![[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 = 100;
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)
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);
170 }
172 int git_diff_basic_config(const char *var, const char *value)
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);
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 fputs(set, file);
518 fwrite(line, len, 1, file);
519 fputs(reset, file);
520 }
522 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
523 {
524 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
525 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
527 if (!*ws)
528 emit_line(ecbdata->file, set, reset, line, len);
529 else {
530 /* Emit just the prefix, then the rest. */
531 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
532 (void)check_and_emit_line(line + ecbdata->nparents,
533 len - ecbdata->nparents, ecbdata->ws_rule,
534 ecbdata->file, set, reset, ws);
535 }
536 }
538 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
539 {
540 const char *cp;
541 unsigned long allot;
542 size_t l = len;
544 if (ecb->truncate)
545 return ecb->truncate(line, len);
546 cp = line;
547 allot = l;
548 while (0 < l) {
549 (void) utf8_width(&cp, &l);
550 if (!cp)
551 break; /* truncated in the middle? */
552 }
553 return allot - l;
554 }
556 static void fn_out_consume(void *priv, char *line, unsigned long len)
557 {
558 int i;
559 int color;
560 struct emit_callback *ecbdata = priv;
561 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
562 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
563 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
565 *(ecbdata->found_changesp) = 1;
567 if (ecbdata->label_path[0]) {
568 const char *name_a_tab, *name_b_tab;
570 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
571 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
573 fprintf(ecbdata->file, "%s--- %s%s%s\n",
574 meta, ecbdata->label_path[0], reset, name_a_tab);
575 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
576 meta, ecbdata->label_path[1], reset, name_b_tab);
577 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
578 }
580 /* This is not really necessary for now because
581 * this codepath only deals with two-way diffs.
582 */
583 for (i = 0; i < len && line[i] == '@'; i++)
584 ;
585 if (2 <= i && i < len && line[i] == ' ') {
586 ecbdata->nparents = i - 1;
587 len = sane_truncate_line(ecbdata, line, len);
588 emit_line(ecbdata->file,
589 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
590 reset, line, len);
591 if (line[len-1] != '\n')
592 putc('\n', ecbdata->file);
593 return;
594 }
596 if (len < ecbdata->nparents) {
597 emit_line(ecbdata->file, reset, reset, line, len);
598 return;
599 }
601 color = DIFF_PLAIN;
602 if (ecbdata->diff_words && ecbdata->nparents != 1)
603 /* fall back to normal diff */
604 free_diff_words_data(ecbdata);
605 if (ecbdata->diff_words) {
606 if (line[0] == '-') {
607 diff_words_append(line, len,
608 &ecbdata->diff_words->minus);
609 return;
610 } else if (line[0] == '+') {
611 diff_words_append(line, len,
612 &ecbdata->diff_words->plus);
613 return;
614 }
615 if (ecbdata->diff_words->minus.text.size ||
616 ecbdata->diff_words->plus.text.size)
617 diff_words_show(ecbdata->diff_words);
618 line++;
619 len--;
620 emit_line(ecbdata->file, plain, reset, line, len);
621 return;
622 }
623 for (i = 0; i < ecbdata->nparents && len; i++) {
624 if (line[i] == '-')
625 color = DIFF_FILE_OLD;
626 else if (line[i] == '+')
627 color = DIFF_FILE_NEW;
628 }
630 if (color != DIFF_FILE_NEW) {
631 emit_line(ecbdata->file,
632 diff_get_color(ecbdata->color_diff, color),
633 reset, line, len);
634 return;
635 }
636 emit_add_line(reset, ecbdata, line, len);
637 }
639 static char *pprint_rename(const char *a, const char *b)
640 {
641 const char *old = a;
642 const char *new = b;
643 struct strbuf name;
644 int pfx_length, sfx_length;
645 int len_a = strlen(a);
646 int len_b = strlen(b);
647 int a_midlen, b_midlen;
648 int qlen_a = quote_c_style(a, NULL, NULL, 0);
649 int qlen_b = quote_c_style(b, NULL, NULL, 0);
651 strbuf_init(&name, 0);
652 if (qlen_a || qlen_b) {
653 quote_c_style(a, &name, NULL, 0);
654 strbuf_addstr(&name, " => ");
655 quote_c_style(b, &name, NULL, 0);
656 return strbuf_detach(&name, NULL);
657 }
659 /* Find common prefix */
660 pfx_length = 0;
661 while (*old && *new && *old == *new) {
662 if (*old == '/')
663 pfx_length = old - a + 1;
664 old++;
665 new++;
666 }
668 /* Find common suffix */
669 old = a + len_a;
670 new = b + len_b;
671 sfx_length = 0;
672 while (a <= old && b <= new && *old == *new) {
673 if (*old == '/')
674 sfx_length = len_a - (old - a);
675 old--;
676 new--;
677 }
679 /*
680 * pfx{mid-a => mid-b}sfx
681 * {pfx-a => pfx-b}sfx
682 * pfx{sfx-a => sfx-b}
683 * name-a => name-b
684 */
685 a_midlen = len_a - pfx_length - sfx_length;
686 b_midlen = len_b - pfx_length - sfx_length;
687 if (a_midlen < 0)
688 a_midlen = 0;
689 if (b_midlen < 0)
690 b_midlen = 0;
692 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
693 if (pfx_length + sfx_length) {
694 strbuf_add(&name, a, pfx_length);
695 strbuf_addch(&name, '{');
696 }
697 strbuf_add(&name, a + pfx_length, a_midlen);
698 strbuf_addstr(&name, " => ");
699 strbuf_add(&name, b + pfx_length, b_midlen);
700 if (pfx_length + sfx_length) {
701 strbuf_addch(&name, '}');
702 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
703 }
704 return strbuf_detach(&name, NULL);
705 }
707 struct diffstat_t {
708 struct xdiff_emit_state xm;
710 int nr;
711 int alloc;
712 struct diffstat_file {
713 char *from_name;
714 char *name;
715 char *print_name;
716 unsigned is_unmerged:1;
717 unsigned is_binary:1;
718 unsigned is_renamed:1;
719 unsigned int added, deleted;
720 } **files;
721 };
723 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
724 const char *name_a,
725 const char *name_b)
726 {
727 struct diffstat_file *x;
728 x = xcalloc(sizeof (*x), 1);
729 if (diffstat->nr == diffstat->alloc) {
730 diffstat->alloc = alloc_nr(diffstat->alloc);
731 diffstat->files = xrealloc(diffstat->files,
732 diffstat->alloc * sizeof(x));
733 }
734 diffstat->files[diffstat->nr++] = x;
735 if (name_b) {
736 x->from_name = xstrdup(name_a);
737 x->name = xstrdup(name_b);
738 x->is_renamed = 1;
739 }
740 else {
741 x->from_name = NULL;
742 x->name = xstrdup(name_a);
743 }
744 return x;
745 }
747 static void diffstat_consume(void *priv, char *line, unsigned long len)
748 {
749 struct diffstat_t *diffstat = priv;
750 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
752 if (line[0] == '+')
753 x->added++;
754 else if (line[0] == '-')
755 x->deleted++;
756 }
758 const char mime_boundary_leader[] = "------------";
760 static int scale_linear(int it, int width, int max_change)
761 {
762 /*
763 * make sure that at least one '-' is printed if there were deletions,
764 * and likewise for '+'.
765 */
766 if (max_change < 2)
767 return it;
768 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
769 }
771 static void show_name(FILE *file,
772 const char *prefix, const char *name, int len,
773 const char *reset, const char *set)
774 {
775 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
776 }
778 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
779 {
780 if (cnt <= 0)
781 return;
782 fprintf(file, "%s", set);
783 while (cnt--)
784 putc(ch, file);
785 fprintf(file, "%s", reset);
786 }
788 static void fill_print_name(struct diffstat_file *file)
789 {
790 char *pname;
792 if (file->print_name)
793 return;
795 if (!file->is_renamed) {
796 struct strbuf buf;
797 strbuf_init(&buf, 0);
798 if (quote_c_style(file->name, &buf, NULL, 0)) {
799 pname = strbuf_detach(&buf, NULL);
800 } else {
801 pname = file->name;
802 strbuf_release(&buf);
803 }
804 } else {
805 pname = pprint_rename(file->from_name, file->name);
806 }
807 file->print_name = pname;
808 }
810 static void show_stats(struct diffstat_t* data, struct diff_options *options)
811 {
812 int i, len, add, del, total, adds = 0, dels = 0;
813 int max_change = 0, max_len = 0;
814 int total_files = data->nr;
815 int width, name_width;
816 const char *reset, *set, *add_c, *del_c;
818 if (data->nr == 0)
819 return;
821 width = options->stat_width ? options->stat_width : 80;
822 name_width = options->stat_name_width ? options->stat_name_width : 50;
824 /* Sanity: give at least 5 columns to the graph,
825 * but leave at least 10 columns for the name.
826 */
827 if (width < name_width + 15) {
828 if (name_width <= 25)
829 width = name_width + 15;
830 else
831 name_width = width - 15;
832 }
834 /* Find the longest filename and max number of changes */
835 reset = diff_get_color_opt(options, DIFF_RESET);
836 set = diff_get_color_opt(options, DIFF_PLAIN);
837 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
838 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
840 for (i = 0; i < data->nr; i++) {
841 struct diffstat_file *file = data->files[i];
842 int change = file->added + file->deleted;
843 fill_print_name(file);
844 len = strlen(file->print_name);
845 if (max_len < len)
846 max_len = len;
848 if (file->is_binary || file->is_unmerged)
849 continue;
850 if (max_change < change)
851 max_change = change;
852 }
854 /* Compute the width of the graph part;
855 * 10 is for one blank at the beginning of the line plus
856 * " | count " between the name and the graph.
857 *
858 * From here on, name_width is the width of the name area,
859 * and width is the width of the graph area.
860 */
861 name_width = (name_width < max_len) ? name_width : max_len;
862 if (width < (name_width + 10) + max_change)
863 width = width - (name_width + 10);
864 else
865 width = max_change;
867 for (i = 0; i < data->nr; i++) {
868 const char *prefix = "";
869 char *name = data->files[i]->print_name;
870 int added = data->files[i]->added;
871 int deleted = data->files[i]->deleted;
872 int name_len;
874 /*
875 * "scale" the filename
876 */
877 len = name_width;
878 name_len = strlen(name);
879 if (name_width < name_len) {
880 char *slash;
881 prefix = "...";
882 len -= 3;
883 name += name_len - len;
884 slash = strchr(name, '/');
885 if (slash)
886 name = slash;
887 }
889 if (data->files[i]->is_binary) {
890 show_name(options->file, prefix, name, len, reset, set);
891 fprintf(options->file, " Bin ");
892 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
893 fprintf(options->file, " -> ");
894 fprintf(options->file, "%s%d%s", add_c, added, reset);
895 fprintf(options->file, " bytes");
896 fprintf(options->file, "\n");
897 continue;
898 }
899 else if (data->files[i]->is_unmerged) {
900 show_name(options->file, prefix, name, len, reset, set);
901 fprintf(options->file, " Unmerged\n");
902 continue;
903 }
904 else if (!data->files[i]->is_renamed &&
905 (added + deleted == 0)) {
906 total_files--;
907 continue;
908 }
910 /*
911 * scale the add/delete
912 */
913 add = added;
914 del = deleted;
915 total = add + del;
916 adds += add;
917 dels += del;
919 if (width <= max_change) {
920 add = scale_linear(add, width, max_change);
921 del = scale_linear(del, width, max_change);
922 total = add + del;
923 }
924 show_name(options->file, prefix, name, len, reset, set);
925 fprintf(options->file, "%5d ", added + deleted);
926 show_graph(options->file, '+', add, add_c, reset);
927 show_graph(options->file, '-', del, del_c, reset);
928 fprintf(options->file, "\n");
929 }
930 fprintf(options->file,
931 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
932 set, total_files, adds, dels, reset);
933 }
935 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
936 {
937 int i, adds = 0, dels = 0, total_files = data->nr;
939 if (data->nr == 0)
940 return;
942 for (i = 0; i < data->nr; i++) {
943 if (!data->files[i]->is_binary &&
944 !data->files[i]->is_unmerged) {
945 int added = data->files[i]->added;
946 int deleted= data->files[i]->deleted;
947 if (!data->files[i]->is_renamed &&
948 (added + deleted == 0)) {
949 total_files--;
950 } else {
951 adds += added;
952 dels += deleted;
953 }
954 }
955 }
956 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
957 total_files, adds, dels);
958 }
960 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
961 {
962 int i;
964 if (data->nr == 0)
965 return;
967 for (i = 0; i < data->nr; i++) {
968 struct diffstat_file *file = data->files[i];
970 if (file->is_binary)
971 fprintf(options->file, "-\t-\t");
972 else
973 fprintf(options->file,
974 "%d\t%d\t", file->added, file->deleted);
975 if (options->line_termination) {
976 fill_print_name(file);
977 if (!file->is_renamed)
978 write_name_quoted(file->name, options->file,
979 options->line_termination);
980 else {
981 fputs(file->print_name, options->file);
982 putc(options->line_termination, options->file);
983 }
984 } else {
985 if (file->is_renamed) {
986 putc('\0', options->file);
987 write_name_quoted(file->from_name, options->file, '\0');
988 }
989 write_name_quoted(file->name, options->file, '\0');
990 }
991 }
992 }
994 struct diffstat_dir {
995 struct diffstat_file **files;
996 int nr, percent, cumulative;
997 };
999 static long gather_dirstat(FILE *file, struct diffstat_dir *dir, unsigned long changed, const char *base, int baselen)
1000 {
1001 unsigned long this_dir = 0;
1002 unsigned int sources = 0;
1004 while (dir->nr) {
1005 struct diffstat_file *f = *dir->files;
1006 int namelen = strlen(f->name);
1007 unsigned long this;
1008 char *slash;
1010 if (namelen < baselen)
1011 break;
1012 if (memcmp(f->name, base, baselen))
1013 break;
1014 slash = strchr(f->name + baselen, '/');
1015 if (slash) {
1016 int newbaselen = slash + 1 - f->name;
1017 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1018 sources++;
1019 } else {
1020 if (f->is_unmerged || f->is_binary)
1021 this = 0;
1022 else
1023 this = f->added + f->deleted;
1024 dir->files++;
1025 dir->nr--;
1026 sources += 2;
1027 }
1028 this_dir += this;
1029 }
1031 /*
1032 * We don't report dirstat's for
1033 * - the top level
1034 * - or cases where everything came from a single directory
1035 * under this directory (sources == 1).
1036 */
1037 if (baselen && sources != 1) {
1038 int permille = this_dir * 1000 / changed;
1039 if (permille) {
1040 int percent = permille / 10;
1041 if (percent >= dir->percent) {
1042 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1043 if (!dir->cumulative)
1044 return 0;
1045 }
1046 }
1047 }
1048 return this_dir;
1049 }
1051 static void show_dirstat(struct diffstat_t *data, struct diff_options *options)
1052 {
1053 int i;
1054 unsigned long changed;
1055 struct diffstat_dir dir;
1057 /* Calculate total changes */
1058 changed = 0;
1059 for (i = 0; i < data->nr; i++) {
1060 if (data->files[i]->is_binary || data->files[i]->is_unmerged)
1061 continue;
1062 changed += data->files[i]->added;
1063 changed += data->files[i]->deleted;
1064 }
1066 /* This can happen even with many files, if everything was renames */
1067 if (!changed)
1068 return;
1070 /* Show all directories with more than x% of the changes */
1071 dir.files = data->files;
1072 dir.nr = data->nr;
1073 dir.percent = options->dirstat_percent;
1074 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1075 gather_dirstat(options->file, &dir, changed, "", 0);
1076 }
1078 static void free_diffstat_info(struct diffstat_t *diffstat)
1079 {
1080 int i;
1081 for (i = 0; i < diffstat->nr; i++) {
1082 struct diffstat_file *f = diffstat->files[i];
1083 if (f->name != f->print_name)
1084 free(f->print_name);
1085 free(f->name);
1086 free(f->from_name);
1087 free(f);
1088 }
1089 free(diffstat->files);
1090 }
1092 struct checkdiff_t {
1093 struct xdiff_emit_state xm;
1094 const char *filename;
1095 int lineno, color_diff;
1096 unsigned ws_rule;
1097 unsigned status;
1098 FILE *file;
1099 };
1101 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1102 {
1103 struct checkdiff_t *data = priv;
1104 const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1105 const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1106 const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1107 char *err;
1109 if (line[0] == '+') {
1110 data->lineno++;
1111 data->status = check_and_emit_line(line + 1, len - 1,
1112 data->ws_rule, NULL, NULL, NULL, NULL);
1113 if (!data->status)
1114 return;
1115 err = whitespace_error_string(data->status);
1116 fprintf(data->file, "%s:%d: %s.\n", data->filename, data->lineno, err);
1117 free(err);
1118 emit_line(data->file, set, reset, line, 1);
1119 (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
1120 data->file, set, reset, ws);
1121 } else if (line[0] == ' ')
1122 data->lineno++;
1123 else if (line[0] == '@') {
1124 char *plus = strchr(line, '+');
1125 if (plus)
1126 data->lineno = strtol(plus, NULL, 10) - 1;
1127 else
1128 die("invalid diff");
1129 }
1130 }
1132 static unsigned char *deflate_it(char *data,
1133 unsigned long size,
1134 unsigned long *result_size)
1135 {
1136 int bound;
1137 unsigned char *deflated;
1138 z_stream stream;
1140 memset(&stream, 0, sizeof(stream));
1141 deflateInit(&stream, zlib_compression_level);
1142 bound = deflateBound(&stream, size);
1143 deflated = xmalloc(bound);
1144 stream.next_out = deflated;
1145 stream.avail_out = bound;
1147 stream.next_in = (unsigned char *)data;
1148 stream.avail_in = size;
1149 while (deflate(&stream, Z_FINISH) == Z_OK)
1150 ; /* nothing */
1151 deflateEnd(&stream);
1152 *result_size = stream.total_out;
1153 return deflated;
1154 }
1156 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1157 {
1158 void *cp;
1159 void *delta;
1160 void *deflated;
1161 void *data;
1162 unsigned long orig_size;
1163 unsigned long delta_size;
1164 unsigned long deflate_size;
1165 unsigned long data_size;
1167 /* We could do deflated delta, or we could do just deflated two,
1168 * whichever is smaller.
1169 */
1170 delta = NULL;
1171 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1172 if (one->size && two->size) {
1173 delta = diff_delta(one->ptr, one->size,
1174 two->ptr, two->size,
1175 &delta_size, deflate_size);
1176 if (delta) {
1177 void *to_free = delta;
1178 orig_size = delta_size;
1179 delta = deflate_it(delta, delta_size, &delta_size);
1180 free(to_free);
1181 }
1182 }
1184 if (delta && delta_size < deflate_size) {
1185 fprintf(file, "delta %lu\n", orig_size);
1186 free(deflated);
1187 data = delta;
1188 data_size = delta_size;
1189 }
1190 else {
1191 fprintf(file, "literal %lu\n", two->size);
1192 free(delta);
1193 data = deflated;
1194 data_size = deflate_size;
1195 }
1197 /* emit data encoded in base85 */
1198 cp = data;
1199 while (data_size) {
1200 int bytes = (52 < data_size) ? 52 : data_size;
1201 char line[70];
1202 data_size -= bytes;
1203 if (bytes <= 26)
1204 line[0] = bytes + 'A' - 1;
1205 else
1206 line[0] = bytes - 26 + 'a' - 1;
1207 encode_85(line + 1, cp, bytes);
1208 cp = (char *) cp + bytes;
1209 fputs(line, file);
1210 fputc('\n', file);
1211 }
1212 fprintf(file, "\n");
1213 free(data);
1214 }
1216 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1217 {
1218 fprintf(file, "GIT binary patch\n");
1219 emit_binary_diff_body(file, one, two);
1220 emit_binary_diff_body(file, two, one);
1221 }
1223 static void setup_diff_attr_check(struct git_attr_check *check)
1224 {
1225 static struct git_attr *attr_diff;
1227 if (!attr_diff) {
1228 attr_diff = git_attr("diff", 4);
1229 }
1230 check[0].attr = attr_diff;
1231 }
1233 static void diff_filespec_check_attr(struct diff_filespec *one)
1234 {
1235 struct git_attr_check attr_diff_check;
1236 int check_from_data = 0;
1238 if (one->checked_attr)
1239 return;
1241 setup_diff_attr_check(&attr_diff_check);
1242 one->is_binary = 0;
1243 one->funcname_pattern_ident = NULL;
1245 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1246 const char *value;
1248 /* binaryness */
1249 value = attr_diff_check.value;
1250 if (ATTR_TRUE(value))
1251 ;
1252 else if (ATTR_FALSE(value))
1253 one->is_binary = 1;
1254 else
1255 check_from_data = 1;
1257 /* funcname pattern ident */
1258 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1259 ;
1260 else
1261 one->funcname_pattern_ident = value;
1262 }
1264 if (check_from_data) {
1265 if (!one->data && DIFF_FILE_VALID(one))
1266 diff_populate_filespec(one, 0);
1268 if (one->data)
1269 one->is_binary = buffer_is_binary(one->data, one->size);
1270 }
1271 }
1273 int diff_filespec_is_binary(struct diff_filespec *one)
1274 {
1275 diff_filespec_check_attr(one);
1276 return one->is_binary;
1277 }
1279 static const char *funcname_pattern(const char *ident)
1280 {
1281 struct funcname_pattern *pp;
1283 for (pp = funcname_pattern_list; pp; pp = pp->next)
1284 if (!strcmp(ident, pp->name))
1285 return pp->pattern;
1286 return NULL;
1287 }
1289 static struct builtin_funcname_pattern {
1290 const char *name;
1291 const char *pattern;
1292 } builtin_funcname_pattern[] = {
1293 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1294 "new\\|return\\|switch\\|throw\\|while\\)\n"
1295 "^[ ]*\\(\\([ ]*"
1296 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1297 "[ ]*([^;]*\\)$" },
1298 { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1299 };
1301 static const char *diff_funcname_pattern(struct diff_filespec *one)
1302 {
1303 const char *ident, *pattern;
1304 int i;
1306 diff_filespec_check_attr(one);
1307 ident = one->funcname_pattern_ident;
1309 if (!ident)
1310 /*
1311 * If the config file has "funcname.default" defined, that
1312 * regexp is used; otherwise NULL is returned and xemit uses
1313 * the built-in default.
1314 */
1315 return funcname_pattern("default");
1317 /* Look up custom "funcname.$ident" regexp from config. */
1318 pattern = funcname_pattern(ident);
1319 if (pattern)
1320 return pattern;
1322 /*
1323 * And define built-in fallback patterns here. Note that
1324 * these can be overridden by the user's config settings.
1325 */
1326 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1327 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1328 return builtin_funcname_pattern[i].pattern;
1330 return NULL;
1331 }
1333 static void builtin_diff(const char *name_a,
1334 const char *name_b,
1335 struct diff_filespec *one,
1336 struct diff_filespec *two,
1337 const char *xfrm_msg,
1338 struct diff_options *o,
1339 int complete_rewrite)
1340 {
1341 mmfile_t mf1, mf2;
1342 const char *lbl[2];
1343 char *a_one, *b_two;
1344 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1345 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1347 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1348 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1349 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1350 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1351 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1352 if (lbl[0][0] == '/') {
1353 /* /dev/null */
1354 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1355 if (xfrm_msg && xfrm_msg[0])
1356 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1357 }
1358 else if (lbl[1][0] == '/') {
1359 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1360 if (xfrm_msg && xfrm_msg[0])
1361 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1362 }
1363 else {
1364 if (one->mode != two->mode) {
1365 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1366 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1367 }
1368 if (xfrm_msg && xfrm_msg[0])
1369 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1370 /*
1371 * we do not run diff between different kind
1372 * of objects.
1373 */
1374 if ((one->mode ^ two->mode) & S_IFMT)
1375 goto free_ab_and_return;
1376 if (complete_rewrite) {
1377 emit_rewrite_diff(name_a, name_b, one, two, o);
1378 o->found_changes = 1;
1379 goto free_ab_and_return;
1380 }
1381 }
1383 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1384 die("unable to read files to diff");
1386 if (!DIFF_OPT_TST(o, TEXT) &&
1387 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1388 /* Quite common confusing case */
1389 if (mf1.size == mf2.size &&
1390 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1391 goto free_ab_and_return;
1392 if (DIFF_OPT_TST(o, BINARY))
1393 emit_binary_diff(o->file, &mf1, &mf2);
1394 else
1395 fprintf(o->file, "Binary files %s and %s differ\n",
1396 lbl[0], lbl[1]);
1397 o->found_changes = 1;
1398 }
1399 else {
1400 /* Crazy xdl interfaces.. */
1401 const char *diffopts = getenv("GIT_DIFF_OPTS");
1402 xpparam_t xpp;
1403 xdemitconf_t xecfg;
1404 xdemitcb_t ecb;
1405 struct emit_callback ecbdata;
1406 const char *funcname_pattern;
1408 funcname_pattern = diff_funcname_pattern(one);
1409 if (!funcname_pattern)
1410 funcname_pattern = diff_funcname_pattern(two);
1412 memset(&xecfg, 0, sizeof(xecfg));
1413 memset(&ecbdata, 0, sizeof(ecbdata));
1414 ecbdata.label_path = lbl;
1415 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1416 ecbdata.found_changesp = &o->found_changes;
1417 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1418 ecbdata.file = o->file;
1419 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1420 xecfg.ctxlen = o->context;
1421 xecfg.flags = XDL_EMIT_FUNCNAMES;
1422 if (funcname_pattern)
1423 xdiff_set_find_func(&xecfg, funcname_pattern);
1424 if (!diffopts)
1425 ;
1426 else if (!prefixcmp(diffopts, "--unified="))
1427 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1428 else if (!prefixcmp(diffopts, "-u"))
1429 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1430 ecb.outf = xdiff_outf;
1431 ecb.priv = &ecbdata;
1432 ecbdata.xm.consume = fn_out_consume;
1433 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1434 ecbdata.diff_words =
1435 xcalloc(1, sizeof(struct diff_words_data));
1436 ecbdata.diff_words->file = o->file;
1437 }
1438 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1439 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1440 free_diff_words_data(&ecbdata);
1441 }
1443 free_ab_and_return:
1444 diff_free_filespec_data(one);
1445 diff_free_filespec_data(two);
1446 free(a_one);
1447 free(b_two);
1448 return;
1449 }
1451 static void builtin_diffstat(const char *name_a, const char *name_b,
1452 struct diff_filespec *one,
1453 struct diff_filespec *two,
1454 struct diffstat_t *diffstat,
1455 struct diff_options *o,
1456 int complete_rewrite)
1457 {
1458 mmfile_t mf1, mf2;
1459 struct diffstat_file *data;
1461 data = diffstat_add(diffstat, name_a, name_b);
1463 if (!one || !two) {
1464 data->is_unmerged = 1;
1465 return;
1466 }
1467 if (complete_rewrite) {
1468 diff_populate_filespec(one, 0);
1469 diff_populate_filespec(two, 0);
1470 data->deleted = count_lines(one->data, one->size);
1471 data->added = count_lines(two->data, two->size);
1472 goto free_and_return;
1473 }
1474 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1475 die("unable to read files to diff");
1477 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1478 data->is_binary = 1;
1479 data->added = mf2.size;
1480 data->deleted = mf1.size;
1481 } else {
1482 /* Crazy xdl interfaces.. */
1483 xpparam_t xpp;
1484 xdemitconf_t xecfg;
1485 xdemitcb_t ecb;
1487 memset(&xecfg, 0, sizeof(xecfg));
1488 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1489 ecb.outf = xdiff_outf;
1490 ecb.priv = diffstat;
1491 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1492 }
1494 free_and_return:
1495 diff_free_filespec_data(one);
1496 diff_free_filespec_data(two);
1497 }
1499 static void builtin_checkdiff(const char *name_a, const char *name_b,
1500 const char *attr_path,
1501 struct diff_filespec *one,
1502 struct diff_filespec *two, struct diff_options *o)
1503 {
1504 mmfile_t mf1, mf2;
1505 struct checkdiff_t data;
1507 if (!two)
1508 return;
1510 memset(&data, 0, sizeof(data));
1511 data.xm.consume = checkdiff_consume;
1512 data.filename = name_b ? name_b : name_a;
1513 data.lineno = 0;
1514 data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1515 data.ws_rule = whitespace_rule(attr_path);
1516 data.file = o->file;
1518 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1519 die("unable to read files to diff");
1521 if (diff_filespec_is_binary(two))
1522 goto free_and_return;
1523 else {
1524 /* Crazy xdl interfaces.. */
1525 xpparam_t xpp;
1526 xdemitconf_t xecfg;
1527 xdemitcb_t ecb;
1529 memset(&xecfg, 0, sizeof(xecfg));
1530 xpp.flags = XDF_NEED_MINIMAL;
1531 ecb.outf = xdiff_outf;
1532 ecb.priv = &data;
1533 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1534 }
1535 free_and_return:
1536 diff_free_filespec_data(one);
1537 diff_free_filespec_data(two);
1538 if (data.status)
1539 DIFF_OPT_SET(o, CHECK_FAILED);
1540 }
1542 struct diff_filespec *alloc_filespec(const char *path)
1543 {
1544 int namelen = strlen(path);
1545 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1547 memset(spec, 0, sizeof(*spec));
1548 spec->path = (char *)(spec + 1);
1549 memcpy(spec->path, path, namelen+1);
1550 spec->count = 1;
1551 return spec;
1552 }
1554 void free_filespec(struct diff_filespec *spec)
1555 {
1556 if (!--spec->count) {
1557 diff_free_filespec_data(spec);
1558 free(spec);
1559 }
1560 }
1562 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1563 unsigned short mode)
1564 {
1565 if (mode) {
1566 spec->mode = canon_mode(mode);
1567 hashcpy(spec->sha1, sha1);
1568 spec->sha1_valid = !is_null_sha1(sha1);
1569 }
1570 }
1572 /*
1573 * Given a name and sha1 pair, if the index tells us the file in
1574 * the work tree has that object contents, return true, so that
1575 * prepare_temp_file() does not have to inflate and extract.
1576 */
1577 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1578 {
1579 struct cache_entry *ce;
1580 struct stat st;
1581 int pos, len;
1583 /* We do not read the cache ourselves here, because the
1584 * benchmark with my previous version that always reads cache
1585 * shows that it makes things worse for diff-tree comparing
1586 * two linux-2.6 kernel trees in an already checked out work
1587 * tree. This is because most diff-tree comparisons deal with
1588 * only a small number of files, while reading the cache is
1589 * expensive for a large project, and its cost outweighs the
1590 * savings we get by not inflating the object to a temporary
1591 * file. Practically, this code only helps when we are used
1592 * by diff-cache --cached, which does read the cache before
1593 * calling us.
1594 */
1595 if (!active_cache)
1596 return 0;
1598 /* We want to avoid the working directory if our caller
1599 * doesn't need the data in a normal file, this system
1600 * is rather slow with its stat/open/mmap/close syscalls,
1601 * and the object is contained in a pack file. The pack
1602 * is probably already open and will be faster to obtain
1603 * the data through than the working directory. Loose
1604 * objects however would tend to be slower as they need
1605 * to be individually opened and inflated.
1606 */
1607 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1608 return 0;
1610 len = strlen(name);
1611 pos = cache_name_pos(name, len);
1612 if (pos < 0)
1613 return 0;
1614 ce = active_cache[pos];
1616 /*
1617 * This is not the sha1 we are looking for, or
1618 * unreusable because it is not a regular file.
1619 */
1620 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1621 return 0;
1623 /*
1624 * If ce matches the file in the work tree, we can reuse it.
1625 */
1626 if (ce_uptodate(ce) ||
1627 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1628 return 1;
1630 return 0;
1631 }
1633 static int populate_from_stdin(struct diff_filespec *s)
1634 {
1635 struct strbuf buf;
1636 size_t size = 0;
1638 strbuf_init(&buf, 0);
1639 if (strbuf_read(&buf, 0, 0) < 0)
1640 return error("error while reading from stdin %s",
1641 strerror(errno));
1643 s->should_munmap = 0;
1644 s->data = strbuf_detach(&buf, &size);
1645 s->size = size;
1646 s->should_free = 1;
1647 return 0;
1648 }
1650 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1651 {
1652 int len;
1653 char *data = xmalloc(100);
1654 len = snprintf(data, 100,
1655 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1656 s->data = data;
1657 s->size = len;
1658 s->should_free = 1;
1659 if (size_only) {
1660 s->data = NULL;
1661 free(data);
1662 }
1663 return 0;
1664 }
1666 /*
1667 * While doing rename detection and pickaxe operation, we may need to
1668 * grab the data for the blob (or file) for our own in-core comparison.
1669 * diff_filespec has data and size fields for this purpose.
1670 */
1671 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1672 {
1673 int err = 0;
1674 if (!DIFF_FILE_VALID(s))
1675 die("internal error: asking to populate invalid file.");
1676 if (S_ISDIR(s->mode))
1677 return -1;
1679 if (s->data)
1680 return 0;
1682 if (size_only && 0 < s->size)
1683 return 0;
1685 if (S_ISGITLINK(s->mode))
1686 return diff_populate_gitlink(s, size_only);
1688 if (!s->sha1_valid ||
1689 reuse_worktree_file(s->path, s->sha1, 0)) {
1690 struct strbuf buf;
1691 struct stat st;
1692 int fd;
1694 if (!strcmp(s->path, "-"))
1695 return populate_from_stdin(s);
1697 if (lstat(s->path, &st) < 0) {
1698 if (errno == ENOENT) {
1699 err_empty:
1700 err = -1;
1701 empty:
1702 s->data = (char *)"";
1703 s->size = 0;
1704 return err;
1705 }
1706 }
1707 s->size = xsize_t(st.st_size);
1708 if (!s->size)
1709 goto empty;
1710 if (size_only)
1711 return 0;
1712 if (S_ISLNK(st.st_mode)) {
1713 int ret;
1714 s->data = xmalloc(s->size);
1715 s->should_free = 1;
1716 ret = readlink(s->path, s->data, s->size);
1717 if (ret < 0) {
1718 free(s->data);
1719 goto err_empty;
1720 }
1721 return 0;
1722 }
1723 fd = open(s->path, O_RDONLY);
1724 if (fd < 0)
1725 goto err_empty;
1726 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1727 close(fd);
1728 s->should_munmap = 1;
1730 /*
1731 * Convert from working tree format to canonical git format
1732 */
1733 strbuf_init(&buf, 0);
1734 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1735 size_t size = 0;
1736 munmap(s->data, s->size);
1737 s->should_munmap = 0;
1738 s->data = strbuf_detach(&buf, &size);
1739 s->size = size;
1740 s->should_free = 1;
1741 }
1742 }
1743 else {
1744 enum object_type type;
1745 if (size_only)
1746 type = sha1_object_info(s->sha1, &s->size);
1747 else {
1748 s->data = read_sha1_file(s->sha1, &type, &s->size);
1749 s->should_free = 1;
1750 }
1751 }
1752 return 0;
1753 }
1755 void diff_free_filespec_blob(struct diff_filespec *s)
1756 {
1757 if (s->should_free)
1758 free(s->data);
1759 else if (s->should_munmap)
1760 munmap(s->data, s->size);
1762 if (s->should_free || s->should_munmap) {
1763 s->should_free = s->should_munmap = 0;
1764 s->data = NULL;
1765 }
1766 }
1768 void diff_free_filespec_data(struct diff_filespec *s)
1769 {
1770 diff_free_filespec_blob(s);
1771 free(s->cnt_data);
1772 s->cnt_data = NULL;
1773 }
1775 static void prep_temp_blob(struct diff_tempfile *temp,
1776 void *blob,
1777 unsigned long size,
1778 const unsigned char *sha1,
1779 int mode)
1780 {
1781 int fd;
1783 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1784 if (fd < 0)
1785 die("unable to create temp-file: %s", strerror(errno));
1786 if (write_in_full(fd, blob, size) != size)
1787 die("unable to write temp-file");
1788 close(fd);
1789 temp->name = temp->tmp_path;
1790 strcpy(temp->hex, sha1_to_hex(sha1));
1791 temp->hex[40] = 0;
1792 sprintf(temp->mode, "%06o", mode);
1793 }
1795 static void prepare_temp_file(const char *name,
1796 struct diff_tempfile *temp,
1797 struct diff_filespec *one)
1798 {
1799 if (!DIFF_FILE_VALID(one)) {
1800 not_a_valid_file:
1801 /* A '-' entry produces this for file-2, and
1802 * a '+' entry produces this for file-1.
1803 */
1804 temp->name = "/dev/null";
1805 strcpy(temp->hex, ".");
1806 strcpy(temp->mode, ".");
1807 return;
1808 }
1810 if (!one->sha1_valid ||
1811 reuse_worktree_file(name, one->sha1, 1)) {
1812 struct stat st;
1813 if (lstat(name, &st) < 0) {
1814 if (errno == ENOENT)
1815 goto not_a_valid_file;
1816 die("stat(%s): %s", name, strerror(errno));
1817 }
1818 if (S_ISLNK(st.st_mode)) {
1819 int ret;
1820 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1821 size_t sz = xsize_t(st.st_size);
1822 if (sizeof(buf) <= st.st_size)
1823 die("symlink too long: %s", name);
1824 ret = readlink(name, buf, sz);
1825 if (ret < 0)
1826 die("readlink(%s)", name);
1827 prep_temp_blob(temp, buf, sz,
1828 (one->sha1_valid ?
1829 one->sha1 : null_sha1),
1830 (one->sha1_valid ?
1831 one->mode : S_IFLNK));
1832 }
1833 else {
1834 /* we can borrow from the file in the work tree */
1835 temp->name = name;
1836 if (!one->sha1_valid)
1837 strcpy(temp->hex, sha1_to_hex(null_sha1));
1838 else
1839 strcpy(temp->hex, sha1_to_hex(one->sha1));
1840 /* Even though we may sometimes borrow the
1841 * contents from the work tree, we always want
1842 * one->mode. mode is trustworthy even when
1843 * !(one->sha1_valid), as long as
1844 * DIFF_FILE_VALID(one).
1845 */
1846 sprintf(temp->mode, "%06o", one->mode);
1847 }
1848 return;
1849 }
1850 else {
1851 if (diff_populate_filespec(one, 0))
1852 die("cannot read data blob for %s", one->path);
1853 prep_temp_blob(temp, one->data, one->size,
1854 one->sha1, one->mode);
1855 }
1856 }
1858 static void remove_tempfile(void)
1859 {
1860 int i;
1862 for (i = 0; i < 2; i++)
1863 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1864 unlink(diff_temp[i].name);
1865 diff_temp[i].name = NULL;
1866 }
1867 }
1869 static void remove_tempfile_on_signal(int signo)
1870 {
1871 remove_tempfile();
1872 signal(SIGINT, SIG_DFL);
1873 raise(signo);
1874 }
1876 /* An external diff command takes:
1877 *
1878 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1879 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1880 *
1881 */
1882 static void run_external_diff(const char *pgm,
1883 const char *name,
1884 const char *other,
1885 struct diff_filespec *one,
1886 struct diff_filespec *two,
1887 const char *xfrm_msg,
1888 int complete_rewrite)
1889 {
1890 const char *spawn_arg[10];
1891 struct diff_tempfile *temp = diff_temp;
1892 int retval;
1893 static int atexit_asked = 0;
1894 const char *othername;
1895 const char **arg = &spawn_arg[0];
1897 othername = (other? other : name);
1898 if (one && two) {
1899 prepare_temp_file(name, &temp[0], one);
1900 prepare_temp_file(othername, &temp[1], two);
1901 if (! atexit_asked &&
1902 (temp[0].name == temp[0].tmp_path ||
1903 temp[1].name == temp[1].tmp_path)) {
1904 atexit_asked = 1;
1905 atexit(remove_tempfile);
1906 }
1907 signal(SIGINT, remove_tempfile_on_signal);
1908 }
1910 if (one && two) {
1911 *arg++ = pgm;
1912 *arg++ = name;
1913 *arg++ = temp[0].name;
1914 *arg++ = temp[0].hex;
1915 *arg++ = temp[0].mode;
1916 *arg++ = temp[1].name;
1917 *arg++ = temp[1].hex;
1918 *arg++ = temp[1].mode;
1919 if (other) {
1920 *arg++ = other;
1921 *arg++ = xfrm_msg;
1922 }
1923 } else {
1924 *arg++ = pgm;
1925 *arg++ = name;
1926 }
1927 *arg = NULL;
1928 fflush(NULL);
1929 retval = run_command_v_opt(spawn_arg, 0);
1930 remove_tempfile();
1931 if (retval) {
1932 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1933 exit(1);
1934 }
1935 }
1937 static const char *external_diff_attr(const char *name)
1938 {
1939 struct git_attr_check attr_diff_check;
1941 if (!name)
1942 return NULL;
1944 setup_diff_attr_check(&attr_diff_check);
1945 if (!git_checkattr(name, 1, &attr_diff_check)) {
1946 const char *value = attr_diff_check.value;
1947 if (!ATTR_TRUE(value) &&
1948 !ATTR_FALSE(value) &&
1949 !ATTR_UNSET(value)) {
1950 struct ll_diff_driver *drv;
1952 for (drv = user_diff; drv; drv = drv->next)
1953 if (!strcmp(drv->name, value))
1954 return drv->cmd;
1955 }
1956 }
1957 return NULL;
1958 }
1960 static void run_diff_cmd(const char *pgm,
1961 const char *name,
1962 const char *other,
1963 const char *attr_path,
1964 struct diff_filespec *one,
1965 struct diff_filespec *two,
1966 const char *xfrm_msg,
1967 struct diff_options *o,
1968 int complete_rewrite)
1969 {
1970 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
1971 pgm = NULL;
1972 else {
1973 const char *cmd = external_diff_attr(attr_path);
1974 if (cmd)
1975 pgm = cmd;
1976 }
1978 if (pgm) {
1979 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1980 complete_rewrite);
1981 return;
1982 }
1983 if (one && two)
1984 builtin_diff(name, other ? other : name,
1985 one, two, xfrm_msg, o, complete_rewrite);
1986 else
1987 fprintf(o->file, "* Unmerged path %s\n", name);
1988 }
1990 static void diff_fill_sha1_info(struct diff_filespec *one)
1991 {
1992 if (DIFF_FILE_VALID(one)) {
1993 if (!one->sha1_valid) {
1994 struct stat st;
1995 if (!strcmp(one->path, "-")) {
1996 hashcpy(one->sha1, null_sha1);
1997 return;
1998 }
1999 if (lstat(one->path, &st) < 0)
2000 die("stat %s", one->path);
2001 if (index_path(one->sha1, one->path, &st, 0))
2002 die("cannot hash %s\n", one->path);
2003 }
2004 }
2005 else
2006 hashclr(one->sha1);
2007 }
2009 static int similarity_index(struct diff_filepair *p)
2010 {
2011 return p->score * 100 / MAX_SCORE;
2012 }
2014 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2015 {
2016 /* Strip the prefix but do not molest /dev/null and absolute paths */
2017 if (*namep && **namep != '/')
2018 *namep += prefix_length;
2019 if (*otherp && **otherp != '/')
2020 *otherp += prefix_length;
2021 }
2023 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2024 {
2025 const char *pgm = external_diff();
2026 struct strbuf msg;
2027 char *xfrm_msg;
2028 struct diff_filespec *one = p->one;
2029 struct diff_filespec *two = p->two;
2030 const char *name;
2031 const char *other;
2032 const char *attr_path;
2033 int complete_rewrite = 0;
2035 name = p->one->path;
2036 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2037 attr_path = name;
2038 if (o->prefix_length)
2039 strip_prefix(o->prefix_length, &name, &other);
2041 if (DIFF_PAIR_UNMERGED(p)) {
2042 run_diff_cmd(pgm, name, NULL, attr_path,
2043 NULL, NULL, NULL, o, 0);
2044 return;
2045 }
2047 diff_fill_sha1_info(one);
2048 diff_fill_sha1_info(two);
2050 strbuf_init(&msg, PATH_MAX * 2 + 300);
2051 switch (p->status) {
2052 case DIFF_STATUS_COPIED:
2053 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2054 strbuf_addstr(&msg, "\ncopy from ");
2055 quote_c_style(name, &msg, NULL, 0);
2056 strbuf_addstr(&msg, "\ncopy to ");
2057 quote_c_style(other, &msg, NULL, 0);
2058 strbuf_addch(&msg, '\n');
2059 break;
2060 case DIFF_STATUS_RENAMED:
2061 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2062 strbuf_addstr(&msg, "\nrename from ");
2063 quote_c_style(name, &msg, NULL, 0);
2064 strbuf_addstr(&msg, "\nrename to ");
2065 quote_c_style(other, &msg, NULL, 0);
2066 strbuf_addch(&msg, '\n');
2067 break;
2068 case DIFF_STATUS_MODIFIED:
2069 if (p->score) {
2070 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2071 similarity_index(p));
2072 complete_rewrite = 1;
2073 break;
2074 }
2075 /* fallthru */
2076 default:
2077 /* nothing */
2078 ;
2079 }
2081 if (hashcmp(one->sha1, two->sha1)) {
2082 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2084 if (DIFF_OPT_TST(o, BINARY)) {
2085 mmfile_t mf;
2086 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2087 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2088 abbrev = 40;
2089 }
2090 strbuf_addf(&msg, "index %.*s..%.*s",
2091 abbrev, sha1_to_hex(one->sha1),
2092 abbrev, sha1_to_hex(two->sha1));
2093 if (one->mode == two->mode)
2094 strbuf_addf(&msg, " %06o", one->mode);
2095 strbuf_addch(&msg, '\n');
2096 }
2098 if (msg.len)
2099 strbuf_setlen(&msg, msg.len - 1);
2100 xfrm_msg = msg.len ? msg.buf : NULL;
2102 if (!pgm &&
2103 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2104 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2105 /* a filepair that changes between file and symlink
2106 * needs to be split into deletion and creation.
2107 */
2108 struct diff_filespec *null = alloc_filespec(two->path);
2109 run_diff_cmd(NULL, name, other, attr_path,
2110 one, null, xfrm_msg, o, 0);
2111 free(null);
2112 null = alloc_filespec(one->path);
2113 run_diff_cmd(NULL, name, other, attr_path,
2114 null, two, xfrm_msg, o, 0);
2115 free(null);
2116 }
2117 else
2118 run_diff_cmd(pgm, name, other, attr_path,
2119 one, two, xfrm_msg, o, complete_rewrite);
2121 strbuf_release(&msg);
2122 }
2124 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2125 struct diffstat_t *diffstat)
2126 {
2127 const char *name;
2128 const char *other;
2129 int complete_rewrite = 0;
2131 if (DIFF_PAIR_UNMERGED(p)) {
2132 /* unmerged */
2133 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2134 return;
2135 }
2137 name = p->one->path;
2138 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2140 if (o->prefix_length)
2141 strip_prefix(o->prefix_length, &name, &other);
2143 diff_fill_sha1_info(p->one);
2144 diff_fill_sha1_info(p->two);
2146 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2147 complete_rewrite = 1;
2148 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2149 }
2151 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2152 {
2153 const char *name;
2154 const char *other;
2155 const char *attr_path;
2157 if (DIFF_PAIR_UNMERGED(p)) {
2158 /* unmerged */
2159 return;
2160 }
2162 name = p->one->path;
2163 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2164 attr_path = other ? other : name;
2166 if (o->prefix_length)
2167 strip_prefix(o->prefix_length, &name, &other);
2169 diff_fill_sha1_info(p->one);
2170 diff_fill_sha1_info(p->two);
2172 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2173 }
2175 void diff_setup(struct diff_options *options)
2176 {
2177 memset(options, 0, sizeof(*options));
2179 options->file = stdout;
2181 options->line_termination = '\n';
2182 options->break_opt = -1;
2183 options->rename_limit = -1;
2184 options->dirstat_percent = 3;
2185 options->context = 3;
2186 options->msg_sep = "";
2188 options->change = diff_change;
2189 options->add_remove = diff_addremove;
2190 if (diff_use_color_default > 0)
2191 DIFF_OPT_SET(options, COLOR_DIFF);
2192 else
2193 DIFF_OPT_CLR(options, COLOR_DIFF);
2194 options->detect_rename = diff_detect_rename_default;
2196 options->a_prefix = "a/";
2197 options->b_prefix = "b/";
2198 }
2200 int diff_setup_done(struct diff_options *options)
2201 {
2202 int count = 0;
2204 if (options->output_format & DIFF_FORMAT_NAME)
2205 count++;
2206 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2207 count++;
2208 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2209 count++;
2210 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2211 count++;
2212 if (count > 1)
2213 die("--name-only, --name-status, --check and -s are mutually exclusive");
2215 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2216 options->detect_rename = DIFF_DETECT_COPY;
2218 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2219 options->prefix = NULL;
2220 if (options->prefix)
2221 options->prefix_length = strlen(options->prefix);
2222 else
2223 options->prefix_length = 0;
2225 if (options->output_format & (DIFF_FORMAT_NAME |
2226 DIFF_FORMAT_NAME_STATUS |
2227 DIFF_FORMAT_CHECKDIFF |
2228 DIFF_FORMAT_NO_OUTPUT))
2229 options->output_format &= ~(DIFF_FORMAT_RAW |
2230 DIFF_FORMAT_NUMSTAT |
2231 DIFF_FORMAT_DIFFSTAT |
2232 DIFF_FORMAT_SHORTSTAT |
2233 DIFF_FORMAT_DIRSTAT |
2234 DIFF_FORMAT_SUMMARY |
2235 DIFF_FORMAT_PATCH);
2237 /*
2238 * These cases always need recursive; we do not drop caller-supplied
2239 * recursive bits for other formats here.
2240 */
2241 if (options->output_format & (DIFF_FORMAT_PATCH |
2242 DIFF_FORMAT_NUMSTAT |
2243 DIFF_FORMAT_DIFFSTAT |
2244 DIFF_FORMAT_SHORTSTAT |
2245 DIFF_FORMAT_DIRSTAT |
2246 DIFF_FORMAT_SUMMARY |
2247 DIFF_FORMAT_CHECKDIFF))
2248 DIFF_OPT_SET(options, RECURSIVE);
2249 /*
2250 * Also pickaxe would not work very well if you do not say recursive
2251 */
2252 if (options->pickaxe)
2253 DIFF_OPT_SET(options, RECURSIVE);
2255 if (options->detect_rename && options->rename_limit < 0)
2256 options->rename_limit = diff_rename_limit_default;
2257 if (options->setup & DIFF_SETUP_USE_CACHE) {
2258 if (!active_cache)
2259 /* read-cache does not die even when it fails
2260 * so it is safe for us to do this here. Also
2261 * it does not smudge active_cache or active_nr
2262 * when it fails, so we do not have to worry about
2263 * cleaning it up ourselves either.
2264 */
2265 read_cache();
2266 }
2267 if (options->abbrev <= 0 || 40 < options->abbrev)
2268 options->abbrev = 40; /* full */
2270 /*
2271 * It does not make sense to show the first hit we happened
2272 * to have found. It does not make sense not to return with
2273 * exit code in such a case either.
2274 */
2275 if (DIFF_OPT_TST(options, QUIET)) {
2276 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2277 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2278 }
2280 /*
2281 * If we postprocess in diffcore, we cannot simply return
2282 * upon the first hit. We need to run diff as usual.
2283 */
2284 if (options->pickaxe || options->filter)
2285 DIFF_OPT_CLR(options, QUIET);
2287 return 0;
2288 }
2290 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2291 {
2292 char c, *eq;
2293 int len;
2295 if (*arg != '-')
2296 return 0;
2297 c = *++arg;
2298 if (!c)
2299 return 0;
2300 if (c == arg_short) {
2301 c = *++arg;
2302 if (!c)
2303 return 1;
2304 if (val && isdigit(c)) {
2305 char *end;
2306 int n = strtoul(arg, &end, 10);
2307 if (*end)
2308 return 0;
2309 *val = n;
2310 return 1;
2311 }
2312 return 0;
2313 }
2314 if (c != '-')
2315 return 0;
2316 arg++;
2317 eq = strchr(arg, '=');
2318 if (eq)
2319 len = eq - arg;
2320 else
2321 len = strlen(arg);
2322 if (!len || strncmp(arg, arg_long, len))
2323 return 0;
2324 if (eq) {
2325 int n;
2326 char *end;
2327 if (!isdigit(*++eq))
2328 return 0;
2329 n = strtoul(eq, &end, 10);
2330 if (*end)
2331 return 0;
2332 *val = n;
2333 }
2334 return 1;
2335 }
2337 static int diff_scoreopt_parse(const char *opt);
2339 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2340 {
2341 const char *arg = av[0];
2343 /* Output format options */
2344 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2345 options->output_format |= DIFF_FORMAT_PATCH;
2346 else if (opt_arg(arg, 'U', "unified", &options->context))
2347 options->output_format |= DIFF_FORMAT_PATCH;
2348 else if (!strcmp(arg, "--raw"))
2349 options->output_format |= DIFF_FORMAT_RAW;
2350 else if (!strcmp(arg, "--patch-with-raw"))
2351 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2352 else if (!strcmp(arg, "--numstat"))
2353 options->output_format |= DIFF_FORMAT_NUMSTAT;
2354 else if (!strcmp(arg, "--shortstat"))
2355 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2356 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2357 options->output_format |= DIFF_FORMAT_DIRSTAT;
2358 else if (!strcmp(arg, "--cumulative"))
2359 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2360 else if (!strcmp(arg, "--check"))
2361 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2362 else if (!strcmp(arg, "--summary"))
2363 options->output_format |= DIFF_FORMAT_SUMMARY;
2364 else if (!strcmp(arg, "--patch-with-stat"))
2365 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2366 else if (!strcmp(arg, "--name-only"))
2367 options->output_format |= DIFF_FORMAT_NAME;
2368 else if (!strcmp(arg, "--name-status"))
2369 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2370 else if (!strcmp(arg, "-s"))
2371 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2372 else if (!prefixcmp(arg, "--stat")) {
2373 char *end;
2374 int width = options->stat_width;
2375 int name_width = options->stat_name_width;
2376 arg += 6;
2377 end = (char *)arg;
2379 switch (*arg) {
2380 case '-':
2381 if (!prefixcmp(arg, "-width="))
2382 width = strtoul(arg + 7, &end, 10);
2383 else if (!prefixcmp(arg, "-name-width="))
2384 name_width = strtoul(arg + 12, &end, 10);
2385 break;
2386 case '=':
2387 width = strtoul(arg+1, &end, 10);
2388 if (*end == ',')
2389 name_width = strtoul(end+1, &end, 10);
2390 }
2392 /* Important! This checks all the error cases! */
2393 if (*end)
2394 return 0;
2395 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2396 options->stat_name_width = name_width;
2397 options->stat_width = width;
2398 }
2400 /* renames options */
2401 else if (!prefixcmp(arg, "-B")) {
2402 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2403 return -1;
2404 }
2405 else if (!prefixcmp(arg, "-M")) {
2406 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2407 return -1;
2408 options->detect_rename = DIFF_DETECT_RENAME;
2409 }
2410 else if (!prefixcmp(arg, "-C")) {
2411 if (options->detect_rename == DIFF_DETECT_COPY)
2412 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2413 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2414 return -1;
2415 options->detect_rename = DIFF_DETECT_COPY;
2416 }
2417 else if (!strcmp(arg, "--no-renames"))
2418 options->detect_rename = 0;
2419 else if (!strcmp(arg, "--relative"))
2420 DIFF_OPT_SET(options, RELATIVE_NAME);
2421 else if (!prefixcmp(arg, "--relative=")) {
2422 DIFF_OPT_SET(options, RELATIVE_NAME);
2423 options->prefix = arg + 11;
2424 }
2426 /* xdiff options */
2427 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2428 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2429 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2430 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2431 else if (!strcmp(arg, "--ignore-space-at-eol"))
2432 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2434 /* flags options */
2435 else if (!strcmp(arg, "--binary")) {
2436 options->output_format |= DIFF_FORMAT_PATCH;
2437 DIFF_OPT_SET(options, BINARY);
2438 }
2439 else if (!strcmp(arg, "--full-index"))
2440 DIFF_OPT_SET(options, FULL_INDEX);
2441 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2442 DIFF_OPT_SET(options, TEXT);
2443 else if (!strcmp(arg, "-R"))
2444 DIFF_OPT_SET(options, REVERSE_DIFF);
2445 else if (!strcmp(arg, "--find-copies-harder"))
2446 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2447 else if (!strcmp(arg, "--follow"))
2448 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2449 else if (!strcmp(arg, "--color"))
2450 DIFF_OPT_SET(options, COLOR_DIFF);
2451 else if (!strcmp(arg, "--no-color"))
2452 DIFF_OPT_CLR(options, COLOR_DIFF);
2453 else if (!strcmp(arg, "--color-words"))
2454 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2455 else if (!strcmp(arg, "--exit-code"))
2456 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2457 else if (!strcmp(arg, "--quiet"))
2458 DIFF_OPT_SET(options, QUIET);
2459 else if (!strcmp(arg, "--ext-diff"))
2460 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2461 else if (!strcmp(arg, "--no-ext-diff"))
2462 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2464 /* misc options */
2465 else if (!strcmp(arg, "-z"))
2466 options->line_termination = 0;
2467 else if (!prefixcmp(arg, "-l"))
2468 options->rename_limit = strtoul(arg+2, NULL, 10);
2469 else if (!prefixcmp(arg, "-S"))
2470 options->pickaxe = arg + 2;
2471 else if (!strcmp(arg, "--pickaxe-all"))
2472 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2473 else if (!strcmp(arg, "--pickaxe-regex"))
2474 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2475 else if (!prefixcmp(arg, "-O"))
2476 options->orderfile = arg + 2;
2477 else if (!prefixcmp(arg, "--diff-filter="))
2478 options->filter = arg + 14;
2479 else if (!strcmp(arg, "--abbrev"))
2480 options->abbrev = DEFAULT_ABBREV;
2481 else if (!prefixcmp(arg, "--abbrev=")) {
2482 options->abbrev = strtoul(arg + 9, NULL, 10);
2483 if (options->abbrev < MINIMUM_ABBREV)
2484 options->abbrev = MINIMUM_ABBREV;
2485 else if (40 < options->abbrev)
2486 options->abbrev = 40;
2487 }
2488 else if (!prefixcmp(arg, "--src-prefix="))
2489 options->a_prefix = arg + 13;
2490 else if (!prefixcmp(arg, "--dst-prefix="))
2491 options->b_prefix = arg + 13;
2492 else if (!strcmp(arg, "--no-prefix"))
2493 options->a_prefix = options->b_prefix = "";
2494 else if (!prefixcmp(arg, "--output=")) {
2495 options->file = fopen(arg + strlen("--output="), "w");
2496 options->close_file = 1;
2497 } else
2498 return 0;
2499 return 1;
2500 }
2502 static int parse_num(const char **cp_p)
2503 {
2504 unsigned long num, scale;
2505 int ch, dot;
2506 const char *cp = *cp_p;
2508 num = 0;
2509 scale = 1;
2510 dot = 0;
2511 for(;;) {
2512 ch = *cp;
2513 if ( !dot && ch == '.' ) {
2514 scale = 1;
2515 dot = 1;
2516 } else if ( ch == '%' ) {
2517 scale = dot ? scale*100 : 100;
2518 cp++; /* % is always at the end */
2519 break;
2520 } else if ( ch >= '0' && ch <= '9' ) {
2521 if ( scale < 100000 ) {
2522 scale *= 10;
2523 num = (num*10) + (ch-'0');
2524 }
2525 } else {
2526 break;
2527 }
2528 cp++;
2529 }
2530 *cp_p = cp;
2532 /* user says num divided by scale and we say internally that
2533 * is MAX_SCORE * num / scale.
2534 */
2535 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2536 }
2538 static int diff_scoreopt_parse(const char *opt)
2539 {
2540 int opt1, opt2, cmd;
2542 if (*opt++ != '-')
2543 return -1;
2544 cmd = *opt++;
2545 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2546 return -1; /* that is not a -M, -C nor -B option */
2548 opt1 = parse_num(&opt);
2549 if (cmd != 'B')
2550 opt2 = 0;
2551 else {
2552 if (*opt == 0)
2553 opt2 = 0;
2554 else if (*opt != '/')
2555 return -1; /* we expect -B80/99 or -B80 */
2556 else {
2557 opt++;
2558 opt2 = parse_num(&opt);
2559 }
2560 }
2561 if (*opt != 0)
2562 return -1;
2563 return opt1 | (opt2 << 16);
2564 }
2566 struct diff_queue_struct diff_queued_diff;
2568 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2569 {
2570 if (queue->alloc <= queue->nr) {
2571 queue->alloc = alloc_nr(queue->alloc);
2572 queue->queue = xrealloc(queue->queue,
2573 sizeof(dp) * queue->alloc);
2574 }
2575 queue->queue[queue->nr++] = dp;
2576 }
2578 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2579 struct diff_filespec *one,
2580 struct diff_filespec *two)
2581 {
2582 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2583 dp->one = one;
2584 dp->two = two;
2585 if (queue)
2586 diff_q(queue, dp);
2587 return dp;
2588 }
2590 void diff_free_filepair(struct diff_filepair *p)
2591 {
2592 free_filespec(p->one);
2593 free_filespec(p->two);
2594 free(p);
2595 }
2597 /* This is different from find_unique_abbrev() in that
2598 * it stuffs the result with dots for alignment.
2599 */
2600 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2601 {
2602 int abblen;
2603 const char *abbrev;
2604 if (len == 40)
2605 return sha1_to_hex(sha1);
2607 abbrev = find_unique_abbrev(sha1, len);
2608 abblen = strlen(abbrev);
2609 if (abblen < 37) {
2610 static char hex[41];
2611 if (len < abblen && abblen <= len + 2)
2612 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2613 else
2614 sprintf(hex, "%s...", abbrev);
2615 return hex;
2616 }
2617 return sha1_to_hex(sha1);
2618 }
2620 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2621 {
2622 int line_termination = opt->line_termination;
2623 int inter_name_termination = line_termination ? '\t' : '\0';
2625 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2626 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2627 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2628 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2629 }
2630 if (p->score) {
2631 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2632 inter_name_termination);
2633 } else {
2634 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2635 }
2637 if (p->status == DIFF_STATUS_COPIED ||
2638 p->status == DIFF_STATUS_RENAMED) {
2639 const char *name_a, *name_b;
2640 name_a = p->one->path;
2641 name_b = p->two->path;
2642 strip_prefix(opt->prefix_length, &name_a, &name_b);
2643 write_name_quoted(name_a, opt->file, inter_name_termination);
2644 write_name_quoted(name_b, opt->file, line_termination);
2645 } else {
2646 const char *name_a, *name_b;
2647 name_a = p->one->mode ? p->one->path : p->two->path;
2648 name_b = NULL;
2649 strip_prefix(opt->prefix_length, &name_a, &name_b);
2650 write_name_quoted(name_a, opt->file, line_termination);
2651 }
2652 }
2654 int diff_unmodified_pair(struct diff_filepair *p)
2655 {
2656 /* This function is written stricter than necessary to support
2657 * the currently implemented transformers, but the idea is to
2658 * let transformers to produce diff_filepairs any way they want,
2659 * and filter and clean them up here before producing the output.
2660 */
2661 struct diff_filespec *one = p->one, *two = p->two;
2663 if (DIFF_PAIR_UNMERGED(p))
2664 return 0; /* unmerged is interesting */
2666 /* deletion, addition, mode or type change
2667 * and rename are all interesting.
2668 */
2669 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2670 DIFF_PAIR_MODE_CHANGED(p) ||
2671 strcmp(one->path, two->path))
2672 return 0;
2674 /* both are valid and point at the same path. that is, we are
2675 * dealing with a change.
2676 */
2677 if (one->sha1_valid && two->sha1_valid &&
2678 !hashcmp(one->sha1, two->sha1))
2679 return 1; /* no change */
2680 if (!one->sha1_valid && !two->sha1_valid)
2681 return 1; /* both look at the same file on the filesystem. */
2682 return 0;
2683 }
2685 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2686 {
2687 if (diff_unmodified_pair(p))
2688 return;
2690 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2691 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2692 return; /* no tree diffs in patch format */
2694 run_diff(p, o);
2695 }
2697 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2698 struct diffstat_t *diffstat)
2699 {
2700 if (diff_unmodified_pair(p))
2701 return;
2703 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2704 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2705 return; /* no tree diffs in patch format */
2707 run_diffstat(p, o, diffstat);
2708 }
2710 static void diff_flush_checkdiff(struct diff_filepair *p,
2711 struct diff_options *o)
2712 {
2713 if (diff_unmodified_pair(p))
2714 return;
2716 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2717 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2718 return; /* no tree diffs in patch format */
2720 run_checkdiff(p, o);
2721 }
2723 int diff_queue_is_empty(void)
2724 {
2725 struct diff_queue_struct *q = &diff_queued_diff;
2726 int i;
2727 for (i = 0; i < q->nr; i++)
2728 if (!diff_unmodified_pair(q->queue[i]))
2729 return 0;
2730 return 1;
2731 }
2733 #if DIFF_DEBUG
2734 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2735 {
2736 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2737 x, one ? one : "",
2738 s->path,
2739 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2740 s->mode,
2741 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2742 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2743 x, one ? one : "",
2744 s->size, s->xfrm_flags);
2745 }
2747 void diff_debug_filepair(const struct diff_filepair *p, int i)
2748 {
2749 diff_debug_filespec(p->one, i, "one");
2750 diff_debug_filespec(p->two, i, "two");
2751 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2752 p->score, p->status ? p->status : '?',
2753 p->one->rename_used, p->broken_pair);
2754 }
2756 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2757 {
2758 int i;
2759 if (msg)
2760 fprintf(stderr, "%s\n", msg);
2761 fprintf(stderr, "q->nr = %d\n", q->nr);
2762 for (i = 0; i < q->nr; i++) {
2763 struct diff_filepair *p = q->queue[i];
2764 diff_debug_filepair(p, i);
2765 }
2766 }
2767 #endif
2769 static void diff_resolve_rename_copy(void)
2770 {
2771 int i;
2772 struct diff_filepair *p;
2773 struct diff_queue_struct *q = &diff_queued_diff;
2775 diff_debug_queue("resolve-rename-copy", q);
2777 for (i = 0; i < q->nr; i++) {
2778 p = q->queue[i];
2779 p->status = 0; /* undecided */
2780 if (DIFF_PAIR_UNMERGED(p))
2781 p->status = DIFF_STATUS_UNMERGED;
2782 else if (!DIFF_FILE_VALID(p->one))
2783 p->status = DIFF_STATUS_ADDED;
2784 else if (!DIFF_FILE_VALID(p->two))
2785 p->status = DIFF_STATUS_DELETED;
2786 else if (DIFF_PAIR_TYPE_CHANGED(p))
2787 p->status = DIFF_STATUS_TYPE_CHANGED;
2789 /* from this point on, we are dealing with a pair
2790 * whose both sides are valid and of the same type, i.e.
2791 * either in-place edit or rename/copy edit.
2792 */
2793 else if (DIFF_PAIR_RENAME(p)) {
2794 /*
2795 * A rename might have re-connected a broken
2796 * pair up, causing the pathnames to be the
2797 * same again. If so, that's not a rename at
2798 * all, just a modification..
2799 *
2800 * Otherwise, see if this source was used for
2801 * multiple renames, in which case we decrement
2802 * the count, and call it a copy.
2803 */
2804 if (!strcmp(p->one->path, p->two->path))
2805 p->status = DIFF_STATUS_MODIFIED;
2806 else if (--p->one->rename_used > 0)
2807 p->status = DIFF_STATUS_COPIED;
2808 else
2809 p->status = DIFF_STATUS_RENAMED;
2810 }
2811 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2812 p->one->mode != p->two->mode ||
2813 is_null_sha1(p->one->sha1))
2814 p->status = DIFF_STATUS_MODIFIED;
2815 else {
2816 /* This is a "no-change" entry and should not
2817 * happen anymore, but prepare for broken callers.
2818 */
2819 error("feeding unmodified %s to diffcore",
2820 p->one->path);
2821 p->status = DIFF_STATUS_UNKNOWN;
2822 }
2823 }
2824 diff_debug_queue("resolve-rename-copy done", q);
2825 }
2827 static int check_pair_status(struct diff_filepair *p)
2828 {
2829 switch (p->status) {
2830 case DIFF_STATUS_UNKNOWN:
2831 return 0;
2832 case 0:
2833 die("internal error in diff-resolve-rename-copy");
2834 default:
2835 return 1;
2836 }
2837 }
2839 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2840 {
2841 int fmt = opt->output_format;
2843 if (fmt & DIFF_FORMAT_CHECKDIFF)
2844 diff_flush_checkdiff(p, opt);
2845 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2846 diff_flush_raw(p, opt);
2847 else if (fmt & DIFF_FORMAT_NAME) {
2848 const char *name_a, *name_b;
2849 name_a = p->two->path;
2850 name_b = NULL;
2851 strip_prefix(opt->prefix_length, &name_a, &name_b);
2852 write_name_quoted(name_a, opt->file, opt->line_termination);
2853 }
2854 }
2856 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2857 {
2858 if (fs->mode)
2859 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2860 else
2861 fprintf(file, " %s ", newdelete);
2862 write_name_quoted(fs->path, file, '\n');
2863 }
2866 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2867 {
2868 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2869 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2870 show_name ? ' ' : '\n');
2871 if (show_name) {
2872 write_name_quoted(p->two->path, file, '\n');
2873 }
2874 }
2875 }
2877 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2878 {
2879 char *names = pprint_rename(p->one->path, p->two->path);
2881 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2882 free(names);
2883 show_mode_change(file, p, 0);
2884 }
2886 static void diff_summary(FILE *file, struct diff_filepair *p)
2887 {
2888 switch(p->status) {
2889 case DIFF_STATUS_DELETED:
2890 show_file_mode_name(file, "delete", p->one);
2891 break;
2892 case DIFF_STATUS_ADDED:
2893 show_file_mode_name(file, "create", p->two);
2894 break;
2895 case DIFF_STATUS_COPIED:
2896 show_rename_copy(file, "copy", p);
2897 break;
2898 case DIFF_STATUS_RENAMED:
2899 show_rename_copy(file, "rename", p);
2900 break;
2901 default:
2902 if (p->score) {
2903 fputs(" rewrite ", file);
2904 write_name_quoted(p->two->path, file, ' ');
2905 fprintf(file, "(%d%%)\n", similarity_index(p));
2906 }
2907 show_mode_change(file, p, !p->score);
2908 break;
2909 }
2910 }
2912 struct patch_id_t {
2913 struct xdiff_emit_state xm;
2914 SHA_CTX *ctx;
2915 int patchlen;
2916 };
2918 static int remove_space(char *line, int len)
2919 {
2920 int i;
2921 char *dst = line;
2922 unsigned char c;
2924 for (i = 0; i < len; i++)
2925 if (!isspace((c = line[i])))
2926 *dst++ = c;
2928 return dst - line;
2929 }
2931 static void patch_id_consume(void *priv, char *line, unsigned long len)
2932 {
2933 struct patch_id_t *data = priv;
2934 int new_len;
2936 /* Ignore line numbers when computing the SHA1 of the patch */
2937 if (!prefixcmp(line, "@@ -"))
2938 return;
2940 new_len = remove_space(line, len);
2942 SHA1_Update(data->ctx, line, new_len);
2943 data->patchlen += new_len;
2944 }
2946 /* returns 0 upon success, and writes result into sha1 */
2947 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2948 {
2949 struct diff_queue_struct *q = &diff_queued_diff;
2950 int i;
2951 SHA_CTX ctx;
2952 struct patch_id_t data;
2953 char buffer[PATH_MAX * 4 + 20];
2955 SHA1_Init(&ctx);
2956 memset(&data, 0, sizeof(struct patch_id_t));
2957 data.ctx = &ctx;
2958 data.xm.consume = patch_id_consume;
2960 for (i = 0; i < q->nr; i++) {
2961 xpparam_t xpp;
2962 xdemitconf_t xecfg;
2963 xdemitcb_t ecb;
2964 mmfile_t mf1, mf2;
2965 struct diff_filepair *p = q->queue[i];
2966 int len1, len2;
2968 memset(&xecfg, 0, sizeof(xecfg));
2969 if (p->status == 0)
2970 return error("internal diff status error");
2971 if (p->status == DIFF_STATUS_UNKNOWN)
2972 continue;
2973 if (diff_unmodified_pair(p))
2974 continue;
2975 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2976 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2977 continue;
2978 if (DIFF_PAIR_UNMERGED(p))
2979 continue;
2981 diff_fill_sha1_info(p->one);
2982 diff_fill_sha1_info(p->two);
2983 if (fill_mmfile(&mf1, p->one) < 0 ||
2984 fill_mmfile(&mf2, p->two) < 0)
2985 return error("unable to read files to diff");
2987 len1 = remove_space(p->one->path, strlen(p->one->path));
2988 len2 = remove_space(p->two->path, strlen(p->two->path));
2989 if (p->one->mode == 0)
2990 len1 = snprintf(buffer, sizeof(buffer),
2991 "diff--gita/%.*sb/%.*s"
2992 "newfilemode%06o"
2993 "---/dev/null"
2994 "+++b/%.*s",
2995 len1, p->one->path,
2996 len2, p->two->path,
2997 p->two->mode,
2998 len2, p->two->path);
2999 else if (p->two->mode == 0)
3000 len1 = snprintf(buffer, sizeof(buffer),
3001 "diff--gita/%.*sb/%.*s"
3002 "deletedfilemode%06o"
3003 "---a/%.*s"
3004 "+++/dev/null",
3005 len1, p->one->path,
3006 len2, p->two->path,
3007 p->one->mode,
3008 len1, p->one->path);
3009 else
3010 len1 = snprintf(buffer, sizeof(buffer),
3011 "diff--gita/%.*sb/%.*s"
3012 "---a/%.*s"
3013 "+++b/%.*s",
3014 len1, p->one->path,
3015 len2, p->two->path,
3016 len1, p->one->path,
3017 len2, p->two->path);
3018 SHA1_Update(&ctx, buffer, len1);
3020 xpp.flags = XDF_NEED_MINIMAL;
3021 xecfg.ctxlen = 3;
3022 xecfg.flags = XDL_EMIT_FUNCNAMES;
3023 ecb.outf = xdiff_outf;
3024 ecb.priv = &data;
3025 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3026 }
3028 SHA1_Final(sha1, &ctx);
3029 return 0;
3030 }
3032 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3033 {
3034 struct diff_queue_struct *q = &diff_queued_diff;
3035 int i;
3036 int result = diff_get_patch_id(options, sha1);
3038 for (i = 0; i < q->nr; i++)
3039 diff_free_filepair(q->queue[i]);
3041 free(q->queue);
3042 q->queue = NULL;
3043 q->nr = q->alloc = 0;
3045 return result;
3046 }
3048 static int is_summary_empty(const struct diff_queue_struct *q)
3049 {
3050 int i;
3052 for (i = 0; i < q->nr; i++) {
3053 const struct diff_filepair *p = q->queue[i];
3055 switch (p->status) {
3056 case DIFF_STATUS_DELETED:
3057 case DIFF_STATUS_ADDED:
3058 case DIFF_STATUS_COPIED:
3059 case DIFF_STATUS_RENAMED:
3060 return 0;
3061 default:
3062 if (p->score)
3063 return 0;
3064 if (p->one->mode && p->two->mode &&
3065 p->one->mode != p->two->mode)
3066 return 0;
3067 break;
3068 }
3069 }
3070 return 1;
3071 }
3073 void diff_flush(struct diff_options *options)
3074 {
3075 struct diff_queue_struct *q = &diff_queued_diff;
3076 int i, output_format = options->output_format;
3077 int separator = 0;
3079 /*
3080 * Order: raw, stat, summary, patch
3081 * or: name/name-status/checkdiff (other bits clear)
3082 */
3083 if (!q->nr)
3084 goto free_queue;
3086 if (output_format & (DIFF_FORMAT_RAW |
3087 DIFF_FORMAT_NAME |
3088 DIFF_FORMAT_NAME_STATUS |
3089 DIFF_FORMAT_CHECKDIFF)) {
3090 for (i = 0; i < q->nr; i++) {
3091 struct diff_filepair *p = q->queue[i];
3092 if (check_pair_status(p))
3093 flush_one_pair(p, options);
3094 }
3095 separator++;
3096 }
3098 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT|DIFF_FORMAT_DIRSTAT)) {
3099 struct diffstat_t diffstat;
3101 memset(&diffstat, 0, sizeof(struct diffstat_t));
3102 diffstat.xm.consume = diffstat_consume;
3103 for (i = 0; i < q->nr; i++) {
3104 struct diff_filepair *p = q->queue[i];
3105 if (check_pair_status(p))
3106 diff_flush_stat(p, options, &diffstat);
3107 }
3108 if (output_format & DIFF_FORMAT_DIRSTAT)
3109 show_dirstat(&diffstat, options);
3110 if (output_format & DIFF_FORMAT_NUMSTAT)
3111 show_numstat(&diffstat, options);
3112 if (output_format & DIFF_FORMAT_DIFFSTAT)
3113 show_stats(&diffstat, options);
3114 if (output_format & DIFF_FORMAT_SHORTSTAT)
3115 show_shortstats(&diffstat, options);
3116 free_diffstat_info(&diffstat);
3117 separator++;
3118 }
3120 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3121 for (i = 0; i < q->nr; i++)
3122 diff_summary(options->file, q->queue[i]);
3123 separator++;
3124 }
3126 if (output_format & DIFF_FORMAT_PATCH) {
3127 if (separator) {
3128 if (options->stat_sep) {
3129 /* attach patch instead of inline */
3130 fputs(options->stat_sep, options->file);
3131 } else {
3132 putc(options->line_termination, options->file);
3133 }
3134 }
3136 for (i = 0; i < q->nr; i++) {
3137 struct diff_filepair *p = q->queue[i];
3138 if (check_pair_status(p))
3139 diff_flush_patch(p, options);
3140 }
3141 }
3143 if (output_format & DIFF_FORMAT_CALLBACK)
3144 options->format_callback(q, options, options->format_callback_data);
3146 for (i = 0; i < q->nr; i++)
3147 diff_free_filepair(q->queue[i]);
3148 free_queue:
3149 free(q->queue);
3150 q->queue = NULL;
3151 q->nr = q->alloc = 0;
3152 if (options->close_file)
3153 fclose(options->file);
3154 }
3156 static void diffcore_apply_filter(const char *filter)
3157 {
3158 int i;
3159 struct diff_queue_struct *q = &diff_queued_diff;
3160 struct diff_queue_struct outq;
3161 outq.queue = NULL;
3162 outq.nr = outq.alloc = 0;
3164 if (!filter)
3165 return;
3167 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3168 int found;
3169 for (i = found = 0; !found && i < q->nr; i++) {
3170 struct diff_filepair *p = q->queue[i];
3171 if (((p->status == DIFF_STATUS_MODIFIED) &&
3172 ((p->score &&
3173 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3174 (!p->score &&
3175 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3176 ((p->status != DIFF_STATUS_MODIFIED) &&
3177 strchr(filter, p->status)))
3178 found++;
3179 }
3180 if (found)
3181 return;
3183 /* otherwise we will clear the whole queue
3184 * by copying the empty outq at the end of this
3185 * function, but first clear the current entries
3186 * in the queue.
3187 */
3188 for (i = 0; i < q->nr; i++)
3189 diff_free_filepair(q->queue[i]);
3190 }
3191 else {
3192 /* Only the matching ones */
3193 for (i = 0; i < q->nr; i++) {
3194 struct diff_filepair *p = q->queue[i];
3196 if (((p->status == DIFF_STATUS_MODIFIED) &&
3197 ((p->score &&
3198 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3199 (!p->score &&
3200 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3201 ((p->status != DIFF_STATUS_MODIFIED) &&
3202 strchr(filter, p->status)))
3203 diff_q(&outq, p);
3204 else
3205 diff_free_filepair(p);
3206 }
3207 }
3208 free(q->queue);
3209 *q = outq;
3210 }
3212 /* Check whether two filespecs with the same mode and size are identical */
3213 static int diff_filespec_is_identical(struct diff_filespec *one,
3214 struct diff_filespec *two)
3215 {
3216 if (S_ISGITLINK(one->mode))
3217 return 0;
3218 if (diff_populate_filespec(one, 0))
3219 return 0;
3220 if (diff_populate_filespec(two, 0))
3221 return 0;
3222 return !memcmp(one->data, two->data, one->size);
3223 }
3225 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3226 {
3227 int i;
3228 struct diff_queue_struct *q = &diff_queued_diff;
3229 struct diff_queue_struct outq;
3230 outq.queue = NULL;
3231 outq.nr = outq.alloc = 0;
3233 for (i = 0; i < q->nr; i++) {
3234 struct diff_filepair *p = q->queue[i];
3236 /*
3237 * 1. Entries that come from stat info dirtyness
3238 * always have both sides (iow, not create/delete),
3239 * one side of the object name is unknown, with
3240 * the same mode and size. Keep the ones that
3241 * do not match these criteria. They have real
3242 * differences.
3243 *
3244 * 2. At this point, the file is known to be modified,
3245 * with the same mode and size, and the object
3246 * name of one side is unknown. Need to inspect
3247 * the identical contents.
3248 */
3249 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3250 !DIFF_FILE_VALID(p->two) ||
3251 (p->one->sha1_valid && p->two->sha1_valid) ||
3252 (p->one->mode != p->two->mode) ||
3253 diff_populate_filespec(p->one, 1) ||
3254 diff_populate_filespec(p->two, 1) ||
3255 (p->one->size != p->two->size) ||
3256 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3257 diff_q(&outq, p);
3258 else {
3259 /*
3260 * The caller can subtract 1 from skip_stat_unmatch
3261 * to determine how many paths were dirty only
3262 * due to stat info mismatch.
3263 */
3264 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3265 diffopt->skip_stat_unmatch++;
3266 diff_free_filepair(p);
3267 }
3268 }
3269 free(q->queue);
3270 *q = outq;
3271 }
3273 void diffcore_std(struct diff_options *options)
3274 {
3275 if (DIFF_OPT_TST(options, QUIET))
3276 return;
3278 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3279 diffcore_skip_stat_unmatch(options);
3280 if (options->break_opt != -1)
3281 diffcore_break(options->break_opt);
3282 if (options->detect_rename)
3283 diffcore_rename(options);
3284 if (options->break_opt != -1)
3285 diffcore_merge_broken();
3286 if (options->pickaxe)
3287 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3288 if (options->orderfile)
3289 diffcore_order(options->orderfile);
3290 diff_resolve_rename_copy();
3291 diffcore_apply_filter(options->filter);
3293 if (diff_queued_diff.nr)
3294 DIFF_OPT_SET(options, HAS_CHANGES);
3295 else
3296 DIFF_OPT_CLR(options, HAS_CHANGES);
3297 }
3299 int diff_result_code(struct diff_options *opt, int status)
3300 {
3301 int result = 0;
3302 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3303 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3304 return status;
3305 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3306 DIFF_OPT_TST(opt, HAS_CHANGES))
3307 result |= 01;
3308 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3309 DIFF_OPT_TST(opt, CHECK_FAILED))
3310 result |= 02;
3311 return result;
3312 }
3314 void diff_addremove(struct diff_options *options,
3315 int addremove, unsigned mode,
3316 const unsigned char *sha1,
3317 const char *base, const char *path)
3318 {
3319 char concatpath[PATH_MAX];
3320 struct diff_filespec *one, *two;
3322 /* This may look odd, but it is a preparation for
3323 * feeding "there are unchanged files which should
3324 * not produce diffs, but when you are doing copy
3325 * detection you would need them, so here they are"
3326 * entries to the diff-core. They will be prefixed
3327 * with something like '=' or '*' (I haven't decided
3328 * which but should not make any difference).
3329 * Feeding the same new and old to diff_change()
3330 * also has the same effect.
3331 * Before the final output happens, they are pruned after
3332 * merged into rename/copy pairs as appropriate.
3333 */
3334 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3335 addremove = (addremove == '+' ? '-' :
3336 addremove == '-' ? '+' : addremove);
3338 if (!path) path = "";
3339 sprintf(concatpath, "%s%s", base, path);
3341 if (options->prefix &&
3342 strncmp(concatpath, options->prefix, options->prefix_length))
3343 return;
3345 one = alloc_filespec(concatpath);
3346 two = alloc_filespec(concatpath);
3348 if (addremove != '+')
3349 fill_filespec(one, sha1, mode);
3350 if (addremove != '-')
3351 fill_filespec(two, sha1, mode);
3353 diff_queue(&diff_queued_diff, one, two);
3354 DIFF_OPT_SET(options, HAS_CHANGES);
3355 }
3357 void diff_change(struct diff_options *options,
3358 unsigned old_mode, unsigned new_mode,
3359 const unsigned char *old_sha1,
3360 const unsigned char *new_sha1,
3361 const char *base, const char *path)
3362 {
3363 char concatpath[PATH_MAX];
3364 struct diff_filespec *one, *two;
3366 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3367 unsigned tmp;
3368 const unsigned char *tmp_c;
3369 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3370 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3371 }
3372 if (!path) path = "";
3373 sprintf(concatpath, "%s%s", base, path);
3375 if (options->prefix &&
3376 strncmp(concatpath, options->prefix, options->prefix_length))
3377 return;
3379 one = alloc_filespec(concatpath);
3380 two = alloc_filespec(concatpath);
3381 fill_filespec(one, old_sha1, old_mode);
3382 fill_filespec(two, new_sha1, new_mode);
3384 diff_queue(&diff_queued_diff, one, two);
3385 DIFF_OPT_SET(options, HAS_CHANGES);
3386 }
3388 void diff_unmerge(struct diff_options *options,
3389 const char *path,
3390 unsigned mode, const unsigned char *sha1)
3391 {
3392 struct diff_filespec *one, *two;
3394 if (options->prefix &&
3395 strncmp(path, options->prefix, options->prefix_length))
3396 return;
3398 one = alloc_filespec(path);
3399 two = alloc_filespec(path);
3400 fill_filespec(one, sha1, mode);
3401 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3402 }