?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'lt/in-core-index'
[git/jnareb-git.git] / diff.c
blobcd8bc4dcc32757dfed9374fa8769599a6a7857ad
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
14
15 #ifdef NO_FAST_WORKING_DIRECTORY
16 #define FAST_WORKING_DIRECTORY 0
17 #else
18 #define FAST_WORKING_DIRECTORY 1
19 #endif
20
21 static int diff_detect_rename_default;
22 static int diff_rename_limit_default = 100;
23 static int diff_use_color_default;
24 static const char *external_diff_cmd_cfg;
25 int diff_auto_refresh_index = 1;
26
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 };
37
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 }
56
57 static struct ll_diff_driver {
58 const char *name;
59 struct ll_diff_driver *next;
60 char *cmd;
61 } *user_diff, **user_diff_tail;
62
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;
74
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 }
88
89 if (!value)
90 return error("%s: lacks value", var);
91 drv->cmd = strdup(value);
92 return 0;
93 }
94
95 /*
96 * 'diff.<what>.funcname' attribute can be specified in the configuration
97 * to define a customized regexp to find the beginning of a function to
98 * be used for hunk header lines of "diff -p" style output.
99 */
100 static struct funcname_pattern {
101 char *name;
102 char *pattern;
103 struct funcname_pattern *next;
104 } *funcname_pattern_list;
105
106 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
107 {
108 const char *name;
109 int namelen;
110 struct funcname_pattern *pp;
111
112 name = var + 5; /* "diff." */
113 namelen = ep - name;
114
115 for (pp = funcname_pattern_list; pp; pp = pp->next)
116 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
117 break;
118 if (!pp) {
119 pp = xcalloc(1, sizeof(*pp));
120 pp->name = xmemdupz(name, namelen);
121 pp->next = funcname_pattern_list;
122 funcname_pattern_list = pp;
123 }
124 if (pp->pattern)
125 free(pp->pattern);
126 pp->pattern = xstrdup(value);
127 return 0;
128 }
129
130 /*
131 * These are to give UI layer defaults.
132 * The core-level commands such as git-diff-files should
133 * never be affected by the setting of diff.renames
134 * the user happens to have in the configuration file.
135 */
136 int git_diff_ui_config(const char *var, const char *value)
137 {
138 if (!strcmp(var, "diff.renamelimit")) {
139 diff_rename_limit_default = git_config_int(var, value);
140 return 0;
141 }
142 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
143 diff_use_color_default = git_config_colorbool(var, value, -1);
144 return 0;
145 }
146 if (!strcmp(var, "diff.renames")) {
147 if (!value)
148 diff_detect_rename_default = DIFF_DETECT_RENAME;
149 else if (!strcasecmp(value, "copies") ||
150 !strcasecmp(value, "copy"))
151 diff_detect_rename_default = DIFF_DETECT_COPY;
152 else if (git_config_bool(var,value))
153 diff_detect_rename_default = DIFF_DETECT_RENAME;
154 return 0;
155 }
156 if (!strcmp(var, "diff.autorefreshindex")) {
157 diff_auto_refresh_index = git_config_bool(var, value);
158 return 0;
159 }
160 if (!strcmp(var, "diff.external")) {
161 if (!value)
162 return config_error_nonbool(var);
163 external_diff_cmd_cfg = xstrdup(value);
164 return 0;
165 }
166 if (!prefixcmp(var, "diff.")) {
167 const char *ep = strrchr(var, '.');
168
169 if (ep != var + 4) {
170 if (!strcmp(ep, ".command")) {
171 if (!value)
172 return config_error_nonbool(var);
173 return parse_lldiff_command(var, ep, value);
174 }
175 }
176 }
177
178 return git_diff_basic_config(var, value);
179 }
180
181 int git_diff_basic_config(const char *var, const char *value)
182 {
183 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
184 int slot = parse_diff_color_slot(var, 11);
185 if (!value)
186 return config_error_nonbool(var);
187 color_parse(value, var, diff_colors[slot]);
188 return 0;
189 }
190
191 if (!prefixcmp(var, "diff.")) {
192 const char *ep = strrchr(var, '.');
193 if (ep != var + 4) {
194 if (!strcmp(ep, ".funcname")) {
195 if (!value)
196 return config_error_nonbool(var);
197 return parse_funcname_pattern(var, ep, value);
198 }
199 }
200 }
201
202 return git_default_config(var, value);
203 }
204
205 static char *quote_two(const char *one, const char *two)
206 {
207 int need_one = quote_c_style(one, NULL, NULL, 1);
208 int need_two = quote_c_style(two, NULL, NULL, 1);
209 struct strbuf res;
210
211 strbuf_init(&res, 0);
212 if (need_one + need_two) {
213 strbuf_addch(&res, '"');
214 quote_c_style(one, &res, NULL, 1);
215 quote_c_style(two, &res, NULL, 1);
216 strbuf_addch(&res, '"');
217 } else {
218 strbuf_addstr(&res, one);
219 strbuf_addstr(&res, two);
220 }
221 return strbuf_detach(&res, NULL);
222 }
223
224 static const char *external_diff(void)
225 {
226 static const char *external_diff_cmd = NULL;
227 static int done_preparing = 0;
228
229 if (done_preparing)
230 return external_diff_cmd;
231 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
232 if (!external_diff_cmd)
233 external_diff_cmd = external_diff_cmd_cfg;
234 done_preparing = 1;
235 return external_diff_cmd;
236 }
237
238 static struct diff_tempfile {
239 const char *name; /* filename external diff should read from */
240 char hex[41];
241 char mode[10];
242 char tmp_path[PATH_MAX];
243 } diff_temp[2];
244
245 static int count_lines(const char *data, int size)
246 {
247 int count, ch, completely_empty = 1, nl_just_seen = 0;
248 count = 0;
249 while (0 < size--) {
250 ch = *data++;
251 if (ch == '\n') {
252 count++;
253 nl_just_seen = 1;
254 completely_empty = 0;
255 }
256 else {
257 nl_just_seen = 0;
258 completely_empty = 0;
259 }
260 }
261 if (completely_empty)
262 return 0;
263 if (!nl_just_seen)
264 count++; /* no trailing newline */
265 return count;
266 }
267
268 static void print_line_count(int count)
269 {
270 switch (count) {
271 case 0:
272 printf("0,0");
273 break;
274 case 1:
275 printf("1");
276 break;
277 default:
278 printf("1,%d", count);
279 break;
280 }
281 }
282
283 static void copy_file(int prefix, const char *data, int size,
284 const char *set, const char *reset)
285 {
286 int ch, nl_just_seen = 1;
287 while (0 < size--) {
288 ch = *data++;
289 if (nl_just_seen) {
290 fputs(set, stdout);
291 putchar(prefix);
292 }
293 if (ch == '\n') {
294 nl_just_seen = 1;
295 fputs(reset, stdout);
296 } else
297 nl_just_seen = 0;
298 putchar(ch);
299 }
300 if (!nl_just_seen)
301 printf("%s\n\\ No newline at end of file\n", reset);
302 }
303
304 static void emit_rewrite_diff(const char *name_a,
305 const char *name_b,
306 struct diff_filespec *one,
307 struct diff_filespec *two,
308 struct diff_options *o)
309 {
310 int lc_a, lc_b;
311 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
312 const char *name_a_tab, *name_b_tab;
313 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
314 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
315 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
316 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
317 const char *reset = diff_get_color(color_diff, DIFF_RESET);
318 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
319
320 name_a += (*name_a == '/');
321 name_b += (*name_b == '/');
322 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
323 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
324
325 strbuf_reset(&a_name);
326 strbuf_reset(&b_name);
327 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
328 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
329
330 diff_populate_filespec(one, 0);
331 diff_populate_filespec(two, 0);
332 lc_a = count_lines(one->data, one->size);
333 lc_b = count_lines(two->data, two->size);
334 printf("%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
335 metainfo, a_name.buf, name_a_tab, reset,
336 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
337 print_line_count(lc_a);
338 printf(" +");
339 print_line_count(lc_b);
340 printf(" @@%s\n", reset);
341 if (lc_a)
342 copy_file('-', one->data, one->size, old, reset);
343 if (lc_b)
344 copy_file('+', two->data, two->size, new, reset);
345 }
346
347 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
348 {
349 if (!DIFF_FILE_VALID(one)) {
350 mf->ptr = (char *)""; /* does not matter */
351 mf->size = 0;
352 return 0;
353 }
354 else if (diff_populate_filespec(one, 0))
355 return -1;
356 mf->ptr = one->data;
357 mf->size = one->size;
358 return 0;
359 }
360
361 struct diff_words_buffer {
362 mmfile_t text;
363 long alloc;
364 long current; /* output pointer */
365 int suppressed_newline;
366 };
367
368 static void diff_words_append(char *line, unsigned long len,
369 struct diff_words_buffer *buffer)
370 {
371 if (buffer->text.size + len > buffer->alloc) {
372 buffer->alloc = (buffer->text.size + len) * 3 / 2;
373 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
374 }
375 line++;
376 len--;
377 memcpy(buffer->text.ptr + buffer->text.size, line, len);
378 buffer->text.size += len;
379 }
380
381 struct diff_words_data {
382 struct xdiff_emit_state xm;
383 struct diff_words_buffer minus, plus;
384 };
385
386 static void print_word(struct diff_words_buffer *buffer, int len, int color,
387 int suppress_newline)
388 {
389 const char *ptr;
390 int eol = 0;
391
392 if (len == 0)
393 return;
394
395 ptr = buffer->text.ptr + buffer->current;
396 buffer->current += len;
397
398 if (ptr[len - 1] == '\n') {
399 eol = 1;
400 len--;
401 }
402
403 fputs(diff_get_color(1, color), stdout);
404 fwrite(ptr, len, 1, stdout);
405 fputs(diff_get_color(1, DIFF_RESET), stdout);
406
407 if (eol) {
408 if (suppress_newline)
409 buffer->suppressed_newline = 1;
410 else
411 putchar('\n');
412 }
413 }
414
415 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
416 {
417 struct diff_words_data *diff_words = priv;
418
419 if (diff_words->minus.suppressed_newline) {
420 if (line[0] != '+')
421 putchar('\n');
422 diff_words->minus.suppressed_newline = 0;
423 }
424
425 len--;
426 switch (line[0]) {
427 case '-':
428 print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
429 break;
430 case '+':
431 print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
432 break;
433 case ' ':
434 print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
435 diff_words->minus.current += len;
436 break;
437 }
438 }
439
440 /* this executes the word diff on the accumulated buffers */
441 static void diff_words_show(struct diff_words_data *diff_words)
442 {
443 xpparam_t xpp;
444 xdemitconf_t xecfg;
445 xdemitcb_t ecb;
446 mmfile_t minus, plus;
447 int i;
448
449 memset(&xecfg, 0, sizeof(xecfg));
450 minus.size = diff_words->minus.text.size;
451 minus.ptr = xmalloc(minus.size);
452 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
453 for (i = 0; i < minus.size; i++)
454 if (isspace(minus.ptr[i]))
455 minus.ptr[i] = '\n';
456 diff_words->minus.current = 0;
457
458 plus.size = diff_words->plus.text.size;
459 plus.ptr = xmalloc(plus.size);
460 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
461 for (i = 0; i < plus.size; i++)
462 if (isspace(plus.ptr[i]))
463 plus.ptr[i] = '\n';
464 diff_words->plus.current = 0;
465
466 xpp.flags = XDF_NEED_MINIMAL;
467 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
468 ecb.outf = xdiff_outf;
469 ecb.priv = diff_words;
470 diff_words->xm.consume = fn_out_diff_words_aux;
471 xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
472
473 free(minus.ptr);
474 free(plus.ptr);
475 diff_words->minus.text.size = diff_words->plus.text.size = 0;
476
477 if (diff_words->minus.suppressed_newline) {
478 putchar('\n');
479 diff_words->minus.suppressed_newline = 0;
480 }
481 }
482
483 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
484
485 struct emit_callback {
486 struct xdiff_emit_state xm;
487 int nparents, color_diff;
488 unsigned ws_rule;
489 sane_truncate_fn truncate;
490 const char **label_path;
491 struct diff_words_data *diff_words;
492 int *found_changesp;
493 };
494
495 static void free_diff_words_data(struct emit_callback *ecbdata)
496 {
497 if (ecbdata->diff_words) {
498 /* flush buffers */
499 if (ecbdata->diff_words->minus.text.size ||
500 ecbdata->diff_words->plus.text.size)
501 diff_words_show(ecbdata->diff_words);
502
503 if (ecbdata->diff_words->minus.text.ptr)
504 free (ecbdata->diff_words->minus.text.ptr);
505 if (ecbdata->diff_words->plus.text.ptr)
506 free (ecbdata->diff_words->plus.text.ptr);
507 free(ecbdata->diff_words);
508 ecbdata->diff_words = NULL;
509 }
510 }
511
512 const char *diff_get_color(int diff_use_color, enum color_diff ix)
513 {
514 if (diff_use_color)
515 return diff_colors[ix];
516 return "";
517 }
518
519 static void emit_line(const char *set, const char *reset, const char *line, int len)
520 {
521 fputs(set, stdout);
522 fwrite(line, len, 1, stdout);
523 fputs(reset, stdout);
524 }
525
526 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
527 {
528 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
529 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
530
531 if (!*ws)
532 emit_line(set, reset, line, len);
533 else {
534 /* Emit just the prefix, then the rest. */
535 emit_line(set, reset, line, ecbdata->nparents);
536 (void)check_and_emit_line(line + ecbdata->nparents,
537 len - ecbdata->nparents, ecbdata->ws_rule,
538 stdout, set, reset, ws);
539 }
540 }
541
542 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
543 {
544 const char *cp;
545 unsigned long allot;
546 size_t l = len;
547
548 if (ecb->truncate)
549 return ecb->truncate(line, len);
550 cp = line;
551 allot = l;
552 while (0 < l) {
553 (void) utf8_width(&cp, &l);
554 if (!cp)
555 break; /* truncated in the middle? */
556 }
557 return allot - l;
558 }
559
560 static void fn_out_consume(void *priv, char *line, unsigned long len)
561 {
562 int i;
563 int color;
564 struct emit_callback *ecbdata = priv;
565 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
566 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
567 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
568
569 *(ecbdata->found_changesp) = 1;
570
571 if (ecbdata->label_path[0]) {
572 const char *name_a_tab, *name_b_tab;
573
574 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
575 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
576
577 printf("%s--- %s%s%s\n",
578 meta, ecbdata->label_path[0], reset, name_a_tab);
579 printf("%s+++ %s%s%s\n",
580 meta, ecbdata->label_path[1], reset, name_b_tab);
581 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
582 }
583
584 /* This is not really necessary for now because
585 * this codepath only deals with two-way diffs.
586 */
587 for (i = 0; i < len && line[i] == '@'; i++)
588 ;
589 if (2 <= i && i < len && line[i] == ' ') {
590 ecbdata->nparents = i - 1;
591 len = sane_truncate_line(ecbdata, line, len);
592 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
593 reset, line, len);
594 if (line[len-1] != '\n')
595 putchar('\n');
596 return;
597 }
598
599 if (len < ecbdata->nparents) {
600 emit_line(reset, reset, line, len);
601 return;
602 }
603
604 color = DIFF_PLAIN;
605 if (ecbdata->diff_words && ecbdata->nparents != 1)
606 /* fall back to normal diff */
607 free_diff_words_data(ecbdata);
608 if (ecbdata->diff_words) {
609 if (line[0] == '-') {
610 diff_words_append(line, len,
611 &ecbdata->diff_words->minus);
612 return;
613 } else if (line[0] == '+') {
614 diff_words_append(line, len,
615 &ecbdata->diff_words->plus);
616 return;
617 }
618 if (ecbdata->diff_words->minus.text.size ||
619 ecbdata->diff_words->plus.text.size)
620 diff_words_show(ecbdata->diff_words);
621 line++;
622 len--;
623 emit_line(plain, reset, line, len);
624 return;
625 }
626 for (i = 0; i < ecbdata->nparents && len; i++) {
627 if (line[i] == '-')
628 color = DIFF_FILE_OLD;
629 else if (line[i] == '+')
630 color = DIFF_FILE_NEW;
631 }
632
633 if (color != DIFF_FILE_NEW) {
634 emit_line(diff_get_color(ecbdata->color_diff, color),
635 reset, line, len);
636 return;
637 }
638 emit_add_line(reset, ecbdata, line, len);
639 }
640
641 static char *pprint_rename(const char *a, const char *b)
642 {
643 const char *old = a;
644 const char *new = b;
645 struct strbuf name;
646 int pfx_length, sfx_length;
647 int len_a = strlen(a);
648 int len_b = strlen(b);
649 int a_midlen, b_midlen;
650 int qlen_a = quote_c_style(a, NULL, NULL, 0);
651 int qlen_b = quote_c_style(b, NULL, NULL, 0);
652
653 strbuf_init(&name, 0);
654 if (qlen_a || qlen_b) {
655 quote_c_style(a, &name, NULL, 0);
656 strbuf_addstr(&name, " => ");
657 quote_c_style(b, &name, NULL, 0);
658 return strbuf_detach(&name, NULL);
659 }
660
661 /* Find common prefix */
662 pfx_length = 0;
663 while (*old && *new && *old == *new) {
664 if (*old == '/')
665 pfx_length = old - a + 1;
666 old++;
667 new++;
668 }
669
670 /* Find common suffix */
671 old = a + len_a;
672 new = b + len_b;
673 sfx_length = 0;
674 while (a <= old && b <= new && *old == *new) {
675 if (*old == '/')
676 sfx_length = len_a - (old - a);
677 old--;
678 new--;
679 }
680
681 /*
682 * pfx{mid-a => mid-b}sfx
683 * {pfx-a => pfx-b}sfx
684 * pfx{sfx-a => sfx-b}
685 * name-a => name-b
686 */
687 a_midlen = len_a - pfx_length - sfx_length;
688 b_midlen = len_b - pfx_length - sfx_length;
689 if (a_midlen < 0)
690 a_midlen = 0;
691 if (b_midlen < 0)
692 b_midlen = 0;
693
694 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
695 if (pfx_length + sfx_length) {
696 strbuf_add(&name, a, pfx_length);
697 strbuf_addch(&name, '{');
698 }
699 strbuf_add(&name, a + pfx_length, a_midlen);
700 strbuf_addstr(&name, " => ");
701 strbuf_add(&name, b + pfx_length, b_midlen);
702 if (pfx_length + sfx_length) {
703 strbuf_addch(&name, '}');
704 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
705 }
706 return strbuf_detach(&name, NULL);
707 }
708
709 struct diffstat_t {
710 struct xdiff_emit_state xm;
711
712 int nr;
713 int alloc;
714 struct diffstat_file {
715 char *from_name;
716 char *name;
717 char *print_name;
718 unsigned is_unmerged:1;
719 unsigned is_binary:1;
720 unsigned is_renamed:1;
721 unsigned int added, deleted;
722 } **files;
723 };
724
725 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
726 const char *name_a,
727 const char *name_b)
728 {
729 struct diffstat_file *x;
730 x = xcalloc(sizeof (*x), 1);
731 if (diffstat->nr == diffstat->alloc) {
732 diffstat->alloc = alloc_nr(diffstat->alloc);
733 diffstat->files = xrealloc(diffstat->files,
734 diffstat->alloc * sizeof(x));
735 }
736 diffstat->files[diffstat->nr++] = x;
737 if (name_b) {
738 x->from_name = xstrdup(name_a);
739 x->name = xstrdup(name_b);
740 x->is_renamed = 1;
741 }
742 else {
743 x->from_name = NULL;
744 x->name = xstrdup(name_a);
745 }
746 return x;
747 }
748
749 static void diffstat_consume(void *priv, char *line, unsigned long len)
750 {
751 struct diffstat_t *diffstat = priv;
752 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
753
754 if (line[0] == '+')
755 x->added++;
756 else if (line[0] == '-')
757 x->deleted++;
758 }
759
760 const char mime_boundary_leader[] = "------------";
761
762 static int scale_linear(int it, int width, int max_change)
763 {
764 /*
765 * make sure that at least one '-' is printed if there were deletions,
766 * and likewise for '+'.
767 */
768 if (max_change < 2)
769 return it;
770 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
771 }
772
773 static void show_name(const char *prefix, const char *name, int len,
774 const char *reset, const char *set)
775 {
776 printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
777 }
778
779 static void show_graph(char ch, int cnt, const char *set, const char *reset)
780 {
781 if (cnt <= 0)
782 return;
783 printf("%s", set);
784 while (cnt--)
785 putchar(ch);
786 printf("%s", reset);
787 }
788
789 static void fill_print_name(struct diffstat_file *file)
790 {
791 char *pname;
792
793 if (file->print_name)
794 return;
795
796 if (!file->is_renamed) {
797 struct strbuf buf;
798 strbuf_init(&buf, 0);
799 if (quote_c_style(file->name, &buf, NULL, 0)) {
800 pname = strbuf_detach(&buf, NULL);
801 } else {
802 pname = file->name;
803 strbuf_release(&buf);
804 }
805 } else {
806 pname = pprint_rename(file->from_name, file->name);
807 }
808 file->print_name = pname;
809 }
810
811 static void show_stats(struct diffstat_t* data, struct diff_options *options)
812 {
813 int i, len, add, del, total, adds = 0, dels = 0;
814 int max_change = 0, max_len = 0;
815 int total_files = data->nr;
816 int width, name_width;
817 const char *reset, *set, *add_c, *del_c;
818
819 if (data->nr == 0)
820 return;
821
822 width = options->stat_width ? options->stat_width : 80;
823 name_width = options->stat_name_width ? options->stat_name_width : 50;
824
825 /* Sanity: give at least 5 columns to the graph,
826 * but leave at least 10 columns for the name.
827 */
828 if (width < name_width + 15) {
829 if (name_width <= 25)
830 width = name_width + 15;
831 else
832 name_width = width - 15;
833 }
834
835 /* Find the longest filename and max number of changes */
836 reset = diff_get_color_opt(options, DIFF_RESET);
837 set = diff_get_color_opt(options, DIFF_PLAIN);
838 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
839 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
840
841 for (i = 0; i < data->nr; i++) {
842 struct diffstat_file *file = data->files[i];
843 int change = file->added + file->deleted;
844 fill_print_name(file);
845 len = strlen(file->print_name);
846 if (max_len < len)
847 max_len = len;
848
849 if (file->is_binary || file->is_unmerged)
850 continue;
851 if (max_change < change)
852 max_change = change;
853 }
854
855 /* Compute the width of the graph part;
856 * 10 is for one blank at the beginning of the line plus
857 * " | count " between the name and the graph.
858 *
859 * From here on, name_width is the width of the name area,
860 * and width is the width of the graph area.
861 */
862 name_width = (name_width < max_len) ? name_width : max_len;
863 if (width < (name_width + 10) + max_change)
864 width = width - (name_width + 10);
865 else
866 width = max_change;
867
868 for (i = 0; i < data->nr; i++) {
869 const char *prefix = "";
870 char *name = data->files[i]->print_name;
871 int added = data->files[i]->added;
872 int deleted = data->files[i]->deleted;
873 int name_len;
874
875 /*
876 * "scale" the filename
877 */
878 len = name_width;
879 name_len = strlen(name);
880 if (name_width < name_len) {
881 char *slash;
882 prefix = "...";
883 len -= 3;
884 name += name_len - len;
885 slash = strchr(name, '/');
886 if (slash)
887 name = slash;
888 }
889
890 if (data->files[i]->is_binary) {
891 show_name(prefix, name, len, reset, set);
892 printf(" Bin ");
893 printf("%s%d%s", del_c, deleted, reset);
894 printf(" -> ");
895 printf("%s%d%s", add_c, added, reset);
896 printf(" bytes");
897 printf("\n");
898 continue;
899 }
900 else if (data->files[i]->is_unmerged) {
901 show_name(prefix, name, len, reset, set);
902 printf(" Unmerged\n");
903 continue;
904 }
905 else if (!data->files[i]->is_renamed &&
906 (added + deleted == 0)) {
907 total_files--;
908 continue;
909 }
910
911 /*
912 * scale the add/delete
913 */
914 add = added;
915 del = deleted;
916 total = add + del;
917 adds += add;
918 dels += del;
919
920 if (width <= max_change) {
921 add = scale_linear(add, width, max_change);
922 del = scale_linear(del, width, max_change);
923 total = add + del;
924 }
925 show_name(prefix, name, len, reset, set);
926 printf("%5d ", added + deleted);
927 show_graph('+', add, add_c, reset);
928 show_graph('-', del, del_c, reset);
929 putchar('\n');
930 }
931 printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
932 set, total_files, adds, dels, reset);
933 }
934
935 static void show_shortstats(struct diffstat_t* data)
936 {
937 int i, adds = 0, dels = 0, total_files = data->nr;
938
939 if (data->nr == 0)
940 return;
941
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 printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
957 total_files, adds, dels);
958 }
959
960 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
961 {
962 int i;
963
964 if (data->nr == 0)
965 return;
966
967 for (i = 0; i < data->nr; i++) {
968 struct diffstat_file *file = data->files[i];
969
970 if (file->is_binary)
971 printf("-\t-\t");
972 else
973 printf("%d\t%d\t", file->added, file->deleted);
974 if (options->line_termination) {
975 fill_print_name(file);
976 if (!file->is_renamed)
977 write_name_quoted(file->name, stdout,
978 options->line_termination);
979 else {
980 fputs(file->print_name, stdout);
981 putchar(options->line_termination);
982 }
983 } else {
984 if (file->is_renamed) {
985 putchar('\0');
986 write_name_quoted(file->from_name, stdout, '\0');
987 }
988 write_name_quoted(file->name, stdout, '\0');
989 }
990 }
991 }
992
993 static void free_diffstat_info(struct diffstat_t *diffstat)
994 {
995 int i;
996 for (i = 0; i < diffstat->nr; i++) {
997 struct diffstat_file *f = diffstat->files[i];
998 if (f->name != f->print_name)
999 free(f->print_name);
1000 free(f->name);
1001 free(f->from_name);
1002 free(f);
1003 }
1004 free(diffstat->files);
1005 }
1006
1007 struct checkdiff_t {
1008 struct xdiff_emit_state xm;
1009 const char *filename;
1010 int lineno, color_diff;
1011 unsigned ws_rule;
1012 unsigned status;
1013 };
1014
1015 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1016 {
1017 struct checkdiff_t *data = priv;
1018 const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1019 const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1020 const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1021 char *err;
1022
1023 if (line[0] == '+') {
1024 data->status = check_and_emit_line(line + 1, len - 1,
1025 data->ws_rule, NULL, NULL, NULL, NULL);
1026 if (!data->status)
1027 return;
1028 err = whitespace_error_string(data->status);
1029 printf("%s:%d: %s.\n", data->filename, data->lineno, err);
1030 free(err);
1031 emit_line(set, reset, line, 1);
1032 (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
1033 stdout, set, reset, ws);
1034 data->lineno++;
1035 } else if (line[0] == ' ')
1036 data->lineno++;
1037 else if (line[0] == '@') {
1038 char *plus = strchr(line, '+');
1039 if (plus)
1040 data->lineno = strtol(plus, NULL, 10);
1041 else
1042 die("invalid diff");
1043 }
1044 }
1045
1046 static unsigned char *deflate_it(char *data,
1047 unsigned long size,
1048 unsigned long *result_size)
1049 {
1050 int bound;
1051 unsigned char *deflated;
1052 z_stream stream;
1053
1054 memset(&stream, 0, sizeof(stream));
1055 deflateInit(&stream, zlib_compression_level);
1056 bound = deflateBound(&stream, size);
1057 deflated = xmalloc(bound);
1058 stream.next_out = deflated;
1059 stream.avail_out = bound;
1060
1061 stream.next_in = (unsigned char *)data;
1062 stream.avail_in = size;
1063 while (deflate(&stream, Z_FINISH) == Z_OK)
1064 ; /* nothing */
1065 deflateEnd(&stream);
1066 *result_size = stream.total_out;
1067 return deflated;
1068 }
1069
1070 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
1071 {
1072 void *cp;
1073 void *delta;
1074 void *deflated;
1075 void *data;
1076 unsigned long orig_size;
1077 unsigned long delta_size;
1078 unsigned long deflate_size;
1079 unsigned long data_size;
1080
1081 /* We could do deflated delta, or we could do just deflated two,
1082 * whichever is smaller.
1083 */
1084 delta = NULL;
1085 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1086 if (one->size && two->size) {
1087 delta = diff_delta(one->ptr, one->size,
1088 two->ptr, two->size,
1089 &delta_size, deflate_size);
1090 if (delta) {
1091 void *to_free = delta;
1092 orig_size = delta_size;
1093 delta = deflate_it(delta, delta_size, &delta_size);
1094 free(to_free);
1095 }
1096 }
1097
1098 if (delta && delta_size < deflate_size) {
1099 printf("delta %lu\n", orig_size);
1100 free(deflated);
1101 data = delta;
1102 data_size = delta_size;
1103 }
1104 else {
1105 printf("literal %lu\n", two->size);
1106 free(delta);
1107 data = deflated;
1108 data_size = deflate_size;
1109 }
1110
1111 /* emit data encoded in base85 */
1112 cp = data;
1113 while (data_size) {
1114 int bytes = (52 < data_size) ? 52 : data_size;
1115 char line[70];
1116 data_size -= bytes;
1117 if (bytes <= 26)
1118 line[0] = bytes + 'A' - 1;
1119 else
1120 line[0] = bytes - 26 + 'a' - 1;
1121 encode_85(line + 1, cp, bytes);
1122 cp = (char *) cp + bytes;
1123 puts(line);
1124 }
1125 printf("\n");
1126 free(data);
1127 }
1128
1129 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1130 {
1131 printf("GIT binary patch\n");
1132 emit_binary_diff_body(one, two);
1133 emit_binary_diff_body(two, one);
1134 }
1135
1136 static void setup_diff_attr_check(struct git_attr_check *check)
1137 {
1138 static struct git_attr *attr_diff;
1139
1140 if (!attr_diff) {
1141 attr_diff = git_attr("diff", 4);
1142 }
1143 check[0].attr = attr_diff;
1144 }
1145
1146 static void diff_filespec_check_attr(struct diff_filespec *one)
1147 {
1148 struct git_attr_check attr_diff_check;
1149 int check_from_data = 0;
1150
1151 if (one->checked_attr)
1152 return;
1153
1154 setup_diff_attr_check(&attr_diff_check);
1155 one->is_binary = 0;
1156 one->funcname_pattern_ident = NULL;
1157
1158 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1159 const char *value;
1160
1161 /* binaryness */
1162 value = attr_diff_check.value;
1163 if (ATTR_TRUE(value))
1164 ;
1165 else if (ATTR_FALSE(value))
1166 one->is_binary = 1;
1167 else
1168 check_from_data = 1;
1169
1170 /* funcname pattern ident */
1171 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1172 ;
1173 else
1174 one->funcname_pattern_ident = value;
1175 }
1176
1177 if (check_from_data) {
1178 if (!one->data && DIFF_FILE_VALID(one))
1179 diff_populate_filespec(one, 0);
1180
1181 if (one->data)
1182 one->is_binary = buffer_is_binary(one->data, one->size);
1183 }
1184 }
1185
1186 int diff_filespec_is_binary(struct diff_filespec *one)
1187 {
1188 diff_filespec_check_attr(one);
1189 return one->is_binary;
1190 }
1191
1192 static const char *funcname_pattern(const char *ident)
1193 {
1194 struct funcname_pattern *pp;
1195
1196 for (pp = funcname_pattern_list; pp; pp = pp->next)
1197 if (!strcmp(ident, pp->name))
1198 return pp->pattern;
1199 return NULL;
1200 }
1201
1202 static struct builtin_funcname_pattern {
1203 const char *name;
1204 const char *pattern;
1205 } builtin_funcname_pattern[] = {
1206 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1207 "new\\|return\\|switch\\|throw\\|while\\)\n"
1208 "^[ ]*\\(\\([ ]*"
1209 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1210 "[ ]*([^;]*$\\)" },
1211 { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1212 };
1213
1214 static const char *diff_funcname_pattern(struct diff_filespec *one)
1215 {
1216 const char *ident, *pattern;
1217 int i;
1218
1219 diff_filespec_check_attr(one);
1220 ident = one->funcname_pattern_ident;
1221
1222 if (!ident)
1223 /*
1224 * If the config file has "funcname.default" defined, that
1225 * regexp is used; otherwise NULL is returned and xemit uses
1226 * the built-in default.
1227 */
1228 return funcname_pattern("default");
1229
1230 /* Look up custom "funcname.$ident" regexp from config. */
1231 pattern = funcname_pattern(ident);
1232 if (pattern)
1233 return pattern;
1234
1235 /*
1236 * And define built-in fallback patterns here. Note that
1237 * these can be overridden by the user's config settings.
1238 */
1239 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1240 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1241 return builtin_funcname_pattern[i].pattern;
1242
1243 return NULL;
1244 }
1245
1246 static void builtin_diff(const char *name_a,
1247 const char *name_b,
1248 struct diff_filespec *one,
1249 struct diff_filespec *two,
1250 const char *xfrm_msg,
1251 struct diff_options *o,
1252 int complete_rewrite)
1253 {
1254 mmfile_t mf1, mf2;
1255 const char *lbl[2];
1256 char *a_one, *b_two;
1257 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1258 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1259
1260 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1261 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1262 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1263 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1264 printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1265 if (lbl[0][0] == '/') {
1266 /* /dev/null */
1267 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1268 if (xfrm_msg && xfrm_msg[0])
1269 printf("%s%s%s\n", set, xfrm_msg, reset);
1270 }
1271 else if (lbl[1][0] == '/') {
1272 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1273 if (xfrm_msg && xfrm_msg[0])
1274 printf("%s%s%s\n", set, xfrm_msg, reset);
1275 }
1276 else {
1277 if (one->mode != two->mode) {
1278 printf("%sold mode %06o%s\n", set, one->mode, reset);
1279 printf("%snew mode %06o%s\n", set, two->mode, reset);
1280 }
1281 if (xfrm_msg && xfrm_msg[0])
1282 printf("%s%s%s\n", set, xfrm_msg, reset);
1283 /*
1284 * we do not run diff between different kind
1285 * of objects.
1286 */
1287 if ((one->mode ^ two->mode) & S_IFMT)
1288 goto free_ab_and_return;
1289 if (complete_rewrite) {
1290 emit_rewrite_diff(name_a, name_b, one, two, o);
1291 o->found_changes = 1;
1292 goto free_ab_and_return;
1293 }
1294 }
1295
1296 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1297 die("unable to read files to diff");
1298
1299 if (!DIFF_OPT_TST(o, TEXT) &&
1300 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1301 /* Quite common confusing case */
1302 if (mf1.size == mf2.size &&
1303 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1304 goto free_ab_and_return;
1305 if (DIFF_OPT_TST(o, BINARY))
1306 emit_binary_diff(&mf1, &mf2);
1307 else
1308 printf("Binary files %s and %s differ\n",
1309 lbl[0], lbl[1]);
1310 o->found_changes = 1;
1311 }
1312 else {
1313 /* Crazy xdl interfaces.. */
1314 const char *diffopts = getenv("GIT_DIFF_OPTS");
1315 xpparam_t xpp;
1316 xdemitconf_t xecfg;
1317 xdemitcb_t ecb;
1318 struct emit_callback ecbdata;
1319 const char *funcname_pattern;
1320
1321 funcname_pattern = diff_funcname_pattern(one);
1322 if (!funcname_pattern)
1323 funcname_pattern = diff_funcname_pattern(two);
1324
1325 memset(&xecfg, 0, sizeof(xecfg));
1326 memset(&ecbdata, 0, sizeof(ecbdata));
1327 ecbdata.label_path = lbl;
1328 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1329 ecbdata.found_changesp = &o->found_changes;
1330 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1331 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1332 xecfg.ctxlen = o->context;
1333 xecfg.flags = XDL_EMIT_FUNCNAMES;
1334 if (funcname_pattern)
1335 xdiff_set_find_func(&xecfg, funcname_pattern);
1336 if (!diffopts)
1337 ;
1338 else if (!prefixcmp(diffopts, "--unified="))
1339 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1340 else if (!prefixcmp(diffopts, "-u"))
1341 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1342 ecb.outf = xdiff_outf;
1343 ecb.priv = &ecbdata;
1344 ecbdata.xm.consume = fn_out_consume;
1345 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1346 ecbdata.diff_words =
1347 xcalloc(1, sizeof(struct diff_words_data));
1348 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1349 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1350 free_diff_words_data(&ecbdata);
1351 }
1352
1353 free_ab_and_return:
1354 diff_free_filespec_data(one);
1355 diff_free_filespec_data(two);
1356 free(a_one);
1357 free(b_two);
1358 return;
1359 }
1360
1361 static void builtin_diffstat(const char *name_a, const char *name_b,
1362 struct diff_filespec *one,
1363 struct diff_filespec *two,
1364 struct diffstat_t *diffstat,
1365 struct diff_options *o,
1366 int complete_rewrite)
1367 {
1368 mmfile_t mf1, mf2;
1369 struct diffstat_file *data;
1370
1371 data = diffstat_add(diffstat, name_a, name_b);
1372
1373 if (!one || !two) {
1374 data->is_unmerged = 1;
1375 return;
1376 }
1377 if (complete_rewrite) {
1378 diff_populate_filespec(one, 0);
1379 diff_populate_filespec(two, 0);
1380 data->deleted = count_lines(one->data, one->size);
1381 data->added = count_lines(two->data, two->size);
1382 goto free_and_return;
1383 }
1384 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1385 die("unable to read files to diff");
1386
1387 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1388 data->is_binary = 1;
1389 data->added = mf2.size;
1390 data->deleted = mf1.size;
1391 } else {
1392 /* Crazy xdl interfaces.. */
1393 xpparam_t xpp;
1394 xdemitconf_t xecfg;
1395 xdemitcb_t ecb;
1396
1397 memset(&xecfg, 0, sizeof(xecfg));
1398 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1399 ecb.outf = xdiff_outf;
1400 ecb.priv = diffstat;
1401 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1402 }
1403
1404 free_and_return:
1405 diff_free_filespec_data(one);
1406 diff_free_filespec_data(two);
1407 }
1408
1409 static void builtin_checkdiff(const char *name_a, const char *name_b,
1410 struct diff_filespec *one,
1411 struct diff_filespec *two, struct diff_options *o)
1412 {
1413 mmfile_t mf1, mf2;
1414 struct checkdiff_t data;
1415
1416 if (!two)
1417 return;
1418
1419 memset(&data, 0, sizeof(data));
1420 data.xm.consume = checkdiff_consume;
1421 data.filename = name_b ? name_b : name_a;
1422 data.lineno = 0;
1423 data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1424 data.ws_rule = whitespace_rule(data.filename);
1425
1426 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1427 die("unable to read files to diff");
1428
1429 if (diff_filespec_is_binary(two))
1430 goto free_and_return;
1431 else {
1432 /* Crazy xdl interfaces.. */
1433 xpparam_t xpp;
1434 xdemitconf_t xecfg;
1435 xdemitcb_t ecb;
1436
1437 memset(&xecfg, 0, sizeof(xecfg));
1438 xpp.flags = XDF_NEED_MINIMAL;
1439 ecb.outf = xdiff_outf;
1440 ecb.priv = &data;
1441 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1442 }
1443 free_and_return:
1444 diff_free_filespec_data(one);
1445 diff_free_filespec_data(two);
1446 if (data.status)
1447 DIFF_OPT_SET(o, CHECK_FAILED);
1448 }
1449
1450 struct diff_filespec *alloc_filespec(const char *path)
1451 {
1452 int namelen = strlen(path);
1453 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1454
1455 memset(spec, 0, sizeof(*spec));
1456 spec->path = (char *)(spec + 1);
1457 memcpy(spec->path, path, namelen+1);
1458 spec->count = 1;
1459 return spec;
1460 }
1461
1462 void free_filespec(struct diff_filespec *spec)
1463 {
1464 if (!--spec->count) {
1465 diff_free_filespec_data(spec);
1466 free(spec);
1467 }
1468 }
1469
1470 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1471 unsigned short mode)
1472 {
1473 if (mode) {
1474 spec->mode = canon_mode(mode);
1475 hashcpy(spec->sha1, sha1);
1476 spec->sha1_valid = !is_null_sha1(sha1);
1477 }
1478 }
1479
1480 /*
1481 * Given a name and sha1 pair, if the index tells us the file in
1482 * the work tree has that object contents, return true, so that
1483 * prepare_temp_file() does not have to inflate and extract.
1484 */
1485 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1486 {
1487 struct cache_entry *ce;
1488 struct stat st;
1489 int pos, len;
1490
1491 /* We do not read the cache ourselves here, because the
1492 * benchmark with my previous version that always reads cache
1493 * shows that it makes things worse for diff-tree comparing
1494 * two linux-2.6 kernel trees in an already checked out work
1495 * tree. This is because most diff-tree comparisons deal with
1496 * only a small number of files, while reading the cache is
1497 * expensive for a large project, and its cost outweighs the
1498 * savings we get by not inflating the object to a temporary
1499 * file. Practically, this code only helps when we are used
1500 * by diff-cache --cached, which does read the cache before
1501 * calling us.
1502 */
1503 if (!active_cache)
1504 return 0;
1505
1506 /* We want to avoid the working directory if our caller
1507 * doesn't need the data in a normal file, this system
1508 * is rather slow with its stat/open/mmap/close syscalls,
1509 * and the object is contained in a pack file. The pack
1510 * is probably already open and will be faster to obtain
1511 * the data through than the working directory. Loose
1512 * objects however would tend to be slower as they need
1513 * to be individually opened and inflated.
1514 */
1515 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1516 return 0;
1517
1518 len = strlen(name);
1519 pos = cache_name_pos(name, len);
1520 if (pos < 0)
1521 return 0;
1522 ce = active_cache[pos];
1523
1524 /*
1525 * This is not the sha1 we are looking for, or
1526 * unreusable because it is not a regular file.
1527 */
1528 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1529 return 0;
1530
1531 /*
1532 * If ce matches the file in the work tree, we can reuse it.
1533 */
1534 if (ce_uptodate(ce) ||
1535 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1536 return 1;
1537
1538 return 0;
1539 }
1540
1541 static int populate_from_stdin(struct diff_filespec *s)
1542 {
1543 struct strbuf buf;
1544 size_t size = 0;
1545
1546 strbuf_init(&buf, 0);
1547 if (strbuf_read(&buf, 0, 0) < 0)
1548 return error("error while reading from stdin %s",
1549 strerror(errno));
1550
1551 s->should_munmap = 0;
1552 s->data = strbuf_detach(&buf, &size);
1553 s->size = size;
1554 s->should_free = 1;
1555 return 0;
1556 }
1557
1558 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1559 {
1560 int len;
1561 char *data = xmalloc(100);
1562 len = snprintf(data, 100,
1563 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1564 s->data = data;
1565 s->size = len;
1566 s->should_free = 1;
1567 if (size_only) {
1568 s->data = NULL;
1569 free(data);
1570 }
1571 return 0;
1572 }
1573
1574 /*
1575 * While doing rename detection and pickaxe operation, we may need to
1576 * grab the data for the blob (or file) for our own in-core comparison.
1577 * diff_filespec has data and size fields for this purpose.
1578 */
1579 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1580 {
1581 int err = 0;
1582 if (!DIFF_FILE_VALID(s))
1583 die("internal error: asking to populate invalid file.");
1584 if (S_ISDIR(s->mode))
1585 return -1;
1586
1587 if (s->data)
1588 return 0;
1589
1590 if (size_only && 0 < s->size)
1591 return 0;
1592
1593 if (S_ISGITLINK(s->mode))
1594 return diff_populate_gitlink(s, size_only);
1595
1596 if (!s->sha1_valid ||
1597 reuse_worktree_file(s->path, s->sha1, 0)) {
1598 struct strbuf buf;
1599 struct stat st;
1600 int fd;
1601
1602 if (!strcmp(s->path, "-"))
1603 return populate_from_stdin(s);
1604
1605 if (lstat(s->path, &st) < 0) {
1606 if (errno == ENOENT) {
1607 err_empty:
1608 err = -1;
1609 empty:
1610 s->data = (char *)"";
1611 s->size = 0;
1612 return err;
1613 }
1614 }
1615 s->size = xsize_t(st.st_size);
1616 if (!s->size)
1617 goto empty;
1618 if (size_only)
1619 return 0;
1620 if (S_ISLNK(st.st_mode)) {
1621 int ret;
1622 s->data = xmalloc(s->size);
1623 s->should_free = 1;
1624 ret = readlink(s->path, s->data, s->size);
1625 if (ret < 0) {
1626 free(s->data);
1627 goto err_empty;
1628 }
1629 return 0;
1630 }
1631 fd = open(s->path, O_RDONLY);
1632 if (fd < 0)
1633 goto err_empty;
1634 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1635 close(fd);
1636 s->should_munmap = 1;
1637
1638 /*
1639 * Convert from working tree format to canonical git format
1640 */
1641 strbuf_init(&buf, 0);
1642 if (convert_to_git(s->path, s->data, s->size, &buf)) {
1643 size_t size = 0;
1644 munmap(s->data, s->size);
1645 s->should_munmap = 0;
1646 s->data = strbuf_detach(&buf, &size);
1647 s->size = size;
1648 s->should_free = 1;
1649 }
1650 }
1651 else {
1652 enum object_type type;
1653 if (size_only)
1654 type = sha1_object_info(s->sha1, &s->size);
1655 else {
1656 s->data = read_sha1_file(s->sha1, &type, &s->size);
1657 s->should_free = 1;
1658 }
1659 }
1660 return 0;
1661 }
1662
1663 void diff_free_filespec_blob(struct diff_filespec *s)
1664 {
1665 if (s->should_free)
1666 free(s->data);
1667 else if (s->should_munmap)
1668 munmap(s->data, s->size);
1669
1670 if (s->should_free || s->should_munmap) {
1671 s->should_free = s->should_munmap = 0;
1672 s->data = NULL;
1673 }
1674 }
1675
1676 void diff_free_filespec_data(struct diff_filespec *s)
1677 {
1678 diff_free_filespec_blob(s);
1679 free(s->cnt_data);
1680 s->cnt_data = NULL;
1681 }
1682
1683 static void prep_temp_blob(struct diff_tempfile *temp,
1684 void *blob,
1685 unsigned long size,
1686 const unsigned char *sha1,
1687 int mode)
1688 {
1689 int fd;
1690
1691 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1692 if (fd < 0)
1693 die("unable to create temp-file: %s", strerror(errno));
1694 if (write_in_full(fd, blob, size) != size)
1695 die("unable to write temp-file");
1696 close(fd);
1697 temp->name = temp->tmp_path;
1698 strcpy(temp->hex, sha1_to_hex(sha1));
1699 temp->hex[40] = 0;
1700 sprintf(temp->mode, "%06o", mode);
1701 }
1702
1703 static void prepare_temp_file(const char *name,
1704 struct diff_tempfile *temp,
1705 struct diff_filespec *one)
1706 {
1707 if (!DIFF_FILE_VALID(one)) {
1708 not_a_valid_file:
1709 /* A '-' entry produces this for file-2, and
1710 * a '+' entry produces this for file-1.
1711 */
1712 temp->name = "/dev/null";
1713 strcpy(temp->hex, ".");
1714 strcpy(temp->mode, ".");
1715 return;
1716 }
1717
1718 if (!one->sha1_valid ||
1719 reuse_worktree_file(name, one->sha1, 1)) {
1720 struct stat st;
1721 if (lstat(name, &st) < 0) {
1722 if (errno == ENOENT)
1723 goto not_a_valid_file;
1724 die("stat(%s): %s", name, strerror(errno));
1725 }
1726 if (S_ISLNK(st.st_mode)) {
1727 int ret;
1728 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1729 size_t sz = xsize_t(st.st_size);
1730 if (sizeof(buf) <= st.st_size)
1731 die("symlink too long: %s", name);
1732 ret = readlink(name, buf, sz);
1733 if (ret < 0)
1734 die("readlink(%s)", name);
1735 prep_temp_blob(temp, buf, sz,
1736 (one->sha1_valid ?
1737 one->sha1 : null_sha1),
1738 (one->sha1_valid ?
1739 one->mode : S_IFLNK));
1740 }
1741 else {
1742 /* we can borrow from the file in the work tree */
1743 temp->name = name;
1744 if (!one->sha1_valid)
1745 strcpy(temp->hex, sha1_to_hex(null_sha1));
1746 else
1747 strcpy(temp->hex, sha1_to_hex(one->sha1));
1748 /* Even though we may sometimes borrow the
1749 * contents from the work tree, we always want
1750 * one->mode. mode is trustworthy even when
1751 * !(one->sha1_valid), as long as
1752 * DIFF_FILE_VALID(one).
1753 */
1754 sprintf(temp->mode, "%06o", one->mode);
1755 }
1756 return;
1757 }
1758 else {
1759 if (diff_populate_filespec(one, 0))
1760 die("cannot read data blob for %s", one->path);
1761 prep_temp_blob(temp, one->data, one->size,
1762 one->sha1, one->mode);
1763 }
1764 }
1765
1766 static void remove_tempfile(void)
1767 {
1768 int i;
1769
1770 for (i = 0; i < 2; i++)
1771 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1772 unlink(diff_temp[i].name);
1773 diff_temp[i].name = NULL;
1774 }
1775 }
1776
1777 static void remove_tempfile_on_signal(int signo)
1778 {
1779 remove_tempfile();
1780 signal(SIGINT, SIG_DFL);
1781 raise(signo);
1782 }
1783
1784 /* An external diff command takes:
1785 *
1786 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1787 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1788 *
1789 */
1790 static void run_external_diff(const char *pgm,
1791 const char *name,
1792 const char *other,
1793 struct diff_filespec *one,
1794 struct diff_filespec *two,
1795 const char *xfrm_msg,
1796 int complete_rewrite)
1797 {
1798 const char *spawn_arg[10];
1799 struct diff_tempfile *temp = diff_temp;
1800 int retval;
1801 static int atexit_asked = 0;
1802 const char *othername;
1803 const char **arg = &spawn_arg[0];
1804
1805 othername = (other? other : name);
1806 if (one && two) {
1807 prepare_temp_file(name, &temp[0], one);
1808 prepare_temp_file(othername, &temp[1], two);
1809 if (! atexit_asked &&
1810 (temp[0].name == temp[0].tmp_path ||
1811 temp[1].name == temp[1].tmp_path)) {
1812 atexit_asked = 1;
1813 atexit(remove_tempfile);
1814 }
1815 signal(SIGINT, remove_tempfile_on_signal);
1816 }
1817
1818 if (one && two) {
1819 *arg++ = pgm;
1820 *arg++ = name;
1821 *arg++ = temp[0].name;
1822 *arg++ = temp[0].hex;
1823 *arg++ = temp[0].mode;
1824 *arg++ = temp[1].name;
1825 *arg++ = temp[1].hex;
1826 *arg++ = temp[1].mode;
1827 if (other) {
1828 *arg++ = other;
1829 *arg++ = xfrm_msg;
1830 }
1831 } else {
1832 *arg++ = pgm;
1833 *arg++ = name;
1834 }
1835 *arg = NULL;
1836 fflush(NULL);
1837 retval = run_command_v_opt(spawn_arg, 0);
1838 remove_tempfile();
1839 if (retval) {
1840 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1841 exit(1);
1842 }
1843 }
1844
1845 static const char *external_diff_attr(const char *name)
1846 {
1847 struct git_attr_check attr_diff_check;
1848
1849 setup_diff_attr_check(&attr_diff_check);
1850 if (!git_checkattr(name, 1, &attr_diff_check)) {
1851 const char *value = attr_diff_check.value;
1852 if (!ATTR_TRUE(value) &&
1853 !ATTR_FALSE(value) &&
1854 !ATTR_UNSET(value)) {
1855 struct ll_diff_driver *drv;
1856
1857 for (drv = user_diff; drv; drv = drv->next)
1858 if (!strcmp(drv->name, value))
1859 return drv->cmd;
1860 }
1861 }
1862 return NULL;
1863 }
1864
1865 static void run_diff_cmd(const char *pgm,
1866 const char *name,
1867 const char *other,
1868 struct diff_filespec *one,
1869 struct diff_filespec *two,
1870 const char *xfrm_msg,
1871 struct diff_options *o,
1872 int complete_rewrite)
1873 {
1874 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
1875 pgm = NULL;
1876 else {
1877 const char *cmd = external_diff_attr(name);
1878 if (cmd)
1879 pgm = cmd;
1880 }
1881
1882 if (pgm) {
1883 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1884 complete_rewrite);
1885 return;
1886 }
1887 if (one && two)
1888 builtin_diff(name, other ? other : name,
1889 one, two, xfrm_msg, o, complete_rewrite);
1890 else
1891 printf("* Unmerged path %s\n", name);
1892 }
1893
1894 static void diff_fill_sha1_info(struct diff_filespec *one)
1895 {
1896 if (DIFF_FILE_VALID(one)) {
1897 if (!one->sha1_valid) {
1898 struct stat st;
1899 if (!strcmp(one->path, "-")) {
1900 hashcpy(one->sha1, null_sha1);
1901 return;
1902 }
1903 if (lstat(one->path, &st) < 0)
1904 die("stat %s", one->path);
1905 if (index_path(one->sha1, one->path, &st, 0))
1906 die("cannot hash %s\n", one->path);
1907 }
1908 }
1909 else
1910 hashclr(one->sha1);
1911 }
1912
1913 static int similarity_index(struct diff_filepair *p)
1914 {
1915 return p->score * 100 / MAX_SCORE;
1916 }
1917
1918 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1919 {
1920 const char *pgm = external_diff();
1921 struct strbuf msg;
1922 char *xfrm_msg;
1923 struct diff_filespec *one = p->one;
1924 struct diff_filespec *two = p->two;
1925 const char *name;
1926 const char *other;
1927 int complete_rewrite = 0;
1928
1929
1930 if (DIFF_PAIR_UNMERGED(p)) {
1931 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1932 return;
1933 }
1934
1935 name = p->one->path;
1936 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1937 diff_fill_sha1_info(one);
1938 diff_fill_sha1_info(two);
1939
1940 strbuf_init(&msg, PATH_MAX * 2 + 300);
1941 switch (p->status) {
1942 case DIFF_STATUS_COPIED:
1943 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1944 strbuf_addstr(&msg, "\ncopy from ");
1945 quote_c_style(name, &msg, NULL, 0);
1946 strbuf_addstr(&msg, "\ncopy to ");
1947 quote_c_style(other, &msg, NULL, 0);
1948 strbuf_addch(&msg, '\n');
1949 break;
1950 case DIFF_STATUS_RENAMED:
1951 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1952 strbuf_addstr(&msg, "\nrename from ");
1953 quote_c_style(name, &msg, NULL, 0);
1954 strbuf_addstr(&msg, "\nrename to ");
1955 quote_c_style(other, &msg, NULL, 0);
1956 strbuf_addch(&msg, '\n');
1957 break;
1958 case DIFF_STATUS_MODIFIED:
1959 if (p->score) {
1960 strbuf_addf(&msg, "dissimilarity index %d%%\n",
1961 similarity_index(p));
1962 complete_rewrite = 1;
1963 break;
1964 }
1965 /* fallthru */
1966 default:
1967 /* nothing */
1968 ;
1969 }
1970
1971 if (hashcmp(one->sha1, two->sha1)) {
1972 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
1973
1974 if (DIFF_OPT_TST(o, BINARY)) {
1975 mmfile_t mf;
1976 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
1977 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
1978 abbrev = 40;
1979 }
1980 strbuf_addf(&msg, "index %.*s..%.*s",
1981 abbrev, sha1_to_hex(one->sha1),
1982 abbrev, sha1_to_hex(two->sha1));
1983 if (one->mode == two->mode)
1984 strbuf_addf(&msg, " %06o", one->mode);
1985 strbuf_addch(&msg, '\n');
1986 }
1987
1988 if (msg.len)
1989 strbuf_setlen(&msg, msg.len - 1);
1990 xfrm_msg = msg.len ? msg.buf : NULL;
1991
1992 if (!pgm &&
1993 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1994 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1995 /* a filepair that changes between file and symlink
1996 * needs to be split into deletion and creation.
1997 */
1998 struct diff_filespec *null = alloc_filespec(two->path);
1999 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
2000 free(null);
2001 null = alloc_filespec(one->path);
2002 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
2003 free(null);
2004 }
2005 else
2006 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
2007 complete_rewrite);
2008
2009 strbuf_release(&msg);
2010 }
2011
2012 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2013 struct diffstat_t *diffstat)
2014 {
2015 const char *name;
2016 const char *other;
2017 int complete_rewrite = 0;
2018
2019 if (DIFF_PAIR_UNMERGED(p)) {
2020 /* unmerged */
2021 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2022 return;
2023 }
2024
2025 name = p->one->path;
2026 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2027
2028 diff_fill_sha1_info(p->one);
2029 diff_fill_sha1_info(p->two);
2030
2031 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2032 complete_rewrite = 1;
2033 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2034 }
2035
2036 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2037 {
2038 const char *name;
2039 const char *other;
2040
2041 if (DIFF_PAIR_UNMERGED(p)) {
2042 /* unmerged */
2043 return;
2044 }
2045
2046 name = p->one->path;
2047 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2048
2049 diff_fill_sha1_info(p->one);
2050 diff_fill_sha1_info(p->two);
2051
2052 builtin_checkdiff(name, other, p->one, p->two, o);
2053 }
2054
2055 void diff_setup(struct diff_options *options)
2056 {
2057 memset(options, 0, sizeof(*options));
2058 options->line_termination = '\n';
2059 options->break_opt = -1;
2060 options->rename_limit = -1;
2061 options->context = 3;
2062 options->msg_sep = "";
2063
2064 options->change = diff_change;
2065 options->add_remove = diff_addremove;
2066 if (diff_use_color_default)
2067 DIFF_OPT_SET(options, COLOR_DIFF);
2068 else
2069 DIFF_OPT_CLR(options, COLOR_DIFF);
2070 options->detect_rename = diff_detect_rename_default;
2071
2072 options->a_prefix = "a/";
2073 options->b_prefix = "b/";
2074 }
2075
2076 int diff_setup_done(struct diff_options *options)
2077 {
2078 int count = 0;
2079
2080 if (options->output_format & DIFF_FORMAT_NAME)
2081 count++;
2082 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2083 count++;
2084 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2085 count++;
2086 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2087 count++;
2088 if (count > 1)
2089 die("--name-only, --name-status, --check and -s are mutually exclusive");
2090
2091 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2092 options->detect_rename = DIFF_DETECT_COPY;
2093
2094 if (options->output_format & (DIFF_FORMAT_NAME |
2095 DIFF_FORMAT_NAME_STATUS |
2096 DIFF_FORMAT_CHECKDIFF |
2097 DIFF_FORMAT_NO_OUTPUT))
2098 options->output_format &= ~(DIFF_FORMAT_RAW |
2099 DIFF_FORMAT_NUMSTAT |
2100 DIFF_FORMAT_DIFFSTAT |
2101 DIFF_FORMAT_SHORTSTAT |
2102 DIFF_FORMAT_SUMMARY |
2103 DIFF_FORMAT_PATCH);
2104
2105 /*
2106 * These cases always need recursive; we do not drop caller-supplied
2107 * recursive bits for other formats here.
2108 */
2109 if (options->output_format & (DIFF_FORMAT_PATCH |
2110 DIFF_FORMAT_NUMSTAT |
2111 DIFF_FORMAT_DIFFSTAT |
2112 DIFF_FORMAT_SHORTSTAT |
2113 DIFF_FORMAT_SUMMARY |
2114 DIFF_FORMAT_CHECKDIFF))
2115 DIFF_OPT_SET(options, RECURSIVE);
2116 /*
2117 * Also pickaxe would not work very well if you do not say recursive
2118 */
2119 if (options->pickaxe)
2120 DIFF_OPT_SET(options, RECURSIVE);
2121
2122 if (options->detect_rename && options->rename_limit < 0)
2123 options->rename_limit = diff_rename_limit_default;
2124 if (options->setup & DIFF_SETUP_USE_CACHE) {
2125 if (!active_cache)
2126 /* read-cache does not die even when it fails
2127 * so it is safe for us to do this here. Also
2128 * it does not smudge active_cache or active_nr
2129 * when it fails, so we do not have to worry about
2130 * cleaning it up ourselves either.
2131 */
2132 read_cache();
2133 }
2134 if (options->abbrev <= 0 || 40 < options->abbrev)
2135 options->abbrev = 40; /* full */
2136
2137 /*
2138 * It does not make sense to show the first hit we happened
2139 * to have found. It does not make sense not to return with
2140 * exit code in such a case either.
2141 */
2142 if (DIFF_OPT_TST(options, QUIET)) {
2143 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2144 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2145 }
2146
2147 /*
2148 * If we postprocess in diffcore, we cannot simply return
2149 * upon the first hit. We need to run diff as usual.
2150 */
2151 if (options->pickaxe || options->filter)
2152 DIFF_OPT_CLR(options, QUIET);
2153
2154 return 0;
2155 }
2156
2157 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2158 {
2159 char c, *eq;
2160 int len;
2161
2162 if (*arg != '-')
2163 return 0;
2164 c = *++arg;
2165 if (!c)
2166 return 0;
2167 if (c == arg_short) {
2168 c = *++arg;
2169 if (!c)
2170 return 1;
2171 if (val && isdigit(c)) {
2172 char *end;
2173 int n = strtoul(arg, &end, 10);
2174 if (*end)
2175 return 0;
2176 *val = n;
2177 return 1;
2178 }
2179 return 0;
2180 }
2181 if (c != '-')
2182 return 0;
2183 arg++;
2184 eq = strchr(arg, '=');
2185 if (eq)
2186 len = eq - arg;
2187 else
2188 len = strlen(arg);
2189 if (!len || strncmp(arg, arg_long, len))
2190 return 0;
2191 if (eq) {
2192 int n;
2193 char *end;
2194 if (!isdigit(*++eq))
2195 return 0;
2196 n = strtoul(eq, &end, 10);
2197 if (*end)
2198 return 0;
2199 *val = n;
2200 }
2201 return 1;
2202 }
2203
2204 static int diff_scoreopt_parse(const char *opt);
2205
2206 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2207 {
2208 const char *arg = av[0];
2209
2210 /* Output format options */
2211 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2212 options->output_format |= DIFF_FORMAT_PATCH;
2213 else if (opt_arg(arg, 'U', "unified", &options->context))
2214 options->output_format |= DIFF_FORMAT_PATCH;
2215 else if (!strcmp(arg, "--raw"))
2216 options->output_format |= DIFF_FORMAT_RAW;
2217 else if (!strcmp(arg, "--patch-with-raw"))
2218 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2219 else if (!strcmp(arg, "--numstat"))
2220 options->output_format |= DIFF_FORMAT_NUMSTAT;
2221 else if (!strcmp(arg, "--shortstat"))
2222 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2223 else if (!strcmp(arg, "--check"))
2224 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2225 else if (!strcmp(arg, "--summary"))
2226 options->output_format |= DIFF_FORMAT_SUMMARY;
2227 else if (!strcmp(arg, "--patch-with-stat"))
2228 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2229 else if (!strcmp(arg, "--name-only"))
2230 options->output_format |= DIFF_FORMAT_NAME;
2231 else if (!strcmp(arg, "--name-status"))
2232 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2233 else if (!strcmp(arg, "-s"))
2234 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2235 else if (!prefixcmp(arg, "--stat")) {
2236 char *end;
2237 int width = options->stat_width;
2238 int name_width = options->stat_name_width;
2239 arg += 6;
2240 end = (char *)arg;
2241
2242 switch (*arg) {
2243 case '-':
2244 if (!prefixcmp(arg, "-width="))
2245 width = strtoul(arg + 7, &end, 10);
2246 else if (!prefixcmp(arg, "-name-width="))
2247 name_width = strtoul(arg + 12, &end, 10);
2248 break;
2249 case '=':
2250 width = strtoul(arg+1, &end, 10);
2251 if (*end == ',')
2252 name_width = strtoul(end+1, &end, 10);
2253 }
2254
2255 /* Important! This checks all the error cases! */
2256 if (*end)
2257 return 0;
2258 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2259 options->stat_name_width = name_width;
2260 options->stat_width = width;
2261 }
2262
2263 /* renames options */
2264 else if (!prefixcmp(arg, "-B")) {
2265 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2266 return -1;
2267 }
2268 else if (!prefixcmp(arg, "-M")) {
2269 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2270 return -1;
2271 options->detect_rename = DIFF_DETECT_RENAME;
2272 }
2273 else if (!prefixcmp(arg, "-C")) {
2274 if (options->detect_rename == DIFF_DETECT_COPY)
2275 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2276 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2277 return -1;
2278 options->detect_rename = DIFF_DETECT_COPY;
2279 }
2280 else if (!strcmp(arg, "--no-renames"))
2281 options->detect_rename = 0;
2282
2283 /* xdiff options */
2284 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2285 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2286 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2287 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2288 else if (!strcmp(arg, "--ignore-space-at-eol"))
2289 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2290
2291 /* flags options */
2292 else if (!strcmp(arg, "--binary")) {
2293 options->output_format |= DIFF_FORMAT_PATCH;
2294 DIFF_OPT_SET(options, BINARY);
2295 }
2296 else if (!strcmp(arg, "--full-index"))
2297 DIFF_OPT_SET(options, FULL_INDEX);
2298 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2299 DIFF_OPT_SET(options, TEXT);
2300 else if (!strcmp(arg, "-R"))
2301 DIFF_OPT_SET(options, REVERSE_DIFF);
2302 else if (!strcmp(arg, "--find-copies-harder"))
2303 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2304 else if (!strcmp(arg, "--follow"))
2305 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2306 else if (!strcmp(arg, "--color"))
2307 DIFF_OPT_SET(options, COLOR_DIFF);
2308 else if (!strcmp(arg, "--no-color"))
2309 DIFF_OPT_CLR(options, COLOR_DIFF);
2310 else if (!strcmp(arg, "--color-words"))
2311 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2312 else if (!strcmp(arg, "--exit-code"))
2313 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2314 else if (!strcmp(arg, "--quiet"))
2315 DIFF_OPT_SET(options, QUIET);
2316 else if (!strcmp(arg, "--ext-diff"))
2317 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2318 else if (!strcmp(arg, "--no-ext-diff"))
2319 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2320
2321 /* misc options */
2322 else if (!strcmp(arg, "-z"))
2323 options->line_termination = 0;
2324 else if (!prefixcmp(arg, "-l"))
2325 options->rename_limit = strtoul(arg+2, NULL, 10);
2326 else if (!prefixcmp(arg, "-S"))
2327 options->pickaxe = arg + 2;
2328 else if (!strcmp(arg, "--pickaxe-all"))
2329 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2330 else if (!strcmp(arg, "--pickaxe-regex"))
2331 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2332 else if (!prefixcmp(arg, "-O"))
2333 options->orderfile = arg + 2;
2334 else if (!prefixcmp(arg, "--diff-filter="))
2335 options->filter = arg + 14;
2336 else if (!strcmp(arg, "--abbrev"))
2337 options->abbrev = DEFAULT_ABBREV;
2338 else if (!prefixcmp(arg, "--abbrev=")) {
2339 options->abbrev = strtoul(arg + 9, NULL, 10);
2340 if (options->abbrev < MINIMUM_ABBREV)
2341 options->abbrev = MINIMUM_ABBREV;
2342 else if (40 < options->abbrev)
2343 options->abbrev = 40;
2344 }
2345 else if (!prefixcmp(arg, "--src-prefix="))
2346 options->a_prefix = arg + 13;
2347 else if (!prefixcmp(arg, "--dst-prefix="))
2348 options->b_prefix = arg + 13;
2349 else if (!strcmp(arg, "--no-prefix"))
2350 options->a_prefix = options->b_prefix = "";
2351 else
2352 return 0;
2353 return 1;
2354 }
2355
2356 static int parse_num(const char **cp_p)
2357 {
2358 unsigned long num, scale;
2359 int ch, dot;
2360 const char *cp = *cp_p;
2361
2362 num = 0;
2363 scale = 1;
2364 dot = 0;
2365 for(;;) {
2366 ch = *cp;
2367 if ( !dot && ch == '.' ) {
2368 scale = 1;
2369 dot = 1;
2370 } else if ( ch == '%' ) {
2371 scale = dot ? scale*100 : 100;
2372 cp++; /* % is always at the end */
2373 break;
2374 } else if ( ch >= '0' && ch <= '9' ) {
2375 if ( scale < 100000 ) {
2376 scale *= 10;
2377 num = (num*10) + (ch-'0');
2378 }
2379 } else {
2380 break;
2381 }
2382 cp++;
2383 }
2384 *cp_p = cp;
2385
2386 /* user says num divided by scale and we say internally that
2387 * is MAX_SCORE * num / scale.
2388 */
2389 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2390 }
2391
2392 static int diff_scoreopt_parse(const char *opt)
2393 {
2394 int opt1, opt2, cmd;
2395
2396 if (*opt++ != '-')
2397 return -1;
2398 cmd = *opt++;
2399 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2400 return -1; /* that is not a -M, -C nor -B option */
2401
2402 opt1 = parse_num(&opt);
2403 if (cmd != 'B')
2404 opt2 = 0;
2405 else {
2406 if (*opt == 0)
2407 opt2 = 0;
2408 else if (*opt != '/')
2409 return -1; /* we expect -B80/99 or -B80 */
2410 else {
2411 opt++;
2412 opt2 = parse_num(&opt);
2413 }
2414 }
2415 if (*opt != 0)
2416 return -1;
2417 return opt1 | (opt2 << 16);
2418 }
2419
2420 struct diff_queue_struct diff_queued_diff;
2421
2422 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2423 {
2424 if (queue->alloc <= queue->nr) {
2425 queue->alloc = alloc_nr(queue->alloc);
2426 queue->queue = xrealloc(queue->queue,
2427 sizeof(dp) * queue->alloc);
2428 }
2429 queue->queue[queue->nr++] = dp;
2430 }
2431
2432 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2433 struct diff_filespec *one,
2434 struct diff_filespec *two)
2435 {
2436 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2437 dp->one = one;
2438 dp->two = two;
2439 if (queue)
2440 diff_q(queue, dp);
2441 return dp;
2442 }
2443
2444 void diff_free_filepair(struct diff_filepair *p)
2445 {
2446 free_filespec(p->one);
2447 free_filespec(p->two);
2448 free(p);
2449 }
2450
2451 /* This is different from find_unique_abbrev() in that
2452 * it stuffs the result with dots for alignment.
2453 */
2454 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2455 {
2456 int abblen;
2457 const char *abbrev;
2458 if (len == 40)
2459 return sha1_to_hex(sha1);
2460
2461 abbrev = find_unique_abbrev(sha1, len);
2462 if (!abbrev)
2463 return sha1_to_hex(sha1);
2464 abblen = strlen(abbrev);
2465 if (abblen < 37) {
2466 static char hex[41];
2467 if (len < abblen && abblen <= len + 2)
2468 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2469 else
2470 sprintf(hex, "%s...", abbrev);
2471 return hex;
2472 }
2473 return sha1_to_hex(sha1);
2474 }
2475
2476 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2477 {
2478 int line_termination = opt->line_termination;
2479 int inter_name_termination = line_termination ? '\t' : '\0';
2480
2481 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2482 printf(":%06o %06o %s ", p->one->mode, p->two->mode,
2483 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2484 printf("%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2485 }
2486 if (p->score) {
2487 printf("%c%03d%c", p->status, similarity_index(p),
2488 inter_name_termination);
2489 } else {
2490 printf("%c%c", p->status, inter_name_termination);
2491 }
2492
2493 if (p->status == DIFF_STATUS_COPIED || p->status == DIFF_STATUS_RENAMED) {
2494 write_name_quoted(p->one->path, stdout, inter_name_termination);
2495 write_name_quoted(p->two->path, stdout, line_termination);
2496 } else {
2497 const char *path = p->one->mode ? p->one->path : p->two->path;
2498 write_name_quoted(path, stdout, line_termination);
2499 }
2500 }
2501
2502 int diff_unmodified_pair(struct diff_filepair *p)
2503 {
2504 /* This function is written stricter than necessary to support
2505 * the currently implemented transformers, but the idea is to
2506 * let transformers to produce diff_filepairs any way they want,
2507 * and filter and clean them up here before producing the output.
2508 */
2509 struct diff_filespec *one = p->one, *two = p->two;
2510
2511 if (DIFF_PAIR_UNMERGED(p))
2512 return 0; /* unmerged is interesting */
2513
2514 /* deletion, addition, mode or type change
2515 * and rename are all interesting.
2516 */
2517 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2518 DIFF_PAIR_MODE_CHANGED(p) ||
2519 strcmp(one->path, two->path))
2520 return 0;
2521
2522 /* both are valid and point at the same path. that is, we are
2523 * dealing with a change.
2524 */
2525 if (one->sha1_valid && two->sha1_valid &&
2526 !hashcmp(one->sha1, two->sha1))
2527 return 1; /* no change */
2528 if (!one->sha1_valid && !two->sha1_valid)
2529 return 1; /* both look at the same file on the filesystem. */
2530 return 0;
2531 }
2532
2533 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2534 {
2535 if (diff_unmodified_pair(p))
2536 return;
2537
2538 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2539 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2540 return; /* no tree diffs in patch format */
2541
2542 run_diff(p, o);
2543 }
2544
2545 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2546 struct diffstat_t *diffstat)
2547 {
2548 if (diff_unmodified_pair(p))
2549 return;
2550
2551 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2552 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2553 return; /* no tree diffs in patch format */
2554
2555 run_diffstat(p, o, diffstat);
2556 }
2557
2558 static void diff_flush_checkdiff(struct diff_filepair *p,
2559 struct diff_options *o)
2560 {
2561 if (diff_unmodified_pair(p))
2562 return;
2563
2564 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2565 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2566 return; /* no tree diffs in patch format */
2567
2568 run_checkdiff(p, o);
2569 }
2570
2571 int diff_queue_is_empty(void)
2572 {
2573 struct diff_queue_struct *q = &diff_queued_diff;
2574 int i;
2575 for (i = 0; i < q->nr; i++)
2576 if (!diff_unmodified_pair(q->queue[i]))
2577 return 0;
2578 return 1;
2579 }
2580
2581 #if DIFF_DEBUG
2582 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2583 {
2584 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2585 x, one ? one : "",
2586 s->path,
2587 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2588 s->mode,
2589 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2590 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2591 x, one ? one : "",
2592 s->size, s->xfrm_flags);
2593 }
2594
2595 void diff_debug_filepair(const struct diff_filepair *p, int i)
2596 {
2597 diff_debug_filespec(p->one, i, "one");
2598 diff_debug_filespec(p->two, i, "two");
2599 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2600 p->score, p->status ? p->status : '?',
2601 p->one->rename_used, p->broken_pair);
2602 }
2603
2604 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2605 {
2606 int i;
2607 if (msg)
2608 fprintf(stderr, "%s\n", msg);
2609 fprintf(stderr, "q->nr = %d\n", q->nr);
2610 for (i = 0; i < q->nr; i++) {
2611 struct diff_filepair *p = q->queue[i];
2612 diff_debug_filepair(p, i);
2613 }
2614 }
2615 #endif
2616
2617 static void diff_resolve_rename_copy(void)
2618 {
2619 int i;
2620 struct diff_filepair *p;
2621 struct diff_queue_struct *q = &diff_queued_diff;
2622
2623 diff_debug_queue("resolve-rename-copy", q);
2624
2625 for (i = 0; i < q->nr; i++) {
2626 p = q->queue[i];
2627 p->status = 0; /* undecided */
2628 if (DIFF_PAIR_UNMERGED(p))
2629 p->status = DIFF_STATUS_UNMERGED;
2630 else if (!DIFF_FILE_VALID(p->one))
2631 p->status = DIFF_STATUS_ADDED;
2632 else if (!DIFF_FILE_VALID(p->two))
2633 p->status = DIFF_STATUS_DELETED;
2634 else if (DIFF_PAIR_TYPE_CHANGED(p))
2635 p->status = DIFF_STATUS_TYPE_CHANGED;
2636
2637 /* from this point on, we are dealing with a pair
2638 * whose both sides are valid and of the same type, i.e.
2639 * either in-place edit or rename/copy edit.
2640 */
2641 else if (DIFF_PAIR_RENAME(p)) {
2642 /*
2643 * A rename might have re-connected a broken
2644 * pair up, causing the pathnames to be the
2645 * same again. If so, that's not a rename at
2646 * all, just a modification..
2647 *
2648 * Otherwise, see if this source was used for
2649 * multiple renames, in which case we decrement
2650 * the count, and call it a copy.
2651 */
2652 if (!strcmp(p->one->path, p->two->path))
2653 p->status = DIFF_STATUS_MODIFIED;
2654 else if (--p->one->rename_used > 0)
2655 p->status = DIFF_STATUS_COPIED;
2656 else
2657 p->status = DIFF_STATUS_RENAMED;
2658 }
2659 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2660 p->one->mode != p->two->mode ||
2661 is_null_sha1(p->one->sha1))
2662 p->status = DIFF_STATUS_MODIFIED;
2663 else {
2664 /* This is a "no-change" entry and should not
2665 * happen anymore, but prepare for broken callers.
2666 */
2667 error("feeding unmodified %s to diffcore",
2668 p->one->path);
2669 p->status = DIFF_STATUS_UNKNOWN;
2670 }
2671 }
2672 diff_debug_queue("resolve-rename-copy done", q);
2673 }
2674
2675 static int check_pair_status(struct diff_filepair *p)
2676 {
2677 switch (p->status) {
2678 case DIFF_STATUS_UNKNOWN:
2679 return 0;
2680 case 0:
2681 die("internal error in diff-resolve-rename-copy");
2682 default:
2683 return 1;
2684 }
2685 }
2686
2687 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2688 {
2689 int fmt = opt->output_format;
2690
2691 if (fmt & DIFF_FORMAT_CHECKDIFF)
2692 diff_flush_checkdiff(p, opt);
2693 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2694 diff_flush_raw(p, opt);
2695 else if (fmt & DIFF_FORMAT_NAME)
2696 write_name_quoted(p->two->path, stdout, opt->line_termination);
2697 }
2698
2699 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2700 {
2701 if (fs->mode)
2702 printf(" %s mode %06o ", newdelete, fs->mode);
2703 else
2704 printf(" %s ", newdelete);
2705 write_name_quoted(fs->path, stdout, '\n');
2706 }
2707
2708
2709 static void show_mode_change(struct diff_filepair *p, int show_name)
2710 {
2711 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2712 printf(" mode change %06o => %06o%c", p->one->mode, p->two->mode,
2713 show_name ? ' ' : '\n');
2714 if (show_name) {
2715 write_name_quoted(p->two->path, stdout, '\n');
2716 }
2717 }
2718 }
2719
2720 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2721 {
2722 char *names = pprint_rename(p->one->path, p->two->path);
2723
2724 printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2725 free(names);
2726 show_mode_change(p, 0);
2727 }
2728
2729 static void diff_summary(struct diff_filepair *p)
2730 {
2731 switch(p->status) {
2732 case DIFF_STATUS_DELETED:
2733 show_file_mode_name("delete", p->one);
2734 break;
2735 case DIFF_STATUS_ADDED:
2736 show_file_mode_name("create", p->two);
2737 break;
2738 case DIFF_STATUS_COPIED:
2739 show_rename_copy("copy", p);
2740 break;
2741 case DIFF_STATUS_RENAMED:
2742 show_rename_copy("rename", p);
2743 break;
2744 default:
2745 if (p->score) {
2746 fputs(" rewrite ", stdout);
2747 write_name_quoted(p->two->path, stdout, ' ');
2748 printf("(%d%%)\n", similarity_index(p));
2749 }
2750 show_mode_change(p, !p->score);
2751 break;
2752 }
2753 }
2754
2755 struct patch_id_t {
2756 struct xdiff_emit_state xm;
2757 SHA_CTX *ctx;
2758 int patchlen;
2759 };
2760
2761 static int remove_space(char *line, int len)
2762 {
2763 int i;
2764 char *dst = line;
2765 unsigned char c;
2766
2767 for (i = 0; i < len; i++)
2768 if (!isspace((c = line[i])))
2769 *dst++ = c;
2770
2771 return dst - line;
2772 }
2773
2774 static void patch_id_consume(void *priv, char *line, unsigned long len)
2775 {
2776 struct patch_id_t *data = priv;
2777 int new_len;
2778
2779 /* Ignore line numbers when computing the SHA1 of the patch */
2780 if (!prefixcmp(line, "@@ -"))
2781 return;
2782
2783 new_len = remove_space(line, len);
2784
2785 SHA1_Update(data->ctx, line, new_len);
2786 data->patchlen += new_len;
2787 }
2788
2789 /* returns 0 upon success, and writes result into sha1 */
2790 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2791 {
2792 struct diff_queue_struct *q = &diff_queued_diff;
2793 int i;
2794 SHA_CTX ctx;
2795 struct patch_id_t data;
2796 char buffer[PATH_MAX * 4 + 20];
2797
2798 SHA1_Init(&ctx);
2799 memset(&data, 0, sizeof(struct patch_id_t));
2800 data.ctx = &ctx;
2801 data.xm.consume = patch_id_consume;
2802
2803 for (i = 0; i < q->nr; i++) {
2804 xpparam_t xpp;
2805 xdemitconf_t xecfg;
2806 xdemitcb_t ecb;
2807 mmfile_t mf1, mf2;
2808 struct diff_filepair *p = q->queue[i];
2809 int len1, len2;
2810
2811 memset(&xecfg, 0, sizeof(xecfg));
2812 if (p->status == 0)
2813 return error("internal diff status error");
2814 if (p->status == DIFF_STATUS_UNKNOWN)
2815 continue;
2816 if (diff_unmodified_pair(p))
2817 continue;
2818 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2819 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2820 continue;
2821 if (DIFF_PAIR_UNMERGED(p))
2822 continue;
2823
2824 diff_fill_sha1_info(p->one);
2825 diff_fill_sha1_info(p->two);
2826 if (fill_mmfile(&mf1, p->one) < 0 ||
2827 fill_mmfile(&mf2, p->two) < 0)
2828 return error("unable to read files to diff");
2829
2830 len1 = remove_space(p->one->path, strlen(p->one->path));
2831 len2 = remove_space(p->two->path, strlen(p->two->path));
2832 if (p->one->mode == 0)
2833 len1 = snprintf(buffer, sizeof(buffer),
2834 "diff--gita/%.*sb/%.*s"
2835 "newfilemode%06o"
2836 "---/dev/null"
2837 "+++b/%.*s",
2838 len1, p->one->path,
2839 len2, p->two->path,
2840 p->two->mode,
2841 len2, p->two->path);
2842 else if (p->two->mode == 0)
2843 len1 = snprintf(buffer, sizeof(buffer),
2844 "diff--gita/%.*sb/%.*s"
2845 "deletedfilemode%06o"
2846 "---a/%.*s"
2847 "+++/dev/null",
2848 len1, p->one->path,
2849 len2, p->two->path,
2850 p->one->mode,
2851 len1, p->one->path);
2852 else
2853 len1 = snprintf(buffer, sizeof(buffer),
2854 "diff--gita/%.*sb/%.*s"
2855 "---a/%.*s"
2856 "+++b/%.*s",
2857 len1, p->one->path,
2858 len2, p->two->path,
2859 len1, p->one->path,
2860 len2, p->two->path);
2861 SHA1_Update(&ctx, buffer, len1);
2862
2863 xpp.flags = XDF_NEED_MINIMAL;
2864 xecfg.ctxlen = 3;
2865 xecfg.flags = XDL_EMIT_FUNCNAMES;
2866 ecb.outf = xdiff_outf;
2867 ecb.priv = &data;
2868 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2869 }
2870
2871 SHA1_Final(sha1, &ctx);
2872 return 0;
2873 }
2874
2875 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2876 {
2877 struct diff_queue_struct *q = &diff_queued_diff;
2878 int i;
2879 int result = diff_get_patch_id(options, sha1);
2880
2881 for (i = 0; i < q->nr; i++)
2882 diff_free_filepair(q->queue[i]);
2883
2884 free(q->queue);
2885 q->queue = NULL;
2886 q->nr = q->alloc = 0;
2887
2888 return result;
2889 }
2890
2891 static int is_summary_empty(const struct diff_queue_struct *q)
2892 {
2893 int i;
2894
2895 for (i = 0; i < q->nr; i++) {
2896 const struct diff_filepair *p = q->queue[i];
2897
2898 switch (p->status) {
2899 case DIFF_STATUS_DELETED:
2900 case DIFF_STATUS_ADDED:
2901 case DIFF_STATUS_COPIED:
2902 case DIFF_STATUS_RENAMED:
2903 return 0;
2904 default:
2905 if (p->score)
2906 return 0;
2907 if (p->one->mode && p->two->mode &&
2908 p->one->mode != p->two->mode)
2909 return 0;
2910 break;
2911 }
2912 }
2913 return 1;
2914 }
2915
2916 void diff_flush(struct diff_options *options)
2917 {
2918 struct diff_queue_struct *q = &diff_queued_diff;
2919 int i, output_format = options->output_format;
2920 int separator = 0;
2921
2922 /*
2923 * Order: raw, stat, summary, patch
2924 * or: name/name-status/checkdiff (other bits clear)
2925 */
2926 if (!q->nr)
2927 goto free_queue;
2928
2929 if (output_format & (DIFF_FORMAT_RAW |
2930 DIFF_FORMAT_NAME |
2931 DIFF_FORMAT_NAME_STATUS |
2932 DIFF_FORMAT_CHECKDIFF)) {
2933 for (i = 0; i < q->nr; i++) {
2934 struct diff_filepair *p = q->queue[i];
2935 if (check_pair_status(p))
2936 flush_one_pair(p, options);
2937 }
2938 separator++;
2939 }
2940
2941 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2942 struct diffstat_t diffstat;
2943
2944 memset(&diffstat, 0, sizeof(struct diffstat_t));
2945 diffstat.xm.consume = diffstat_consume;
2946 for (i = 0; i < q->nr; i++) {
2947 struct diff_filepair *p = q->queue[i];
2948 if (check_pair_status(p))
2949 diff_flush_stat(p, options, &diffstat);
2950 }
2951 if (output_format & DIFF_FORMAT_NUMSTAT)
2952 show_numstat(&diffstat, options);
2953 if (output_format & DIFF_FORMAT_DIFFSTAT)
2954 show_stats(&diffstat, options);
2955 if (output_format & DIFF_FORMAT_SHORTSTAT)
2956 show_shortstats(&diffstat);
2957 free_diffstat_info(&diffstat);
2958 separator++;
2959 }
2960
2961 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2962 for (i = 0; i < q->nr; i++)
2963 diff_summary(q->queue[i]);
2964 separator++;
2965 }
2966
2967 if (output_format & DIFF_FORMAT_PATCH) {
2968 if (separator) {
2969 if (options->stat_sep) {
2970 /* attach patch instead of inline */
2971 fputs(options->stat_sep, stdout);
2972 } else {
2973 putchar(options->line_termination);
2974 }
2975 }
2976
2977 for (i = 0; i < q->nr; i++) {
2978 struct diff_filepair *p = q->queue[i];
2979 if (check_pair_status(p))
2980 diff_flush_patch(p, options);
2981 }
2982 }
2983
2984 if (output_format & DIFF_FORMAT_CALLBACK)
2985 options->format_callback(q, options, options->format_callback_data);
2986
2987 for (i = 0; i < q->nr; i++)
2988 diff_free_filepair(q->queue[i]);
2989 free_queue:
2990 free(q->queue);
2991 q->queue = NULL;
2992 q->nr = q->alloc = 0;
2993 }
2994
2995 static void diffcore_apply_filter(const char *filter)
2996 {
2997 int i;
2998 struct diff_queue_struct *q = &diff_queued_diff;
2999 struct diff_queue_struct outq;
3000 outq.queue = NULL;
3001 outq.nr = outq.alloc = 0;
3002
3003 if (!filter)
3004 return;
3005
3006 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3007 int found;
3008 for (i = found = 0; !found && i < q->nr; i++) {
3009 struct diff_filepair *p = q->queue[i];
3010 if (((p->status == DIFF_STATUS_MODIFIED) &&
3011 ((p->score &&
3012 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3013 (!p->score &&
3014 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3015 ((p->status != DIFF_STATUS_MODIFIED) &&
3016 strchr(filter, p->status)))
3017 found++;
3018 }
3019 if (found)
3020 return;
3021
3022 /* otherwise we will clear the whole queue
3023 * by copying the empty outq at the end of this
3024 * function, but first clear the current entries
3025 * in the queue.
3026 */
3027 for (i = 0; i < q->nr; i++)
3028 diff_free_filepair(q->queue[i]);
3029 }
3030 else {
3031 /* Only the matching ones */
3032 for (i = 0; i < q->nr; i++) {
3033 struct diff_filepair *p = q->queue[i];
3034
3035 if (((p->status == DIFF_STATUS_MODIFIED) &&
3036 ((p->score &&
3037 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3038 (!p->score &&
3039 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3040 ((p->status != DIFF_STATUS_MODIFIED) &&
3041 strchr(filter, p->status)))
3042 diff_q(&outq, p);
3043 else
3044 diff_free_filepair(p);
3045 }
3046 }
3047 free(q->queue);
3048 *q = outq;
3049 }
3050
3051 /* Check whether two filespecs with the same mode and size are identical */
3052 static int diff_filespec_is_identical(struct diff_filespec *one,
3053 struct diff_filespec *two)
3054 {
3055 if (S_ISGITLINK(one->mode)) {
3056 diff_fill_sha1_info(one);
3057 diff_fill_sha1_info(two);
3058 return !hashcmp(one->sha1, two->sha1);
3059 }
3060 if (diff_populate_filespec(one, 0))
3061 return 0;
3062 if (diff_populate_filespec(two, 0))
3063 return 0;
3064 return !memcmp(one->data, two->data, one->size);
3065 }
3066
3067 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3068 {
3069 int i;
3070 struct diff_queue_struct *q = &diff_queued_diff;
3071 struct diff_queue_struct outq;
3072 outq.queue = NULL;
3073 outq.nr = outq.alloc = 0;
3074
3075 for (i = 0; i < q->nr; i++) {
3076 struct diff_filepair *p = q->queue[i];
3077
3078 /*
3079 * 1. Entries that come from stat info dirtyness
3080 * always have both sides (iow, not create/delete),
3081 * one side of the object name is unknown, with
3082 * the same mode and size. Keep the ones that
3083 * do not match these criteria. They have real
3084 * differences.
3085 *
3086 * 2. At this point, the file is known to be modified,
3087 * with the same mode and size, and the object
3088 * name of one side is unknown. Need to inspect
3089 * the identical contents.
3090 */
3091 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3092 !DIFF_FILE_VALID(p->two) ||
3093 (p->one->sha1_valid && p->two->sha1_valid) ||
3094 (p->one->mode != p->two->mode) ||
3095 diff_populate_filespec(p->one, 1) ||
3096 diff_populate_filespec(p->two, 1) ||
3097 (p->one->size != p->two->size) ||
3098 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3099 diff_q(&outq, p);
3100 else {
3101 /*
3102 * The caller can subtract 1 from skip_stat_unmatch
3103 * to determine how many paths were dirty only
3104 * due to stat info mismatch.
3105 */
3106 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3107 diffopt->skip_stat_unmatch++;
3108 diff_free_filepair(p);
3109 }
3110 }
3111 free(q->queue);
3112 *q = outq;
3113 }
3114
3115 void diffcore_std(struct diff_options *options)
3116 {
3117 if (DIFF_OPT_TST(options, QUIET))
3118 return;
3119
3120 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3121 diffcore_skip_stat_unmatch(options);
3122 if (options->break_opt != -1)
3123 diffcore_break(options->break_opt);
3124 if (options->detect_rename)
3125 diffcore_rename(options);
3126 if (options->break_opt != -1)
3127 diffcore_merge_broken();
3128 if (options->pickaxe)
3129 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3130 if (options->orderfile)
3131 diffcore_order(options->orderfile);
3132 diff_resolve_rename_copy();
3133 diffcore_apply_filter(options->filter);
3134
3135 if (diff_queued_diff.nr)
3136 DIFF_OPT_SET(options, HAS_CHANGES);
3137 else
3138 DIFF_OPT_CLR(options, HAS_CHANGES);
3139 }
3140
3141 int diff_result_code(struct diff_options *opt, int status)
3142 {
3143 int result = 0;
3144 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3145 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3146 return status;
3147 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3148 DIFF_OPT_TST(opt, HAS_CHANGES))
3149 result |= 01;
3150 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3151 DIFF_OPT_TST(opt, CHECK_FAILED))
3152 result |= 02;
3153 return result;
3154 }
3155
3156 void diff_addremove(struct diff_options *options,
3157 int addremove, unsigned mode,
3158 const unsigned char *sha1,
3159 const char *base, const char *path)
3160 {
3161 char concatpath[PATH_MAX];
3162 struct diff_filespec *one, *two;
3163
3164 /* This may look odd, but it is a preparation for
3165 * feeding "there are unchanged files which should
3166 * not produce diffs, but when you are doing copy
3167 * detection you would need them, so here they are"
3168 * entries to the diff-core. They will be prefixed
3169 * with something like '=' or '*' (I haven't decided
3170 * which but should not make any difference).
3171 * Feeding the same new and old to diff_change()
3172 * also has the same effect.
3173 * Before the final output happens, they are pruned after
3174 * merged into rename/copy pairs as appropriate.
3175 */
3176 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3177 addremove = (addremove == '+' ? '-' :
3178 addremove == '-' ? '+' : addremove);
3179
3180 if (!path) path = "";
3181 sprintf(concatpath, "%s%s", base, path);
3182 one = alloc_filespec(concatpath);
3183 two = alloc_filespec(concatpath);
3184
3185 if (addremove != '+')
3186 fill_filespec(one, sha1, mode);
3187 if (addremove != '-')
3188 fill_filespec(two, sha1, mode);
3189
3190 diff_queue(&diff_queued_diff, one, two);
3191 DIFF_OPT_SET(options, HAS_CHANGES);
3192 }
3193
3194 void diff_change(struct diff_options *options,
3195 unsigned old_mode, unsigned new_mode,
3196 const unsigned char *old_sha1,
3197 const unsigned char *new_sha1,
3198 const char *base, const char *path)
3199 {
3200 char concatpath[PATH_MAX];
3201 struct diff_filespec *one, *two;
3202
3203 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3204 unsigned tmp;
3205 const unsigned char *tmp_c;
3206 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3207 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3208 }
3209 if (!path) path = "";
3210 sprintf(concatpath, "%s%s", base, path);
3211 one = alloc_filespec(concatpath);
3212 two = alloc_filespec(concatpath);
3213 fill_filespec(one, old_sha1, old_mode);
3214 fill_filespec(two, new_sha1, new_mode);
3215
3216 diff_queue(&diff_queued_diff, one, two);
3217 DIFF_OPT_SET(options, HAS_CHANGES);
3218 }
3219
3220 void diff_unmerge(struct diff_options *options,
3221 const char *path,
3222 unsigned mode, const unsigned char *sha1)
3223 {
3224 struct diff_filespec *one, *two;
3225 one = alloc_filespec(path);
3226 two = alloc_filespec(path);
3227 fill_filespec(one, sha1, mode);
3228 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3229 }
git with Jakub Narębski modifications