if (!prefixcmp(arg, "--output=")) {
2531 options->file = fopen(arg + strlen("--output="), "w");
2532 options->close_file = 1;
2533 } else
2534 return 0;
2535 return 1;
2538 static int parse_num(const char **cp_p)
2540 unsigned long num, scale;
2541 int ch, dot;
2542 const char *cp = *cp_p;
2544 num = 0;
2545 scale = 1;
2546 dot = 0;
2547 for(;;) {
2548 ch = *cp;
2549 if ( !dot && ch == '.' ) {
2550 scale = 1;
2551 dot = 1;
2552 } else if ( ch == '%' ) {
2553 scale = dot ? scale*100 : 100;
2554 cp++; /* % is always at the end */
2555 break;
2556 } else if ( ch >= '0' && ch <= '9' ) {
2557 if ( scale < 100000 ) {
2558 scale *= 10;
2559 num = (num*10) + (ch-'0');
2561 } else {
2562 break;
2564 cp++;
2566 *cp_p = cp;
2568 /* user says num divided by scale and we say internally that
2569 * is MAX_SCORE * num / scale.
2571 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2574 static int diff_scoreopt_parse(const char *opt)
2576 int opt1, opt2, cmd;
2578 if (*opt++ != '-')
2579 return -1;
2580 cmd = *opt++;
2581 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2582 return -1; /* that is not a -M, -C nor -B option */
2584 opt1 = parse_num(&opt);
2585 if (cmd != 'B')
2586 opt2 = 0;
2587 else {
2588 if (*opt == 0)
2589 opt2 = 0;
2590 else if (*opt != '/')
2591 return -1; /* we expect -B80/99 or -B80 */
2592 else {
2593 opt++;
2594 opt2 = parse_num(&opt);
2597 if (*opt != 0)
2598 return -1;
2599 return opt1 | (opt2 << 16);
2602 struct diff_queue_struct diff_queued_diff;
2604 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2606 if (queue->alloc <= queue->nr) {
2607 queue->alloc = alloc_nr(queue->alloc);
2608 queue->queue = xrealloc(queue->queue,
2609 sizeof(dp) * queue->alloc);
2611 queue->queue[queue->nr++] = dp;
2614 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2615 struct diff_filespec *one,
2616 struct diff_filespec *two)
2618 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2619 dp->one = one;
2620 dp->two = two;
2621 if (queue)
2622 diff_q(queue, dp);
2623 return dp;
2626 void diff_free_filepair(struct diff_filepair *p)
2628 free_filespec(p->one);
2629 free_filespec(p->two);
2630 free(p);
2633 /* This is different from find_unique_abbrev() in that
2634 * it stuffs the result with dots for alignment.
2636 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2638 int abblen;
2639 const char *abbrev;
2640 if (len == 40)
2641 return sha1_to_hex(sha1);
2643 abbrev = find_unique_abbrev(sha1, len);
2644 abblen = strlen(abbrev);
2645 if (abblen < 37) {
2646 static char hex[41];
2647 if (len < abblen && abblen <= len + 2)
2648 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2649 else
2650 sprintf(hex, "%s...", abbrev);
2651 return hex;
2653 return sha1_to_hex(sha1);
2656 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2658 int line_termination = opt->line_termination;
2659 int inter_name_termination = line_termination ? '\t' : '\0';
2661 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2662 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2663 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2664 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2666 if (p->score) {
2667 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2668 inter_name_termination);
2669 } else {
2670 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2673 if (p->status == DIFF_STATUS_COPIED ||
2674 p->status == DIFF_STATUS_RENAMED) {
2675 const char *name_a, *name_b;
2676 name_a = p->one->path;
2677 name_b = p->two->path;
2678 strip_prefix(opt->prefix_length, &name_a, &name_b);
2679 write_name_quoted(name_a, opt->file, inter_name_termination);
2680 write_name_quoted(name_b, opt->file, line_termination);
2681 } else {
2682 const char *name_a, *name_b;
2683 name_a = p->one->mode ? p->one->path : p->two->path;
2684 name_b = NULL;
2685 strip_prefix(opt->prefix_length, &name_a, &name_b);
2686 write_name_quoted(name_a, opt->file, line_termination);
2690 int diff_unmodified_pair(struct diff_filepair *p)
2692 /* This function is written stricter than necessary to support
2693 * the currently implemented transformers, but the idea is to
2694 * let transformers to produce diff_filepairs any way they want,
2695 * and filter and clean them up here before producing the output.
2697 struct diff_filespec *one = p->one, *two = p->two;
2699 if (DIFF_PAIR_UNMERGED(p))
2700 return 0; /* unmerged is interesting */
2702 /* deletion, addition, mode or type change
2703 * and rename are all interesting.
2705 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2706 DIFF_PAIR_MODE_CHANGED(p) ||
2707 strcmp(one->path, two->path))
2708 return 0;
2710 /* both are valid and point at the same path. that is, we are
2711 * dealing with a change.
2713 if (one->sha1_valid && two->sha1_valid &&
2714 !hashcmp(one->sha1, two->sha1))
2715 return 1; /* no change */
2716 if (!one->sha1_valid && !two->sha1_valid)
2717 return 1; /* both look at the same file on the filesystem. */
2718 return 0;
2721 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2723 if (diff_unmodified_pair(p))
2724 return;
2726 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2727 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2728 return; /* no tree diffs in patch format */
2730 run_diff(p, o);
2733 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2734 struct diffstat_t *diffstat)
2736 if (diff_unmodified_pair(p))
2737 return;
2739 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2740 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2741 return; /* no tree diffs in patch format */
2743 run_diffstat(p, o, diffstat);
2746 static void diff_flush_checkdiff(struct diff_filepair *p,
2747 struct diff_options *o)
2749 if (diff_unmodified_pair(p))
2750 return;
2752 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2753 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2754 return; /* no tree diffs in patch format */
2756 run_checkdiff(p, o);
2759 int diff_queue_is_empty(void)
2761 struct diff_queue_struct *q = &diff_queued_diff;
2762 int i;
2763 for (i = 0; i < q->nr; i++)
2764 if (!diff_unmodified_pair(q->queue[i]))
2765 return 0;
2766 return 1;
2769 #if DIFF_DEBUG
2770 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2772 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2773 x, one ? one : "",
2774 s->path,
2775 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2776 s->mode,
2777 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2778 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2779 x, one ? one : "",
2780 s->size, s->xfrm_flags);
2783 void diff_debug_filepair(const struct diff_filepair *p, int i)
2785 diff_debug_filespec(p->one, i, "one");
2786 diff_debug_filespec(p->two, i, "two");
2787 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2788 p->score, p->status ? p->status : '?',
2789 p->one->rename_used, p->broken_pair);
2792 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2794 int i;
2795 if (msg)
2796 fprintf(stderr, "%s\n", msg);
2797 fprintf(stderr, "q->nr = %d\n", q->nr);
2798 for (i = 0; i < q->nr; i++) {
2799 struct diff_filepair *p = q->queue[i];
2800 diff_debug_filepair(p, i);
2803 #endif
2805 static void diff_resolve_rename_copy(void)
2807 int i;
2808 struct diff_filepair *p;
2809 struct diff_queue_struct *q = &diff_queued_diff;
2811 diff_debug_queue("resolve-rename-copy", q);
2813 for (i = 0; i < q->nr; i++) {
2814 p = q->queue[i];
2815 p->status = 0; /* undecided */
2816 if (DIFF_PAIR_UNMERGED(p))
2817 p->status = DIFF_STATUS_UNMERGED;
2818 else if (!DIFF_FILE_VALID(p->one))
2819 p->status = DIFF_STATUS_ADDED;
2820 else if (!DIFF_FILE_VALID(p->two))
2821 p->status = DIFF_STATUS_DELETED;
2822 else if (DIFF_PAIR_TYPE_CHANGED(p))
2823 p->status = DIFF_STATUS_TYPE_CHANGED;
2825 /* from this point on, we are dealing with a pair
2826 * whose both sides are valid and of the same type, i.e.
2827 * either in-place edit or rename/copy edit.
2829 else if (DIFF_PAIR_RENAME(p)) {
2831 * A rename might have re-connected a broken
2832 * pair up, causing the pathnames to be the
2833 * same again. If so, that's not a rename at
2834 * all, just a modification..
2836 * Otherwise, see if this source was used for
2837 * multiple renames, in which case we decrement
2838 * the count, and call it a copy.
2840 if (!strcmp(p->one->path, p->two->path))
2841 p->status = DIFF_STATUS_MODIFIED;
2842 else if (--p->one->rename_used > 0)
2843 p->status = DIFF_STATUS_COPIED;
2844 else
2845 p->status = DIFF_STATUS_RENAMED;
2847 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2848 p->one->mode != p->two->mode ||
2849 is_null_sha1(p->one->sha1))
2850 p->status = DIFF_STATUS_MODIFIED;
2851 else {
2852 /* This is a "no-change" entry and should not
2853 * happen anymore, but prepare for broken callers.
2855 error("feeding unmodified %s to diffcore",
2856 p->one->path);
2857 p->status = DIFF_STATUS_UNKNOWN;
2860 diff_debug_queue("resolve-rename-copy done", q);
2863 static int check_pair_status(struct diff_filepair *p)
2865 switch (p->status) {
2866 case DIFF_STATUS_UNKNOWN:
2867 return 0;
2868 case 0:
2869 die("internal error in diff-resolve-rename-copy");
2870 default:
2871 return 1;
2875 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2877 int fmt = opt->output_format;
2879 if (fmt & DIFF_FORMAT_CHECKDIFF)
2880 diff_flush_checkdiff(p, opt);
2881 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2882 diff_flush_raw(p, opt);
2883 else if (fmt & DIFF_FORMAT_NAME) {
2884 const char *name_a, *name_b;
2885 name_a = p->two->path;
2886 name_b = NULL;
2887 strip_prefix(opt->prefix_length, &name_a, &name_b);
2888 write_name_quoted(name_a, opt->file, opt->line_termination);
2892 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2894 if (fs->mode)
2895 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2896 else
2897 fprintf(file, " %s ", newdelete);
2898 write_name_quoted(fs->path, file, '\n');
2902 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2904 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2905 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2906 show_name ? ' ' : '\n');
2907 if (show_name) {
2908 write_name_quoted(p->two->path, file, '\n');
2913 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2915 char *names = pprint_rename(p->one->path, p->two->path);
2917 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2918 free(names);
2919 show_mode_change(file, p, 0);
2922 static void diff_summary(FILE *file, struct diff_filepair *p)
2924 switch(p->status) {
2925 case DIFF_STATUS_DELETED:
2926 show_file_mode_name(file, "delete", p->one);
2927 break;
2928 case DIFF_STATUS_ADDED:
2929 show_file_mode_name(file, "create", p->two);
2930 break;
2931 case DIFF_STATUS_COPIED:
2932 show_rename_copy(file, "copy", p);
2933 break;
2934 case DIFF_STATUS_RENAMED:
2935 show_rename_copy(file, "rename", p);
2936 break;
2937 default:
2938 if (p->score) {
2939 fputs(" rewrite ", file);
2940 write_name_quoted(p->two->path, file, ' ');
2941 fprintf(file, "(%d%%)\n", similarity_index(p));
2943 show_mode_change(file, p, !p->score);
2944 break;
2948 struct patch_id_t {
2949 struct xdiff_emit_state xm;
2950 SHA_CTX *ctx;
2951 int patchlen;
2954 static int remove_space(char *line, int len)
2956 int i;
2957 char *dst = line;
2958 unsigned char c;
2960 for (i = 0; i < len; i++)
2961 if (!isspace((c = line[i])))
2962 *dst++ = c;
2964 return dst - line;
2967 static void patch_id_consume(void *priv, char *line, unsigned long len)
2969 struct patch_id_t *data = priv;
2970 int new_len;
2972 /* Ignore line numbers when computing the SHA1 of the patch */
2973 if (!prefixcmp(line, "@@ -"))
2974 return;
2976 new_len = remove_space(line, len);
2978 SHA1_Update(data->ctx, line, new_len);
2979 data->patchlen += new_len;
2982 /* returns 0 upon success, and writes result into sha1 */
2983 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2985 struct diff_queue_struct *q = &diff_queued_diff;
2986 int i;
2987 SHA_CTX ctx;
2988 struct patch_id_t data;
2989 char buffer[PATH_MAX * 4 + 20];
2991 SHA1_Init(&ctx);
2992 memset(&data, 0, sizeof(struct patch_id_t));
2993 data.ctx = &ctx;
2994 data.xm.consume = patch_id_consume;
2996 for (i = 0; i < q->nr; i++) {
2997 xpparam_t xpp;
2998 xdemitconf_t xecfg;
2999 xdemitcb_t ecb;
3000 mmfile_t mf1, mf2;
3001 struct diff_filepair *p = q->queue[i];
3002 int len1, len2;
3004 memset(&xecfg, 0, sizeof(xecfg));
3005 if (p->status == 0)
3006 return error("internal diff status error");
3007 if (p->status == DIFF_STATUS_UNKNOWN)
3008 continue;
3009 if (diff_unmodified_pair(p))
3010 continue;
3011 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3012 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3013 continue;
3014 if (DIFF_PAIR_UNMERGED(p))
3015 continue;
3017 diff_fill_sha1_info(p->one);
3018 diff_fill_sha1_info(p->two);
3019 if (fill_mmfile(&mf1, p->one) < 0 ||
3020 fill_mmfile(&mf2, p->two) < 0)
3021 return error("unable to read files to diff");
3023 len1 = remove_space(p->one->path, strlen(p->one->path));
3024 len2 = remove_space(p->two->path, strlen(p->two->path));
3025 if (p->one->mode == 0)
3026 len1 = snprintf(buffer, sizeof(buffer),
3027 "diff--gita/%.*sb/%.*s"
3028 "newfilemode%06o"
3029 "---/dev/null"
3030 "+++b/%.*s",
3031 len1, p->one->path,
3032 len2, p->two->path,
3033 p->two->mode,
3034 len2, p->two->path);
3035 else if (p->two->mode == 0)
3036 len1 = snprintf(buffer, sizeof(buffer),
3037 "diff--gita/%.*sb/%.*s"
3038 "deletedfilemode%06o"
3039 "---a/%.*s"
3040 "+++/dev/null",
3041 len1, p->one->path,
3042 len2, p->two->path,
3043 p->one->mode,
3044 len1, p->one->path);
3045 else
3046 len1 = snprintf(buffer, sizeof(buffer),
3047 "diff--gita/%.*sb/%.*s"
3048 "---a/%.*s"
3049 "+++b/%.*s",
3050 len1, p->one->path,
3051 len2, p->two->path,
3052 len1, p->one->path,
3053 len2, p->two->path);
3054 SHA1_Update(&ctx, buffer, len1);
3056 xpp.flags = XDF_NEED_MINIMAL;
3057 xecfg.ctxlen = 3;
3058 xecfg.flags = XDL_EMIT_FUNCNAMES;
3059 ecb.outf = xdiff_outf;
3060 ecb.priv = &data;
3061 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3064 SHA1_Final(sha1, &ctx);
3065 return 0;
3068 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3070 struct diff_queue_struct *q = &diff_queued_diff;
3071 int i;
3072 int result = diff_get_patch_id(options, sha1);
3074 for (i = 0; i < q->nr; i++)
3075 diff_free_filepair(q->queue[i]);
3077 free(q->queue);
3078 q->queue = NULL;
3079 q->nr = q->alloc = 0;
3081 return result;
3084 static int is_summary_empty(const struct diff_queue_struct *q)
3086 int i;
3088 for (i = 0; i < q->nr; i++) {
3089 const struct diff_filepair *p = q->queue[i];
3091 switch (p->status) {
3092 case DIFF_STATUS_DELETED:
3093 case DIFF_STATUS_ADDED:
3094 case DIFF_STATUS_COPIED:
3095 case DIFF_STATUS_RENAMED:
3096 return 0;
3097 default:
3098 if (p->score)
3099 return 0;
3100 if (p->one->mode && p->two->mode &&
3101 p->one->mode != p->two->mode)
3102 return 0;
3103 break;
3106 return 1;
3109 void diff_flush(struct diff_options *options)
3111 struct diff_queue_struct *q = &diff_queued_diff;
3112 int i, output_format = options->output_format;
3113 int separator = 0;
3116 * Order: raw, stat, summary, patch
3117 * or: name/name-status/checkdiff (other bits clear)
3119 if (!q->nr)
3120 goto free_queue;
3122 if (output_format & (DIFF_FORMAT_RAW |
3123 DIFF_FORMAT_NAME |
3124 DIFF_FORMAT_NAME_STATUS |
3125 DIFF_FORMAT_CHECKDIFF)) {
3126 for (i = 0; i < q->nr; i++) {
3127 struct diff_filepair *p = q->queue[i];
3128 if (check_pair_status(p))
3129 flush_one_pair(p, options);
3131 separator++;
3134 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3135 struct diffstat_t diffstat;
3137 memset(&diffstat, 0, sizeof(struct diffstat_t));
3138 diffstat.xm.consume = diffstat_consume;
3139 for (i = 0; i < q->nr; i++) {
3140 struct diff_filepair *p = q->queue[i];
3141 if (check_pair_status(p))
3142 diff_flush_stat(p, options, &diffstat);
3144 if (output_format & DIFF_FORMAT_NUMSTAT)
3145 show_numstat(&diffstat, options);
3146 if (output_format & DIFF_FORMAT_DIFFSTAT)
3147 show_stats(&diffstat, options);
3148 if (output_format & DIFF_FORMAT_SHORTSTAT)
3149 show_shortstats(&diffstat, options);
3150 free_diffstat_info(&diffstat);
3151 separator++;
3153 if (output_format & DIFF_FORMAT_DIRSTAT)
3154 show_dirstat(options);
3156 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3157 for (i = 0; i < q->nr; i++)
3158 diff_summary(options->file, q->queue[i]);
3159 separator++;
3162 if (output_format & DIFF_FORMAT_PATCH) {
3163 if (separator) {
3164 if (options->stat_sep) {
3165 /* attach patch instead of inline */
3166 fputs(options->stat_sep, options->file);
3167 } else {
3168 putc(options->line_termination, options->file);
3172 for (i = 0; i < q->nr; i++) {
3173 struct diff_filepair *p = q->queue[i];
3174 if (check_pair_status(p))
3175 diff_flush_patch(p, options);
3179 if (output_format & DIFF_FORMAT_CALLBACK)
3180 options->format_callback(q, options, options->format_callback_data);
3182 for (i = 0; i < q->nr; i++)
3183 diff_free_filepair(q->queue[i]);
3184 free_queue:
3185 free(q->queue);
3186 q->queue = NULL;
3187 q->nr = q->alloc = 0;
3188 if (options->close_file)
3189 fclose(options->file);
3192 static void diffcore_apply_filter(const char *filter)
3194 int i;
3195 struct diff_queue_struct *q = &diff_queued_diff;
3196 struct diff_queue_struct outq;
3197 outq.queue = NULL;
3198 outq.nr = outq.alloc = 0;
3200 if (!filter)
3201 return;
3203 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3204 int found;
3205 for (i = found = 0; !found && i < q->nr; i++) {
3206 struct diff_filepair *p = q->queue[i];
3207 if (((p->status == DIFF_STATUS_MODIFIED) &&
3208 ((p->score &&
3209 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3210 (!p->score &&
3211 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3212 ((p->status != DIFF_STATUS_MODIFIED) &&
3213 strchr(filter, p->status)))
3214 found++;
3216 if (found)
3217 return;
3219 /* otherwise we will clear the whole queue
3220 * by copying the empty outq at the end of this
3221 * function, but first clear the current entries
3222 * in the queue.
3224 for (i = 0; i < q->nr; i++)
3225 diff_free_filepair(q->queue[i]);
3227 else {
3228 /* Only the matching ones */
3229 for (i = 0; i < q->nr; i++) {
3230 struct diff_filepair *p = q->queue[i];
3232 if (((p->status == DIFF_STATUS_MODIFIED) &&
3233 ((p->score &&
3234 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3235 (!p->score &&
3236 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3237 ((p->status != DIFF_STATUS_MODIFIED) &&
3238 strchr(filter, p->status)))
3239 diff_q(&outq, p);
3240 else
3241 diff_free_filepair(p);
3244 free(q->queue);
3245 *q = outq;
3248 /* Check whether two filespecs with the same mode and size are identical */
3249 static int diff_filespec_is_identical(struct diff_filespec *one,
3250 struct diff_filespec *two)
3252 if (S_ISGITLINK(one->mode))
3253 return 0;
3254 if (diff_populate_filespec(one, 0))
3255 return 0;
3256 if (diff_populate_filespec(two, 0))
3257 return 0;
3258 return !memcmp(one->data, two->data, one->size);
3261 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3263 int i;
3264 struct diff_queue_struct *q = &diff_queued_diff;
3265 struct diff_queue_struct outq;
3266 outq.queue = NULL;
3267 outq.nr = outq.alloc = 0;
3269 for (i = 0; i < q->nr; i++) {
3270 struct diff_filepair *p = q->queue[i];
3273 * 1. Entries that come from stat info dirtyness
3274 * always have both sides (iow, not create/delete),
3275 * one side of the object name is unknown, with
3276 * the same mode and size. Keep the ones that
3277 * do not match these criteria. They have real
3278 * differences.
3280 * 2. At this point, the file is known to be modified,
3281 * with the same mode and size, and the object
3282 * name of one side is unknown. Need to inspect
3283 * the identical contents.
3285 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3286 !DIFF_FILE_VALID(p->two) ||
3287 (p->one->sha1_valid && p->two->sha1_valid) ||
3288 (p->one->mode != p->two->mode) ||
3289 diff_populate_filespec(p->one, 1) ||
3290 diff_populate_filespec(p->two, 1) ||
3291 (p->one->size != p->two->size) ||
3292 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3293 diff_q(&outq, p);
3294 else {
3296 * The caller can subtract 1 from skip_stat_unmatch
3297 * to determine how many paths were dirty only
3298 * due to stat info mismatch.
3300 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3301 diffopt->skip_stat_unmatch++;
3302 diff_free_filepair(p);
3305 free(q->queue);
3306 *q = outq;
3309 void diffcore_std(struct diff_options *options)
3311 if (DIFF_OPT_TST(options, QUIET))
3312 return;
3314 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3315 diffcore_skip_stat_unmatch(options);
3316 if (options->break_opt != -1)
3317 diffcore_break(options->break_opt);
3318 if (options->detect_rename)
3319 diffcore_rename(options);
3320 if (options->break_opt != -1)
3321 diffcore_merge_broken();
3322 if (options->pickaxe)
3323 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3324 if (options->orderfile)
3325 diffcore_order(options->orderfile);
3326 diff_resolve_rename_copy();
3327 diffcore_apply_filter(options->filter);
3329 if (diff_queued_diff.nr)
3330 DIFF_OPT_SET(options, HAS_CHANGES);
3331 else
3332 DIFF_OPT_CLR(options, HAS_CHANGES);
3335 int diff_result_code(struct diff_options *opt, int status)
3337 int result = 0;
3338 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3339 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3340 return status;
3341 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3342 DIFF_OPT_TST(opt, HAS_CHANGES))
3343 result |= 01;
3344 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3345 DIFF_OPT_TST(opt, CHECK_FAILED))
3346 result |= 02;
3347 return result;
3350 void diff_addremove(struct diff_options *options,
3351 int addremove, unsigned mode,
3352 const unsigned char *sha1,
3353 const char *base, const char *path)
3355 char concatpath[PATH_MAX];
3356 struct diff_filespec *one, *two;
3358 /* This may look odd, but it is a preparation for
3359 * feeding "there are unchanged files which should
3360 * not produce diffs, but when you are doing copy
3361 * detection you would need them, so here they are"
3362 * entries to the diff-core. They will be prefixed
3363 * with something like '=' or '*' (I haven't decided
3364 * which but should not make any difference).
3365 * Feeding the same new and old to diff_change()
3366 * also has the same effect.
3367 * Before the final output happens, they are pruned after
3368 * merged into rename/copy pairs as appropriate.
3370 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3371 addremove = (addremove == '+' ? '-' :
3372 addremove == '-' ? '+' : addremove);
3374 if (!path) path = "";
3375 sprintf(concatpath, "%s%s", base, path);
3377 if (options->prefix &&
3378 strncmp(concatpath, options->prefix, options->prefix_length))
3379 return;
3381 one = alloc_filespec(concatpath);
3382 two = alloc_filespec(concatpath);
3384 if (addremove != '+')
3385 fill_filespec(one, sha1, mode);
3386 if (addremove != '-')
3387 fill_filespec(two, sha1, mode);
3389 diff_queue(&diff_queued_diff, one, two);
3390 DIFF_OPT_SET(options, HAS_CHANGES);
3393 void diff_change(struct diff_options *options,
3394 unsigned old_mode, unsigned new_mode,
3395 const unsigned char *old_sha1,
3396 const unsigned char *new_sha1,
3397 const char *base, const char *path)
3399 char concatpath[PATH_MAX];
3400 struct diff_filespec *one, *two;
3402 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3403 unsigned tmp;
3404 const unsigned char *tmp_c;
3405 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3406 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3408 if (!path) path = "";
3409 sprintf(concatpath, "%s%s", base, path);
3411 if (options->prefix &&
3412 strncmp(concatpath, options->prefix, options->prefix_length))
3413 return;
3415 one = alloc_filespec(concatpath);
3416 two = alloc_filespec(concatpath);
3417 fill_filespec(one, old_sha1, old_mode);
3418 fill_filespec(two, new_sha1, new_mode);
3420 diff_queue(&diff_queued_diff, one, two);
3421 DIFF_OPT_SET(options, HAS_CHANGES);
3424 void diff_unmerge(struct diff_options *options,
3425 const char *path,
3426 unsigned mode, const unsigned char *sha1)
3428 struct diff_filespec *one, *two;
3430 if (options->prefix &&
3431 strncmp(path, options->prefix, options->prefix_length))
3432 return;
3434 one = alloc_filespec(path);
3435 two = alloc_filespec(path);
3436 fill_filespec(one, sha1, mode);
3437 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;