color-words: refactor word splitting and use ALLOC_GROW()
[4msysgit-hv.git] / diff.c
blobc111eef13ea7c527c9ac7e0d47aad7bd772bca7b
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
14 #include "userdiff.h"
16 #ifdef NO_FAST_WORKING_DIRECTORY
17 #define FAST_WORKING_DIRECTORY 0
18 #else
19 #define FAST_WORKING_DIRECTORY 1
20 #endif
22 static int diff_detect_rename_default;
23 static int diff_rename_limit_default = 200;
24 static int diff_suppress_blank_empty;
25 int diff_use_color_default = -1;
26 static const char *external_diff_cmd_cfg;
27 int diff_auto_refresh_index = 1;
28 static int diff_mnemonic_prefix;
30 static char diff_colors[][COLOR_MAXLEN] = {
31 "\033[m", /* reset */
32 "", /* PLAIN (normal) */
33 "\033[1m", /* METAINFO (bold) */
34 "\033[36m", /* FRAGINFO (cyan) */
35 "\033[31m", /* OLD (red) */
36 "\033[32m", /* NEW (green) */
37 "\033[33m", /* COMMIT (yellow) */
38 "\033[41m", /* WHITESPACE (red background) */
41 static void diff_filespec_load_driver(struct diff_filespec *one);
42 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
44 static int parse_diff_color_slot(const char *var, int ofs)
46 if (!strcasecmp(var+ofs, "plain"))
47 return DIFF_PLAIN;
48 if (!strcasecmp(var+ofs, "meta"))
49 return DIFF_METAINFO;
50 if (!strcasecmp(var+ofs, "frag"))
51 return DIFF_FRAGINFO;
52 if (!strcasecmp(var+ofs, "old"))
53 return DIFF_FILE_OLD;
54 if (!strcasecmp(var+ofs, "new"))
55 return DIFF_FILE_NEW;
56 if (!strcasecmp(var+ofs, "commit"))
57 return DIFF_COMMIT;
58 if (!strcasecmp(var+ofs, "whitespace"))
59 return DIFF_WHITESPACE;
60 die("bad config variable '%s'", var);
64 * These are to give UI layer defaults.
65 * The core-level commands such as git-diff-files should
66 * never be affected by the setting of diff.renames
67 * the user happens to have in the configuration file.
69 int git_diff_ui_config(const char *var, const char *value, void *cb)
71 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
72 diff_use_color_default = git_config_colorbool(var, value, -1);
73 return 0;
75 if (!strcmp(var, "diff.renames")) {
76 if (!value)
77 diff_detect_rename_default = DIFF_DETECT_RENAME;
78 else if (!strcasecmp(value, "copies") ||
79 !strcasecmp(value, "copy"))
80 diff_detect_rename_default = DIFF_DETECT_COPY;
81 else if (git_config_bool(var,value))
82 diff_detect_rename_default = DIFF_DETECT_RENAME;
83 return 0;
85 if (!strcmp(var, "diff.autorefreshindex")) {
86 diff_auto_refresh_index = git_config_bool(var, value);
87 return 0;
89 if (!strcmp(var, "diff.mnemonicprefix")) {
90 diff_mnemonic_prefix = git_config_bool(var, value);
91 return 0;
93 if (!strcmp(var, "diff.external"))
94 return git_config_string(&external_diff_cmd_cfg, var, value);
96 return git_diff_basic_config(var, value, cb);
99 int git_diff_basic_config(const char *var, const char *value, void *cb)
101 if (!strcmp(var, "diff.renamelimit")) {
102 diff_rename_limit_default = git_config_int(var, value);
103 return 0;
106 switch (userdiff_config(var, value)) {
107 case 0: break;
108 case -1: return -1;
109 default: return 0;
112 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
113 int slot = parse_diff_color_slot(var, 11);
114 if (!value)
115 return config_error_nonbool(var);
116 color_parse(value, var, diff_colors[slot]);
117 return 0;
120 /* like GNU diff's --suppress-blank-empty option */
121 if (!strcmp(var, "diff.suppress-blank-empty")) {
122 diff_suppress_blank_empty = git_config_bool(var, value);
123 return 0;
126 return git_color_default_config(var, value, cb);
129 static char *quote_two(const char *one, const char *two)
131 int need_one = quote_c_style(one, NULL, NULL, 1);
132 int need_two = quote_c_style(two, NULL, NULL, 1);
133 struct strbuf res = STRBUF_INIT;
135 if (need_one + need_two) {
136 strbuf_addch(&res, '"');
137 quote_c_style(one, &res, NULL, 1);
138 quote_c_style(two, &res, NULL, 1);
139 strbuf_addch(&res, '"');
140 } else {
141 strbuf_addstr(&res, one);
142 strbuf_addstr(&res, two);
144 return strbuf_detach(&res, NULL);
147 static const char *external_diff(void)
149 static const char *external_diff_cmd = NULL;
150 static int done_preparing = 0;
152 if (done_preparing)
153 return external_diff_cmd;
154 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
155 if (!external_diff_cmd)
156 external_diff_cmd = external_diff_cmd_cfg;
157 done_preparing = 1;
158 return external_diff_cmd;
161 static struct diff_tempfile {
162 const char *name; /* filename external diff should read from */
163 char hex[41];
164 char mode[10];
165 char tmp_path[PATH_MAX];
166 } diff_temp[2];
168 static int count_lines(const char *data, int size)
170 int count, ch, completely_empty = 1, nl_just_seen = 0;
171 count = 0;
172 while (0 < size--) {
173 ch = *data++;
174 if (ch == '\n') {
175 count++;
176 nl_just_seen = 1;
177 completely_empty = 0;
179 else {
180 nl_just_seen = 0;
181 completely_empty = 0;
184 if (completely_empty)
185 return 0;
186 if (!nl_just_seen)
187 count++; /* no trailing newline */
188 return count;
191 static void print_line_count(FILE *file, int count)
193 switch (count) {
194 case 0:
195 fprintf(file, "0,0");
196 break;
197 case 1:
198 fprintf(file, "1");
199 break;
200 default:
201 fprintf(file, "1,%d", count);
202 break;
206 static void copy_file_with_prefix(FILE *file,
207 int prefix, const char *data, int size,
208 const char *set, const char *reset)
210 int ch, nl_just_seen = 1;
211 while (0 < size--) {
212 ch = *data++;
213 if (nl_just_seen) {
214 fputs(set, file);
215 putc(prefix, file);
217 if (ch == '\n') {
218 nl_just_seen = 1;
219 fputs(reset, file);
220 } else
221 nl_just_seen = 0;
222 putc(ch, file);
224 if (!nl_just_seen)
225 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
228 static void emit_rewrite_diff(const char *name_a,
229 const char *name_b,
230 struct diff_filespec *one,
231 struct diff_filespec *two,
232 const char *textconv_one,
233 const char *textconv_two,
234 struct diff_options *o)
236 int lc_a, lc_b;
237 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
238 const char *name_a_tab, *name_b_tab;
239 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
240 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
241 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
242 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
243 const char *reset = diff_get_color(color_diff, DIFF_RESET);
244 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
245 const char *a_prefix, *b_prefix;
246 const char *data_one, *data_two;
247 size_t size_one, size_two;
249 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
250 a_prefix = o->b_prefix;
251 b_prefix = o->a_prefix;
252 } else {
253 a_prefix = o->a_prefix;
254 b_prefix = o->b_prefix;
257 name_a += (*name_a == '/');
258 name_b += (*name_b == '/');
259 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
260 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
262 strbuf_reset(&a_name);
263 strbuf_reset(&b_name);
264 quote_two_c_style(&a_name, a_prefix, name_a, 0);
265 quote_two_c_style(&b_name, b_prefix, name_b, 0);
267 diff_populate_filespec(one, 0);
268 diff_populate_filespec(two, 0);
269 if (textconv_one) {
270 data_one = run_textconv(textconv_one, one, &size_one);
271 if (!data_one)
272 die("unable to read files to diff");
274 else {
275 data_one = one->data;
276 size_one = one->size;
278 if (textconv_two) {
279 data_two = run_textconv(textconv_two, two, &size_two);
280 if (!data_two)
281 die("unable to read files to diff");
283 else {
284 data_two = two->data;
285 size_two = two->size;
288 lc_a = count_lines(data_one, size_one);
289 lc_b = count_lines(data_two, size_two);
290 fprintf(o->file,
291 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
292 metainfo, a_name.buf, name_a_tab, reset,
293 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
294 print_line_count(o->file, lc_a);
295 fprintf(o->file, " +");
296 print_line_count(o->file, lc_b);
297 fprintf(o->file, " @@%s\n", reset);
298 if (lc_a)
299 copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
300 if (lc_b)
301 copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
304 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
306 if (!DIFF_FILE_VALID(one)) {
307 mf->ptr = (char *)""; /* does not matter */
308 mf->size = 0;
309 return 0;
311 else if (diff_populate_filespec(one, 0))
312 return -1;
314 mf->ptr = one->data;
315 mf->size = one->size;
316 return 0;
319 struct diff_words_buffer {
320 mmfile_t text;
321 long alloc;
322 long current; /* output pointer */
323 int suppressed_newline;
326 static void diff_words_append(char *line, unsigned long len,
327 struct diff_words_buffer *buffer)
329 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
330 line++;
331 len--;
332 memcpy(buffer->text.ptr + buffer->text.size, line, len);
333 buffer->text.size += len;
336 struct diff_words_data {
337 struct diff_words_buffer minus, plus;
338 FILE *file;
341 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
342 int suppress_newline)
344 const char *ptr;
345 int eol = 0;
347 if (len == 0)
348 return;
350 ptr = buffer->text.ptr + buffer->current;
351 buffer->current += len;
353 if (ptr[len - 1] == '\n') {
354 eol = 1;
355 len--;
358 fputs(diff_get_color(1, color), file);
359 fwrite(ptr, len, 1, file);
360 fputs(diff_get_color(1, DIFF_RESET), file);
362 if (eol) {
363 if (suppress_newline)
364 buffer->suppressed_newline = 1;
365 else
366 putc('\n', file);
370 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
372 struct diff_words_data *diff_words = priv;
374 if (diff_words->minus.suppressed_newline) {
375 if (line[0] != '+')
376 putc('\n', diff_words->file);
377 diff_words->minus.suppressed_newline = 0;
380 len--;
381 switch (line[0]) {
382 case '-':
383 print_word(diff_words->file,
384 &diff_words->minus, len, DIFF_FILE_OLD, 1);
385 break;
386 case '+':
387 print_word(diff_words->file,
388 &diff_words->plus, len, DIFF_FILE_NEW, 0);
389 break;
390 case ' ':
391 print_word(diff_words->file,
392 &diff_words->plus, len, DIFF_PLAIN, 0);
393 diff_words->minus.current += len;
394 break;
399 * This function splits the words in buffer->text, and stores the list with
400 * newline separator into out.
402 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out)
404 int i;
405 out->size = buffer->text.size;
406 out->ptr = xmalloc(out->size);
407 memcpy(out->ptr, buffer->text.ptr, out->size);
408 for (i = 0; i < out->size; i++)
409 if (isspace(out->ptr[i]))
410 out->ptr[i] = '\n';
411 buffer->current = 0;
414 /* this executes the word diff on the accumulated buffers */
415 static void diff_words_show(struct diff_words_data *diff_words)
417 xpparam_t xpp;
418 xdemitconf_t xecfg;
419 xdemitcb_t ecb;
420 mmfile_t minus, plus;
422 memset(&xpp, 0, sizeof(xpp));
423 memset(&xecfg, 0, sizeof(xecfg));
424 diff_words_fill(&diff_words->minus, &minus);
425 diff_words_fill(&diff_words->plus, &plus);
426 xpp.flags = XDF_NEED_MINIMAL;
427 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
428 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
429 &xpp, &xecfg, &ecb);
430 free(minus.ptr);
431 free(plus.ptr);
432 diff_words->minus.text.size = diff_words->plus.text.size = 0;
434 if (diff_words->minus.suppressed_newline) {
435 putc('\n', diff_words->file);
436 diff_words->minus.suppressed_newline = 0;
440 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
442 struct emit_callback {
443 int nparents, color_diff;
444 unsigned ws_rule;
445 sane_truncate_fn truncate;
446 const char **label_path;
447 struct diff_words_data *diff_words;
448 int *found_changesp;
449 FILE *file;
452 static void free_diff_words_data(struct emit_callback *ecbdata)
454 if (ecbdata->diff_words) {
455 /* flush buffers */
456 if (ecbdata->diff_words->minus.text.size ||
457 ecbdata->diff_words->plus.text.size)
458 diff_words_show(ecbdata->diff_words);
460 free (ecbdata->diff_words->minus.text.ptr);
461 free (ecbdata->diff_words->plus.text.ptr);
462 free(ecbdata->diff_words);
463 ecbdata->diff_words = NULL;
467 const char *diff_get_color(int diff_use_color, enum color_diff ix)
469 if (diff_use_color)
470 return diff_colors[ix];
471 return "";
474 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
476 int has_trailing_newline, has_trailing_carriage_return;
478 has_trailing_newline = (len > 0 && line[len-1] == '\n');
479 if (has_trailing_newline)
480 len--;
481 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
482 if (has_trailing_carriage_return)
483 len--;
485 fputs(set, file);
486 fwrite(line, len, 1, file);
487 fputs(reset, file);
488 if (has_trailing_carriage_return)
489 fputc('\r', file);
490 if (has_trailing_newline)
491 fputc('\n', file);
494 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
496 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
497 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
499 if (!*ws)
500 emit_line(ecbdata->file, set, reset, line, len);
501 else {
502 /* Emit just the prefix, then the rest. */
503 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
504 ws_check_emit(line + ecbdata->nparents,
505 len - ecbdata->nparents, ecbdata->ws_rule,
506 ecbdata->file, set, reset, ws);
510 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
512 const char *cp;
513 unsigned long allot;
514 size_t l = len;
516 if (ecb->truncate)
517 return ecb->truncate(line, len);
518 cp = line;
519 allot = l;
520 while (0 < l) {
521 (void) utf8_width(&cp, &l);
522 if (!cp)
523 break; /* truncated in the middle? */
525 return allot - l;
528 static void fn_out_consume(void *priv, char *line, unsigned long len)
530 int i;
531 int color;
532 struct emit_callback *ecbdata = priv;
533 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
534 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
535 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
537 *(ecbdata->found_changesp) = 1;
539 if (ecbdata->label_path[0]) {
540 const char *name_a_tab, *name_b_tab;
542 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
543 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
545 fprintf(ecbdata->file, "%s--- %s%s%s\n",
546 meta, ecbdata->label_path[0], reset, name_a_tab);
547 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
548 meta, ecbdata->label_path[1], reset, name_b_tab);
549 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
552 if (diff_suppress_blank_empty
553 && len == 2 && line[0] == ' ' && line[1] == '\n') {
554 line[0] = '\n';
555 len = 1;
558 /* This is not really necessary for now because
559 * this codepath only deals with two-way diffs.
561 for (i = 0; i < len && line[i] == '@'; i++)
563 if (2 <= i && i < len && line[i] == ' ') {
564 ecbdata->nparents = i - 1;
565 len = sane_truncate_line(ecbdata, line, len);
566 emit_line(ecbdata->file,
567 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
568 reset, line, len);
569 if (line[len-1] != '\n')
570 putc('\n', ecbdata->file);
571 return;
574 if (len < ecbdata->nparents) {
575 emit_line(ecbdata->file, reset, reset, line, len);
576 return;
579 color = DIFF_PLAIN;
580 if (ecbdata->diff_words && ecbdata->nparents != 1)
581 /* fall back to normal diff */
582 free_diff_words_data(ecbdata);
583 if (ecbdata->diff_words) {
584 if (line[0] == '-') {
585 diff_words_append(line, len,
586 &ecbdata->diff_words->minus);
587 return;
588 } else if (line[0] == '+') {
589 diff_words_append(line, len,
590 &ecbdata->diff_words->plus);
591 return;
593 if (ecbdata->diff_words->minus.text.size ||
594 ecbdata->diff_words->plus.text.size)
595 diff_words_show(ecbdata->diff_words);
596 line++;
597 len--;
598 emit_line(ecbdata->file, plain, reset, line, len);
599 return;
601 for (i = 0; i < ecbdata->nparents && len; i++) {
602 if (line[i] == '-')
603 color = DIFF_FILE_OLD;
604 else if (line[i] == '+')
605 color = DIFF_FILE_NEW;
608 if (color != DIFF_FILE_NEW) {
609 emit_line(ecbdata->file,
610 diff_get_color(ecbdata->color_diff, color),
611 reset, line, len);
612 return;
614 emit_add_line(reset, ecbdata, line, len);
617 static char *pprint_rename(const char *a, const char *b)
619 const char *old = a;
620 const char *new = b;
621 struct strbuf name = STRBUF_INIT;
622 int pfx_length, sfx_length;
623 int len_a = strlen(a);
624 int len_b = strlen(b);
625 int a_midlen, b_midlen;
626 int qlen_a = quote_c_style(a, NULL, NULL, 0);
627 int qlen_b = quote_c_style(b, NULL, NULL, 0);
629 if (qlen_a || qlen_b) {
630 quote_c_style(a, &name, NULL, 0);
631 strbuf_addstr(&name, " => ");
632 quote_c_style(b, &name, NULL, 0);
633 return strbuf_detach(&name, NULL);
636 /* Find common prefix */
637 pfx_length = 0;
638 while (*old && *new && *old == *new) {
639 if (*old == '/')
640 pfx_length = old - a + 1;
641 old++;
642 new++;
645 /* Find common suffix */
646 old = a + len_a;
647 new = b + len_b;
648 sfx_length = 0;
649 while (a <= old && b <= new && *old == *new) {
650 if (*old == '/')
651 sfx_length = len_a - (old - a);
652 old--;
653 new--;
657 * pfx{mid-a => mid-b}sfx
658 * {pfx-a => pfx-b}sfx
659 * pfx{sfx-a => sfx-b}
660 * name-a => name-b
662 a_midlen = len_a - pfx_length - sfx_length;
663 b_midlen = len_b - pfx_length - sfx_length;
664 if (a_midlen < 0)
665 a_midlen = 0;
666 if (b_midlen < 0)
667 b_midlen = 0;
669 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
670 if (pfx_length + sfx_length) {
671 strbuf_add(&name, a, pfx_length);
672 strbuf_addch(&name, '{');
674 strbuf_add(&name, a + pfx_length, a_midlen);
675 strbuf_addstr(&name, " => ");
676 strbuf_add(&name, b + pfx_length, b_midlen);
677 if (pfx_length + sfx_length) {
678 strbuf_addch(&name, '}');
679 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
681 return strbuf_detach(&name, NULL);
684 struct diffstat_t {
685 int nr;
686 int alloc;
687 struct diffstat_file {
688 char *from_name;
689 char *name;
690 char *print_name;
691 unsigned is_unmerged:1;
692 unsigned is_binary:1;
693 unsigned is_renamed:1;
694 unsigned int added, deleted;
695 } **files;
698 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
699 const char *name_a,
700 const char *name_b)
702 struct diffstat_file *x;
703 x = xcalloc(sizeof (*x), 1);
704 if (diffstat->nr == diffstat->alloc) {
705 diffstat->alloc = alloc_nr(diffstat->alloc);
706 diffstat->files = xrealloc(diffstat->files,
707 diffstat->alloc * sizeof(x));
709 diffstat->files[diffstat->nr++] = x;
710 if (name_b) {
711 x->from_name = xstrdup(name_a);
712 x->name = xstrdup(name_b);
713 x->is_renamed = 1;
715 else {
716 x->from_name = NULL;
717 x->name = xstrdup(name_a);
719 return x;
722 static void diffstat_consume(void *priv, char *line, unsigned long len)
724 struct diffstat_t *diffstat = priv;
725 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
727 if (line[0] == '+')
728 x->added++;
729 else if (line[0] == '-')
730 x->deleted++;
733 const char mime_boundary_leader[] = "------------";
735 static int scale_linear(int it, int width, int max_change)
738 * make sure that at least one '-' is printed if there were deletions,
739 * and likewise for '+'.
741 if (max_change < 2)
742 return it;
743 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
746 static void show_name(FILE *file,
747 const char *prefix, const char *name, int len,
748 const char *reset, const char *set)
750 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
753 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
755 if (cnt <= 0)
756 return;
757 fprintf(file, "%s", set);
758 while (cnt--)
759 putc(ch, file);
760 fprintf(file, "%s", reset);
763 static void fill_print_name(struct diffstat_file *file)
765 char *pname;
767 if (file->print_name)
768 return;
770 if (!file->is_renamed) {
771 struct strbuf buf = STRBUF_INIT;
772 if (quote_c_style(file->name, &buf, NULL, 0)) {
773 pname = strbuf_detach(&buf, NULL);
774 } else {
775 pname = file->name;
776 strbuf_release(&buf);
778 } else {
779 pname = pprint_rename(file->from_name, file->name);
781 file->print_name = pname;
784 static void show_stats(struct diffstat_t* data, struct diff_options *options)
786 int i, len, add, del, total, adds = 0, dels = 0;
787 int max_change = 0, max_len = 0;
788 int total_files = data->nr;
789 int width, name_width;
790 const char *reset, *set, *add_c, *del_c;
792 if (data->nr == 0)
793 return;
795 width = options->stat_width ? options->stat_width : 80;
796 name_width = options->stat_name_width ? options->stat_name_width : 50;
798 /* Sanity: give at least 5 columns to the graph,
799 * but leave at least 10 columns for the name.
801 if (width < 25)
802 width = 25;
803 if (name_width < 10)
804 name_width = 10;
805 else if (width < name_width + 15)
806 name_width = width - 15;
808 /* Find the longest filename and max number of changes */
809 reset = diff_get_color_opt(options, DIFF_RESET);
810 set = diff_get_color_opt(options, DIFF_PLAIN);
811 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
812 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
814 for (i = 0; i < data->nr; i++) {
815 struct diffstat_file *file = data->files[i];
816 int change = file->added + file->deleted;
817 fill_print_name(file);
818 len = strlen(file->print_name);
819 if (max_len < len)
820 max_len = len;
822 if (file->is_binary || file->is_unmerged)
823 continue;
824 if (max_change < change)
825 max_change = change;
828 /* Compute the width of the graph part;
829 * 10 is for one blank at the beginning of the line plus
830 * " | count " between the name and the graph.
832 * From here on, name_width is the width of the name area,
833 * and width is the width of the graph area.
835 name_width = (name_width < max_len) ? name_width : max_len;
836 if (width < (name_width + 10) + max_change)
837 width = width - (name_width + 10);
838 else
839 width = max_change;
841 for (i = 0; i < data->nr; i++) {
842 const char *prefix = "";
843 char *name = data->files[i]->print_name;
844 int added = data->files[i]->added;
845 int deleted = data->files[i]->deleted;
846 int name_len;
849 * "scale" the filename
851 len = name_width;
852 name_len = strlen(name);
853 if (name_width < name_len) {
854 char *slash;
855 prefix = "...";
856 len -= 3;
857 name += name_len - len;
858 slash = strchr(name, '/');
859 if (slash)
860 name = slash;
863 if (data->files[i]->is_binary) {
864 show_name(options->file, prefix, name, len, reset, set);
865 fprintf(options->file, " Bin ");
866 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
867 fprintf(options->file, " -> ");
868 fprintf(options->file, "%s%d%s", add_c, added, reset);
869 fprintf(options->file, " bytes");
870 fprintf(options->file, "\n");
871 continue;
873 else if (data->files[i]->is_unmerged) {
874 show_name(options->file, prefix, name, len, reset, set);
875 fprintf(options->file, " Unmerged\n");
876 continue;
878 else if (!data->files[i]->is_renamed &&
879 (added + deleted == 0)) {
880 total_files--;
881 continue;
885 * scale the add/delete
887 add = added;
888 del = deleted;
889 total = add + del;
890 adds += add;
891 dels += del;
893 if (width <= max_change) {
894 add = scale_linear(add, width, max_change);
895 del = scale_linear(del, width, max_change);
896 total = add + del;
898 show_name(options->file, prefix, name, len, reset, set);
899 fprintf(options->file, "%5d%s", added + deleted,
900 added + deleted ? " " : "");
901 show_graph(options->file, '+', add, add_c, reset);
902 show_graph(options->file, '-', del, del_c, reset);
903 fprintf(options->file, "\n");
905 fprintf(options->file,
906 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
907 set, total_files, adds, dels, reset);
910 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
912 int i, adds = 0, dels = 0, total_files = data->nr;
914 if (data->nr == 0)
915 return;
917 for (i = 0; i < data->nr; i++) {
918 if (!data->files[i]->is_binary &&
919 !data->files[i]->is_unmerged) {
920 int added = data->files[i]->added;
921 int deleted= data->files[i]->deleted;
922 if (!data->files[i]->is_renamed &&
923 (added + deleted == 0)) {
924 total_files--;
925 } else {
926 adds += added;
927 dels += deleted;
931 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
932 total_files, adds, dels);
935 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
937 int i;
939 if (data->nr == 0)
940 return;
942 for (i = 0; i < data->nr; i++) {
943 struct diffstat_file *file = data->files[i];
945 if (file->is_binary)
946 fprintf(options->file, "-\t-\t");
947 else
948 fprintf(options->file,
949 "%d\t%d\t", file->added, file->deleted);
950 if (options->line_termination) {
951 fill_print_name(file);
952 if (!file->is_renamed)
953 write_name_quoted(file->name, options->file,
954 options->line_termination);
955 else {
956 fputs(file->print_name, options->file);
957 putc(options->line_termination, options->file);
959 } else {
960 if (file->is_renamed) {
961 putc('\0', options->file);
962 write_name_quoted(file->from_name, options->file, '\0');
964 write_name_quoted(file->name, options->file, '\0');
969 struct dirstat_file {
970 const char *name;
971 unsigned long changed;
974 struct dirstat_dir {
975 struct dirstat_file *files;
976 int alloc, nr, percent, cumulative;
979 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
981 unsigned long this_dir = 0;
982 unsigned int sources = 0;
984 while (dir->nr) {
985 struct dirstat_file *f = dir->files;
986 int namelen = strlen(f->name);
987 unsigned long this;
988 char *slash;
990 if (namelen < baselen)
991 break;
992 if (memcmp(f->name, base, baselen))
993 break;
994 slash = strchr(f->name + baselen, '/');
995 if (slash) {
996 int newbaselen = slash + 1 - f->name;
997 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
998 sources++;
999 } else {
1000 this = f->changed;
1001 dir->files++;
1002 dir->nr--;
1003 sources += 2;
1005 this_dir += this;
1009 * We don't report dirstat's for
1010 * - the top level
1011 * - or cases where everything came from a single directory
1012 * under this directory (sources == 1).
1014 if (baselen && sources != 1) {
1015 int permille = this_dir * 1000 / changed;
1016 if (permille) {
1017 int percent = permille / 10;
1018 if (percent >= dir->percent) {
1019 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1020 if (!dir->cumulative)
1021 return 0;
1025 return this_dir;
1028 static int dirstat_compare(const void *_a, const void *_b)
1030 const struct dirstat_file *a = _a;
1031 const struct dirstat_file *b = _b;
1032 return strcmp(a->name, b->name);
1035 static void show_dirstat(struct diff_options *options)
1037 int i;
1038 unsigned long changed;
1039 struct dirstat_dir dir;
1040 struct diff_queue_struct *q = &diff_queued_diff;
1042 dir.files = NULL;
1043 dir.alloc = 0;
1044 dir.nr = 0;
1045 dir.percent = options->dirstat_percent;
1046 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1048 changed = 0;
1049 for (i = 0; i < q->nr; i++) {
1050 struct diff_filepair *p = q->queue[i];
1051 const char *name;
1052 unsigned long copied, added, damage;
1054 name = p->one->path ? p->one->path : p->two->path;
1056 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1057 diff_populate_filespec(p->one, 0);
1058 diff_populate_filespec(p->two, 0);
1059 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1060 &copied, &added);
1061 diff_free_filespec_data(p->one);
1062 diff_free_filespec_data(p->two);
1063 } else if (DIFF_FILE_VALID(p->one)) {
1064 diff_populate_filespec(p->one, 1);
1065 copied = added = 0;
1066 diff_free_filespec_data(p->one);
1067 } else if (DIFF_FILE_VALID(p->two)) {
1068 diff_populate_filespec(p->two, 1);
1069 copied = 0;
1070 added = p->two->size;
1071 diff_free_filespec_data(p->two);
1072 } else
1073 continue;
1076 * Original minus copied is the removed material,
1077 * added is the new material. They are both damages
1078 * made to the preimage. In --dirstat-by-file mode, count
1079 * damaged files, not damaged lines. This is done by
1080 * counting only a single damaged line per file.
1082 damage = (p->one->size - copied) + added;
1083 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1084 damage = 1;
1086 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1087 dir.files[dir.nr].name = name;
1088 dir.files[dir.nr].changed = damage;
1089 changed += damage;
1090 dir.nr++;
1093 /* This can happen even with many files, if everything was renames */
1094 if (!changed)
1095 return;
1097 /* Show all directories with more than x% of the changes */
1098 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1099 gather_dirstat(options->file, &dir, changed, "", 0);
1102 static void free_diffstat_info(struct diffstat_t *diffstat)
1104 int i;
1105 for (i = 0; i < diffstat->nr; i++) {
1106 struct diffstat_file *f = diffstat->files[i];
1107 if (f->name != f->print_name)
1108 free(f->print_name);
1109 free(f->name);
1110 free(f->from_name);
1111 free(f);
1113 free(diffstat->files);
1116 struct checkdiff_t {
1117 const char *filename;
1118 int lineno;
1119 struct diff_options *o;
1120 unsigned ws_rule;
1121 unsigned status;
1122 int trailing_blanks_start;
1125 static int is_conflict_marker(const char *line, unsigned long len)
1127 char firstchar;
1128 int cnt;
1130 if (len < 8)
1131 return 0;
1132 firstchar = line[0];
1133 switch (firstchar) {
1134 case '=': case '>': case '<':
1135 break;
1136 default:
1137 return 0;
1139 for (cnt = 1; cnt < 7; cnt++)
1140 if (line[cnt] != firstchar)
1141 return 0;
1142 /* line[0] thru line[6] are same as firstchar */
1143 if (firstchar == '=') {
1144 /* divider between ours and theirs? */
1145 if (len != 8 || line[7] != '\n')
1146 return 0;
1147 } else if (len < 8 || !isspace(line[7])) {
1148 /* not divider before ours nor after theirs */
1149 return 0;
1151 return 1;
1154 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1156 struct checkdiff_t *data = priv;
1157 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1158 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1159 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1160 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1161 char *err;
1163 if (line[0] == '+') {
1164 unsigned bad;
1165 data->lineno++;
1166 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1167 data->trailing_blanks_start = 0;
1168 else if (!data->trailing_blanks_start)
1169 data->trailing_blanks_start = data->lineno;
1170 if (is_conflict_marker(line + 1, len - 1)) {
1171 data->status |= 1;
1172 fprintf(data->o->file,
1173 "%s:%d: leftover conflict marker\n",
1174 data->filename, data->lineno);
1176 bad = ws_check(line + 1, len - 1, data->ws_rule);
1177 if (!bad)
1178 return;
1179 data->status |= bad;
1180 err = whitespace_error_string(bad);
1181 fprintf(data->o->file, "%s:%d: %s.\n",
1182 data->filename, data->lineno, err);
1183 free(err);
1184 emit_line(data->o->file, set, reset, line, 1);
1185 ws_check_emit(line + 1, len - 1, data->ws_rule,
1186 data->o->file, set, reset, ws);
1187 } else if (line[0] == ' ') {
1188 data->lineno++;
1189 data->trailing_blanks_start = 0;
1190 } else if (line[0] == '@') {
1191 char *plus = strchr(line, '+');
1192 if (plus)
1193 data->lineno = strtol(plus, NULL, 10) - 1;
1194 else
1195 die("invalid diff");
1196 data->trailing_blanks_start = 0;
1200 static unsigned char *deflate_it(char *data,
1201 unsigned long size,
1202 unsigned long *result_size)
1204 int bound;
1205 unsigned char *deflated;
1206 z_stream stream;
1208 memset(&stream, 0, sizeof(stream));
1209 deflateInit(&stream, zlib_compression_level);
1210 bound = deflateBound(&stream, size);
1211 deflated = xmalloc(bound);
1212 stream.next_out = deflated;
1213 stream.avail_out = bound;
1215 stream.next_in = (unsigned char *)data;
1216 stream.avail_in = size;
1217 while (deflate(&stream, Z_FINISH) == Z_OK)
1218 ; /* nothing */
1219 deflateEnd(&stream);
1220 *result_size = stream.total_out;
1221 return deflated;
1224 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1226 void *cp;
1227 void *delta;
1228 void *deflated;
1229 void *data;
1230 unsigned long orig_size;
1231 unsigned long delta_size;
1232 unsigned long deflate_size;
1233 unsigned long data_size;
1235 /* We could do deflated delta, or we could do just deflated two,
1236 * whichever is smaller.
1238 delta = NULL;
1239 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1240 if (one->size && two->size) {
1241 delta = diff_delta(one->ptr, one->size,
1242 two->ptr, two->size,
1243 &delta_size, deflate_size);
1244 if (delta) {
1245 void *to_free = delta;
1246 orig_size = delta_size;
1247 delta = deflate_it(delta, delta_size, &delta_size);
1248 free(to_free);
1252 if (delta && delta_size < deflate_size) {
1253 fprintf(file, "delta %lu\n", orig_size);
1254 free(deflated);
1255 data = delta;
1256 data_size = delta_size;
1258 else {
1259 fprintf(file, "literal %lu\n", two->size);
1260 free(delta);
1261 data = deflated;
1262 data_size = deflate_size;
1265 /* emit data encoded in base85 */
1266 cp = data;
1267 while (data_size) {
1268 int bytes = (52 < data_size) ? 52 : data_size;
1269 char line[70];
1270 data_size -= bytes;
1271 if (bytes <= 26)
1272 line[0] = bytes + 'A' - 1;
1273 else
1274 line[0] = bytes - 26 + 'a' - 1;
1275 encode_85(line + 1, cp, bytes);
1276 cp = (char *) cp + bytes;
1277 fputs(line, file);
1278 fputc('\n', file);
1280 fprintf(file, "\n");
1281 free(data);
1284 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1286 fprintf(file, "GIT binary patch\n");
1287 emit_binary_diff_body(file, one, two);
1288 emit_binary_diff_body(file, two, one);
1291 static void diff_filespec_load_driver(struct diff_filespec *one)
1293 if (!one->driver)
1294 one->driver = userdiff_find_by_path(one->path);
1295 if (!one->driver)
1296 one->driver = userdiff_find_by_name("default");
1299 int diff_filespec_is_binary(struct diff_filespec *one)
1301 if (one->is_binary == -1) {
1302 diff_filespec_load_driver(one);
1303 if (one->driver->binary != -1)
1304 one->is_binary = one->driver->binary;
1305 else {
1306 if (!one->data && DIFF_FILE_VALID(one))
1307 diff_populate_filespec(one, 0);
1308 if (one->data)
1309 one->is_binary = buffer_is_binary(one->data,
1310 one->size);
1311 if (one->is_binary == -1)
1312 one->is_binary = 0;
1315 return one->is_binary;
1318 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1320 diff_filespec_load_driver(one);
1321 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1324 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1326 if (!options->a_prefix)
1327 options->a_prefix = a;
1328 if (!options->b_prefix)
1329 options->b_prefix = b;
1332 static const char *get_textconv(struct diff_filespec *one)
1334 if (!DIFF_FILE_VALID(one))
1335 return NULL;
1336 if (!S_ISREG(one->mode))
1337 return NULL;
1338 diff_filespec_load_driver(one);
1339 return one->driver->textconv;
1342 static void builtin_diff(const char *name_a,
1343 const char *name_b,
1344 struct diff_filespec *one,
1345 struct diff_filespec *two,
1346 const char *xfrm_msg,
1347 struct diff_options *o,
1348 int complete_rewrite)
1350 mmfile_t mf1, mf2;
1351 const char *lbl[2];
1352 char *a_one, *b_two;
1353 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1354 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1355 const char *a_prefix, *b_prefix;
1356 const char *textconv_one = NULL, *textconv_two = NULL;
1358 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1359 textconv_one = get_textconv(one);
1360 textconv_two = get_textconv(two);
1363 diff_set_mnemonic_prefix(o, "a/", "b/");
1364 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1365 a_prefix = o->b_prefix;
1366 b_prefix = o->a_prefix;
1367 } else {
1368 a_prefix = o->a_prefix;
1369 b_prefix = o->b_prefix;
1372 /* Never use a non-valid filename anywhere if at all possible */
1373 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1374 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1376 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1377 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1378 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1379 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1380 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1381 if (lbl[0][0] == '/') {
1382 /* /dev/null */
1383 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1384 if (xfrm_msg && xfrm_msg[0])
1385 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1387 else if (lbl[1][0] == '/') {
1388 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1389 if (xfrm_msg && xfrm_msg[0])
1390 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1392 else {
1393 if (one->mode != two->mode) {
1394 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1395 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1397 if (xfrm_msg && xfrm_msg[0])
1398 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1400 * we do not run diff between different kind
1401 * of objects.
1403 if ((one->mode ^ two->mode) & S_IFMT)
1404 goto free_ab_and_return;
1405 if (complete_rewrite &&
1406 (textconv_one || !diff_filespec_is_binary(one)) &&
1407 (textconv_two || !diff_filespec_is_binary(two))) {
1408 emit_rewrite_diff(name_a, name_b, one, two,
1409 textconv_one, textconv_two, o);
1410 o->found_changes = 1;
1411 goto free_ab_and_return;
1415 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1416 die("unable to read files to diff");
1418 if (!DIFF_OPT_TST(o, TEXT) &&
1419 ( (diff_filespec_is_binary(one) && !textconv_one) ||
1420 (diff_filespec_is_binary(two) && !textconv_two) )) {
1421 /* Quite common confusing case */
1422 if (mf1.size == mf2.size &&
1423 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1424 goto free_ab_and_return;
1425 if (DIFF_OPT_TST(o, BINARY))
1426 emit_binary_diff(o->file, &mf1, &mf2);
1427 else
1428 fprintf(o->file, "Binary files %s and %s differ\n",
1429 lbl[0], lbl[1]);
1430 o->found_changes = 1;
1432 else {
1433 /* Crazy xdl interfaces.. */
1434 const char *diffopts = getenv("GIT_DIFF_OPTS");
1435 xpparam_t xpp;
1436 xdemitconf_t xecfg;
1437 xdemitcb_t ecb;
1438 struct emit_callback ecbdata;
1439 const struct userdiff_funcname *pe;
1441 if (textconv_one) {
1442 size_t size;
1443 mf1.ptr = run_textconv(textconv_one, one, &size);
1444 if (!mf1.ptr)
1445 die("unable to read files to diff");
1446 mf1.size = size;
1448 if (textconv_two) {
1449 size_t size;
1450 mf2.ptr = run_textconv(textconv_two, two, &size);
1451 if (!mf2.ptr)
1452 die("unable to read files to diff");
1453 mf2.size = size;
1456 pe = diff_funcname_pattern(one);
1457 if (!pe)
1458 pe = diff_funcname_pattern(two);
1460 memset(&xpp, 0, sizeof(xpp));
1461 memset(&xecfg, 0, sizeof(xecfg));
1462 memset(&ecbdata, 0, sizeof(ecbdata));
1463 ecbdata.label_path = lbl;
1464 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1465 ecbdata.found_changesp = &o->found_changes;
1466 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1467 ecbdata.file = o->file;
1468 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1469 xecfg.ctxlen = o->context;
1470 xecfg.interhunkctxlen = o->interhunkcontext;
1471 xecfg.flags = XDL_EMIT_FUNCNAMES;
1472 if (pe)
1473 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1474 if (!diffopts)
1476 else if (!prefixcmp(diffopts, "--unified="))
1477 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1478 else if (!prefixcmp(diffopts, "-u"))
1479 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1480 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1481 ecbdata.diff_words =
1482 xcalloc(1, sizeof(struct diff_words_data));
1483 ecbdata.diff_words->file = o->file;
1485 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1486 &xpp, &xecfg, &ecb);
1487 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1488 free_diff_words_data(&ecbdata);
1489 if (textconv_one)
1490 free(mf1.ptr);
1491 if (textconv_two)
1492 free(mf2.ptr);
1495 free_ab_and_return:
1496 diff_free_filespec_data(one);
1497 diff_free_filespec_data(two);
1498 free(a_one);
1499 free(b_two);
1500 return;
1503 static void builtin_diffstat(const char *name_a, const char *name_b,
1504 struct diff_filespec *one,
1505 struct diff_filespec *two,
1506 struct diffstat_t *diffstat,
1507 struct diff_options *o,
1508 int complete_rewrite)
1510 mmfile_t mf1, mf2;
1511 struct diffstat_file *data;
1513 data = diffstat_add(diffstat, name_a, name_b);
1515 if (!one || !two) {
1516 data->is_unmerged = 1;
1517 return;
1519 if (complete_rewrite) {
1520 diff_populate_filespec(one, 0);
1521 diff_populate_filespec(two, 0);
1522 data->deleted = count_lines(one->data, one->size);
1523 data->added = count_lines(two->data, two->size);
1524 goto free_and_return;
1526 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1527 die("unable to read files to diff");
1529 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1530 data->is_binary = 1;
1531 data->added = mf2.size;
1532 data->deleted = mf1.size;
1533 } else {
1534 /* Crazy xdl interfaces.. */
1535 xpparam_t xpp;
1536 xdemitconf_t xecfg;
1537 xdemitcb_t ecb;
1539 memset(&xpp, 0, sizeof(xpp));
1540 memset(&xecfg, 0, sizeof(xecfg));
1541 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1542 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1543 &xpp, &xecfg, &ecb);
1546 free_and_return:
1547 diff_free_filespec_data(one);
1548 diff_free_filespec_data(two);
1551 static void builtin_checkdiff(const char *name_a, const char *name_b,
1552 const char *attr_path,
1553 struct diff_filespec *one,
1554 struct diff_filespec *two,
1555 struct diff_options *o)
1557 mmfile_t mf1, mf2;
1558 struct checkdiff_t data;
1560 if (!two)
1561 return;
1563 memset(&data, 0, sizeof(data));
1564 data.filename = name_b ? name_b : name_a;
1565 data.lineno = 0;
1566 data.o = o;
1567 data.ws_rule = whitespace_rule(attr_path);
1569 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1570 die("unable to read files to diff");
1573 * All the other codepaths check both sides, but not checking
1574 * the "old" side here is deliberate. We are checking the newly
1575 * introduced changes, and as long as the "new" side is text, we
1576 * can and should check what it introduces.
1578 if (diff_filespec_is_binary(two))
1579 goto free_and_return;
1580 else {
1581 /* Crazy xdl interfaces.. */
1582 xpparam_t xpp;
1583 xdemitconf_t xecfg;
1584 xdemitcb_t ecb;
1586 memset(&xpp, 0, sizeof(xpp));
1587 memset(&xecfg, 0, sizeof(xecfg));
1588 xecfg.ctxlen = 1; /* at least one context line */
1589 xpp.flags = XDF_NEED_MINIMAL;
1590 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1591 &xpp, &xecfg, &ecb);
1593 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1594 data.trailing_blanks_start) {
1595 fprintf(o->file, "%s:%d: ends with blank lines.\n",
1596 data.filename, data.trailing_blanks_start);
1597 data.status = 1; /* report errors */
1600 free_and_return:
1601 diff_free_filespec_data(one);
1602 diff_free_filespec_data(two);
1603 if (data.status)
1604 DIFF_OPT_SET(o, CHECK_FAILED);
1607 struct diff_filespec *alloc_filespec(const char *path)
1609 int namelen = strlen(path);
1610 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1612 memset(spec, 0, sizeof(*spec));
1613 spec->path = (char *)(spec + 1);
1614 memcpy(spec->path, path, namelen+1);
1615 spec->count = 1;
1616 spec->is_binary = -1;
1617 return spec;
1620 void free_filespec(struct diff_filespec *spec)
1622 if (!--spec->count) {
1623 diff_free_filespec_data(spec);
1624 free(spec);
1628 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1629 unsigned short mode)
1631 if (mode) {
1632 spec->mode = canon_mode(mode);
1633 hashcpy(spec->sha1, sha1);
1634 spec->sha1_valid = !is_null_sha1(sha1);
1639 * Given a name and sha1 pair, if the index tells us the file in
1640 * the work tree has that object contents, return true, so that
1641 * prepare_temp_file() does not have to inflate and extract.
1643 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1645 struct cache_entry *ce;
1646 struct stat st;
1647 int pos, len;
1649 /* We do not read the cache ourselves here, because the
1650 * benchmark with my previous version that always reads cache
1651 * shows that it makes things worse for diff-tree comparing
1652 * two linux-2.6 kernel trees in an already checked out work
1653 * tree. This is because most diff-tree comparisons deal with
1654 * only a small number of files, while reading the cache is
1655 * expensive for a large project, and its cost outweighs the
1656 * savings we get by not inflating the object to a temporary
1657 * file. Practically, this code only helps when we are used
1658 * by diff-cache --cached, which does read the cache before
1659 * calling us.
1661 if (!active_cache)
1662 return 0;
1664 /* We want to avoid the working directory if our caller
1665 * doesn't need the data in a normal file, this system
1666 * is rather slow with its stat/open/mmap/close syscalls,
1667 * and the object is contained in a pack file. The pack
1668 * is probably already open and will be faster to obtain
1669 * the data through than the working directory. Loose
1670 * objects however would tend to be slower as they need
1671 * to be individually opened and inflated.
1673 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1674 return 0;
1676 len = strlen(name);
1677 pos = cache_name_pos(name, len);
1678 if (pos < 0)
1679 return 0;
1680 ce = active_cache[pos];
1683 * This is not the sha1 we are looking for, or
1684 * unreusable because it is not a regular file.
1686 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1687 return 0;
1690 * If ce matches the file in the work tree, we can reuse it.
1692 if (ce_uptodate(ce) ||
1693 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1694 return 1;
1696 return 0;
1699 static int populate_from_stdin(struct diff_filespec *s)
1701 struct strbuf buf = STRBUF_INIT;
1702 size_t size = 0;
1704 if (strbuf_read(&buf, 0, 0) < 0)
1705 return error("error while reading from stdin %s",
1706 strerror(errno));
1708 s->should_munmap = 0;
1709 s->data = strbuf_detach(&buf, &size);
1710 s->size = size;
1711 s->should_free = 1;
1712 return 0;
1715 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1717 int len;
1718 char *data = xmalloc(100);
1719 len = snprintf(data, 100,
1720 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1721 s->data = data;
1722 s->size = len;
1723 s->should_free = 1;
1724 if (size_only) {
1725 s->data = NULL;
1726 free(data);
1728 return 0;
1732 * While doing rename detection and pickaxe operation, we may need to
1733 * grab the data for the blob (or file) for our own in-core comparison.
1734 * diff_filespec has data and size fields for this purpose.
1736 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1738 int err = 0;
1739 if (!DIFF_FILE_VALID(s))
1740 die("internal error: asking to populate invalid file.");
1741 if (S_ISDIR(s->mode))
1742 return -1;
1744 if (s->data)
1745 return 0;
1747 if (size_only && 0 < s->size)
1748 return 0;
1750 if (S_ISGITLINK(s->mode))
1751 return diff_populate_gitlink(s, size_only);
1753 if (!s->sha1_valid ||
1754 reuse_worktree_file(s->path, s->sha1, 0)) {
1755 struct strbuf buf = STRBUF_INIT;
1756 struct stat st;
1757 int fd;
1759 if (!strcmp(s->path, "-"))
1760 return populate_from_stdin(s);
1762 if (lstat(s->path, &st) < 0) {
1763 if (errno == ENOENT) {
1764 err_empty:
1765 err = -1;
1766 empty:
1767 s->data = (char *)"";
1768 s->size = 0;
1769 return err;
1772 s->size = xsize_t(st.st_size);
1773 if (!s->size)
1774 goto empty;
1775 if (S_ISLNK(st.st_mode)) {
1776 struct strbuf sb = STRBUF_INIT;
1778 if (strbuf_readlink(&sb, s->path, s->size))
1779 goto err_empty;
1780 s->size = sb.len;
1781 s->data = strbuf_detach(&sb, NULL);
1782 s->should_free = 1;
1783 return 0;
1785 if (size_only)
1786 return 0;
1787 fd = open(s->path, O_RDONLY);
1788 if (fd < 0)
1789 goto err_empty;
1790 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1791 close(fd);
1792 s->should_munmap = 1;
1795 * Convert from working tree format to canonical git format
1797 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1798 size_t size = 0;
1799 munmap(s->data, s->size);
1800 s->should_munmap = 0;
1801 s->data = strbuf_detach(&buf, &size);
1802 s->size = size;
1803 s->should_free = 1;
1806 else {
1807 enum object_type type;
1808 if (size_only)
1809 type = sha1_object_info(s->sha1, &s->size);
1810 else {
1811 s->data = read_sha1_file(s->sha1, &type, &s->size);
1812 s->should_free = 1;
1815 return 0;
1818 void diff_free_filespec_blob(struct diff_filespec *s)
1820 if (s->should_free)
1821 free(s->data);
1822 else if (s->should_munmap)
1823 munmap(s->data, s->size);
1825 if (s->should_free || s->should_munmap) {
1826 s->should_free = s->should_munmap = 0;
1827 s->data = NULL;
1831 void diff_free_filespec_data(struct diff_filespec *s)
1833 diff_free_filespec_blob(s);
1834 free(s->cnt_data);
1835 s->cnt_data = NULL;
1838 static void prep_temp_blob(struct diff_tempfile *temp,
1839 void *blob,
1840 unsigned long size,
1841 const unsigned char *sha1,
1842 int mode)
1844 int fd;
1846 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1847 if (fd < 0)
1848 die("unable to create temp-file: %s", strerror(errno));
1849 if (write_in_full(fd, blob, size) != size)
1850 die("unable to write temp-file");
1851 close(fd);
1852 temp->name = temp->tmp_path;
1853 strcpy(temp->hex, sha1_to_hex(sha1));
1854 temp->hex[40] = 0;
1855 sprintf(temp->mode, "%06o", mode);
1858 static void prepare_temp_file(const char *name,
1859 struct diff_tempfile *temp,
1860 struct diff_filespec *one)
1862 if (!DIFF_FILE_VALID(one)) {
1863 not_a_valid_file:
1864 /* A '-' entry produces this for file-2, and
1865 * a '+' entry produces this for file-1.
1867 temp->name = "/dev/null";
1868 strcpy(temp->hex, ".");
1869 strcpy(temp->mode, ".");
1870 return;
1873 if (!one->sha1_valid ||
1874 reuse_worktree_file(name, one->sha1, 1)) {
1875 struct stat st;
1876 if (lstat(name, &st) < 0) {
1877 if (errno == ENOENT)
1878 goto not_a_valid_file;
1879 die("stat(%s): %s", name, strerror(errno));
1881 if (S_ISLNK(st.st_mode)) {
1882 int ret;
1883 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1884 ret = readlink(name, buf, sizeof(buf));
1885 if (ret < 0)
1886 die("readlink(%s)", name);
1887 if (ret == sizeof(buf))
1888 die("symlink too long: %s", name);
1889 prep_temp_blob(temp, buf, ret,
1890 (one->sha1_valid ?
1891 one->sha1 : null_sha1),
1892 (one->sha1_valid ?
1893 one->mode : S_IFLNK));
1895 else {
1896 /* we can borrow from the file in the work tree */
1897 temp->name = name;
1898 if (!one->sha1_valid)
1899 strcpy(temp->hex, sha1_to_hex(null_sha1));
1900 else
1901 strcpy(temp->hex, sha1_to_hex(one->sha1));
1902 /* Even though we may sometimes borrow the
1903 * contents from the work tree, we always want
1904 * one->mode. mode is trustworthy even when
1905 * !(one->sha1_valid), as long as
1906 * DIFF_FILE_VALID(one).
1908 sprintf(temp->mode, "%06o", one->mode);
1910 return;
1912 else {
1913 if (diff_populate_filespec(one, 0))
1914 die("cannot read data blob for %s", one->path);
1915 prep_temp_blob(temp, one->data, one->size,
1916 one->sha1, one->mode);
1920 static void remove_tempfile(void)
1922 int i;
1924 for (i = 0; i < 2; i++)
1925 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1926 unlink(diff_temp[i].name);
1927 diff_temp[i].name = NULL;
1931 static void remove_tempfile_on_signal(int signo)
1933 remove_tempfile();
1934 signal(SIGINT, SIG_DFL);
1935 raise(signo);
1938 /* An external diff command takes:
1940 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1941 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1944 static void run_external_diff(const char *pgm,
1945 const char *name,
1946 const char *other,
1947 struct diff_filespec *one,
1948 struct diff_filespec *two,
1949 const char *xfrm_msg,
1950 int complete_rewrite)
1952 const char *spawn_arg[10];
1953 struct diff_tempfile *temp = diff_temp;
1954 int retval;
1955 static int atexit_asked = 0;
1956 const char *othername;
1957 const char **arg = &spawn_arg[0];
1959 othername = (other? other : name);
1960 if (one && two) {
1961 prepare_temp_file(name, &temp[0], one);
1962 prepare_temp_file(othername, &temp[1], two);
1963 if (! atexit_asked &&
1964 (temp[0].name == temp[0].tmp_path ||
1965 temp[1].name == temp[1].tmp_path)) {
1966 atexit_asked = 1;
1967 atexit(remove_tempfile);
1969 signal(SIGINT, remove_tempfile_on_signal);
1972 if (one && two) {
1973 *arg++ = pgm;
1974 *arg++ = name;
1975 *arg++ = temp[0].name;
1976 *arg++ = temp[0].hex;
1977 *arg++ = temp[0].mode;
1978 *arg++ = temp[1].name;
1979 *arg++ = temp[1].hex;
1980 *arg++ = temp[1].mode;
1981 if (other) {
1982 *arg++ = other;
1983 *arg++ = xfrm_msg;
1985 } else {
1986 *arg++ = pgm;
1987 *arg++ = name;
1989 *arg = NULL;
1990 fflush(NULL);
1991 retval = run_command_v_opt(spawn_arg, 0);
1992 remove_tempfile();
1993 if (retval) {
1994 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1995 exit(1);
1999 static void run_diff_cmd(const char *pgm,
2000 const char *name,
2001 const char *other,
2002 const char *attr_path,
2003 struct diff_filespec *one,
2004 struct diff_filespec *two,
2005 const char *xfrm_msg,
2006 struct diff_options *o,
2007 int complete_rewrite)
2009 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2010 pgm = NULL;
2011 else {
2012 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2013 if (drv && drv->external)
2014 pgm = drv->external;
2017 if (pgm) {
2018 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2019 complete_rewrite);
2020 return;
2022 if (one && two)
2023 builtin_diff(name, other ? other : name,
2024 one, two, xfrm_msg, o, complete_rewrite);
2025 else
2026 fprintf(o->file, "* Unmerged path %s\n", name);
2029 static void diff_fill_sha1_info(struct diff_filespec *one)
2031 if (DIFF_FILE_VALID(one)) {
2032 if (!one->sha1_valid) {
2033 struct stat st;
2034 if (!strcmp(one->path, "-")) {
2035 hashcpy(one->sha1, null_sha1);
2036 return;
2038 if (lstat(one->path, &st) < 0)
2039 die("stat %s", one->path);
2040 if (index_path(one->sha1, one->path, &st, 0))
2041 die("cannot hash %s", one->path);
2044 else
2045 hashclr(one->sha1);
2048 static int similarity_index(struct diff_filepair *p)
2050 return p->score * 100 / MAX_SCORE;
2053 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2055 /* Strip the prefix but do not molest /dev/null and absolute paths */
2056 if (*namep && **namep != '/')
2057 *namep += prefix_length;
2058 if (*otherp && **otherp != '/')
2059 *otherp += prefix_length;
2062 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2064 const char *pgm = external_diff();
2065 struct strbuf msg;
2066 char *xfrm_msg;
2067 struct diff_filespec *one = p->one;
2068 struct diff_filespec *two = p->two;
2069 const char *name;
2070 const char *other;
2071 const char *attr_path;
2072 int complete_rewrite = 0;
2074 name = p->one->path;
2075 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2076 attr_path = name;
2077 if (o->prefix_length)
2078 strip_prefix(o->prefix_length, &name, &other);
2080 if (DIFF_PAIR_UNMERGED(p)) {
2081 run_diff_cmd(pgm, name, NULL, attr_path,
2082 NULL, NULL, NULL, o, 0);
2083 return;
2086 diff_fill_sha1_info(one);
2087 diff_fill_sha1_info(two);
2089 strbuf_init(&msg, PATH_MAX * 2 + 300);
2090 switch (p->status) {
2091 case DIFF_STATUS_COPIED:
2092 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2093 strbuf_addstr(&msg, "\ncopy from ");
2094 quote_c_style(name, &msg, NULL, 0);
2095 strbuf_addstr(&msg, "\ncopy to ");
2096 quote_c_style(other, &msg, NULL, 0);
2097 strbuf_addch(&msg, '\n');
2098 break;
2099 case DIFF_STATUS_RENAMED:
2100 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2101 strbuf_addstr(&msg, "\nrename from ");
2102 quote_c_style(name, &msg, NULL, 0);
2103 strbuf_addstr(&msg, "\nrename to ");
2104 quote_c_style(other, &msg, NULL, 0);
2105 strbuf_addch(&msg, '\n');
2106 break;
2107 case DIFF_STATUS_MODIFIED:
2108 if (p->score) {
2109 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2110 similarity_index(p));
2111 complete_rewrite = 1;
2112 break;
2114 /* fallthru */
2115 default:
2116 /* nothing */
2120 if (hashcmp(one->sha1, two->sha1)) {
2121 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2123 if (DIFF_OPT_TST(o, BINARY)) {
2124 mmfile_t mf;
2125 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2126 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2127 abbrev = 40;
2129 strbuf_addf(&msg, "index %.*s..%.*s",
2130 abbrev, sha1_to_hex(one->sha1),
2131 abbrev, sha1_to_hex(two->sha1));
2132 if (one->mode == two->mode)
2133 strbuf_addf(&msg, " %06o", one->mode);
2134 strbuf_addch(&msg, '\n');
2137 if (msg.len)
2138 strbuf_setlen(&msg, msg.len - 1);
2139 xfrm_msg = msg.len ? msg.buf : NULL;
2141 if (!pgm &&
2142 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2143 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2144 /* a filepair that changes between file and symlink
2145 * needs to be split into deletion and creation.
2147 struct diff_filespec *null = alloc_filespec(two->path);
2148 run_diff_cmd(NULL, name, other, attr_path,
2149 one, null, xfrm_msg, o, 0);
2150 free(null);
2151 null = alloc_filespec(one->path);
2152 run_diff_cmd(NULL, name, other, attr_path,
2153 null, two, xfrm_msg, o, 0);
2154 free(null);
2156 else
2157 run_diff_cmd(pgm, name, other, attr_path,
2158 one, two, xfrm_msg, o, complete_rewrite);
2160 strbuf_release(&msg);
2163 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2164 struct diffstat_t *diffstat)
2166 const char *name;
2167 const char *other;
2168 int complete_rewrite = 0;
2170 if (DIFF_PAIR_UNMERGED(p)) {
2171 /* unmerged */
2172 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2173 return;
2176 name = p->one->path;
2177 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2179 if (o->prefix_length)
2180 strip_prefix(o->prefix_length, &name, &other);
2182 diff_fill_sha1_info(p->one);
2183 diff_fill_sha1_info(p->two);
2185 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2186 complete_rewrite = 1;
2187 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2190 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2192 const char *name;
2193 const char *other;
2194 const char *attr_path;
2196 if (DIFF_PAIR_UNMERGED(p)) {
2197 /* unmerged */
2198 return;
2201 name = p->one->path;
2202 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2203 attr_path = other ? other : name;
2205 if (o->prefix_length)
2206 strip_prefix(o->prefix_length, &name, &other);
2208 diff_fill_sha1_info(p->one);
2209 diff_fill_sha1_info(p->two);
2211 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2214 void diff_setup(struct diff_options *options)
2216 memset(options, 0, sizeof(*options));
2218 options->file = stdout;
2220 options->line_termination = '\n';
2221 options->break_opt = -1;
2222 options->rename_limit = -1;
2223 options->dirstat_percent = 3;
2224 DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2225 options->context = 3;
2227 options->change = diff_change;
2228 options->add_remove = diff_addremove;
2229 if (diff_use_color_default > 0)
2230 DIFF_OPT_SET(options, COLOR_DIFF);
2231 else
2232 DIFF_OPT_CLR(options, COLOR_DIFF);
2233 options->detect_rename = diff_detect_rename_default;
2235 if (!diff_mnemonic_prefix) {
2236 options->a_prefix = "a/";
2237 options->b_prefix = "b/";
2241 int diff_setup_done(struct diff_options *options)
2243 int count = 0;
2245 if (options->output_format & DIFF_FORMAT_NAME)
2246 count++;
2247 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2248 count++;
2249 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2250 count++;
2251 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2252 count++;
2253 if (count > 1)
2254 die("--name-only, --name-status, --check and -s are mutually exclusive");
2256 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2257 options->detect_rename = DIFF_DETECT_COPY;
2259 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2260 options->prefix = NULL;
2261 if (options->prefix)
2262 options->prefix_length = strlen(options->prefix);
2263 else
2264 options->prefix_length = 0;
2266 if (options->output_format & (DIFF_FORMAT_NAME |
2267 DIFF_FORMAT_NAME_STATUS |
2268 DIFF_FORMAT_CHECKDIFF |
2269 DIFF_FORMAT_NO_OUTPUT))
2270 options->output_format &= ~(DIFF_FORMAT_RAW |
2271 DIFF_FORMAT_NUMSTAT |
2272 DIFF_FORMAT_DIFFSTAT |
2273 DIFF_FORMAT_SHORTSTAT |
2274 DIFF_FORMAT_DIRSTAT |
2275 DIFF_FORMAT_SUMMARY |
2276 DIFF_FORMAT_PATCH);
2279 * These cases always need recursive; we do not drop caller-supplied
2280 * recursive bits for other formats here.
2282 if (options->output_format & (DIFF_FORMAT_PATCH |
2283 DIFF_FORMAT_NUMSTAT |
2284 DIFF_FORMAT_DIFFSTAT |
2285 DIFF_FORMAT_SHORTSTAT |
2286 DIFF_FORMAT_DIRSTAT |
2287 DIFF_FORMAT_SUMMARY |
2288 DIFF_FORMAT_CHECKDIFF))
2289 DIFF_OPT_SET(options, RECURSIVE);
2291 * Also pickaxe would not work very well if you do not say recursive
2293 if (options->pickaxe)
2294 DIFF_OPT_SET(options, RECURSIVE);
2296 if (options->detect_rename && options->rename_limit < 0)
2297 options->rename_limit = diff_rename_limit_default;
2298 if (options->setup & DIFF_SETUP_USE_CACHE) {
2299 if (!active_cache)
2300 /* read-cache does not die even when it fails
2301 * so it is safe for us to do this here. Also
2302 * it does not smudge active_cache or active_nr
2303 * when it fails, so we do not have to worry about
2304 * cleaning it up ourselves either.
2306 read_cache();
2308 if (options->abbrev <= 0 || 40 < options->abbrev)
2309 options->abbrev = 40; /* full */
2312 * It does not make sense to show the first hit we happened
2313 * to have found. It does not make sense not to return with
2314 * exit code in such a case either.
2316 if (DIFF_OPT_TST(options, QUIET)) {
2317 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2318 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2321 return 0;
2324 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2326 char c, *eq;
2327 int len;
2329 if (*arg != '-')
2330 return 0;
2331 c = *++arg;
2332 if (!c)
2333 return 0;
2334 if (c == arg_short) {
2335 c = *++arg;
2336 if (!c)
2337 return 1;
2338 if (val && isdigit(c)) {
2339 char *end;
2340 int n = strtoul(arg, &end, 10);
2341 if (*end)
2342 return 0;
2343 *val = n;
2344 return 1;
2346 return 0;
2348 if (c != '-')
2349 return 0;
2350 arg++;
2351 eq = strchr(arg, '=');
2352 if (eq)
2353 len = eq - arg;
2354 else
2355 len = strlen(arg);
2356 if (!len || strncmp(arg, arg_long, len))
2357 return 0;
2358 if (eq) {
2359 int n;
2360 char *end;
2361 if (!isdigit(*++eq))
2362 return 0;
2363 n = strtoul(eq, &end, 10);
2364 if (*end)
2365 return 0;
2366 *val = n;
2368 return 1;
2371 static int diff_scoreopt_parse(const char *opt);
2373 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2375 const char *arg = av[0];
2377 /* Output format options */
2378 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2379 options->output_format |= DIFF_FORMAT_PATCH;
2380 else if (opt_arg(arg, 'U', "unified", &options->context))
2381 options->output_format |= DIFF_FORMAT_PATCH;
2382 else if (!strcmp(arg, "--raw"))
2383 options->output_format |= DIFF_FORMAT_RAW;
2384 else if (!strcmp(arg, "--patch-with-raw"))
2385 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2386 else if (!strcmp(arg, "--numstat"))
2387 options->output_format |= DIFF_FORMAT_NUMSTAT;
2388 else if (!strcmp(arg, "--shortstat"))
2389 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2390 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2391 options->output_format |= DIFF_FORMAT_DIRSTAT;
2392 else if (!strcmp(arg, "--cumulative")) {
2393 options->output_format |= DIFF_FORMAT_DIRSTAT;
2394 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2395 } else if (opt_arg(arg, 0, "dirstat-by-file",
2396 &options->dirstat_percent)) {
2397 options->output_format |= DIFF_FORMAT_DIRSTAT;
2398 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2400 else if (!strcmp(arg, "--check"))
2401 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2402 else if (!strcmp(arg, "--summary"))
2403 options->output_format |= DIFF_FORMAT_SUMMARY;
2404 else if (!strcmp(arg, "--patch-with-stat"))
2405 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2406 else if (!strcmp(arg, "--name-only"))
2407 options->output_format |= DIFF_FORMAT_NAME;
2408 else if (!strcmp(arg, "--name-status"))
2409 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2410 else if (!strcmp(arg, "-s"))
2411 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2412 else if (!prefixcmp(arg, "--stat")) {
2413 char *end;
2414 int width = options->stat_width;
2415 int name_width = options->stat_name_width;
2416 arg += 6;
2417 end = (char *)arg;
2419 switch (*arg) {
2420 case '-':
2421 if (!prefixcmp(arg, "-width="))
2422 width = strtoul(arg + 7, &end, 10);
2423 else if (!prefixcmp(arg, "-name-width="))
2424 name_width = strtoul(arg + 12, &end, 10);
2425 break;
2426 case '=':
2427 width = strtoul(arg+1, &end, 10);
2428 if (*end == ',')
2429 name_width = strtoul(end+1, &end, 10);
2432 /* Important! This checks all the error cases! */
2433 if (*end)
2434 return 0;
2435 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2436 options->stat_name_width = name_width;
2437 options->stat_width = width;
2440 /* renames options */
2441 else if (!prefixcmp(arg, "-B")) {
2442 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2443 return -1;
2445 else if (!prefixcmp(arg, "-M")) {
2446 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2447 return -1;
2448 options->detect_rename = DIFF_DETECT_RENAME;
2450 else if (!prefixcmp(arg, "-C")) {
2451 if (options->detect_rename == DIFF_DETECT_COPY)
2452 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2453 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2454 return -1;
2455 options->detect_rename = DIFF_DETECT_COPY;
2457 else if (!strcmp(arg, "--no-renames"))
2458 options->detect_rename = 0;
2459 else if (!strcmp(arg, "--relative"))
2460 DIFF_OPT_SET(options, RELATIVE_NAME);
2461 else if (!prefixcmp(arg, "--relative=")) {
2462 DIFF_OPT_SET(options, RELATIVE_NAME);
2463 options->prefix = arg + 11;
2466 /* xdiff options */
2467 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2468 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2469 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2470 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2471 else if (!strcmp(arg, "--ignore-space-at-eol"))
2472 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2474 /* flags options */
2475 else if (!strcmp(arg, "--binary")) {
2476 options->output_format |= DIFF_FORMAT_PATCH;
2477 DIFF_OPT_SET(options, BINARY);
2479 else if (!strcmp(arg, "--full-index"))
2480 DIFF_OPT_SET(options, FULL_INDEX);
2481 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2482 DIFF_OPT_SET(options, TEXT);
2483 else if (!strcmp(arg, "-R"))
2484 DIFF_OPT_SET(options, REVERSE_DIFF);
2485 else if (!strcmp(arg, "--find-copies-harder"))
2486 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2487 else if (!strcmp(arg, "--follow"))
2488 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2489 else if (!strcmp(arg, "--color"))
2490 DIFF_OPT_SET(options, COLOR_DIFF);
2491 else if (!strcmp(arg, "--no-color"))
2492 DIFF_OPT_CLR(options, COLOR_DIFF);
2493 else if (!strcmp(arg, "--color-words"))
2494 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2495 else if (!strcmp(arg, "--exit-code"))
2496 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2497 else if (!strcmp(arg, "--quiet"))
2498 DIFF_OPT_SET(options, QUIET);
2499 else if (!strcmp(arg, "--ext-diff"))
2500 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2501 else if (!strcmp(arg, "--no-ext-diff"))
2502 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2503 else if (!strcmp(arg, "--textconv"))
2504 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2505 else if (!strcmp(arg, "--no-textconv"))
2506 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2507 else if (!strcmp(arg, "--ignore-submodules"))
2508 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2510 /* misc options */
2511 else if (!strcmp(arg, "-z"))
2512 options->line_termination = 0;
2513 else if (!prefixcmp(arg, "-l"))
2514 options->rename_limit = strtoul(arg+2, NULL, 10);
2515 else if (!prefixcmp(arg, "-S"))
2516 options->pickaxe = arg + 2;
2517 else if (!strcmp(arg, "--pickaxe-all"))
2518 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2519 else if (!strcmp(arg, "--pickaxe-regex"))
2520 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2521 else if (!prefixcmp(arg, "-O"))
2522 options->orderfile = arg + 2;
2523 else if (!prefixcmp(arg, "--diff-filter="))
2524 options->filter = arg + 14;
2525 else if (!strcmp(arg, "--abbrev"))
2526 options->abbrev = DEFAULT_ABBREV;
2527 else if (!prefixcmp(arg, "--abbrev=")) {
2528 options->abbrev = strtoul(arg + 9, NULL, 10);
2529 if (options->abbrev < MINIMUM_ABBREV)
2530 options->abbrev = MINIMUM_ABBREV;
2531 else if (40 < options->abbrev)
2532 options->abbrev = 40;
2534 else if (!prefixcmp(arg, "--src-prefix="))
2535 options->a_prefix = arg + 13;
2536 else if (!prefixcmp(arg, "--dst-prefix="))
2537 options->b_prefix = arg + 13;
2538 else if (!strcmp(arg, "--no-prefix"))
2539 options->a_prefix = options->b_prefix = "";
2540 else if (opt_arg(arg, '\0', "inter-hunk-context",
2541 &options->interhunkcontext))
2543 else if (!prefixcmp(arg, "--output=")) {
2544 options->file = fopen(arg + strlen("--output="), "w");
2545 options->close_file = 1;
2546 } else
2547 return 0;
2548 return 1;
2551 static int parse_num(const char **cp_p)
2553 unsigned long num, scale;
2554 int ch, dot;
2555 const char *cp = *cp_p;
2557 num = 0;
2558 scale = 1;
2559 dot = 0;
2560 for(;;) {
2561 ch = *cp;
2562 if ( !dot && ch == '.' ) {
2563 scale = 1;
2564 dot = 1;
2565 } else if ( ch == '%' ) {
2566 scale = dot ? scale*100 : 100;
2567 cp++; /* % is always at the end */
2568 break;
2569 } else if ( ch >= '0' && ch <= '9' ) {
2570 if ( scale < 100000 ) {
2571 scale *= 10;
2572 num = (num*10) + (ch-'0');
2574 } else {
2575 break;
2577 cp++;
2579 *cp_p = cp;
2581 /* user says num divided by scale and we say internally that
2582 * is MAX_SCORE * num / scale.
2584 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2587 static int diff_scoreopt_parse(const char *opt)
2589 int opt1, opt2, cmd;
2591 if (*opt++ != '-')
2592 return -1;
2593 cmd = *opt++;
2594 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2595 return -1; /* that is not a -M, -C nor -B option */
2597 opt1 = parse_num(&opt);
2598 if (cmd != 'B')
2599 opt2 = 0;
2600 else {
2601 if (*opt == 0)
2602 opt2 = 0;
2603 else if (*opt != '/')
2604 return -1; /* we expect -B80/99 or -B80 */
2605 else {
2606 opt++;
2607 opt2 = parse_num(&opt);
2610 if (*opt != 0)
2611 return -1;
2612 return opt1 | (opt2 << 16);
2615 struct diff_queue_struct diff_queued_diff;
2617 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2619 if (queue->alloc <= queue->nr) {
2620 queue->alloc = alloc_nr(queue->alloc);
2621 queue->queue = xrealloc(queue->queue,
2622 sizeof(dp) * queue->alloc);
2624 queue->queue[queue->nr++] = dp;
2627 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2628 struct diff_filespec *one,
2629 struct diff_filespec *two)
2631 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2632 dp->one = one;
2633 dp->two = two;
2634 if (queue)
2635 diff_q(queue, dp);
2636 return dp;
2639 void diff_free_filepair(struct diff_filepair *p)
2641 free_filespec(p->one);
2642 free_filespec(p->two);
2643 free(p);
2646 /* This is different from find_unique_abbrev() in that
2647 * it stuffs the result with dots for alignment.
2649 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2651 int abblen;
2652 const char *abbrev;
2653 if (len == 40)
2654 return sha1_to_hex(sha1);
2656 abbrev = find_unique_abbrev(sha1, len);
2657 abblen = strlen(abbrev);
2658 if (abblen < 37) {
2659 static char hex[41];
2660 if (len < abblen && abblen <= len + 2)
2661 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2662 else
2663 sprintf(hex, "%s...", abbrev);
2664 return hex;
2666 return sha1_to_hex(sha1);
2669 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2671 int line_termination = opt->line_termination;
2672 int inter_name_termination = line_termination ? '\t' : '\0';
2674 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2675 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2676 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2677 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2679 if (p->score) {
2680 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2681 inter_name_termination);
2682 } else {
2683 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2686 if (p->status == DIFF_STATUS_COPIED ||
2687 p->status == DIFF_STATUS_RENAMED) {
2688 const char *name_a, *name_b;
2689 name_a = p->one->path;
2690 name_b = p->two->path;
2691 strip_prefix(opt->prefix_length, &name_a, &name_b);
2692 write_name_quoted(name_a, opt->file, inter_name_termination);
2693 write_name_quoted(name_b, opt->file, line_termination);
2694 } else {
2695 const char *name_a, *name_b;
2696 name_a = p->one->mode ? p->one->path : p->two->path;
2697 name_b = NULL;
2698 strip_prefix(opt->prefix_length, &name_a, &name_b);
2699 write_name_quoted(name_a, opt->file, line_termination);
2703 int diff_unmodified_pair(struct diff_filepair *p)
2705 /* This function is written stricter than necessary to support
2706 * the currently implemented transformers, but the idea is to
2707 * let transformers to produce diff_filepairs any way they want,
2708 * and filter and clean them up here before producing the output.
2710 struct diff_filespec *one = p->one, *two = p->two;
2712 if (DIFF_PAIR_UNMERGED(p))
2713 return 0; /* unmerged is interesting */
2715 /* deletion, addition, mode or type change
2716 * and rename are all interesting.
2718 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2719 DIFF_PAIR_MODE_CHANGED(p) ||
2720 strcmp(one->path, two->path))
2721 return 0;
2723 /* both are valid and point at the same path. that is, we are
2724 * dealing with a change.
2726 if (one->sha1_valid && two->sha1_valid &&
2727 !hashcmp(one->sha1, two->sha1))
2728 return 1; /* no change */
2729 if (!one->sha1_valid && !two->sha1_valid)
2730 return 1; /* both look at the same file on the filesystem. */
2731 return 0;
2734 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
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_diff(p, o);
2746 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2747 struct diffstat_t *diffstat)
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_diffstat(p, o, diffstat);
2759 static void diff_flush_checkdiff(struct diff_filepair *p,
2760 struct diff_options *o)
2762 if (diff_unmodified_pair(p))
2763 return;
2765 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2766 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2767 return; /* no tree diffs in patch format */
2769 run_checkdiff(p, o);
2772 int diff_queue_is_empty(void)
2774 struct diff_queue_struct *q = &diff_queued_diff;
2775 int i;
2776 for (i = 0; i < q->nr; i++)
2777 if (!diff_unmodified_pair(q->queue[i]))
2778 return 0;
2779 return 1;
2782 #if DIFF_DEBUG
2783 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2785 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2786 x, one ? one : "",
2787 s->path,
2788 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2789 s->mode,
2790 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2791 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2792 x, one ? one : "",
2793 s->size, s->xfrm_flags);
2796 void diff_debug_filepair(const struct diff_filepair *p, int i)
2798 diff_debug_filespec(p->one, i, "one");
2799 diff_debug_filespec(p->two, i, "two");
2800 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2801 p->score, p->status ? p->status : '?',
2802 p->one->rename_used, p->broken_pair);
2805 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2807 int i;
2808 if (msg)
2809 fprintf(stderr, "%s\n", msg);
2810 fprintf(stderr, "q->nr = %d\n", q->nr);
2811 for (i = 0; i < q->nr; i++) {
2812 struct diff_filepair *p = q->queue[i];
2813 diff_debug_filepair(p, i);
2816 #endif
2818 static void diff_resolve_rename_copy(void)
2820 int i;
2821 struct diff_filepair *p;
2822 struct diff_queue_struct *q = &diff_queued_diff;
2824 diff_debug_queue("resolve-rename-copy", q);
2826 for (i = 0; i < q->nr; i++) {
2827 p = q->queue[i];
2828 p->status = 0; /* undecided */
2829 if (DIFF_PAIR_UNMERGED(p))
2830 p->status = DIFF_STATUS_UNMERGED;
2831 else if (!DIFF_FILE_VALID(p->one))
2832 p->status = DIFF_STATUS_ADDED;
2833 else if (!DIFF_FILE_VALID(p->two))
2834 p->status = DIFF_STATUS_DELETED;
2835 else if (DIFF_PAIR_TYPE_CHANGED(p))
2836 p->status = DIFF_STATUS_TYPE_CHANGED;
2838 /* from this point on, we are dealing with a pair
2839 * whose both sides are valid and of the same type, i.e.
2840 * either in-place edit or rename/copy edit.
2842 else if (DIFF_PAIR_RENAME(p)) {
2844 * A rename might have re-connected a broken
2845 * pair up, causing the pathnames to be the
2846 * same again. If so, that's not a rename at
2847 * all, just a modification..
2849 * Otherwise, see if this source was used for
2850 * multiple renames, in which case we decrement
2851 * the count, and call it a copy.
2853 if (!strcmp(p->one->path, p->two->path))
2854 p->status = DIFF_STATUS_MODIFIED;
2855 else if (--p->one->rename_used > 0)
2856 p->status = DIFF_STATUS_COPIED;
2857 else
2858 p->status = DIFF_STATUS_RENAMED;
2860 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2861 p->one->mode != p->two->mode ||
2862 is_null_sha1(p->one->sha1))
2863 p->status = DIFF_STATUS_MODIFIED;
2864 else {
2865 /* This is a "no-change" entry and should not
2866 * happen anymore, but prepare for broken callers.
2868 error("feeding unmodified %s to diffcore",
2869 p->one->path);
2870 p->status = DIFF_STATUS_UNKNOWN;
2873 diff_debug_queue("resolve-rename-copy done", q);
2876 static int check_pair_status(struct diff_filepair *p)
2878 switch (p->status) {
2879 case DIFF_STATUS_UNKNOWN:
2880 return 0;
2881 case 0:
2882 die("internal error in diff-resolve-rename-copy");
2883 default:
2884 return 1;
2888 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2890 int fmt = opt->output_format;
2892 if (fmt & DIFF_FORMAT_CHECKDIFF)
2893 diff_flush_checkdiff(p, opt);
2894 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2895 diff_flush_raw(p, opt);
2896 else if (fmt & DIFF_FORMAT_NAME) {
2897 const char *name_a, *name_b;
2898 name_a = p->two->path;
2899 name_b = NULL;
2900 strip_prefix(opt->prefix_length, &name_a, &name_b);
2901 write_name_quoted(name_a, opt->file, opt->line_termination);
2905 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2907 if (fs->mode)
2908 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2909 else
2910 fprintf(file, " %s ", newdelete);
2911 write_name_quoted(fs->path, file, '\n');
2915 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2917 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2918 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2919 show_name ? ' ' : '\n');
2920 if (show_name) {
2921 write_name_quoted(p->two->path, file, '\n');
2926 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2928 char *names = pprint_rename(p->one->path, p->two->path);
2930 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2931 free(names);
2932 show_mode_change(file, p, 0);
2935 static void diff_summary(FILE *file, struct diff_filepair *p)
2937 switch(p->status) {
2938 case DIFF_STATUS_DELETED:
2939 show_file_mode_name(file, "delete", p->one);
2940 break;
2941 case DIFF_STATUS_ADDED:
2942 show_file_mode_name(file, "create", p->two);
2943 break;
2944 case DIFF_STATUS_COPIED:
2945 show_rename_copy(file, "copy", p);
2946 break;
2947 case DIFF_STATUS_RENAMED:
2948 show_rename_copy(file, "rename", p);
2949 break;
2950 default:
2951 if (p->score) {
2952 fputs(" rewrite ", file);
2953 write_name_quoted(p->two->path, file, ' ');
2954 fprintf(file, "(%d%%)\n", similarity_index(p));
2956 show_mode_change(file, p, !p->score);
2957 break;
2961 struct patch_id_t {
2962 git_SHA_CTX *ctx;
2963 int patchlen;
2966 static int remove_space(char *line, int len)
2968 int i;
2969 char *dst = line;
2970 unsigned char c;
2972 for (i = 0; i < len; i++)
2973 if (!isspace((c = line[i])))
2974 *dst++ = c;
2976 return dst - line;
2979 static void patch_id_consume(void *priv, char *line, unsigned long len)
2981 struct patch_id_t *data = priv;
2982 int new_len;
2984 /* Ignore line numbers when computing the SHA1 of the patch */
2985 if (!prefixcmp(line, "@@ -"))
2986 return;
2988 new_len = remove_space(line, len);
2990 git_SHA1_Update(data->ctx, line, new_len);
2991 data->patchlen += new_len;
2994 /* returns 0 upon success, and writes result into sha1 */
2995 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2997 struct diff_queue_struct *q = &diff_queued_diff;
2998 int i;
2999 git_SHA_CTX ctx;
3000 struct patch_id_t data;
3001 char buffer[PATH_MAX * 4 + 20];
3003 git_SHA1_Init(&ctx);
3004 memset(&data, 0, sizeof(struct patch_id_t));
3005 data.ctx = &ctx;
3007 for (i = 0; i < q->nr; i++) {
3008 xpparam_t xpp;
3009 xdemitconf_t xecfg;
3010 xdemitcb_t ecb;
3011 mmfile_t mf1, mf2;
3012 struct diff_filepair *p = q->queue[i];
3013 int len1, len2;
3015 memset(&xpp, 0, sizeof(xpp));
3016 memset(&xecfg, 0, sizeof(xecfg));
3017 if (p->status == 0)
3018 return error("internal diff status error");
3019 if (p->status == DIFF_STATUS_UNKNOWN)
3020 continue;
3021 if (diff_unmodified_pair(p))
3022 continue;
3023 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3024 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3025 continue;
3026 if (DIFF_PAIR_UNMERGED(p))
3027 continue;
3029 diff_fill_sha1_info(p->one);
3030 diff_fill_sha1_info(p->two);
3031 if (fill_mmfile(&mf1, p->one) < 0 ||
3032 fill_mmfile(&mf2, p->two) < 0)
3033 return error("unable to read files to diff");
3035 len1 = remove_space(p->one->path, strlen(p->one->path));
3036 len2 = remove_space(p->two->path, strlen(p->two->path));
3037 if (p->one->mode == 0)
3038 len1 = snprintf(buffer, sizeof(buffer),
3039 "diff--gita/%.*sb/%.*s"
3040 "newfilemode%06o"
3041 "---/dev/null"
3042 "+++b/%.*s",
3043 len1, p->one->path,
3044 len2, p->two->path,
3045 p->two->mode,
3046 len2, p->two->path);
3047 else if (p->two->mode == 0)
3048 len1 = snprintf(buffer, sizeof(buffer),
3049 "diff--gita/%.*sb/%.*s"
3050 "deletedfilemode%06o"
3051 "---a/%.*s"
3052 "+++/dev/null",
3053 len1, p->one->path,
3054 len2, p->two->path,
3055 p->one->mode,
3056 len1, p->one->path);
3057 else
3058 len1 = snprintf(buffer, sizeof(buffer),
3059 "diff--gita/%.*sb/%.*s"
3060 "---a/%.*s"
3061 "+++b/%.*s",
3062 len1, p->one->path,
3063 len2, p->two->path,
3064 len1, p->one->path,
3065 len2, p->two->path);
3066 git_SHA1_Update(&ctx, buffer, len1);
3068 xpp.flags = XDF_NEED_MINIMAL;
3069 xecfg.ctxlen = 3;
3070 xecfg.flags = XDL_EMIT_FUNCNAMES;
3071 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3072 &xpp, &xecfg, &ecb);
3075 git_SHA1_Final(sha1, &ctx);
3076 return 0;
3079 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3081 struct diff_queue_struct *q = &diff_queued_diff;
3082 int i;
3083 int result = diff_get_patch_id(options, sha1);
3085 for (i = 0; i < q->nr; i++)
3086 diff_free_filepair(q->queue[i]);
3088 free(q->queue);
3089 q->queue = NULL;
3090 q->nr = q->alloc = 0;
3092 return result;
3095 static int is_summary_empty(const struct diff_queue_struct *q)
3097 int i;
3099 for (i = 0; i < q->nr; i++) {
3100 const struct diff_filepair *p = q->queue[i];
3102 switch (p->status) {
3103 case DIFF_STATUS_DELETED:
3104 case DIFF_STATUS_ADDED:
3105 case DIFF_STATUS_COPIED:
3106 case DIFF_STATUS_RENAMED:
3107 return 0;
3108 default:
3109 if (p->score)
3110 return 0;
3111 if (p->one->mode && p->two->mode &&
3112 p->one->mode != p->two->mode)
3113 return 0;
3114 break;
3117 return 1;
3120 void diff_flush(struct diff_options *options)
3122 struct diff_queue_struct *q = &diff_queued_diff;
3123 int i, output_format = options->output_format;
3124 int separator = 0;
3127 * Order: raw, stat, summary, patch
3128 * or: name/name-status/checkdiff (other bits clear)
3130 if (!q->nr)
3131 goto free_queue;
3133 if (output_format & (DIFF_FORMAT_RAW |
3134 DIFF_FORMAT_NAME |
3135 DIFF_FORMAT_NAME_STATUS |
3136 DIFF_FORMAT_CHECKDIFF)) {
3137 for (i = 0; i < q->nr; i++) {
3138 struct diff_filepair *p = q->queue[i];
3139 if (check_pair_status(p))
3140 flush_one_pair(p, options);
3142 separator++;
3145 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3146 struct diffstat_t diffstat;
3148 memset(&diffstat, 0, sizeof(struct diffstat_t));
3149 for (i = 0; i < q->nr; i++) {
3150 struct diff_filepair *p = q->queue[i];
3151 if (check_pair_status(p))
3152 diff_flush_stat(p, options, &diffstat);
3154 if (output_format & DIFF_FORMAT_NUMSTAT)
3155 show_numstat(&diffstat, options);
3156 if (output_format & DIFF_FORMAT_DIFFSTAT)
3157 show_stats(&diffstat, options);
3158 if (output_format & DIFF_FORMAT_SHORTSTAT)
3159 show_shortstats(&diffstat, options);
3160 free_diffstat_info(&diffstat);
3161 separator++;
3163 if (output_format & DIFF_FORMAT_DIRSTAT)
3164 show_dirstat(options);
3166 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3167 for (i = 0; i < q->nr; i++)
3168 diff_summary(options->file, q->queue[i]);
3169 separator++;
3172 if (output_format & DIFF_FORMAT_PATCH) {
3173 if (separator) {
3174 putc(options->line_termination, options->file);
3175 if (options->stat_sep) {
3176 /* attach patch instead of inline */
3177 fputs(options->stat_sep, options->file);
3181 for (i = 0; i < q->nr; i++) {
3182 struct diff_filepair *p = q->queue[i];
3183 if (check_pair_status(p))
3184 diff_flush_patch(p, options);
3188 if (output_format & DIFF_FORMAT_CALLBACK)
3189 options->format_callback(q, options, options->format_callback_data);
3191 for (i = 0; i < q->nr; i++)
3192 diff_free_filepair(q->queue[i]);
3193 free_queue:
3194 free(q->queue);
3195 q->queue = NULL;
3196 q->nr = q->alloc = 0;
3197 if (options->close_file)
3198 fclose(options->file);
3201 static void diffcore_apply_filter(const char *filter)
3203 int i;
3204 struct diff_queue_struct *q = &diff_queued_diff;
3205 struct diff_queue_struct outq;
3206 outq.queue = NULL;
3207 outq.nr = outq.alloc = 0;
3209 if (!filter)
3210 return;
3212 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3213 int found;
3214 for (i = found = 0; !found && i < q->nr; i++) {
3215 struct diff_filepair *p = q->queue[i];
3216 if (((p->status == DIFF_STATUS_MODIFIED) &&
3217 ((p->score &&
3218 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3219 (!p->score &&
3220 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3221 ((p->status != DIFF_STATUS_MODIFIED) &&
3222 strchr(filter, p->status)))
3223 found++;
3225 if (found)
3226 return;
3228 /* otherwise we will clear the whole queue
3229 * by copying the empty outq at the end of this
3230 * function, but first clear the current entries
3231 * in the queue.
3233 for (i = 0; i < q->nr; i++)
3234 diff_free_filepair(q->queue[i]);
3236 else {
3237 /* Only the matching ones */
3238 for (i = 0; i < q->nr; i++) {
3239 struct diff_filepair *p = q->queue[i];
3241 if (((p->status == DIFF_STATUS_MODIFIED) &&
3242 ((p->score &&
3243 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3244 (!p->score &&
3245 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3246 ((p->status != DIFF_STATUS_MODIFIED) &&
3247 strchr(filter, p->status)))
3248 diff_q(&outq, p);
3249 else
3250 diff_free_filepair(p);
3253 free(q->queue);
3254 *q = outq;
3257 /* Check whether two filespecs with the same mode and size are identical */
3258 static int diff_filespec_is_identical(struct diff_filespec *one,
3259 struct diff_filespec *two)
3261 if (S_ISGITLINK(one->mode))
3262 return 0;
3263 if (diff_populate_filespec(one, 0))
3264 return 0;
3265 if (diff_populate_filespec(two, 0))
3266 return 0;
3267 return !memcmp(one->data, two->data, one->size);
3270 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3272 int i;
3273 struct diff_queue_struct *q = &diff_queued_diff;
3274 struct diff_queue_struct outq;
3275 outq.queue = NULL;
3276 outq.nr = outq.alloc = 0;
3278 for (i = 0; i < q->nr; i++) {
3279 struct diff_filepair *p = q->queue[i];
3282 * 1. Entries that come from stat info dirtyness
3283 * always have both sides (iow, not create/delete),
3284 * one side of the object name is unknown, with
3285 * the same mode and size. Keep the ones that
3286 * do not match these criteria. They have real
3287 * differences.
3289 * 2. At this point, the file is known to be modified,
3290 * with the same mode and size, and the object
3291 * name of one side is unknown. Need to inspect
3292 * the identical contents.
3294 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3295 !DIFF_FILE_VALID(p->two) ||
3296 (p->one->sha1_valid && p->two->sha1_valid) ||
3297 (p->one->mode != p->two->mode) ||
3298 diff_populate_filespec(p->one, 1) ||
3299 diff_populate_filespec(p->two, 1) ||
3300 (p->one->size != p->two->size) ||
3301 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3302 diff_q(&outq, p);
3303 else {
3305 * The caller can subtract 1 from skip_stat_unmatch
3306 * to determine how many paths were dirty only
3307 * due to stat info mismatch.
3309 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3310 diffopt->skip_stat_unmatch++;
3311 diff_free_filepair(p);
3314 free(q->queue);
3315 *q = outq;
3318 void diffcore_std(struct diff_options *options)
3320 if (options->skip_stat_unmatch)
3321 diffcore_skip_stat_unmatch(options);
3322 if (options->break_opt != -1)
3323 diffcore_break(options->break_opt);
3324 if (options->detect_rename)
3325 diffcore_rename(options);
3326 if (options->break_opt != -1)
3327 diffcore_merge_broken();
3328 if (options->pickaxe)
3329 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3330 if (options->orderfile)
3331 diffcore_order(options->orderfile);
3332 diff_resolve_rename_copy();
3333 diffcore_apply_filter(options->filter);
3335 if (diff_queued_diff.nr)
3336 DIFF_OPT_SET(options, HAS_CHANGES);
3337 else
3338 DIFF_OPT_CLR(options, HAS_CHANGES);
3341 int diff_result_code(struct diff_options *opt, int status)
3343 int result = 0;
3344 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3345 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3346 return status;
3347 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3348 DIFF_OPT_TST(opt, HAS_CHANGES))
3349 result |= 01;
3350 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3351 DIFF_OPT_TST(opt, CHECK_FAILED))
3352 result |= 02;
3353 return result;
3356 void diff_addremove(struct diff_options *options,
3357 int addremove, unsigned mode,
3358 const unsigned char *sha1,
3359 const char *concatpath)
3361 struct diff_filespec *one, *two;
3363 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3364 return;
3366 /* This may look odd, but it is a preparation for
3367 * feeding "there are unchanged files which should
3368 * not produce diffs, but when you are doing copy
3369 * detection you would need them, so here they are"
3370 * entries to the diff-core. They will be prefixed
3371 * with something like '=' or '*' (I haven't decided
3372 * which but should not make any difference).
3373 * Feeding the same new and old to diff_change()
3374 * also has the same effect.
3375 * Before the final output happens, they are pruned after
3376 * merged into rename/copy pairs as appropriate.
3378 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3379 addremove = (addremove == '+' ? '-' :
3380 addremove == '-' ? '+' : addremove);
3382 if (options->prefix &&
3383 strncmp(concatpath, options->prefix, options->prefix_length))
3384 return;
3386 one = alloc_filespec(concatpath);
3387 two = alloc_filespec(concatpath);
3389 if (addremove != '+')
3390 fill_filespec(one, sha1, mode);
3391 if (addremove != '-')
3392 fill_filespec(two, sha1, mode);
3394 diff_queue(&diff_queued_diff, one, two);
3395 DIFF_OPT_SET(options, HAS_CHANGES);
3398 void diff_change(struct diff_options *options,
3399 unsigned old_mode, unsigned new_mode,
3400 const unsigned char *old_sha1,
3401 const unsigned char *new_sha1,
3402 const char *concatpath)
3404 struct diff_filespec *one, *two;
3406 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3407 && S_ISGITLINK(new_mode))
3408 return;
3410 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3411 unsigned tmp;
3412 const unsigned char *tmp_c;
3413 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3414 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3417 if (options->prefix &&
3418 strncmp(concatpath, options->prefix, options->prefix_length))
3419 return;
3421 one = alloc_filespec(concatpath);
3422 two = alloc_filespec(concatpath);
3423 fill_filespec(one, old_sha1, old_mode);
3424 fill_filespec(two, new_sha1, new_mode);
3426 diff_queue(&diff_queued_diff, one, two);
3427 DIFF_OPT_SET(options, HAS_CHANGES);
3430 void diff_unmerge(struct diff_options *options,
3431 const char *path,
3432 unsigned mode, const unsigned char *sha1)
3434 struct diff_filespec *one, *two;
3436 if (options->prefix &&
3437 strncmp(path, options->prefix, options->prefix_length))
3438 return;
3440 one = alloc_filespec(path);
3441 two = alloc_filespec(path);
3442 fill_filespec(one, sha1, mode);
3443 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3446 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3447 size_t *outsize)
3449 struct diff_tempfile temp;
3450 const char *argv[3];
3451 const char **arg = argv;
3452 struct child_process child;
3453 struct strbuf buf = STRBUF_INIT;
3455 prepare_temp_file(spec->path, &temp, spec);
3456 *arg++ = pgm;
3457 *arg++ = temp.name;
3458 *arg = NULL;
3460 memset(&child, 0, sizeof(child));
3461 child.argv = argv;
3462 child.out = -1;
3463 if (start_command(&child) != 0 ||
3464 strbuf_read(&buf, child.out, 0) < 0 ||
3465 finish_command(&child) != 0) {
3466 if (temp.name == temp.tmp_path)
3467 unlink(temp.name);
3468 error("error running textconv command '%s'", pgm);
3469 return NULL;
3471 if (temp.name == temp.tmp_path)
3472 unlink(temp.name);
3474 return strbuf_detach(&buf, outsize);