![[tokkee]](http://tokkee.org/images/avatar.png){kind=avatar}
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
15 #ifdef NO_FAST_WORKING_DIRECTORY
16 #define FAST_WORKING_DIRECTORY 0
17 #else
18 #define FAST_WORKING_DIRECTORY 1
19 #endif
21 static int diff_detect_rename_default;
22 static int diff_rename_limit_default = 200;
23 int diff_use_color_default = -1;
24 static const char *external_diff_cmd_cfg;
25 int diff_auto_refresh_index = 1;
27 static char diff_colors[][COLOR_MAXLEN] = {
28 "\033[m", /* reset */
29 "", /* PLAIN (normal) */
30 "\033[1m", /* METAINFO (bold) */
31 "\033[36m", /* FRAGINFO (cyan) */
32 "\033[31m", /* OLD (red) */
33 "\033[32m", /* NEW (green) */
34 "\033[33m", /* COMMIT (yellow) */
35 "\033[41m", /* WHITESPACE (red background) */
36 };
38 static int parse_diff_color_slot(const char *var, int ofs)
39 {
40 if (!strcasecmp(var+ofs, "plain"))
41 return DIFF_PLAIN;
42 if (!strcasecmp(var+ofs, "meta"))
43 return DIFF_METAINFO;
44 if (!strcasecmp(var+ofs, "frag"))
45 return DIFF_FRAGINFO;
46 if (!strcasecmp(var+ofs, "old"))
47 return DIFF_FILE_OLD;
48 if (!strcasecmp(var+ofs, "new"))
49 return DIFF_FILE_NEW;
50 if (!strcasecmp(var+ofs, "commit"))
51 return DIFF_COMMIT;
52 if (!strcasecmp(var+ofs, "whitespace"))
53 return DIFF_WHITESPACE;
54 die("bad config variable '%s'", var);
55 }
57 static struct ll_diff_driver {
58 const char *name;
59 struct ll_diff_driver *next;
60 const char *cmd;
61 } *user_diff, **user_diff_tail;
63 /*
64 * Currently there is only "diff.<drivername>.command" variable;
65 * because there are "diff.color.<slot>" variables, we are parsing
66 * this in a bit convoluted way to allow low level diff driver
67 * called "color".
68 */
69 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
70 {
71 const char *name;
72 int namelen;
73 struct ll_diff_driver *drv;
75 name = var + 5;
76 namelen = ep - name;
77 for (drv = user_diff; drv; drv = drv->next)
78 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
79 break;
80 if (!drv) {
81 drv = xcalloc(1, sizeof(struct ll_diff_driver));
82 drv->name = xmemdupz(name, namelen);
83 if (!user_diff_tail)
84 user_diff_tail = &user_diff;
85 *user_diff_tail = drv;
86 user_diff_tail = &(drv->next);
87 }
89 return git_config_string(&(drv->cmd), var, value);
90 }
92 /*
93 * 'diff.<what>.funcname' attribute can be specified in the configuration
94 * to define a customized regexp to find the beginning of a function to
95 * be used for hunk header lines of "diff -p" style output.
96 */
97 static struct funcname_pattern {
98 char *name;
99 char *pattern;
100 struct funcname_pattern *next;
101 } *funcname_pattern_list;
103 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
104 {
105 const char *name;
106 int namelen;
107 struct funcname_pattern *pp;
109 name = var + 5; /* "diff." */
110 namelen = ep - name;
112 for (pp = funcname_pattern_list; pp; pp = pp->next)
113 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
114 break;
115 if (!pp) {
116 pp = xcalloc(1, sizeof(*pp));
117 pp->name = xmemdupz(name, namelen);
118 pp->next = funcname_pattern_list;
119 funcname_pattern_list = pp;
120 }
121 free(pp->pattern);
122 pp->pattern = xstrdup(value);
123 return 0;
124 }
126 /*
127 * These are to give UI layer defaults.
128 * The core-level commands such as git-diff-files should
129 * never be affected by the setting of diff.renames
130 * the user happens to have in the configuration file.
131 */
132 int git_diff_ui_config(const char *var, const char *value, void *cb)
133 {
134 if (!strcmp(var, "diff.renamelimit")) {
135 diff_rename_limit_default = git_config_int(var, value);
136 return 0;
137 }
138 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
139 diff_use_color_default = git_config_colorbool(var, value, -1);
140 return 0;
141 }
142 if (!strcmp(var, "diff.renames")) {
143 if (!value)
144 diff_detect_rename_default = DIFF_DETECT_RENAME;
145 else if (!strcasecmp(value, "copies") ||
146 !strcasecmp(value, "copy"))
147 diff_detect_rename_default = DIFF_DETECT_COPY;
148 else if (git_config_bool(var,value))
149 diff_detect_rename_default = DIFF_DETECT_RENAME;
150 return 0;
151 }
152 if (!strcmp(var, "diff.autorefreshindex")) {
153 diff_auto_refresh_index = git_config_bool(var, value);
154 return 0;
155 }
156 if (!strcmp(var, "diff.external")) {
157 if (!value)
158 return config_error_nonbool(var);
159 external_diff_cmd_cfg = xstrdup(value);
160 return 0;
161 }
162 if (!prefixcmp(var, "diff.")) {
163 const char *ep = strrchr(var, '.');
165 if (ep != var + 4 && !strcmp(ep, ".command"))
166 return parse_lldiff_command(var, ep, value);
167 }
169 return git_diff_basic_config(var, value, cb);
170 }
172 int git_diff_basic_config(const char *var, const char *value, void *cb)
173 {
174 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
175 int slot = parse_diff_color_slot(var, 11);
176 if (!value)
177 return config_error_nonbool(var);
178 color_parse(value, var, diff_colors[slot]);
179 return 0;
180 }
182 if (!prefixcmp(var, "diff.")) {
183 const char *ep = strrchr(var, '.');
184 if (ep != var + 4) {
185 if (!strcmp(ep, ".funcname")) {
186 if (!value)
187 return config_error_nonbool(var);
188 return parse_funcname_pattern(var, ep, value);
189 }
190 }
191 }
193 return git_color_default_config(var, value, cb);
194 }
196 static char *quote_two(const char *one, const char *two)
197 {
198 int need_one = quote_c_style(one, NULL, NULL, 1);
199 int need_two = quote_c_style(two, NULL, NULL, 1);
200 struct strbuf res;
202 strbuf_init(&res, 0);
203 if (need_one + need_two) {
204 strbuf_addch(&res, '"');
205 quote_c_style(one, &res, NULL, 1);
206 quote_c_style(two, &res, NULL, 1);
207 strbuf_addch(&res, '"');
208 } else {
209 strbuf_addstr(&res, one);
210 strbuf_addstr(&res, two);
211 }
212 return strbuf_detach(&res, NULL);
213 }
215 static const char *external_diff(void)
216 {
217 static const char *external_diff_cmd = NULL;
218 static int done_preparing = 0;
220 if (done_preparing)
221 return external_diff_cmd;
222 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
223 if (!external_diff_cmd)
224 external_diff_cmd = external_diff_cmd_cfg;
225 done_preparing = 1;
226 return external_diff_cmd;
227 }
229 static struct diff_tempfile {
230 const char *name; /* filename external diff should read from */
231 char hex[41];
232 char mode[10];
233 char tmp_path[PATH_MAX];
234 } diff_temp[2];
236 static int count_lines(const char *data, int size)
237 {
238 int count, ch, completely_empty = 1, nl_just_seen = 0;
239 count = 0;
240 while (0 < size--) {
241 ch = *data++;
242 if (ch == '\n') {
243 count++;
244 nl_just_seen = 1;
245 completely_empty = 0;
246 }
247 else {
248 nl_just_seen = 0;
249 completely_empty = 0;
250 }
251 }
252 if (completely_empty)
253 return 0;
254 if (!nl_just_seen)
255 count++; /* no trailing newline */
256 return count;
257 }
259 static void print_line_count(FILE *file, int count)
260 {
261 switch (count) {
262 case 0:
263 fprintf(file, "0,0");
264 break;
265 case 1:
266 fprintf(file, "1");
267 break;
268 default:
269 fprintf(file, "1,%d", count);
270 break;
271 }
272 }
274 static void copy_file_with_prefix(FILE *file,
275 int prefix, const char *data, int size,
276 const char *set, const char *reset)
277 {
278 int ch, nl_just_seen = 1;
279 while (0 < size--) {
280 ch = *data++;
281 if (nl_just_seen) {
282 fputs(set, file);
283 putc(prefix, file);
284 }
285 if (ch == '\n') {
286 nl_just_seen = 1;
287 fputs(reset, file);
288 } else
289 nl_just_seen = 0;
290 putc(ch, file);
291 }
292 if (!nl_just_seen)
293 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
294 }
296 static void emit_rewrite_diff(const char *name_a,
297 const char *name_b,
298 struct diff_filespec *one,
299 struct diff_filespec *two,
300 struct diff_options *o)
301 {
302 int lc_a, lc_b;
303 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
304 const char *name_a_tab, *name_b_tab;
305 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
306 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
307 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
308 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
309 const char *reset = diff_get_color(color_diff, DIFF_RESET);
310 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
312 name_a += (*name_a == '/');
313 name_b += (*name_b == '/');
314 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
315 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
317 strbuf_reset(&a_name);
318 strbuf_reset(&b_name);
319 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
320 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
322 diff_populate_filespec(one, 0);
323 diff_populate_filespec(two, 0);
324 lc_a = count_lines(one->data, one->size);
325 lc_b = count_lines(two->data, two->size);
326 fprintf(o->file,
327 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
328 metainfo, a_name.buf, name_a_tab, reset,
329 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
330 print_line_count(o->file, lc_a);
331 fprintf(o->file, " +");
332 print_line_count(o->file, lc_b);
333 fprintf(o->file, " @@%s\n", reset);
334 if (lc_a)
335 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
336 if (lc_b)
337 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
338 }
340 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
341 {
342 if (!DIFF_FILE_VALID(one)) {
343 mf->ptr = (char *)""; /* does not matter */
344 mf->size = 0;
345 return 0;
346 }
347 else if (diff_populate_filespec(one, 0))
348 return -1;
349 mf->ptr = one->data;
350 mf->size = one->size;
351 return 0;
352 }
354 struct diff_words_buffer {
355 mmfile_t text;
356 long alloc;
357 long current; /* output pointer */
358 int suppressed_newline;
359 };
361 static void diff_words_append(char *line, unsigned long len,
362 struct diff_words_buffer *buffer)
363 {
364 if (buffer->text.size + len > buffer->alloc) {
365 buffer->alloc = (buffer->text.size + len) * 3 / 2;
366 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
367 }
368 line++;
369 len--;
370 memcpy(buffer->text.ptr + buffer->text.size, line, len);
371 buffer->text.size += len;
372 }
374 struct diff_words_data {
375 struct xdiff_emit_state xm;
376 struct diff_words_buffer minus, plus;
377 FILE *file;
378 };
380 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
381 int suppress_newline)
382 {
383 const char *ptr;
384 int eol = 0;
386 if (len == 0)
387 return;
389 ptr = buffer->text.ptr + buffer->current;
390 buffer->current += len;
392 if (ptr[len - 1] == '\n') {
393 eol = 1;
394 len--;
395 }
397 fputs(diff_get_color(1, color), file);
398 fwrite(ptr, len, 1, file);
399 fputs(diff_get_color(1, DIFF_RESET), file);
401 if (eol) {
402 if (suppress_newline)
403 buffer->suppressed_newline = 1;
404 else
405 putc('\n', file);
406 }
407 }
409 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
410 {
411 struct diff_words_data *diff_words = priv;
413 if (diff_words->minus.suppressed_newline) {
414 if (line[0] != '+')
415 putc('\n', diff_words->file);
416 diff_words->minus.suppressed_newline = 0;
417 }
419 len--;
420 switch (line[0]) {
421 case '-':
422 print_word(diff_words->file,
423 &diff_words->minus, len, DIFF_FILE_OLD, 1);
424 break;
425 case '+':
426 print_word(diff_words->file,
427 &diff_words->plus, len, DIFF_FILE_NEW, 0);
428 break;
429 case ' ':
430 print_word(diff_words->file,
431 &diff_words->plus, len, DIFF_PLAIN, 0);
432 diff_words->minus.current += len;
433 break;
434 }
435 }
437 /* this executes the word diff on the accumulated buffers */
438 static void diff_words_show(struct diff_words_data *diff_words)
439 {
440 xpparam_t xpp;
441 xdemitconf_t xecfg;
442 xdemitcb_t ecb;
443 mmfile_t minus, plus;
444 int i;
446 memset(&xecfg, 0, sizeof(xecfg));
447 minus.size = diff_words->minus.text.size;
448 minus.ptr = xmalloc(minus.size);
449 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
450 for (i = 0; i < minus.size; i++)
451 if (isspace(minus.ptr[i]))
452 minus.ptr[i] = '\n';
453 diff_words->minus.current = 0;
455 plus.size = diff_words->plus.text.size;
456 plus.ptr = xmalloc(plus.size);
457 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
458 for (i = 0; i < plus.size; i++)
459 if (isspace(plus.ptr[i]))
460 plus.ptr[i] = '\n';
461 diff_words->plus.current = 0;
463 xpp.flags = XDF_NEED_MINIMAL;
464 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
465 ecb.outf = xdiff_outf;
466 ecb.priv = diff_words;
467 diff_words->xm.consume = fn_out_diff_words_aux;
468 xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
470 free(minus.ptr);
471 free(plus.ptr);
472 diff_words->minus.text.size = diff_words->plus.text.size = 0;
474 if (diff_words->minus.suppressed_newline) {
475 putc('\n', diff_words->file);
476 diff_words->minus.suppressed_newline = 0;
477 }
478 }
480 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
482 struct emit_callback {
483 struct xdiff_emit_state xm;
484 int nparents, color_diff;
485 unsigned ws_rule;
486 sane_truncate_fn truncate;
487 const char **label_path;
488 struct diff_words_data *diff_words;
489 int *found_changesp;
490 FILE *file;
491 };
493 static void free_diff_words_data(struct emit_callback *ecbdata)
494 {
495 if (ecbdata->diff_words) {
496 /* flush buffers */
497 if (ecbdata->diff_words->minus.text.size ||
498 ecbdata->diff_words->plus.text.size)
499 diff_words_show(ecbdata->diff_words);
501 free (ecbdata->diff_words->minus.text.ptr);
502 free (ecbdata->diff_words->plus.text.ptr);
503 free(ecbdata->diff_words);
504 ecbdata->diff_words = NULL;
505 }
506 }
508 const char *diff_get_color(int diff_use_color, enum color_diff ix)
509 {
510 if (diff_use_color)
511 return diff_colors[ix];
512 return "";
513 }
515 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
516 {
517 int has_trailing_newline = (len > 0 && line[len-1] == '\n');
518 if (has_trailing_newline)
519 len--;
521 fputs(set, file);
522 fwrite(line, len, 1, file);
523 fputs(reset, file);
524 if (has_trailing_newline)
525 fputc('\n', file);
526 }
528 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
529 {
530 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
531 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
533 if (!*ws)
534 emit_line(ecbdata->file, set, reset, line, len);
535 else {
536 /* Emit just the prefix, then the rest. */
537 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
538 ws_check_emit(line + ecbdata->nparents,
539 len - ecbdata->nparents, ecbdata->ws_rule,
540 ecbdata->file, set, reset, ws);
541 }
542 }
544 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
545 {
546 const char *cp;
547 unsigned long allot;
548 size_t l = len;
550 if (ecb->truncate)
551 return ecb->truncate(line, len);
552 cp = line;
553 allot = l;
554 while (0 < l) {
555 (void) utf8_width(&cp, &l);
556 if (!cp)
557 break; /* truncated in the middle? */
558 }
559 return allot - l;
560 }
562 static void fn_out_consume(void *priv, char *line, unsigned long len)
563 {
564 int i;
565 int color;
566 struct emit_callback *ecbdata = priv;
567 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
568 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
569 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
571 *(ecbdata->found_changesp) = 1;
573 if (ecbdata->label_path[0]) {
574 const char *name_a_tab, *name_b_tab;
576 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
577 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
579 fprintf(ecbdata->file, "%s--- %s%s%s\n",
580 meta, ecbdata->label_path[0], reset, name_a_tab);
581 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
582 meta, ecbdata->label_path[1], reset, name_b_tab);
583 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
584 }
586 /* This is not really necessary for now because
587 * this codepath only deals with two-way diffs.
588 */
589 for (i = 0; i < len && line[i] == '@'; i++)
590 ;
591 if (2 <= i && i < len && line[i] == ' ') {
592 ecbdata->nparents = i - 1;
593 len = sane_truncate_line(ecbdata, line, len);
594 emit_line(ecbdata->file,
595 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
596 reset, line, len);
597 if (line[len-1] != '\n')
598 putc('\n', ecbdata->file);
599 return;
600 }
602 if (len < ecbdata->nparents) {
603 emit_line(ecbdata->file, reset, reset, line, len);
604 return;
605 }
607 color = DIFF_PLAIN;
608 if (ecbdata->diff_words && ecbdata->nparents != 1)
609 /* fall back to normal diff */
610 free_diff_words_data(ecbdata);
611 if (ecbdata->diff_words) {
612 if (line[0] == '-') {
613 diff_words_append(line, len,
614 &ecbdata->diff_words->minus);
615 return;
616 } else if (line[0] == '+') {
617 diff_words_append(line, len,
618 &ecbdata->diff_words->plus);
619 return;
620 }
621 if (ecbdata->diff_words->minus.text.size ||
622 ecbdata->diff_words->plus.text.size)
623 diff_words_show(ecbdata->diff_words);
624 line++;
625 len--;
626 emit_line(ecbdata->file, plain, reset, line, len);
627 return;
628 }
629 for (i = 0; i < ecbdata->nparents && len; i++) {
630 if (line[i] == '-')
631 color = DIFF_FILE_OLD;
632 else if (line[i] == '+')
633 color = DIFF_FILE_NEW;
634 }
636 if (color != DIFF_FILE_NEW) {
637 emit_line(ecbdata->file,
638 diff_get_color(ecbdata->color_diff, color),
639 reset, line, len);
640 return;
641 }
642 emit_add_line(reset, ecbdata, line, len);
643 }
645 static char *pprint_rename(const char *a, const char *b)
646 {
647 const char *old = a;
648 const char *new = b;
649 struct strbuf name;
650 int pfx_length, sfx_length;
651 int len_a = strlen(a);
652 int len_b = strlen(b);
653 int a_midlen, b_midlen;
654 int qlen_a = quote_c_style(a, NULL, NULL, 0);
655 int qlen_b = quote_c_style(b, NULL, NULL, 0);
657 strbuf_init(&name, 0);
658 if (qlen_a || qlen_b) {
659 quote_c_style(a, &name, NULL, 0);
660 strbuf_addstr(&name, " => ");
661 quote_c_style(b, &name, NULL, 0);
662 return strbuf_detach(&name, NULL);
663 }
665 /* Find common prefix */
666 pfx_length = 0;
667 while (*old && *new && *old == *new) {
668 if (*old == '/')
669 pfx_length = old - a + 1;
670 old++;
671 new++;
672 }
674 /* Find common suffix */
675 old = a + len_a;
676 new = b + len_b;
677 sfx_length = 0;
678 while (a <= old && b <= new && *old == *new) {
679 if (*old == '/')
680 sfx_length = len_a - (old - a);
681 old--;
682 new--;
683 }
685 /*
686 * pfx{mid-a => mid-b}sfx
687 * {pfx-a => pfx-b}sfx
688 * pfx{sfx-a => sfx-b}
689 * name-a => name-b
690 */
691 a_midlen = len_a - pfx_length - sfx_length;
692 b_midlen = len_b - pfx_length - sfx_length;
693 if (a_midlen < 0)
694 a_midlen = 0;
695 if (b_midlen < 0)
696 b_midlen = 0;
698 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
699 if (pfx_length + sfx_length) {
700 strbuf_add(&name, a, pfx_length);
701 strbuf_addch(&name, '{');
702 }
703 strbuf_add(&name, a + pfx_length, a_midlen);
704 strbuf_addstr(&name, " => ");
705 strbuf_add(&name, b + pfx_length, b_midlen);
706 if (pfx_length + sfx_length) {
707 strbuf_addch(&name, '}');
708 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
709 }
710 return strbuf_detach(&name, NULL);
711 }
713 struct diffstat_t {
714 struct xdiff_emit_state xm;
716 int nr;
717 int alloc;
718 struct diffstat_file {
719 char *from_name;
720 char *name;
721 char *print_name;
722 unsigned is_unmerged:1;
723 unsigned is_binary:1;
724 unsigned is_renamed:1;
725 unsigned int added, deleted;
726 } **files;
727 };
729 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
730 const char *name_a,
731 const char *name_b)
732 {
733 struct diffstat_file *x;
734 x = xcalloc(sizeof (*x), 1);
735 if (diffstat->nr == diffstat->alloc) {
736 diffstat->alloc = alloc_nr(diffstat->alloc);
737 diffstat->files = xrealloc(diffstat->files,
738 diffstat->alloc * sizeof(x));
739 }
740 diffstat->files[diffstat->nr++] = x;
741 if (name_b) {
742 x->from_name = xstrdup(name_a);
743 x->name = xstrdup(name_b);
744 x->is_renamed = 1;
745 }
746 else {
747 x->from_name = NULL;
748 x->name = xstrdup(name_a);
749 }
750 return x;
751 }
753 static void diffstat_consume(void *priv, char *line, unsigned long len)
754 {
755 struct diffstat_t *diffstat = priv;
756 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
758 if (line[0] == '+')
759 x->added++;
760 else if (line[0] == '-')
761 x->deleted++;
762 }
764 const char mime_boundary_leader[] = "------------";
766 static int scale_linear(int it, int width, int max_change)
767 {
768 /*
769 * make sure that at least one '-' is printed if there were deletions,
770 * and likewise for '+'.
771 */
772 if (max_change < 2)
773 return it;
774 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
775 }
777 static void show_name(FILE *file,
778 const char *prefix, const char *name, int len,
779 const char *reset, const char *set)
780 {
781 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
782 }
784 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
785 {
786 if (cnt <= 0)
787 return;
788 fprintf(file, "%s", set);
789 while (cnt--)
790 putc(ch, file);
791 fprintf(file, "%s", reset);
792 }
794 static void fill_print_name(struct diffstat_file *file)
795 {
796 char *pname;
798 if (file->print_name)
799 return;
801 if (!file->is_renamed) {
802 struct strbuf buf;
803 strbuf_init(&buf, 0);
804 if (quote_c_style(file->name, &buf, NULL, 0)) {
805 pname = strbuf_detach(&buf, NULL);
806 } else {
807 pname = file->name;
808 strbuf_release(&buf);
809 }
810 } else {
811 pname = pprint_rename(file->from_name, file->name);
812 }
813 file->print_name = pname;
814 }
816 static void show_stats(struct diffstat_t* data, struct diff_options *options)
817 {
818 int i, len, add, del, total, adds = 0, dels = 0;
819 int max_change = 0, max_len = 0;
820 int total_files = data->nr;
821 int width, name_width;
822 const char *reset, *set, *add_c, *del_c;
824 if (data->nr == 0)
825 return;
827 width = options->stat_width ? options->stat_width : 80;
828 name_width = options->stat_name_width ? options->stat_name_width : 50;
830 /* Sanity: give at least 5 columns to the graph,
831 * but leave at least 10 columns for the name.
832 */
833 if (width < name_width + 15) {
834 if (name_width <= 25)
835 width = name_width + 15;
836 else
837 name_width = width - 15;
838 }
840 /* Find the longest filename and max number of changes */
841 reset = diff_get_color_opt(options, DIFF_RESET);
842 set = diff_get_color_opt(options, DIFF_PLAIN);
843 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
844 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
846 for (i = 0; i < data->nr; i++) {
847 struct diffstat_file *file = data->files[i];
848 int change = file->added + file->deleted;
849 fill_print_name(file);
850 len = strlen(file->print_name);
851 if (max_len < len)
852 max_len = len;
854 if (file->is_binary || file->is_unmerged)
855 continue;
856 if (max_change < change)
857 max_change = change;
858 }
860 /* Compute the width of the graph part;
861 * 10 is for one blank at the beginning of the line plus
862 * " | count " between the name and the graph.
863 *
864 * From here on, name_width is the width of the name area,
865 * and width is the width of the graph area.
866 */
867 name_width = (name_width < max_len) ? name_width : max_len;
868 if (width < (name_width + 10) + max_change)
869 width = width - (name_width + 10);
870 else
871 width = max_change;
873 for (i = 0; i < data->nr; i++) {
874 const char *prefix = "";
875 char *name = data->files[i]->print_name;
876 int added = data->files[i]->added;
877 int deleted = data->files[i]->deleted;
878 int name_len;
880 /*
881 * "scale" the filename
882 */
883 len = name_width;
884 name_len = strlen(name);
885 if (name_width < name_len) {
886 char *slash;
887 prefix = "...";
888 len -= 3;
889 name += name_len - len;
890 slash = strchr(name, '/');
891 if (slash)
892 name = slash;
893 }
895 if (data->files[i]->is_binary) {
896 show_name(options->file, prefix, name, len, reset, set);
897 fprintf(options->file, " Bin ");
898 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
899 fprintf(options->file, " -> ");
900 fprintf(options->file, "%s%d%s", add_c, added, reset);
901 fprintf(options->file, " bytes");
902 fprintf(options->file, "\n");
903 continue;
904 }
905 else if (data->files[i]->is_unmerged) {
906 show_name(options->file, prefix, name, len, reset, set);
907 fprintf(options->file, " Unmerged\n");
908 continue;
909 }
910 else if (!data->files[i]->is_renamed &&
911 (added + deleted == 0)) {
912 total_files--;
913 continue;
914 }
916 /*
917 * scale the add/delete
918 */
919 add = added;
920 del = deleted;
921 total = add + del;
922 adds += add;
923 dels += del;
925 if (width <= max_change) {
926 add = scale_linear(add, width, max_change);
927 del = scale_linear(del, width, max_change);
928 total = add + del;
929 }
930 show_name(options->file, prefix, name, len, reset, set);
931 fprintf(options->file, "%5d%s", added + deleted,
932 added + deleted ? " " : "");
933 show_graph(options->file, '+', add, add_c, reset);
934 show_graph(options->file, '-', del, del_c, reset);
935 fprintf(options->file, "\n");
936 }
937 fprintf(options->file,
938 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
939 set, total_files, adds, dels, reset);
940 }
942 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
943 {
944 int i, adds = 0, dels = 0, total_files = data->nr;
946 if (data->nr == 0)
947 return;
949 for (i = 0; i < data->nr; i++) {
950 if (!data->files[i]->is_binary &&
951 !data->files[i]->is_unmerged) {
952 int added = data->files[i]->added;
953 int deleted= data->files[i]->deleted;
954 if (!data->files[i]->is_renamed &&
955 (added + deleted == 0)) {
956 total_files--;
957 } else {
958 adds += added;
959 dels += deleted;
960 }
961 }
962 }
963 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
964 total_files, adds, dels);
965 }
967 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
968 {
969 int i;
971 if (data->nr == 0)
972 return;
974 for (i = 0; i < data->nr; i++) {
975 struct diffstat_file *file = data->files[i];
977 if (file->is_binary)
978 fprintf(options->file, "-\t-\t");
979 else
980 fprintf(options->file,
981 "%d\t%d\t", file->added, file->deleted);
982 if (options->line_termination) {
983 fill_print_name(file);
984 if (!file->is_renamed)
985 write_name_quoted(file->name, options->file,
986 options->line_termination);
987 else {
988 fputs(file->print_name, options->file);
989 putc(options->line_termination, options->file);
990 }
991 } else {
992 if (file->is_renamed) {
993 putc('\0', options->file);
994 write_name_quoted(file->from_name, options->file, '\0');
995 }
996 write_name_quoted(file->name, options->file, '\0');
997 }
998 }
999 }
1001 struct dirstat_file {
1002 const char *name;
1003 unsigned long changed;
1004 };
1006 struct dirstat_dir {
1007 struct dirstat_file *files;
1008 int alloc, nr, percent, cumulative;
1009 };
1011 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1012 {
1013 unsigned long this_dir = 0;
1014 unsigned int sources = 0;
1016 while (dir->nr) {
1017 struct dirstat_file *f = dir->files;
1018 int namelen = strlen(f->name);
1019 unsigned long this;
1020 char *slash;
1022 if (namelen < baselen)
1023 break;
1024 if (memcmp(f->name, base, baselen))
1025 break;
1026 slash = strchr(f->name + baselen, '/');
1027 if (slash) {
1028 int newbaselen = slash + 1 - f->name;
1029 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1030 sources++;
1031 } else {
1032 this = f->changed;
1033 dir->files++;
1034 dir->nr--;
1035 sources += 2;
1036 }
1037 this_dir += this;
1038 }
1040 /*
1041 * We don't report dirstat's for
1042 * - the top level
1043 * - or cases where everything came from a single directory
1044 * under this directory (sources == 1).
1045 */
1046 if (baselen && sources != 1) {
1047 int permille = this_dir * 1000 / changed;
1048 if (permille) {
1049 int percent = permille / 10;
1050 if (percent >= dir->percent) {
1051 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1052 if (!dir->cumulative)
1053 return 0;
1054 }
1055 }
1056 }
1057 return this_dir;
1058 }
1060 static void show_dirstat(struct diff_options *options)
1061 {
1062 int i;
1063 unsigned long changed;
1064 struct dirstat_dir dir;
1065 struct diff_queue_struct *q = &diff_queued_diff;
1067 dir.files = NULL;
1068 dir.alloc = 0;
1069 dir.nr = 0;
1070 dir.percent = options->dirstat_percent;
1071 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1073 changed = 0;
1074 for (i = 0; i < q->nr; i++) {
1075 struct diff_filepair *p = q->queue[i];
1076 const char *name;
1077 unsigned long copied, added, damage;
1079 name = p->one->path ? p->one->path : p->two->path;
1081 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1082 diff_populate_filespec(p->one, 0);
1083 diff_populate_filespec(p->two, 0);
1084 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1085 &copied, &added);
1086 diff_free_filespec_data(p->one);
1087 diff_free_filespec_data(p->two);
1088 } else if (DIFF_FILE_VALID(p->one)) {
1089 diff_populate_filespec(p->one, 1);
1090 copied = added = 0;
1091 diff_free_filespec_data(p->one);
1092 } else if (DIFF_FILE_VALID(p->two)) {
1093 diff_populate_filespec(p->two, 1);
1094 copied = 0;
1095 added = p->two->size;
1096 diff_free_filespec_data(p->two);
1097 } else
1098 continue;
1100 /*
1101 * Original minus copied is the removed material,
1102 * added is the new material. They are both damages
1103 * made to the preimage.
1104 */
1105 damage = (p->one->size - copied) + added;
1107 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1108 dir.files[dir.nr].name = name;
1109 dir.files[dir.nr].changed = damage;
1110 changed += damage;
1111 dir.nr++;
1112 }
1114 /* This can happen even with many files, if everything was renames */
1115 if (!changed)
1116 return;
1118 /* Show all directories with more than x% of the changes */
1119 gather_dirstat(options->file, &dir, changed, "", 0);
1120 }
1122 static void free_diffstat_info(struct diffstat_t *diffstat)
1123 {
1124 int i;
1125 for (i = 0; i < diffstat->nr; i++) {
1126 struct diffstat_file *f = diffstat->files[i];
1127 if (f->name != f->print_name)
1128 free(f->print_name);
1129 free(f->name);
1130 free(f->from_name);
1131 free(f);
1132 }
1133 free(diffstat->files);
1134 }
1136 struct checkdiff_t {
1137 struct xdiff_emit_state xm;
1138 const char *filename;
1139 int lineno, color_diff;
1140 unsigned ws_rule;
1141 unsigned status;
1142 FILE *file;
1143 };
1145 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1146 {
1147 struct checkdiff_t *data = priv;
1148 const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1149 const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1150 const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1151 char *err;
1153 if (line[0] == '+') {
1154 unsigned bad;
1155 data->lineno++;
1156 bad = ws_check(line + 1, len - 1, data->ws_rule);
1157 if (!bad)
1158 return;
1159 data->status |= bad;
1160 err = whitespace_error_string(bad);
1161 fprintf(data->file, "%s:%d: %s.\n", data->filename, data->lineno, err);
1162 free(err);
1163 emit_line(data->file, set, reset, line, 1);
1164 ws_check_emit(line + 1, len - 1, data->ws_rule,
1165 data->file, set, reset, ws);
1166 } else if (line[0] == ' ')
1167 data->lineno++;
1168 else if (line[0] == '@') {
1169 char *plus = strchr(line, '+');
1170 if (plus)
1171 data->lineno = strtol(plus, NULL, 10) - 1;
1172 else
1173 die("invalid diff");
1174 }
1175 }
1177 static unsigned char *deflate_it(char *data,
1178 unsigned long size,
1179 unsigned long *result_size)
1180 {
1181 int bound;
1182 unsigned char *deflated;
1183 z_stream stream;
1185 memset(&stream, 0, sizeof(stream));
1186 deflateInit(&stream, zlib_compression_level);
1187 bound = deflateBound(&stream, size);
1188 deflated = xmalloc(bound);
1189 stream.next_out = deflated;
1190 stream.avail_out = bound;
1192 stream.next_in = (unsigned char *)data;
1193 stream.avail_in = size;
1194 while (deflate(&stream, Z_FINISH) == Z_OK)
1195 ; /* nothing */
1196 deflateEnd(&stream);
1197 *result_size = stream.total_out;
1198 return deflated;
1199 }
1201 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1202 {
1203 void *cp;
1204 void *delta;
1205 void *deflated;
1206 void *data;
1207 unsigned long orig_size;
1208 unsigned long delta_size;
1209 unsigned long deflate_size;
1210 unsigned long data_size;
1212 /* We could do deflated delta, or we could do just deflated two,
1213 * whichever is smaller.
1214 */
1215 delta = NULL;
1216 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1217 if (one->size && two->size) {
1218 delta = diff_delta(one->ptr, one->size,
1219 two->ptr, two->size,
1220 &delta_size, deflate_size);
1221 if (delta) {
1222 void *to_free = delta;
1223 orig_size = delta_size;
1224 delta = deflate_it(delta, delta_size, &delta_size);
1225 free(to_free);
1226 }
1227 }
1229 if (delta && delta_size < deflate_size) {
1230 fprintf(file, "delta %lu\n", orig_size);
1231 free(deflated);
1232 data = delta;
1233 data_size = delta_size;
1234 }
1235 else {
1236 fprintf(file, "literal %lu\n", two->size);
1237 free(delta);
1238 data = deflated;
1239 data_size = deflate_size;
1240 }
1242 /* emit data encoded in base85 */
1243 cp = data;
1244 while (data_size) {
1245 int bytes = (52 < data_size) ? 52 : data_size;
1246 char line[70];
1247 data_size -= bytes;
1248 if (bytes <= 26)
1249 line[0] = bytes + 'A' - 1;
1250 else
1251 line[0] = bytes - 26 + 'a' - 1;
1252 encode_85(line + 1, cp, bytes);
1253 cp = (char *) cp + bytes;
1254 fputs(line, file);
1255 fputc('\n', file);
1256 }
1257 fprintf(file, "\n");
1258 free(data);
1259 }
1261 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1262 {
1263 fprintf(file, "GIT binary patch\n");
1264 emit_binary_diff_body(file, one, two);
1265 emit_binary_diff_body(file, two, one);
1266 }
1268 static void setup_diff_attr_check(struct git_attr_check *check)
1269 {
1270 static struct git_attr *attr_diff;
1272 if (!attr_diff) {
1273 attr_diff = git_attr("diff", 4);
1274 }
1275 check[0].attr = attr_diff;
1276 }
1278 static void diff_filespec_check_attr(struct diff_filespec *one)
1279 {
1280 struct git_attr_check attr_diff_check;
1281 int check_from_data = 0;
1283 if (one->checked_attr)
1284 return;
1286 setup_diff_attr_check(&attr_diff_check);
1287 one->is_binary = 0;
1288 one->funcname_pattern_ident = NULL;
1290 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1291 const char *value;
1293 /* binaryness */
1294 value = attr_diff_check.value;
1295 if (ATTR_TRUE(value))
1296 ;
1297 else if (ATTR_FALSE(value))
1298 one->is_binary = 1;
1299 else
1300 check_from_data = 1;
1302 /* funcname pattern ident */
1303 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1304 ;
1305 else
1306 one->funcname_pattern_ident = value;
1307 }
1309 if (check_from_data) {
1310 if (!one->data && DIFF_FILE_VALID(one))
1311 diff_populate_filespec(one, 0);
1313 if (one->data)
1314 one->is_binary = buffer_is_binary(one->data, one->size);
1315 }
1316 }
1318 int diff_filespec_is_binary(struct diff_filespec *one)
1319 {
1320 diff_filespec_check_attr(one);
1321 return one->is_binary;
1322 }
1324 static const char *funcname_pattern(const char *ident)
1325 {
1326 struct funcname_pattern *pp;
1328 for (pp = funcname_pattern_list; pp; pp = pp->next)
1329 if (!strcmp(ident, pp->name))
1330 return pp->pattern;
1331 return NULL;
1332 }
1334 static struct builtin_funcname_pattern {
1335 const char *name;
1336 const char *pattern;
1337 } builtin_funcname_pattern[] = {
1338 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1339 "new\\|return\\|switch\\|throw\\|while\\)\n"
1340 "^[ ]*\\(\\([ ]*"
1341 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1342 "[ ]*([^;]*\\)$" },
1343 { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1344 };
1346 static const char *diff_funcname_pattern(struct diff_filespec *one)
1347 {
1348 const char *ident, *pattern;
1349 int i;
1351 diff_filespec_check_attr(one);
1352 ident = one->funcname_pattern_ident;
1354 if (!ident)
1355 /*
1356 * If the config file has "funcname.default" defined, that
1357 * regexp is used; otherwise NULL is returned and xemit uses
1358 * the built-in default.
1359 */
1360 return funcname_pattern("default");
1362 /* Look up custom "funcname.$ident" regexp from config. */
1363 pattern = funcname_pattern(ident);
1364 if (pattern)
1365 return pattern;
1367 /*
1368 * And define built-in fallback patterns here. Note that
1369 * these can be overridden by the user's config settings.
1370 */
1371 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1372 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1373 return builtin_funcname_pattern[i].pattern;
1375 return NULL;
1376 }
1378 static void builtin_diff(const char *name_a,
1379 const char *name_b,
1380 struct diff_filespec *one,
1381 struct diff_filespec *two,
1382 const char *xfrm_msg,
1383 struct diff_options *o,
1384 int complete_rewrite)
1385 {
1386 mmfile_t mf1, mf2;
1387 const char *lbl[2];
1388 char *a_one, *b_two;
1389 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1390 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1392 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1393 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1394 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1395 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1396 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1397 if (lbl[0][0] == '/') {
1398 /* /dev/null */
1399 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1400 if (xfrm_msg && xfrm_msg[0])
1401 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1402 }
1403 else if (lbl[1][0] == '/') {
1404 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1405 if (xfrm_msg && xfrm_msg[0])
1406 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1407 }
1408 else {
1409 if (one->mode != two->mode) {
1410 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1411 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1412 }
1413 if (xfrm_msg && xfrm_msg[0])
1414 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1415 /*
1416 * we do not run diff between different kind
1417 * of objects.
1418 */
1419 if ((one->mode ^ two->mode) & S_IFMT)
1420 goto free_ab_and_return;
1421 if (complete_rewrite) {
1422 emit_rewrite_diff(name_a, name_b, one, two, o);
1423 o->found_changes = 1;
1424 goto free_ab_and_return;
1425 }
1426 }
1428 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1429 die("unable to read files to diff");
1431 if (!DIFF_OPT_TST(o, TEXT) &&
1432 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1433 /* Quite common confusing case */
1434 if (mf1.size == mf2.size &&
1435 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1436 goto free_ab_and_return;
1437 if (DIFF_OPT_TST(o, BINARY))
1438 emit_binary_diff(o->file, &mf1, &mf2);
1439 else
1440 fprintf(o->file, "Binary files %s and %s differ\n",
1441 lbl[0], lbl[1]);
1442 o->found_changes = 1;
1443 }
1444 else {
1445 /* Crazy xdl interfaces.. */
1446 const char *diffopts = getenv("GIT_DIFF_OPTS");
1447 xpparam_t xpp;
1448 xdemitconf_t xecfg;
1449 xdemitcb_t ecb;
1450 struct emit_callback ecbdata;
1451 const char *funcname_pattern;
1453 funcname_pattern = diff_funcname_pattern(one);
1454 if (!funcname_pattern)
1455 funcname_pattern = diff_funcname_pattern(two);
1457 memset(&xecfg, 0, sizeof(xecfg));
1458 memset(&ecbdata, 0, sizeof(ecbdata));
1459 ecbdata.label_path = lbl;
1460 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1461 ecbdata.found_changesp = &o->found_changes;
1462 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1463 ecbdata.file = o->file;
1464 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1465 xecfg.ctxlen = o->context;
1466 xecfg.flags = XDL_EMIT_FUNCNAMES;
1467 if (funcname_pattern)
1468 xdiff_set_find_func(&xecfg, funcname_pattern);
1469 if (!diffopts)
1470 ;
1471 else if (!prefixcmp(diffopts, "--unified="))
1472 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1473 else if (!prefixcmp(diffopts, "-u"))
1474 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1475 ecb.outf = xdiff_outf;
1476 ecb.priv = &ecbdata;
1477 ecbdata.xm.consume = fn_out_consume;
1478 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1479 ecbdata.diff_words =
1480 xcalloc(1, sizeof(struct diff_words_data));
1481 ecbdata.diff_words->file = o->file;
1482 }
1483 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1484 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1485 free_diff_words_data(&ecbdata);
1486 }
1488 free_ab_and_return:
1489 diff_free_filespec_data(one);
1490 diff_free_filespec_data(two);
1491 free(a_one);
1492 free(b_two);
1493 return;
1494 }
1496 static void builtin_diffstat(const char *name_a, const char *name_b,
1497 struct diff_filespec *one,
1498 struct diff_filespec *two,
1499 struct diffstat_t *diffstat,
1500 struct diff_options *o,
1501 int complete_rewrite)
1502 {
1503 mmfile_t mf1, mf2;
1504 struct diffstat_file *data;
1506 data = diffstat_add(diffstat, name_a, name_b);
1508 if (!one || !two) {
1509 data->is_unmerged = 1;
1510 return;
1511 }
1512 if (complete_rewrite) {
1513 diff_populate_filespec(one, 0);
1514 diff_populate_filespec(two, 0);
1515 data->deleted = count_lines(one->data, one->size);
1516 data->added = count_lines(two->data, two->size);
1517 goto free_and_return;
1518 }
1519 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1520 die("unable to read files to diff");
1522 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1523 data->is_binary = 1;
1524 data->added = mf2.size;
1525 data->deleted = mf1.size;
1526 } else {
1527 /* Crazy xdl interfaces.. */
1528 xpparam_t xpp;
1529 xdemitconf_t xecfg;
1530 xdemitcb_t ecb;
1532 memset(&xecfg, 0, sizeof(xecfg));
1533 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1534 ecb.outf = xdiff_outf;
1535 ecb.priv = diffstat;
1536 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1537 }
1539 free_and_return:
1540 diff_free_filespec_data(one);
1541 diff_free_filespec_data(two);
1542 }
1544 static void builtin_checkdiff(const char *name_a, const char *name_b,
1545 const char *attr_path,
1546 struct diff_filespec *one,
1547 struct diff_filespec *two,
1548 struct diff_options *o)
1549 {
1550 mmfile_t mf1, mf2;
1551 struct checkdiff_t data;
1553 if (!two)
1554 return;
1556 memset(&data, 0, sizeof(data));
1557 data.xm.consume = checkdiff_consume;
1558 data.filename = name_b ? name_b : name_a;
1559 data.lineno = 0;
1560 data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1561 data.ws_rule = whitespace_rule(attr_path);
1562 data.file = o->file;
1564 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1565 die("unable to read files to diff");
1567 /*
1568 * All the other codepaths check both sides, but not checking
1569 * the "old" side here is deliberate. We are checking the newly
1570 * introduced changes, and as long as the "new" side is text, we
1571 * can and should check what it introduces.
1572 */
1573 if (diff_filespec_is_binary(two))
1574 goto free_and_return;
1575 else {
1576 /* Crazy xdl interfaces.. */
1577 xpparam_t xpp;
1578 xdemitconf_t xecfg;
1579 xdemitcb_t ecb;
1581 memset(&xecfg, 0, sizeof(xecfg));
1582 xpp.flags = XDF_NEED_MINIMAL;
1583 ecb.outf = xdiff_outf;
1584 ecb.priv = &data;
1585 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1586 }
1587 free_and_return:
1588 diff_free_filespec_data(one);
1589 diff_free_filespec_data(two);
1590 if (data.status)
1591 DIFF_OPT_SET(o, CHECK_FAILED);
1592 }
1594 struct diff_filespec *alloc_filespec(const char *path)
1595 {
1596 int namelen = strlen(path);
1597 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1599 memset(spec, 0, sizeof(*spec));
1600 spec->path = (char *)(spec + 1);
1601 memcpy(spec->path, path, namelen+1);
1602 spec->count = 1;
1603 return spec;
1604 }
1606 void free_filespec(struct diff_filespec *spec)
1607 {
1608 if (!--spec->count) {
1609 diff_free_filespec_data(spec);
1610 free(spec);
1611 }
1612 }
1614 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1615 unsigned short mode)
1616 {
1617 if (mode) {
1618 spec->mode = canon_mode(mode);
1619 hashcpy(spec->sha1, sha1);
1620 spec->sha1_valid = !is_null_sha1(sha1);
1621 }
1622 }
1624 /*
1625 * Given a name and sha1 pair, if the index tells us the file in
1626 * the work tree has that object contents, return true, so that
1627 * prepare_temp_file() does not have to inflate and extract.
1628 */
1629 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1630 {
1631 struct cache_entry *ce;
1632 struct stat st;
1633 int pos, len;
1635 /* We do not read the cache ourselves here, because the
1636 * benchmark with my previous version that always reads cache
1637 * shows that it makes things worse for diff-tree comparing
1638 * two linux-2.6 kernel trees in an already checked out work
1639 * tree. This is because most diff-tree comparisons deal with
1640 * only a small number of files, while reading the cache is
1641 * expensive for a large project, and its cost outweighs the
1642 * savings we get by not inflating the object to a temporary
1643 * file. Practically, this code only helps when we are used
1644 * by diff-cache --cached, which does read the cache before
1645 * calling us.
1646 */
1647 if (!active_cache)
1648 return 0;
1650 /* We want to avoid the working directory if our caller
1651 * doesn't need the data in a normal file, this system
1652 * is rather slow with its stat/open/mmap/close syscalls,
1653 * and the object is contained in a pack file. The pack
1654 * is probably already open and will be faster to obtain
1655 * the data through than the working directory. Loose
1656 * objects however would tend to be slower as they need
1657 * to be individually opened and inflated.
1658 */
1659 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1660 return 0;
1662 len = strlen(name);
1663 pos = cache_name_pos(name, len);
1664 if (pos < 0)
1665 return 0;
1666 ce = active_cache[pos];
1668 /*
1669 * This is not the sha1 we are looking for, or
1670 * unreusable because it is not a regular file.
1671 */
1672 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1673 return 0;
1675 /*
1676 * If ce matches the file in the work tree, we can reuse it.
1677 */
1678 if (ce_uptodate(ce) ||
1679 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1680 return 1;
1682 return 0;
1683 }
1685 static int populate_from_stdin(struct diff_filespec *s)
1686 {
1687 struct strbuf buf;
1688 size_t size = 0;
1690 strbuf_init(&buf, 0);
1691 if (strbuf_read(&buf, 0, 0) < 0)
1692 return error("error while reading from stdin %s",
1693 strerror(errno));
1695 s->should_munmap = 0;
1696 s->data = strbuf_detach(&buf, &size);
1697 s->size = size;
1698 s->should_free = 1;
1699 return 0;
1700 }
1702 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1703 {
1704 int len;
1705 char *data = xmalloc(100);
1706 len = snprintf(data, 100,
1707 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1708 s->data = data;
1709 s->size = len;
1710 s->should_free = 1;
1711 if (size_only) {
1712 s->data = NULL;
1713 free(data);
1714 }
1715 return 0;
1716 }
1718 /*
1719 * While doing rename detection and pickaxe operation, we may need to
1720 * grab the data for the blob (or file) for our own in-core comparison.
1721 * diff_filespec has data and size fields for this purpose.
1722 */
1723 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1724 {
1725 int err = 0;
1726 if (!DIFF_FILE_VALID(s))
1727 die("internal error: asking to populate invalid file.");
1728 if (S_ISDIR(s->mode))
1729 return -1;
1731 if (s->data)
1732 return 0;
1734 if (size_only && 0 < s->size)
1735 return 0;
1737 if (S_ISGITLINK(s->mode))
1738 return diff_populate_gitlink(s, size_only);
1740 if (!s->sha1_valid ||
1741 reuse_worktree_file(s->path, s->sha1, 0)) {
1742 struct strbuf buf;
1743 struct stat st;
1744 int fd;
1746 if (!strcmp(s->path, "-"))
1747 return populate_from_stdin(s);
1749 if (lstat(s->path, &st) < 0) {
1750 if (errno == ENOENT) {
1751 err_empty:
1752 err = -1;
1753 empty:
1754 s->data = (char *)"";
1755 s->size = 0;
1756 return err;
1757 }
1758 }
1759 s->size = xsize_t(st.st_size);
1760 if (!s->size)
1761 goto empty;
1762 if (size_only)
1763 return 0;
1764 if (S_ISLNK(st.st_mode)) {
1765 int ret;
1766 s->data = xmalloc(s->size);
1767 s->should_free = 1;
1768 ret = readlink(s->path, s->data, s->size);
1769 if (ret < 0) {
1770 free(s->data);
1771 goto err_empty;
1772 }
1773 return 0;
1774 }
1775 fd = open(s->path, O_RDONLY);
1776 if (fd < 0)
1777 goto err_empty;
1778 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1779 close(fd);
1780 s->should_munmap = 1;
1782 /*
1783 * Convert from working tree format to canonical git format
1784 */
1785 strbuf_init(&buf, 0);
1786 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1787 size_t size = 0;
1788 munmap(s->data, s->size);
1789 s->should_munmap = 0;
1790 s->data = strbuf_detach(&buf, &size);
1791 s->size = size;
1792 s->should_free = 1;
1793 }
1794 }
1795 else {
1796 enum object_type type;
1797 if (size_only)
1798 type = sha1_object_info(s->sha1, &s->size);
1799 else {
1800 s->data = read_sha1_file(s->sha1, &type, &s->size);
1801 s->should_free = 1;
1802 }
1803 }
1804 return 0;
1805 }
1807 void diff_free_filespec_blob(struct diff_filespec *s)
1808 {
1809 if (s->should_free)
1810 free(s->data);
1811 else if (s->should_munmap)
1812 munmap(s->data, s->size);
1814 if (s->should_free || s->should_munmap) {
1815 s->should_free = s->should_munmap = 0;
1816 s->data = NULL;
1817 }
1818 }
1820 void diff_free_filespec_data(struct diff_filespec *s)
1821 {
1822 diff_free_filespec_blob(s);
1823 free(s->cnt_data);
1824 s->cnt_data = NULL;
1825 }
1827 static void prep_temp_blob(struct diff_tempfile *temp,
1828 void *blob,
1829 unsigned long size,
1830 const unsigned char *sha1,
1831 int mode)
1832 {
1833 int fd;
1835 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1836 if (fd < 0)
1837 die("unable to create temp-file: %s", strerror(errno));
1838 if (write_in_full(fd, blob, size) != size)
1839 die("unable to write temp-file");
1840 close(fd);
1841 temp->name = temp->tmp_path;
1842 strcpy(temp->hex, sha1_to_hex(sha1));
1843 temp->hex[40] = 0;
1844 sprintf(temp->mode, "%06o", mode);
1845 }
1847 static void prepare_temp_file(const char *name,
1848 struct diff_tempfile *temp,
1849 struct diff_filespec *one)
1850 {
1851 if (!DIFF_FILE_VALID(one)) {
1852 not_a_valid_file:
1853 /* A '-' entry produces this for file-2, and
1854 * a '+' entry produces this for file-1.
1855 */
1856 temp->name = "/dev/null";
1857 strcpy(temp->hex, ".");
1858 strcpy(temp->mode, ".");
1859 return;
1860 }
1862 if (!one->sha1_valid ||
1863 reuse_worktree_file(name, one->sha1, 1)) {
1864 struct stat st;
1865 if (lstat(name, &st) < 0) {
1866 if (errno == ENOENT)
1867 goto not_a_valid_file;
1868 die("stat(%s): %s", name, strerror(errno));
1869 }
1870 if (S_ISLNK(st.st_mode)) {
1871 int ret;
1872 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1873 size_t sz = xsize_t(st.st_size);
1874 if (sizeof(buf) <= st.st_size)
1875 die("symlink too long: %s", name);
1876 ret = readlink(name, buf, sz);
1877 if (ret < 0)
1878 die("readlink(%s)", name);
1879 prep_temp_blob(temp, buf, sz,
1880 (one->sha1_valid ?
1881 one->sha1 : null_sha1),
1882 (one->sha1_valid ?
1883 one->mode : S_IFLNK));
1884 }
1885 else {
1886 /* we can borrow from the file in the work tree */
1887 temp->name = name;
1888 if (!one->sha1_valid)
1889 strcpy(temp->hex, sha1_to_hex(null_sha1));
1890 else
1891 strcpy(temp->hex, sha1_to_hex(one->sha1));
1892 /* Even though we may sometimes borrow the
1893 * contents from the work tree, we always want
1894 * one->mode. mode is trustworthy even when
1895 * !(one->sha1_valid), as long as
1896 * DIFF_FILE_VALID(one).
1897 */
1898 sprintf(temp->mode, "%06o", one->mode);
1899 }
1900 return;
1901 }
1902 else {
1903 if (diff_populate_filespec(one, 0))
1904 die("cannot read data blob for %s", one->path);
1905 prep_temp_blob(temp, one->data, one->size,
1906 one->sha1, one->mode);
1907 }
1908 }
1910 static void remove_tempfile(void)
1911 {
1912 int i;
1914 for (i = 0; i < 2; i++)
1915 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1916 unlink(diff_temp[i].name);
1917 diff_temp[i].name = NULL;
1918 }
1919 }
1921 static void remove_tempfile_on_signal(int signo)
1922 {
1923 remove_tempfile();
1924 signal(SIGINT, SIG_DFL);
1925 raise(signo);
1926 }
1928 /* An external diff command takes:
1929 *
1930 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1931 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1932 *
1933 */
1934 static void run_external_diff(const char *pgm,
1935 const char *name,
1936 const char *other,
1937 struct diff_filespec *one,
1938 struct diff_filespec *two,
1939 const char *xfrm_msg,
1940 int complete_rewrite)
1941 {
1942 const char *spawn_arg[10];
1943 struct diff_tempfile *temp = diff_temp;
1944 int retval;
1945 static int atexit_asked = 0;
1946 const char *othername;
1947 const char **arg = &spawn_arg[0];
1949 othername = (other? other : name);
1950 if (one && two) {
1951 prepare_temp_file(name, &temp[0], one);
1952 prepare_temp_file(othername, &temp[1], two);
1953 if (! atexit_asked &&
1954 (temp[0].name == temp[0].tmp_path ||
1955 temp[1].name == temp[1].tmp_path)) {
1956 atexit_asked = 1;
1957 atexit(remove_tempfile);
1958 }
1959 signal(SIGINT, remove_tempfile_on_signal);
1960 }
1962 if (one && two) {
1963 *arg++ = pgm;
1964 *arg++ = name;
1965 *arg++ = temp[0].name;
1966 *arg++ = temp[0].hex;
1967 *arg++ = temp[0].mode;
1968 *arg++ = temp[1].name;
1969 *arg++ = temp[1].hex;
1970 *arg++ = temp[1].mode;
1971 if (other) {
1972 *arg++ = other;
1973 *arg++ = xfrm_msg;
1974 }
1975 } else {
1976 *arg++ = pgm;
1977 *arg++ = name;
1978 }
1979 *arg = NULL;
1980 fflush(NULL);
1981 retval = run_command_v_opt(spawn_arg, 0);
1982 remove_tempfile();
1983 if (retval) {
1984 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1985 exit(1);
1986 }
1987 }
1989 static const char *external_diff_attr(const char *name)
1990 {
1991 struct git_attr_check attr_diff_check;
1993 if (!name)
1994 return NULL;
1996 setup_diff_attr_check(&attr_diff_check);
1997 if (!git_checkattr(name, 1, &attr_diff_check)) {
1998 const char *value = attr_diff_check.value;
1999 if (!ATTR_TRUE(value) &&
2000 !ATTR_FALSE(value) &&
2001 !ATTR_UNSET(value)) {
2002 struct ll_diff_driver *drv;
2004 for (drv = user_diff; drv; drv = drv->next)
2005 if (!strcmp(drv->name, value))
2006 return drv->cmd;
2007 }
2008 }
2009 return NULL;
2010 }
2012 static void run_diff_cmd(const char *pgm,
2013 const char *name,
2014 const char *other,
2015 const char *attr_path,
2016 struct diff_filespec *one,
2017 struct diff_filespec *two,
2018 const char *xfrm_msg,
2019 struct diff_options *o,
2020 int complete_rewrite)
2021 {
2022 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2023 pgm = NULL;
2024 else {
2025 const char *cmd = external_diff_attr(attr_path);
2026 if (cmd)
2027 pgm = cmd;
2028 }
2030 if (pgm) {
2031 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2032 complete_rewrite);
2033 return;
2034 }
2035 if (one && two)
2036 builtin_diff(name, other ? other : name,
2037 one, two, xfrm_msg, o, complete_rewrite);
2038 else
2039 fprintf(o->file, "* Unmerged path %s\n", name);
2040 }
2042 static void diff_fill_sha1_info(struct diff_filespec *one)
2043 {
2044 if (DIFF_FILE_VALID(one)) {
2045 if (!one->sha1_valid) {
2046 struct stat st;
2047 if (!strcmp(one->path, "-")) {
2048 hashcpy(one->sha1, null_sha1);
2049 return;
2050 }
2051 if (lstat(one->path, &st) < 0)
2052 die("stat %s", one->path);
2053 if (index_path(one->sha1, one->path, &st, 0))
2054 die("cannot hash %s\n", one->path);
2055 }
2056 }
2057 else
2058 hashclr(one->sha1);
2059 }
2061 static int similarity_index(struct diff_filepair *p)
2062 {
2063 return p->score * 100 / MAX_SCORE;
2064 }
2066 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2067 {
2068 /* Strip the prefix but do not molest /dev/null and absolute paths */
2069 if (*namep && **namep != '/')
2070 *namep += prefix_length;
2071 if (*otherp && **otherp != '/')
2072 *otherp += prefix_length;
2073 }
2075 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2076 {
2077 const char *pgm = external_diff();
2078 struct strbuf msg;
2079 char *xfrm_msg;
2080 struct diff_filespec *one = p->one;
2081 struct diff_filespec *two = p->two;
2082 const char *name;
2083 const char *other;
2084 const char *attr_path;
2085 int complete_rewrite = 0;
2087 name = p->one->path;
2088 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2089 attr_path = name;
2090 if (o->prefix_length)
2091 strip_prefix(o->prefix_length, &name, &other);
2093 if (DIFF_PAIR_UNMERGED(p)) {
2094 run_diff_cmd(pgm, name, NULL, attr_path,
2095 NULL, NULL, NULL, o, 0);
2096 return;
2097 }
2099 diff_fill_sha1_info(one);
2100 diff_fill_sha1_info(two);
2102 strbuf_init(&msg, PATH_MAX * 2 + 300);
2103 switch (p->status) {
2104 case DIFF_STATUS_COPIED:
2105 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2106 strbuf_addstr(&msg, "\ncopy from ");
2107 quote_c_style(name, &msg, NULL, 0);
2108 strbuf_addstr(&msg, "\ncopy to ");
2109 quote_c_style(other, &msg, NULL, 0);
2110 strbuf_addch(&msg, '\n');
2111 break;
2112 case DIFF_STATUS_RENAMED:
2113 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2114 strbuf_addstr(&msg, "\nrename from ");
2115 quote_c_style(name, &msg, NULL, 0);
2116 strbuf_addstr(&msg, "\nrename to ");
2117 quote_c_style(other, &msg, NULL, 0);
2118 strbuf_addch(&msg, '\n');
2119 break;
2120 case DIFF_STATUS_MODIFIED:
2121 if (p->score) {
2122 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2123 similarity_index(p));
2124 complete_rewrite = 1;
2125 break;
2126 }
2127 /* fallthru */
2128 default:
2129 /* nothing */
2130 ;
2131 }
2133 if (hashcmp(one->sha1, two->sha1)) {
2134 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2136 if (DIFF_OPT_TST(o, BINARY)) {
2137 mmfile_t mf;
2138 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2139 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2140 abbrev = 40;
2141 }
2142 strbuf_addf(&msg, "index %.*s..%.*s",
2143 abbrev, sha1_to_hex(one->sha1),
2144 abbrev, sha1_to_hex(two->sha1));
2145 if (one->mode == two->mode)
2146 strbuf_addf(&msg, " %06o", one->mode);
2147 strbuf_addch(&msg, '\n');
2148 }
2150 if (msg.len)
2151 strbuf_setlen(&msg, msg.len - 1);
2152 xfrm_msg = msg.len ? msg.buf : NULL;
2154 if (!pgm &&
2155 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2156 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2157 /* a filepair that changes between file and symlink
2158 * needs to be split into deletion and creation.
2159 */
2160 struct diff_filespec *null = alloc_filespec(two->path);
2161 run_diff_cmd(NULL, name, other, attr_path,
2162 one, null, xfrm_msg, o, 0);
2163 free(null);
2164 null = alloc_filespec(one->path);
2165 run_diff_cmd(NULL, name, other, attr_path,
2166 null, two, xfrm_msg, o, 0);
2167 free(null);
2168 }
2169 else
2170 run_diff_cmd(pgm, name, other, attr_path,
2171 one, two, xfrm_msg, o, complete_rewrite);
2173 strbuf_release(&msg);
2174 }
2176 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2177 struct diffstat_t *diffstat)
2178 {
2179 const char *name;
2180 const char *other;
2181 int complete_rewrite = 0;
2183 if (DIFF_PAIR_UNMERGED(p)) {
2184 /* unmerged */
2185 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2186 return;
2187 }
2189 name = p->one->path;
2190 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2192 if (o->prefix_length)
2193 strip_prefix(o->prefix_length, &name, &other);
2195 diff_fill_sha1_info(p->one);
2196 diff_fill_sha1_info(p->two);
2198 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2199 complete_rewrite = 1;
2200 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2201 }
2203 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2204 {
2205 const char *name;
2206 const char *other;
2207 const char *attr_path;
2209 if (DIFF_PAIR_UNMERGED(p)) {
2210 /* unmerged */
2211 return;
2212 }
2214 name = p->one->path;
2215 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2216 attr_path = other ? other : name;
2218 if (o->prefix_length)
2219 strip_prefix(o->prefix_length, &name, &other);
2221 diff_fill_sha1_info(p->one);
2222 diff_fill_sha1_info(p->two);
2224 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2225 }
2227 void diff_setup(struct diff_options *options)
2228 {
2229 memset(options, 0, sizeof(*options));
2231 options->file = stdout;
2233 options->line_termination = '\n';
2234 options->break_opt = -1;
2235 options->rename_limit = -1;
2236 options->dirstat_percent = 3;
2237 options->context = 3;
2239 options->change = diff_change;
2240 options->add_remove = diff_addremove;
2241 if (diff_use_color_default > 0)
2242 DIFF_OPT_SET(options, COLOR_DIFF);
2243 else
2244 DIFF_OPT_CLR(options, COLOR_DIFF);
2245 options->detect_rename = diff_detect_rename_default;
2247 options->a_prefix = "a/";
2248 options->b_prefix = "b/";
2249 }
2251 int diff_setup_done(struct diff_options *options)
2252 {
2253 int count = 0;
2255 if (options->output_format & DIFF_FORMAT_NAME)
2256 count++;
2257 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2258 count++;
2259 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2260 count++;
2261 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2262 count++;
2263 if (count > 1)
2264 die("--name-only, --name-status, --check and -s are mutually exclusive");
2266 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2267 options->detect_rename = DIFF_DETECT_COPY;
2269 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2270 options->prefix = NULL;
2271 if (options->prefix)
2272 options->prefix_length = strlen(options->prefix);
2273 else
2274 options->prefix_length = 0;
2276 if (options->output_format & (DIFF_FORMAT_NAME |
2277 DIFF_FORMAT_NAME_STATUS |
2278 DIFF_FORMAT_CHECKDIFF |
2279 DIFF_FORMAT_NO_OUTPUT))
2280 options->output_format &= ~(DIFF_FORMAT_RAW |
2281 DIFF_FORMAT_NUMSTAT |
2282 DIFF_FORMAT_DIFFSTAT |
2283 DIFF_FORMAT_SHORTSTAT |
2284 DIFF_FORMAT_DIRSTAT |
2285 DIFF_FORMAT_SUMMARY |
2286 DIFF_FORMAT_PATCH);
2288 /*
2289 * These cases always need recursive; we do not drop caller-supplied
2290 * recursive bits for other formats here.
2291 */
2292 if (options->output_format & (DIFF_FORMAT_PATCH |
2293 DIFF_FORMAT_NUMSTAT |
2294 DIFF_FORMAT_DIFFSTAT |
2295 DIFF_FORMAT_SHORTSTAT |
2296 DIFF_FORMAT_DIRSTAT |
2297 DIFF_FORMAT_SUMMARY |
2298 DIFF_FORMAT_CHECKDIFF))
2299 DIFF_OPT_SET(options, RECURSIVE);
2300 /*
2301 * Also pickaxe would not work very well if you do not say recursive
2302 */
2303 if (options->pickaxe)
2304 DIFF_OPT_SET(options, RECURSIVE);
2306 if (options->detect_rename && options->rename_limit < 0)
2307 options->rename_limit = diff_rename_limit_default;
2308 if (options->setup & DIFF_SETUP_USE_CACHE) {
2309 if (!active_cache)
2310 /* read-cache does not die even when it fails
2311 * so it is safe for us to do this here. Also
2312 * it does not smudge active_cache or active_nr
2313 * when it fails, so we do not have to worry about
2314 * cleaning it up ourselves either.
2315 */
2316 read_cache();
2317 }
2318 if (options->abbrev <= 0 || 40 < options->abbrev)
2319 options->abbrev = 40; /* full */
2321 /*
2322 * It does not make sense to show the first hit we happened
2323 * to have found. It does not make sense not to return with
2324 * exit code in such a case either.
2325 */
2326 if (DIFF_OPT_TST(options, QUIET)) {
2327 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2328 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2329 }
2331 /*
2332 * If we postprocess in diffcore, we cannot simply return
2333 * upon the first hit. We need to run diff as usual.
2334 */
2335 if (options->pickaxe || options->filter)
2336 DIFF_OPT_CLR(options, QUIET);
2338 return 0;
2339 }
2341 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2342 {
2343 char c, *eq;
2344 int len;
2346 if (*arg != '-')
2347 return 0;
2348 c = *++arg;
2349 if (!c)
2350 return 0;
2351 if (c == arg_short) {
2352 c = *++arg;
2353 if (!c)
2354 return 1;
2355 if (val && isdigit(c)) {
2356 char *end;
2357 int n = strtoul(arg, &end, 10);
2358 if (*end)
2359 return 0;
2360 *val = n;
2361 return 1;
2362 }
2363 return 0;
2364 }
2365 if (c != '-')
2366 return 0;
2367 arg++;
2368 eq = strchr(arg, '=');
2369 if (eq)
2370 len = eq - arg;
2371 else
2372 len = strlen(arg);
2373 if (!len || strncmp(arg, arg_long, len))
2374 return 0;
2375 if (eq) {
2376 int n;
2377 char *end;
2378 if (!isdigit(*++eq))
2379 return 0;
2380 n = strtoul(eq, &end, 10);
2381 if (*end)
2382 return 0;
2383 *val = n;
2384 }
2385 return 1;
2386 }
2388 static int diff_scoreopt_parse(const char *opt);
2390 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2391 {
2392 const char *arg = av[0];
2394 /* Output format options */
2395 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2396 options->output_format |= DIFF_FORMAT_PATCH;
2397 else if (opt_arg(arg, 'U', "unified", &options->context))
2398 options->output_format |= DIFF_FORMAT_PATCH;
2399 else if (!strcmp(arg, "--raw"))
2400 options->output_format |= DIFF_FORMAT_RAW;
2401 else if (!strcmp(arg, "--patch-with-raw"))
2402 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2403 else if (!strcmp(arg, "--numstat"))
2404 options->output_format |= DIFF_FORMAT_NUMSTAT;
2405 else if (!strcmp(arg, "--shortstat"))
2406 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2407 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2408 options->output_format |= DIFF_FORMAT_DIRSTAT;
2409 else if (!strcmp(arg, "--cumulative"))
2410 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2411 else if (!strcmp(arg, "--check"))
2412 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2413 else if (!strcmp(arg, "--summary"))
2414 options->output_format |= DIFF_FORMAT_SUMMARY;
2415 else if (!strcmp(arg, "--patch-with-stat"))
2416 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2417 else if (!strcmp(arg, "--name-only"))
2418 options->output_format |= DIFF_FORMAT_NAME;
2419 else if (!strcmp(arg, "--name-status"))
2420 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2421 else if (!strcmp(arg, "-s"))
2422 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2423 else if (!prefixcmp(arg, "--stat")) {
2424 char *end;
2425 int width = options->stat_width;
2426 int name_width = options->stat_name_width;
2427 arg += 6;
2428 end = (char *)arg;
2430 switch (*arg) {
2431 case '-':
2432 if (!prefixcmp(arg, "-width="))
2433 width = strtoul(arg + 7, &end, 10);
2434 else if (!prefixcmp(arg, "-name-width="))
2435 name_width = strtoul(arg + 12, &end, 10);
2436 break;
2437 case '=':
2438 width = strtoul(arg+1, &end, 10);
2439 if (*end == ',')
2440 name_width = strtoul(end+1, &end, 10);
2441 }
2443 /* Important! This checks all the error cases! */
2444 if (*end)
2445 return 0;
2446 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2447 options->stat_name_width = name_width;
2448 options->stat_width = width;
2449 }
2451 /* renames options */
2452 else if (!prefixcmp(arg, "-B")) {
2453 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2454 return -1;
2455 }
2456 else if (!prefixcmp(arg, "-M")) {
2457 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2458 return -1;
2459 options->detect_rename = DIFF_DETECT_RENAME;
2460 }
2461 else if (!prefixcmp(arg, "-C")) {
2462 if (options->detect_rename == DIFF_DETECT_COPY)
2463 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2464 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2465 return -1;
2466 options->detect_rename = DIFF_DETECT_COPY;
2467 }
2468 else if (!strcmp(arg, "--no-renames"))
2469 options->detect_rename = 0;
2470 else if (!strcmp(arg, "--relative"))
2471 DIFF_OPT_SET(options, RELATIVE_NAME);
2472 else if (!prefixcmp(arg, "--relative=")) {
2473 DIFF_OPT_SET(options, RELATIVE_NAME);
2474 options->prefix = arg + 11;
2475 }
2477 /* xdiff options */
2478 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2479 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2480 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2481 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2482 else if (!strcmp(arg, "--ignore-space-at-eol"))
2483 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2485 /* flags options */
2486 else if (!strcmp(arg, "--binary")) {
2487 options->output_format |= DIFF_FORMAT_PATCH;
2488 DIFF_OPT_SET(options, BINARY);
2489 }
2490 else if (!strcmp(arg, "--full-index"))
2491 DIFF_OPT_SET(options, FULL_INDEX);
2492 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2493 DIFF_OPT_SET(options, TEXT);
2494 else if (!strcmp(arg, "-R"))
2495 DIFF_OPT_SET(options, REVERSE_DIFF);
2496 else if (!strcmp(arg, "--find-copies-harder"))
2497 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2498 else if (!strcmp(arg, "--follow"))
2499 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2500 else if (!strcmp(arg, "--color"))
2501 DIFF_OPT_SET(options, COLOR_DIFF);
2502 else if (!strcmp(arg, "--no-color"))
2503 DIFF_OPT_CLR(options, COLOR_DIFF);
2504 else if (!strcmp(arg, "--color-words"))
2505 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2506 else if (!strcmp(arg, "--exit-code"))
2507 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2508 else if (!strcmp(arg, "--quiet"))
2509 DIFF_OPT_SET(options, QUIET);
2510 else if (!strcmp(arg, "--ext-diff"))
2511 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2512 else if (!strcmp(arg, "--no-ext-diff"))
2513 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2514 else if (!strcmp(arg, "--ignore-submodules"))
2515 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2517 /* misc options */
2518 else if (!strcmp(arg, "-z"))
2519 options->line_termination = 0;
2520 else if (!prefixcmp(arg, "-l"))
2521 options->rename_limit = strtoul(arg+2, NULL, 10);
2522 else if (!prefixcmp(arg, "-S"))
2523 options->pickaxe = arg + 2;
2524 else if (!strcmp(arg, "--pickaxe-all"))
2525 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2526 else if (!strcmp(arg, "--pickaxe-regex"))
2527 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2528 else if (!prefixcmp(arg, "-O"))
2529 options->orderfile = arg + 2;
2530 else if (!prefixcmp(arg, "--diff-filter="))
2531 options->filter = arg + 14;
2532 else if (!strcmp(arg, "--abbrev"))
2533 options->abbrev = DEFAULT_ABBREV;
2534 else if (!prefixcmp(arg, "--abbrev=")) {
2535 options->abbrev = strtoul(arg + 9, NULL, 10);
2536 if (options->abbrev < MINIMUM_ABBREV)
2537 options->abbrev = MINIMUM_ABBREV;
2538 else if (40 < options->abbrev)
2539 options->abbrev = 40;
2540 }
2541 else if (!prefixcmp(arg, "--src-prefix="))
2542 options->a_prefix = arg + 13;
2543 else if (!prefixcmp(arg, "--dst-prefix="))
2544 options->b_prefix = arg + 13;
2545 else if (!strcmp(arg, "--no-prefix"))
2546 options->a_prefix = options->b_prefix = "";
2547 else if (!prefixcmp(arg, "--output=")) {
2548 options->file = fopen(arg + strlen("--output="), "w");
2549 options->close_file = 1;
2550 } else
2551 return 0;
2552 return 1;
2553 }
2555 static int parse_num(const char **cp_p)
2556 {
2557 unsigned long num, scale;
2558 int ch, dot;
2559 const char *cp = *cp_p;
2561 num = 0;
2562 scale = 1;
2563 dot = 0;
2564 for(;;) {
2565 ch = *cp;
2566 if ( !dot && ch == '.' ) {
2567 scale = 1;
2568 dot = 1;
2569 } else if ( ch == '%' ) {
2570 scale = dot ? scale*100 : 100;
2571 cp++; /* % is always at the end */
2572 break;
2573 } else if ( ch >= '0' && ch <= '9' ) {
2574 if ( scale < 100000 ) {
2575 scale *= 10;
2576 num = (num*10) + (ch-'0');
2577 }
2578 } else {
2579 break;
2580 }
2581 cp++;
2582 }
2583 *cp_p = cp;
2585 /* user says num divided by scale and we say internally that
2586 * is MAX_SCORE * num / scale.
2587 */
2588 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2589 }
2591 static int diff_scoreopt_parse(const char *opt)
2592 {
2593 int opt1, opt2, cmd;
2595 if (*opt++ != '-')
2596 return -1;
2597 cmd = *opt++;
2598 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2599 return -1; /* that is not a -M, -C nor -B option */
2601 opt1 = parse_num(&opt);
2602 if (cmd != 'B')
2603 opt2 = 0;
2604 else {
2605 if (*opt == 0)
2606 opt2 = 0;
2607 else if (*opt != '/')
2608 return -1; /* we expect -B80/99 or -B80 */
2609 else {
2610 opt++;
2611 opt2 = parse_num(&opt);
2612 }
2613 }
2614 if (*opt != 0)
2615 return -1;
2616 return opt1 | (opt2 << 16);
2617 }
2619 struct diff_queue_struct diff_queued_diff;
2621 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2622 {
2623 if (queue->alloc <= queue->nr) {
2624 queue->alloc = alloc_nr(queue->alloc);
2625 queue->queue = xrealloc(queue->queue,
2626 sizeof(dp) * queue->alloc);
2627 }
2628 queue->queue[queue->nr++] = dp;
2629 }
2631 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2632 struct diff_filespec *one,
2633 struct diff_filespec *two)
2634 {
2635 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2636 dp->one = one;
2637 dp->two = two;
2638 if (queue)
2639 diff_q(queue, dp);
2640 return dp;
2641 }
2643 void diff_free_filepair(struct diff_filepair *p)
2644 {
2645 free_filespec(p->one);
2646 free_filespec(p->two);
2647 free(p);
2648 }
2650 /* This is different from find_unique_abbrev() in that
2651 * it stuffs the result with dots for alignment.
2652 */
2653 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2654 {
2655 int abblen;
2656 const char *abbrev;
2657 if (len == 40)
2658 return sha1_to_hex(sha1);
2660 abbrev = find_unique_abbrev(sha1, len);
2661 abblen = strlen(abbrev);
2662 if (abblen < 37) {
2663 static char hex[41];
2664 if (len < abblen && abblen <= len + 2)
2665 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2666 else
2667 sprintf(hex, "%s...", abbrev);
2668 return hex;
2669 }
2670 return sha1_to_hex(sha1);
2671 }
2673 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2674 {
2675 int line_termination = opt->line_termination;
2676 int inter_name_termination = line_termination ? '\t' : '\0';
2678 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2679 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2680 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2681 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2682 }
2683 if (p->score) {
2684 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2685 inter_name_termination);
2686 } else {
2687 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2688 }
2690 if (p->status == DIFF_STATUS_COPIED ||
2691 p->status == DIFF_STATUS_RENAMED) {
2692 const char *name_a, *name_b;
2693 name_a = p->one->path;
2694 name_b = p->two->path;
2695 strip_prefix(opt->prefix_length, &name_a, &name_b);
2696 write_name_quoted(name_a, opt->file, inter_name_termination);
2697 write_name_quoted(name_b, opt->file, line_termination);
2698 } else {
2699 const char *name_a, *name_b;
2700 name_a = p->one->mode ? p->one->path : p->two->path;
2701 name_b = NULL;
2702 strip_prefix(opt->prefix_length, &name_a, &name_b);
2703 write_name_quoted(name_a, opt->file, line_termination);
2704 }
2705 }
2707 int diff_unmodified_pair(struct diff_filepair *p)
2708 {
2709 /* This function is written stricter than necessary to support
2710 * the currently implemented transformers, but the idea is to
2711 * let transformers to produce diff_filepairs any way they want,
2712 * and filter and clean them up here before producing the output.
2713 */
2714 struct diff_filespec *one = p->one, *two = p->two;
2716 if (DIFF_PAIR_UNMERGED(p))
2717 return 0; /* unmerged is interesting */
2719 /* deletion, addition, mode or type change
2720 * and rename are all interesting.
2721 */
2722 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2723 DIFF_PAIR_MODE_CHANGED(p) ||
2724 strcmp(one->path, two->path))
2725 return 0;
2727 /* both are valid and point at the same path. that is, we are
2728 * dealing with a change.
2729 */
2730 if (one->sha1_valid && two->sha1_valid &&
2731 !hashcmp(one->sha1, two->sha1))
2732 return 1; /* no change */
2733 if (!one->sha1_valid && !two->sha1_valid)
2734 return 1; /* both look at the same file on the filesystem. */
2735 return 0;
2736 }
2738 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2739 {
2740 if (diff_unmodified_pair(p))
2741 return;
2743 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2744 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2745 return; /* no tree diffs in patch format */
2747 run_diff(p, o);
2748 }
2750 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2751 struct diffstat_t *diffstat)
2752 {
2753 if (diff_unmodified_pair(p))
2754 return;
2756 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2757 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2758 return; /* no tree diffs in patch format */
2760 run_diffstat(p, o, diffstat);
2761 }
2763 static void diff_flush_checkdiff(struct diff_filepair *p,
2764 struct diff_options *o)
2765 {
2766 if (diff_unmodified_pair(p))
2767 return;
2769 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2770 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2771 return; /* no tree diffs in patch format */
2773 run_checkdiff(p, o);
2774 }
2776 int diff_queue_is_empty(void)
2777 {
2778 struct diff_queue_struct *q = &diff_queued_diff;
2779 int i;
2780 for (i = 0; i < q->nr; i++)
2781 if (!diff_unmodified_pair(q->queue[i]))
2782 return 0;
2783 return 1;
2784 }
2786 #if DIFF_DEBUG
2787 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2788 {
2789 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2790 x, one ? one : "",
2791 s->path,
2792 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2793 s->mode,
2794 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2795 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2796 x, one ? one : "",
2797 s->size, s->xfrm_flags);
2798 }
2800 void diff_debug_filepair(const struct diff_filepair *p, int i)
2801 {
2802 diff_debug_filespec(p->one, i, "one");
2803 diff_debug_filespec(p->two, i, "two");
2804 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2805 p->score, p->status ? p->status : '?',
2806 p->one->rename_used, p->broken_pair);
2807 }
2809 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2810 {
2811 int i;
2812 if (msg)
2813 fprintf(stderr, "%s\n", msg);
2814 fprintf(stderr, "q->nr = %d\n", q->nr);
2815 for (i = 0; i < q->nr; i++) {
2816 struct diff_filepair *p = q->queue[i];
2817 diff_debug_filepair(p, i);
2818 }
2819 }
2820 #endif
2822 static void diff_resolve_rename_copy(void)
2823 {
2824 int i;
2825 struct diff_filepair *p;
2826 struct diff_queue_struct *q = &diff_queued_diff;
2828 diff_debug_queue("resolve-rename-copy", q);
2830 for (i = 0; i < q->nr; i++) {
2831 p = q->queue[i];
2832 p->status = 0; /* undecided */
2833 if (DIFF_PAIR_UNMERGED(p))
2834 p->status = DIFF_STATUS_UNMERGED;
2835 else if (!DIFF_FILE_VALID(p->one))
2836 p->status = DIFF_STATUS_ADDED;
2837 else if (!DIFF_FILE_VALID(p->two))
2838 p->status = DIFF_STATUS_DELETED;
2839 else if (DIFF_PAIR_TYPE_CHANGED(p))
2840 p->status = DIFF_STATUS_TYPE_CHANGED;
2842 /* from this point on, we are dealing with a pair
2843 * whose both sides are valid and of the same type, i.e.
2844 * either in-place edit or rename/copy edit.
2845 */
2846 else if (DIFF_PAIR_RENAME(p)) {
2847 /*
2848 * A rename might have re-connected a broken
2849 * pair up, causing the pathnames to be the
2850 * same again. If so, that's not a rename at
2851 * all, just a modification..
2852 *
2853 * Otherwise, see if this source was used for
2854 * multiple renames, in which case we decrement
2855 * the count, and call it a copy.
2856 */
2857 if (!strcmp(p->one->path, p->two->path))
2858 p->status = DIFF_STATUS_MODIFIED;
2859 else if (--p->one->rename_used > 0)
2860 p->status = DIFF_STATUS_COPIED;
2861 else
2862 p->status = DIFF_STATUS_RENAMED;
2863 }
2864 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2865 p->one->mode != p->two->mode ||
2866 is_null_sha1(p->one->sha1))
2867 p->status = DIFF_STATUS_MODIFIED;
2868 else {
2869 /* This is a "no-change" entry and should not
2870 * happen anymore, but prepare for broken callers.
2871 */
2872 error("feeding unmodified %s to diffcore",
2873 p->one->path);
2874 p->status = DIFF_STATUS_UNKNOWN;
2875 }
2876 }
2877 diff_debug_queue("resolve-rename-copy done", q);
2878 }
2880 static int check_pair_status(struct diff_filepair *p)
2881 {
2882 switch (p->status) {
2883 case DIFF_STATUS_UNKNOWN:
2884 return 0;
2885 case 0:
2886 die("internal error in diff-resolve-rename-copy");
2887 default:
2888 return 1;
2889 }
2890 }
2892 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2893 {
2894 int fmt = opt->output_format;
2896 if (fmt & DIFF_FORMAT_CHECKDIFF)
2897 diff_flush_checkdiff(p, opt);
2898 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2899 diff_flush_raw(p, opt);
2900 else if (fmt & DIFF_FORMAT_NAME) {
2901 const char *name_a, *name_b;
2902 name_a = p->two->path;
2903 name_b = NULL;
2904 strip_prefix(opt->prefix_length, &name_a, &name_b);
2905 write_name_quoted(name_a, opt->file, opt->line_termination);
2906 }
2907 }
2909 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2910 {
2911 if (fs->mode)
2912 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2913 else
2914 fprintf(file, " %s ", newdelete);
2915 write_name_quoted(fs->path, file, '\n');
2916 }
2919 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2920 {
2921 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2922 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2923 show_name ? ' ' : '\n');
2924 if (show_name) {
2925 write_name_quoted(p->two->path, file, '\n');
2926 }
2927 }
2928 }
2930 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2931 {
2932 char *names = pprint_rename(p->one->path, p->two->path);
2934 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2935 free(names);
2936 show_mode_change(file, p, 0);
2937 }
2939 static void diff_summary(FILE *file, struct diff_filepair *p)
2940 {
2941 switch(p->status) {
2942 case DIFF_STATUS_DELETED:
2943 show_file_mode_name(file, "delete", p->one);
2944 break;
2945 case DIFF_STATUS_ADDED:
2946 show_file_mode_name(file, "create", p->two);
2947 break;
2948 case DIFF_STATUS_COPIED:
2949 show_rename_copy(file, "copy", p);
2950 break;
2951 case DIFF_STATUS_RENAMED:
2952 show_rename_copy(file, "rename", p);
2953 break;
2954 default:
2955 if (p->score) {
2956 fputs(" rewrite ", file);
2957 write_name_quoted(p->two->path, file, ' ');
2958 fprintf(file, "(%d%%)\n", similarity_index(p));
2959 }
2960 show_mode_change(file, p, !p->score);
2961 break;
2962 }
2963 }
2965 struct patch_id_t {
2966 struct xdiff_emit_state xm;
2967 SHA_CTX *ctx;
2968 int patchlen;
2969 };
2971 static int remove_space(char *line, int len)
2972 {
2973 int i;
2974 char *dst = line;
2975 unsigned char c;
2977 for (i = 0; i < len; i++)
2978 if (!isspace((c = line[i])))
2979 *dst++ = c;
2981 return dst - line;
2982 }
2984 static void patch_id_consume(void *priv, char *line, unsigned long len)
2985 {
2986 struct patch_id_t *data = priv;
2987 int new_len;
2989 /* Ignore line numbers when computing the SHA1 of the patch */
2990 if (!prefixcmp(line, "@@ -"))
2991 return;
2993 new_len = remove_space(line, len);
2995 SHA1_Update(data->ctx, line, new_len);
2996 data->patchlen += new_len;
2997 }
2999 /* returns 0 upon success, and writes result into sha1 */
3000 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3001 {
3002 struct diff_queue_struct *q = &diff_queued_diff;
3003 int i;
3004 SHA_CTX ctx;
3005 struct patch_id_t data;
3006 char buffer[PATH_MAX * 4 + 20];
3008 SHA1_Init(&ctx);
3009 memset(&data, 0, sizeof(struct patch_id_t));
3010 data.ctx = &ctx;
3011 data.xm.consume = patch_id_consume;
3013 for (i = 0; i < q->nr; i++) {
3014 xpparam_t xpp;
3015 xdemitconf_t xecfg;
3016 xdemitcb_t ecb;
3017 mmfile_t mf1, mf2;
3018 struct diff_filepair *p = q->queue[i];
3019 int len1, len2;
3021 memset(&xecfg, 0, sizeof(xecfg));
3022 if (p->status == 0)
3023 return error("internal diff status error");
3024 if (p->status == DIFF_STATUS_UNKNOWN)
3025 continue;
3026 if (diff_unmodified_pair(p))
3027 continue;
3028 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3029 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3030 continue;
3031 if (DIFF_PAIR_UNMERGED(p))
3032 continue;
3034 diff_fill_sha1_info(p->one);
3035 diff_fill_sha1_info(p->two);
3036 if (fill_mmfile(&mf1, p->one) < 0 ||
3037 fill_mmfile(&mf2, p->two) < 0)
3038 return error("unable to read files to diff");
3040 len1 = remove_space(p->one->path, strlen(p->one->path));
3041 len2 = remove_space(p->two->path, strlen(p->two->path));
3042 if (p->one->mode == 0)
3043 len1 = snprintf(buffer, sizeof(buffer),
3044 "diff--gita/%.*sb/%.*s"
3045 "newfilemode%06o"
3046 "---/dev/null"
3047 "+++b/%.*s",
3048 len1, p->one->path,
3049 len2, p->two->path,
3050 p->two->mode,
3051 len2, p->two->path);
3052 else if (p->two->mode == 0)
3053 len1 = snprintf(buffer, sizeof(buffer),
3054 "diff--gita/%.*sb/%.*s"
3055 "deletedfilemode%06o"
3056 "---a/%.*s"
3057 "+++/dev/null",
3058 len1, p->one->path,
3059 len2, p->two->path,
3060 p->one->mode,
3061 len1, p->one->path);
3062 else
3063 len1 = snprintf(buffer, sizeof(buffer),
3064 "diff--gita/%.*sb/%.*s"
3065 "---a/%.*s"
3066 "+++b/%.*s",
3067 len1, p->one->path,
3068 len2, p->two->path,
3069 len1, p->one->path,
3070 len2, p->two->path);
3071 SHA1_Update(&ctx, buffer, len1);
3073 xpp.flags = XDF_NEED_MINIMAL;
3074 xecfg.ctxlen = 3;
3075 xecfg.flags = XDL_EMIT_FUNCNAMES;
3076 ecb.outf = xdiff_outf;
3077 ecb.priv = &data;
3078 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3079 }
3081 SHA1_Final(sha1, &ctx);
3082 return 0;
3083 }
3085 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3086 {
3087 struct diff_queue_struct *q = &diff_queued_diff;
3088 int i;
3089 int result = diff_get_patch_id(options, sha1);
3091 for (i = 0; i < q->nr; i++)
3092 diff_free_filepair(q->queue[i]);
3094 free(q->queue);
3095 q->queue = NULL;
3096 q->nr = q->alloc = 0;
3098 return result;
3099 }
3101 static int is_summary_empty(const struct diff_queue_struct *q)
3102 {
3103 int i;
3105 for (i = 0; i < q->nr; i++) {
3106 const struct diff_filepair *p = q->queue[i];
3108 switch (p->status) {
3109 case DIFF_STATUS_DELETED:
3110 case DIFF_STATUS_ADDED:
3111 case DIFF_STATUS_COPIED:
3112 case DIFF_STATUS_RENAMED:
3113 return 0;
3114 default:
3115 if (p->score)
3116 return 0;
3117 if (p->one->mode && p->two->mode &&
3118 p->one->mode != p->two->mode)
3119 return 0;
3120 break;
3121 }
3122 }
3123 return 1;
3124 }
3126 void diff_flush(struct diff_options *options)
3127 {
3128 struct diff_queue_struct *q = &diff_queued_diff;
3129 int i, output_format = options->output_format;
3130 int separator = 0;
3132 /*
3133 * Order: raw, stat, summary, patch
3134 * or: name/name-status/checkdiff (other bits clear)
3135 */
3136 if (!q->nr)
3137 goto free_queue;
3139 if (output_format & (DIFF_FORMAT_RAW |
3140 DIFF_FORMAT_NAME |
3141 DIFF_FORMAT_NAME_STATUS |
3142 DIFF_FORMAT_CHECKDIFF)) {
3143 for (i = 0; i < q->nr; i++) {
3144 struct diff_filepair *p = q->queue[i];
3145 if (check_pair_status(p))
3146 flush_one_pair(p, options);
3147 }
3148 separator++;
3149 }
3151 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3152 struct diffstat_t diffstat;
3154 memset(&diffstat, 0, sizeof(struct diffstat_t));
3155 diffstat.xm.consume = diffstat_consume;
3156 for (i = 0; i < q->nr; i++) {
3157 struct diff_filepair *p = q->queue[i];
3158 if (check_pair_status(p))
3159 diff_flush_stat(p, options, &diffstat);
3160 }
3161 if (output_format & DIFF_FORMAT_NUMSTAT)
3162 show_numstat(&diffstat, options);
3163 if (output_format & DIFF_FORMAT_DIFFSTAT)
3164 show_stats(&diffstat, options);
3165 if (output_format & DIFF_FORMAT_SHORTSTAT)
3166 show_shortstats(&diffstat, options);
3167 free_diffstat_info(&diffstat);
3168 separator++;
3169 }
3170 if (output_format & DIFF_FORMAT_DIRSTAT)
3171 show_dirstat(options);
3173 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3174 for (i = 0; i < q->nr; i++)
3175 diff_summary(options->file, q->queue[i]);
3176 separator++;
3177 }
3179 if (output_format & DIFF_FORMAT_PATCH) {
3180 if (separator) {
3181 if (options->stat_sep) {
3182 /* attach patch instead of inline */
3183 fputs(options->stat_sep, options->file);
3184 } else {
3185 putc(options->line_termination, options->file);
3186 }
3187 }
3189 for (i = 0; i < q->nr; i++) {
3190 struct diff_filepair *p = q->queue[i];
3191 if (check_pair_status(p))
3192 diff_flush_patch(p, options);
3193 }
3194 }
3196 if (output_format & DIFF_FORMAT_CALLBACK)
3197 options->format_callback(q, options, options->format_callback_data);
3199 for (i = 0; i < q->nr; i++)
3200 diff_free_filepair(q->queue[i]);
3201 free_queue:
3202 free(q->queue);
3203 q->queue = NULL;
3204 q->nr = q->alloc = 0;
3205 if (options->close_file)
3206 fclose(options->file);
3207 }
3209 static void diffcore_apply_filter(const char *filter)
3210 {
3211 int i;
3212 struct diff_queue_struct *q = &diff_queued_diff;
3213 struct diff_queue_struct outq;
3214 outq.queue = NULL;
3215 outq.nr = outq.alloc = 0;
3217 if (!filter)
3218 return;
3220 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3221 int found;
3222 for (i = found = 0; !found && i < q->nr; i++) {
3223 struct diff_filepair *p = q->queue[i];
3224 if (((p->status == DIFF_STATUS_MODIFIED) &&
3225 ((p->score &&
3226 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3227 (!p->score &&
3228 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3229 ((p->status != DIFF_STATUS_MODIFIED) &&
3230 strchr(filter, p->status)))
3231 found++;
3232 }
3233 if (found)
3234 return;
3236 /* otherwise we will clear the whole queue
3237 * by copying the empty outq at the end of this
3238 * function, but first clear the current entries
3239 * in the queue.
3240 */
3241 for (i = 0; i < q->nr; i++)
3242 diff_free_filepair(q->queue[i]);
3243 }
3244 else {
3245 /* Only the matching ones */
3246 for (i = 0; i < q->nr; i++) {
3247 struct diff_filepair *p = q->queue[i];
3249 if (((p->status == DIFF_STATUS_MODIFIED) &&
3250 ((p->score &&
3251 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3252 (!p->score &&
3253 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3254 ((p->status != DIFF_STATUS_MODIFIED) &&
3255 strchr(filter, p->status)))
3256 diff_q(&outq, p);
3257 else
3258 diff_free_filepair(p);
3259 }
3260 }
3261 free(q->queue);
3262 *q = outq;
3263 }
3265 /* Check whether two filespecs with the same mode and size are identical */
3266 static int diff_filespec_is_identical(struct diff_filespec *one,
3267 struct diff_filespec *two)
3268 {
3269 if (S_ISGITLINK(one->mode))
3270 return 0;
3271 if (diff_populate_filespec(one, 0))
3272 return 0;
3273 if (diff_populate_filespec(two, 0))
3274 return 0;
3275 return !memcmp(one->data, two->data, one->size);
3276 }
3278 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3279 {
3280 int i;
3281 struct diff_queue_struct *q = &diff_queued_diff;
3282 struct diff_queue_struct outq;
3283 outq.queue = NULL;
3284 outq.nr = outq.alloc = 0;
3286 for (i = 0; i < q->nr; i++) {
3287 struct diff_filepair *p = q->queue[i];
3289 /*
3290 * 1. Entries that come from stat info dirtyness
3291 * always have both sides (iow, not create/delete),
3292 * one side of the object name is unknown, with
3293 * the same mode and size. Keep the ones that
3294 * do not match these criteria. They have real
3295 * differences.
3296 *
3297 * 2. At this point, the file is known to be modified,
3298 * with the same mode and size, and the object
3299 * name of one side is unknown. Need to inspect
3300 * the identical contents.
3301 */
3302 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3303 !DIFF_FILE_VALID(p->two) ||
3304 (p->one->sha1_valid && p->two->sha1_valid) ||
3305 (p->one->mode != p->two->mode) ||
3306 diff_populate_filespec(p->one, 1) ||
3307 diff_populate_filespec(p->two, 1) ||
3308 (p->one->size != p->two->size) ||
3309 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3310 diff_q(&outq, p);
3311 else {
3312 /*
3313 * The caller can subtract 1 from skip_stat_unmatch
3314 * to determine how many paths were dirty only
3315 * due to stat info mismatch.
3316 */
3317 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3318 diffopt->skip_stat_unmatch++;
3319 diff_free_filepair(p);
3320 }
3321 }
3322 free(q->queue);
3323 *q = outq;
3324 }
3326 void diffcore_std(struct diff_options *options)
3327 {
3328 if (DIFF_OPT_TST(options, QUIET))
3329 return;
3331 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3332 diffcore_skip_stat_unmatch(options);
3333 if (options->break_opt != -1)
3334 diffcore_break(options->break_opt);
3335 if (options->detect_rename)
3336 diffcore_rename(options);
3337 if (options->break_opt != -1)
3338 diffcore_merge_broken();
3339 if (options->pickaxe)
3340 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3341 if (options->orderfile)
3342 diffcore_order(options->orderfile);
3343 diff_resolve_rename_copy();
3344 diffcore_apply_filter(options->filter);
3346 if (diff_queued_diff.nr)
3347 DIFF_OPT_SET(options, HAS_CHANGES);
3348 else
3349 DIFF_OPT_CLR(options, HAS_CHANGES);
3350 }
3352 int diff_result_code(struct diff_options *opt, int status)
3353 {
3354 int result = 0;
3355 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3356 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3357 return status;
3358 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3359 DIFF_OPT_TST(opt, HAS_CHANGES))
3360 result |= 01;
3361 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3362 DIFF_OPT_TST(opt, CHECK_FAILED))
3363 result |= 02;
3364 return result;
3365 }
3367 void diff_addremove(struct diff_options *options,
3368 int addremove, unsigned mode,
3369 const unsigned char *sha1,
3370 const char *base, const char *path)
3371 {
3372 char concatpath[PATH_MAX];
3373 struct diff_filespec *one, *two;
3375 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3376 return;
3378 /* This may look odd, but it is a preparation for
3379 * feeding "there are unchanged files which should
3380 * not produce diffs, but when you are doing copy
3381 * detection you would need them, so here they are"
3382 * entries to the diff-core. They will be prefixed
3383 * with something like '=' or '*' (I haven't decided
3384 * which but should not make any difference).
3385 * Feeding the same new and old to diff_change()
3386 * also has the same effect.
3387 * Before the final output happens, they are pruned after
3388 * merged into rename/copy pairs as appropriate.
3389 */
3390 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3391 addremove = (addremove == '+' ? '-' :
3392 addremove == '-' ? '+' : addremove);
3394 if (!path) path = "";
3395 sprintf(concatpath, "%s%s", base, path);
3397 if (options->prefix &&
3398 strncmp(concatpath, options->prefix, options->prefix_length))
3399 return;
3401 one = alloc_filespec(concatpath);
3402 two = alloc_filespec(concatpath);
3404 if (addremove != '+')
3405 fill_filespec(one, sha1, mode);
3406 if (addremove != '-')
3407 fill_filespec(two, sha1, mode);
3409 diff_queue(&diff_queued_diff, one, two);
3410 DIFF_OPT_SET(options, HAS_CHANGES);
3411 }
3413 void diff_change(struct diff_options *options,
3414 unsigned old_mode, unsigned new_mode,
3415 const unsigned char *old_sha1,
3416 const unsigned char *new_sha1,
3417 const char *base, const char *path)
3418 {
3419 char concatpath[PATH_MAX];
3420 struct diff_filespec *one, *two;
3422 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3423 && S_ISGITLINK(new_mode))
3424 return;
3426 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3427 unsigned tmp;
3428 const unsigned char *tmp_c;
3429 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3430 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3431 }
3432 if (!path) path = "";
3433 sprintf(concatpath, "%s%s", base, path);
3435 if (options->prefix &&
3436 strncmp(concatpath, options->prefix, options->prefix_length))
3437 return;
3439 one = alloc_filespec(concatpath);
3440 two = alloc_filespec(concatpath);
3441 fill_filespec(one, old_sha1, old_mode);
3442 fill_filespec(two, new_sha1, new_mode);
3444 diff_queue(&diff_queued_diff, one, two);
3445 DIFF_OPT_SET(options, HAS_CHANGES);
3446 }
3448 void diff_unmerge(struct diff_options *options,
3449 const char *path,
3450 unsigned mode, const unsigned char *sha1)
3451 {
3452 struct diff_filespec *one, *two;
3454 if (options->prefix &&
3455 strncmp(path, options->prefix, options->prefix_length))
3456 return;
3458 one = alloc_filespec(path);
3459 two = alloc_filespec(path);
3460 fill_filespec(one, sha1, mode);
3461 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3462 }