?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
diff: don't retrieve binary blobs for diffstat
[git.git] / diff.c
blob0d2ed00dad9879ef9209495f78930e0a066452fb
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
14 #include "userdiff.h"
15 #include "sigchain.h"
16 #include "submodule.h"
17
18 #ifdef NO_FAST_WORKING_DIRECTORY
19 #define FAST_WORKING_DIRECTORY 0
20 #else
21 #define FAST_WORKING_DIRECTORY 1
22 #endif
23
24 static int diff_detect_rename_default;
25 static int diff_rename_limit_default = 200;
26 static int diff_suppress_blank_empty;
27 int diff_use_color_default = -1;
28 static const char *diff_word_regex_cfg;
29 static const char *external_diff_cmd_cfg;
30 int diff_auto_refresh_index = 1;
31 static int diff_mnemonic_prefix;
32
33 static char diff_colors[][COLOR_MAXLEN] = {
34 GIT_COLOR_RESET,
35 GIT_COLOR_NORMAL, /* PLAIN */
36 GIT_COLOR_BOLD, /* METAINFO */
37 GIT_COLOR_CYAN, /* FRAGINFO */
38 GIT_COLOR_RED, /* OLD */
39 GIT_COLOR_GREEN, /* NEW */
40 GIT_COLOR_YELLOW, /* COMMIT */
41 GIT_COLOR_BG_RED, /* WHITESPACE */
42 GIT_COLOR_NORMAL, /* FUNCINFO */
43 };
44
45 static void diff_filespec_load_driver(struct diff_filespec *one);
46 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
47
48 static int parse_diff_color_slot(const char *var, int ofs)
49 {
50 if (!strcasecmp(var+ofs, "plain"))
51 return DIFF_PLAIN;
52 if (!strcasecmp(var+ofs, "meta"))
53 return DIFF_METAINFO;
54 if (!strcasecmp(var+ofs, "frag"))
55 return DIFF_FRAGINFO;
56 if (!strcasecmp(var+ofs, "old"))
57 return DIFF_FILE_OLD;
58 if (!strcasecmp(var+ofs, "new"))
59 return DIFF_FILE_NEW;
60 if (!strcasecmp(var+ofs, "commit"))
61 return DIFF_COMMIT;
62 if (!strcasecmp(var+ofs, "whitespace"))
63 return DIFF_WHITESPACE;
64 if (!strcasecmp(var+ofs, "func"))
65 return DIFF_FUNCINFO;
66 return -1;
67 }
68
69 static int git_config_rename(const char *var, const char *value)
70 {
71 if (!value)
72 return DIFF_DETECT_RENAME;
73 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
74 return DIFF_DETECT_COPY;
75 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
76 }
77
78 /*
79 * These are to give UI layer defaults.
80 * The core-level commands such as git-diff-files should
81 * never be affected by the setting of diff.renames
82 * the user happens to have in the configuration file.
83 */
84 int git_diff_ui_config(const char *var, const char *value, void *cb)
85 {
86 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
87 diff_use_color_default = git_config_colorbool(var, value, -1);
88 return 0;
89 }
90 if (!strcmp(var, "diff.renames")) {
91 diff_detect_rename_default = git_config_rename(var, value);
92 return 0;
93 }
94 if (!strcmp(var, "diff.autorefreshindex")) {
95 diff_auto_refresh_index = git_config_bool(var, value);
96 return 0;
97 }
98 if (!strcmp(var, "diff.mnemonicprefix")) {
99 diff_mnemonic_prefix = git_config_bool(var, value);
100 return 0;
101 }
102 if (!strcmp(var, "diff.external"))
103 return git_config_string(&external_diff_cmd_cfg, var, value);
104 if (!strcmp(var, "diff.wordregex"))
105 return git_config_string(&diff_word_regex_cfg, var, value);
106
107 return git_diff_basic_config(var, value, cb);
108 }
109
110 int git_diff_basic_config(const char *var, const char *value, void *cb)
111 {
112 if (!strcmp(var, "diff.renamelimit")) {
113 diff_rename_limit_default = git_config_int(var, value);
114 return 0;
115 }
116
117 switch (userdiff_config(var, value)) {
118 case 0: break;
119 case -1: return -1;
120 default: return 0;
121 }
122
123 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
124 int slot = parse_diff_color_slot(var, 11);
125 if (slot < 0)
126 return 0;
127 if (!value)
128 return config_error_nonbool(var);
129 color_parse(value, var, diff_colors[slot]);
130 return 0;
131 }
132
133 /* like GNU diff's --suppress-blank-empty option */
134 if (!strcmp(var, "diff.suppressblankempty") ||
135 /* for backwards compatibility */
136 !strcmp(var, "diff.suppress-blank-empty")) {
137 diff_suppress_blank_empty = git_config_bool(var, value);
138 return 0;
139 }
140
141 return git_color_default_config(var, value, cb);
142 }
143
144 static char *quote_two(const char *one, const char *two)
145 {
146 int need_one = quote_c_style(one, NULL, NULL, 1);
147 int need_two = quote_c_style(two, NULL, NULL, 1);
148 struct strbuf res = STRBUF_INIT;
149
150 if (need_one + need_two) {
151 strbuf_addch(&res, '"');
152 quote_c_style(one, &res, NULL, 1);
153 quote_c_style(two, &res, NULL, 1);
154 strbuf_addch(&res, '"');
155 } else {
156 strbuf_addstr(&res, one);
157 strbuf_addstr(&res, two);
158 }
159 return strbuf_detach(&res, NULL);
160 }
161
162 static const char *external_diff(void)
163 {
164 static const char *external_diff_cmd = NULL;
165 static int done_preparing = 0;
166
167 if (done_preparing)
168 return external_diff_cmd;
169 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
170 if (!external_diff_cmd)
171 external_diff_cmd = external_diff_cmd_cfg;
172 done_preparing = 1;
173 return external_diff_cmd;
174 }
175
176 static struct diff_tempfile {
177 const char *name; /* filename external diff should read from */
178 char hex[41];
179 char mode[10];
180 char tmp_path[PATH_MAX];
181 } diff_temp[2];
182
183 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
184
185 struct emit_callback {
186 int color_diff;
187 unsigned ws_rule;
188 int blank_at_eof_in_preimage;
189 int blank_at_eof_in_postimage;
190 int lno_in_preimage;
191 int lno_in_postimage;
192 sane_truncate_fn truncate;
193 const char **label_path;
194 struct diff_words_data *diff_words;
195 int *found_changesp;
196 FILE *file;
197 struct strbuf *header;
198 };
199
200 static int count_lines(const char *data, int size)
201 {
202 int count, ch, completely_empty = 1, nl_just_seen = 0;
203 count = 0;
204 while (0 < size--) {
205 ch = *data++;
206 if (ch == '\n') {
207 count++;
208 nl_just_seen = 1;
209 completely_empty = 0;
210 }
211 else {
212 nl_just_seen = 0;
213 completely_empty = 0;
214 }
215 }
216 if (completely_empty)
217 return 0;
218 if (!nl_just_seen)
219 count++; /* no trailing newline */
220 return count;
221 }
222
223 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
224 {
225 if (!DIFF_FILE_VALID(one)) {
226 mf->ptr = (char *)""; /* does not matter */
227 mf->size = 0;
228 return 0;
229 }
230 else if (diff_populate_filespec(one, 0))
231 return -1;
232
233 mf->ptr = one->data;
234 mf->size = one->size;
235 return 0;
236 }
237
238 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
239 static unsigned long diff_filespec_size(struct diff_filespec *one)
240 {
241 if (!DIFF_FILE_VALID(one))
242 return 0;
243 diff_populate_filespec(one, 1);
244 return one->size;
245 }
246
247 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
248 {
249 char *ptr = mf->ptr;
250 long size = mf->size;
251 int cnt = 0;
252
253 if (!size)
254 return cnt;
255 ptr += size - 1; /* pointing at the very end */
256 if (*ptr != '\n')
257 ; /* incomplete line */
258 else
259 ptr--; /* skip the last LF */
260 while (mf->ptr < ptr) {
261 char *prev_eol;
262 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
263 if (*prev_eol == '\n')
264 break;
265 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
266 break;
267 cnt++;
268 ptr = prev_eol - 1;
269 }
270 return cnt;
271 }
272
273 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
274 struct emit_callback *ecbdata)
275 {
276 int l1, l2, at;
277 unsigned ws_rule = ecbdata->ws_rule;
278 l1 = count_trailing_blank(mf1, ws_rule);
279 l2 = count_trailing_blank(mf2, ws_rule);
280 if (l2 <= l1) {
281 ecbdata->blank_at_eof_in_preimage = 0;
282 ecbdata->blank_at_eof_in_postimage = 0;
283 return;
284 }
285 at = count_lines(mf1->ptr, mf1->size);
286 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
287
288 at = count_lines(mf2->ptr, mf2->size);
289 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
290 }
291
292 static void emit_line_0(FILE *file, const char *set, const char *reset,
293 int first, const char *line, int len)
294 {
295 int has_trailing_newline, has_trailing_carriage_return;
296 int nofirst;
297
298 if (len == 0) {
299 has_trailing_newline = (first == '\n');
300 has_trailing_carriage_return = (!has_trailing_newline &&
301 (first == '\r'));
302 nofirst = has_trailing_newline || has_trailing_carriage_return;
303 } else {
304 has_trailing_newline = (len > 0 && line[len-1] == '\n');
305 if (has_trailing_newline)
306 len--;
307 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
308 if (has_trailing_carriage_return)
309 len--;
310 nofirst = 0;
311 }
312
313 if (len || !nofirst) {
314 fputs(set, file);
315 if (!nofirst)
316 fputc(first, file);
317 fwrite(line, len, 1, file);
318 fputs(reset, file);
319 }
320 if (has_trailing_carriage_return)
321 fputc('\r', file);
322 if (has_trailing_newline)
323 fputc('\n', file);
324 }
325
326 static void emit_line(FILE *file, const char *set, const char *reset,
327 const char *line, int len)
328 {
329 emit_line_0(file, set, reset, line[0], line+1, len-1);
330 }
331
332 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
333 {
334 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
335 ecbdata->blank_at_eof_in_preimage &&
336 ecbdata->blank_at_eof_in_postimage &&
337 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
338 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
339 return 0;
340 return ws_blank_line(line, len, ecbdata->ws_rule);
341 }
342
343 static void emit_add_line(const char *reset,
344 struct emit_callback *ecbdata,
345 const char *line, int len)
346 {
347 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
348 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
349
350 if (!*ws)
351 emit_line_0(ecbdata->file, set, reset, '+', line, len);
352 else if (new_blank_line_at_eof(ecbdata, line, len))
353 /* Blank line at EOF - paint '+' as well */
354 emit_line_0(ecbdata->file, ws, reset, '+', line, len);
355 else {
356 /* Emit just the prefix, then the rest. */
357 emit_line_0(ecbdata->file, set, reset, '+', "", 0);
358 ws_check_emit(line, len, ecbdata->ws_rule,
359 ecbdata->file, set, reset, ws);
360 }
361 }
362
363 static void emit_hunk_header(struct emit_callback *ecbdata,
364 const char *line, int len)
365 {
366 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
367 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
368 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
369 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
370 static const char atat[2] = { '@', '@' };
371 const char *cp, *ep;
372
373 /*
374 * As a hunk header must begin with "@@ -<old>, +<new> @@",
375 * it always is at least 10 bytes long.
376 */
377 if (len < 10 ||
378 memcmp(line, atat, 2) ||
379 !(ep = memmem(line + 2, len - 2, atat, 2))) {
380 emit_line(ecbdata->file, plain, reset, line, len);
381 return;
382 }
383 ep += 2; /* skip over @@ */
384
385 /* The hunk header in fraginfo color */
386 emit_line(ecbdata->file, frag, reset, line, ep - line);
387
388 /* blank before the func header */
389 for (cp = ep; ep - line < len; ep++)
390 if (*ep != ' ' && *ep != '\t')
391 break;
392 if (ep != cp)
393 emit_line(ecbdata->file, plain, reset, cp, ep - cp);
394
395 if (ep < line + len)
396 emit_line(ecbdata->file, func, reset, ep, line + len - ep);
397 }
398
399 static struct diff_tempfile *claim_diff_tempfile(void) {
400 int i;
401 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
402 if (!diff_temp[i].name)
403 return diff_temp + i;
404 die("BUG: diff is failing to clean up its tempfiles");
405 }
406
407 static int remove_tempfile_installed;
408
409 static void remove_tempfile(void)
410 {
411 int i;
412 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
413 if (diff_temp[i].name == diff_temp[i].tmp_path)
414 unlink_or_warn(diff_temp[i].name);
415 diff_temp[i].name = NULL;
416 }
417 }
418
419 static void remove_tempfile_on_signal(int signo)
420 {
421 remove_tempfile();
422 sigchain_pop(signo);
423 raise(signo);
424 }
425
426 static void print_line_count(FILE *file, int count)
427 {
428 switch (count) {
429 case 0:
430 fprintf(file, "0,0");
431 break;
432 case 1:
433 fprintf(file, "1");
434 break;
435 default:
436 fprintf(file, "1,%d", count);
437 break;
438 }
439 }
440
441 static void emit_rewrite_lines(struct emit_callback *ecb,
442 int prefix, const char *data, int size)
443 {
444 const char *endp = NULL;
445 static const char *nneof = " No newline at end of file\n";
446 const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
447 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
448
449 while (0 < size) {
450 int len;
451
452 endp = memchr(data, '\n', size);
453 len = endp ? (endp - data + 1) : size;
454 if (prefix != '+') {
455 ecb->lno_in_preimage++;
456 emit_line_0(ecb->file, old, reset, '-',
457 data, len);
458 } else {
459 ecb->lno_in_postimage++;
460 emit_add_line(reset, ecb, data, len);
461 }
462 size -= len;
463 data += len;
464 }
465 if (!endp) {
466 const char *plain = diff_get_color(ecb->color_diff,
467 DIFF_PLAIN);
468 emit_line_0(ecb->file, plain, reset, '\\',
469 nneof, strlen(nneof));
470 }
471 }
472
473 static void emit_rewrite_diff(const char *name_a,
474 const char *name_b,
475 struct diff_filespec *one,
476 struct diff_filespec *two,
477 const char *textconv_one,
478 const char *textconv_two,
479 struct diff_options *o)
480 {
481 int lc_a, lc_b;
482 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
483 const char *name_a_tab, *name_b_tab;
484 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
485 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
486 const char *reset = diff_get_color(color_diff, DIFF_RESET);
487 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
488 const char *a_prefix, *b_prefix;
489 const char *data_one, *data_two;
490 size_t size_one, size_two;
491 struct emit_callback ecbdata;
492
493 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
494 a_prefix = o->b_prefix;
495 b_prefix = o->a_prefix;
496 } else {
497 a_prefix = o->a_prefix;
498 b_prefix = o->b_prefix;
499 }
500
501 name_a += (*name_a == '/');
502 name_b += (*name_b == '/');
503 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
504 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
505
506 strbuf_reset(&a_name);
507 strbuf_reset(&b_name);
508 quote_two_c_style(&a_name, a_prefix, name_a, 0);
509 quote_two_c_style(&b_name, b_prefix, name_b, 0);
510
511 diff_populate_filespec(one, 0);
512 diff_populate_filespec(two, 0);
513 if (textconv_one) {
514 data_one = run_textconv(textconv_one, one, &size_one);
515 if (!data_one)
516 die("unable to read files to diff");
517 }
518 else {
519 data_one = one->data;
520 size_one = one->size;
521 }
522 if (textconv_two) {
523 data_two = run_textconv(textconv_two, two, &size_two);
524 if (!data_two)
525 die("unable to read files to diff");
526 }
527 else {
528 data_two = two->data;
529 size_two = two->size;
530 }
531
532 memset(&ecbdata, 0, sizeof(ecbdata));
533 ecbdata.color_diff = color_diff;
534 ecbdata.found_changesp = &o->found_changes;
535 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
536 ecbdata.file = o->file;
537 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
538 mmfile_t mf1, mf2;
539 mf1.ptr = (char *)data_one;
540 mf2.ptr = (char *)data_two;
541 mf1.size = size_one;
542 mf2.size = size_two;
543 check_blank_at_eof(&mf1, &mf2, &ecbdata);
544 }
545 ecbdata.lno_in_preimage = 1;
546 ecbdata.lno_in_postimage = 1;
547
548 lc_a = count_lines(data_one, size_one);
549 lc_b = count_lines(data_two, size_two);
550 fprintf(o->file,
551 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
552 metainfo, a_name.buf, name_a_tab, reset,
553 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
554 print_line_count(o->file, lc_a);
555 fprintf(o->file, " +");
556 print_line_count(o->file, lc_b);
557 fprintf(o->file, " @@%s\n", reset);
558 if (lc_a)
559 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
560 if (lc_b)
561 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
562 }
563
564 struct diff_words_buffer {
565 mmfile_t text;
566 long alloc;
567 struct diff_words_orig {
568 const char *begin, *end;
569 } *orig;
570 int orig_nr, orig_alloc;
571 };
572
573 static void diff_words_append(char *line, unsigned long len,
574 struct diff_words_buffer *buffer)
575 {
576 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
577 line++;
578 len--;
579 memcpy(buffer->text.ptr + buffer->text.size, line, len);
580 buffer->text.size += len;
581 buffer->text.ptr[buffer->text.size] = '\0';
582 }
583
584 struct diff_words_data {
585 struct diff_words_buffer minus, plus;
586 const char *current_plus;
587 FILE *file;
588 regex_t *word_regex;
589 };
590
591 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
592 {
593 struct diff_words_data *diff_words = priv;
594 int minus_first, minus_len, plus_first, plus_len;
595 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
596
597 if (line[0] != '@' || parse_hunk_header(line, len,
598 &minus_first, &minus_len, &plus_first, &plus_len))
599 return;
600
601 /* POSIX requires that first be decremented by one if len == 0... */
602 if (minus_len) {
603 minus_begin = diff_words->minus.orig[minus_first].begin;
604 minus_end =
605 diff_words->minus.orig[minus_first + minus_len - 1].end;
606 } else
607 minus_begin = minus_end =
608 diff_words->minus.orig[minus_first].end;
609
610 if (plus_len) {
611 plus_begin = diff_words->plus.orig[plus_first].begin;
612 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
613 } else
614 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
615
616 if (diff_words->current_plus != plus_begin)
617 fwrite(diff_words->current_plus,
618 plus_begin - diff_words->current_plus, 1,
619 diff_words->file);
620 if (minus_begin != minus_end)
621 color_fwrite_lines(diff_words->file,
622 diff_get_color(1, DIFF_FILE_OLD),
623 minus_end - minus_begin, minus_begin);
624 if (plus_begin != plus_end)
625 color_fwrite_lines(diff_words->file,
626 diff_get_color(1, DIFF_FILE_NEW),
627 plus_end - plus_begin, plus_begin);
628
629 diff_words->current_plus = plus_end;
630 }
631
632 /* This function starts looking at *begin, and returns 0 iff a word was found. */
633 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
634 int *begin, int *end)
635 {
636 if (word_regex && *begin < buffer->size) {
637 regmatch_t match[1];
638 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
639 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
640 '\n', match[0].rm_eo - match[0].rm_so);
641 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
642 *begin += match[0].rm_so;
643 return *begin >= *end;
644 }
645 return -1;
646 }
647
648 /* find the next word */
649 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
650 (*begin)++;
651 if (*begin >= buffer->size)
652 return -1;
653
654 /* find the end of the word */
655 *end = *begin + 1;
656 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
657 (*end)++;
658
659 return 0;
660 }
661
662 /*
663 * This function splits the words in buffer->text, stores the list with
664 * newline separator into out, and saves the offsets of the original words
665 * in buffer->orig.
666 */
667 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
668 regex_t *word_regex)
669 {
670 int i, j;
671 long alloc = 0;
672
673 out->size = 0;
674 out->ptr = NULL;
675
676 /* fake an empty "0th" word */
677 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
678 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
679 buffer->orig_nr = 1;
680
681 for (i = 0; i < buffer->text.size; i++) {
682 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
683 return;
684
685 /* store original boundaries */
686 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
687 buffer->orig_alloc);
688 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
689 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
690 buffer->orig_nr++;
691
692 /* store one word */
693 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
694 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
695 out->ptr[out->size + j - i] = '\n';
696 out->size += j - i + 1;
697
698 i = j - 1;
699 }
700 }
701
702 /* this executes the word diff on the accumulated buffers */
703 static void diff_words_show(struct diff_words_data *diff_words)
704 {
705 xpparam_t xpp;
706 xdemitconf_t xecfg;
707 xdemitcb_t ecb;
708 mmfile_t minus, plus;
709
710 /* special case: only removal */
711 if (!diff_words->plus.text.size) {
712 color_fwrite_lines(diff_words->file,
713 diff_get_color(1, DIFF_FILE_OLD),
714 diff_words->minus.text.size, diff_words->minus.text.ptr);
715 diff_words->minus.text.size = 0;
716 return;
717 }
718
719 diff_words->current_plus = diff_words->plus.text.ptr;
720
721 memset(&xpp, 0, sizeof(xpp));
722 memset(&xecfg, 0, sizeof(xecfg));
723 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
724 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
725 xpp.flags = XDF_NEED_MINIMAL;
726 /* as only the hunk header will be parsed, we need a 0-context */
727 xecfg.ctxlen = 0;
728 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
729 &xpp, &xecfg, &ecb);
730 free(minus.ptr);
731 free(plus.ptr);
732 if (diff_words->current_plus != diff_words->plus.text.ptr +
733 diff_words->plus.text.size)
734 fwrite(diff_words->current_plus,
735 diff_words->plus.text.ptr + diff_words->plus.text.size
736 - diff_words->current_plus, 1,
737 diff_words->file);
738 diff_words->minus.text.size = diff_words->plus.text.size = 0;
739 }
740
741 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
742 static void diff_words_flush(struct emit_callback *ecbdata)
743 {
744 if (ecbdata->diff_words->minus.text.size ||
745 ecbdata->diff_words->plus.text.size)
746 diff_words_show(ecbdata->diff_words);
747 }
748
749 static void free_diff_words_data(struct emit_callback *ecbdata)
750 {
751 if (ecbdata->diff_words) {
752 diff_words_flush(ecbdata);
753 free (ecbdata->diff_words->minus.text.ptr);
754 free (ecbdata->diff_words->minus.orig);
755 free (ecbdata->diff_words->plus.text.ptr);
756 free (ecbdata->diff_words->plus.orig);
757 free(ecbdata->diff_words->word_regex);
758 free(ecbdata->diff_words);
759 ecbdata->diff_words = NULL;
760 }
761 }
762
763 const char *diff_get_color(int diff_use_color, enum color_diff ix)
764 {
765 if (diff_use_color)
766 return diff_colors[ix];
767 return "";
768 }
769
770 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
771 {
772 const char *cp;
773 unsigned long allot;
774 size_t l = len;
775
776 if (ecb->truncate)
777 return ecb->truncate(line, len);
778 cp = line;
779 allot = l;
780 while (0 < l) {
781 (void) utf8_width(&cp, &l);
782 if (!cp)
783 break; /* truncated in the middle? */
784 }
785 return allot - l;
786 }
787
788 static void find_lno(const char *line, struct emit_callback *ecbdata)
789 {
790 const char *p;
791 ecbdata->lno_in_preimage = 0;
792 ecbdata->lno_in_postimage = 0;
793 p = strchr(line, '-');
794 if (!p)
795 return; /* cannot happen */
796 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
797 p = strchr(p, '+');
798 if (!p)
799 return; /* cannot happen */
800 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
801 }
802
803 static void fn_out_consume(void *priv, char *line, unsigned long len)
804 {
805 struct emit_callback *ecbdata = priv;
806 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
807 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
808 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
809
810 if (ecbdata->header) {
811 fprintf(ecbdata->file, "%s", ecbdata->header->buf);
812 strbuf_reset(ecbdata->header);
813 ecbdata->header = NULL;
814 }
815 *(ecbdata->found_changesp) = 1;
816
817 if (ecbdata->label_path[0]) {
818 const char *name_a_tab, *name_b_tab;
819
820 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
821 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
822
823 fprintf(ecbdata->file, "%s--- %s%s%s\n",
824 meta, ecbdata->label_path[0], reset, name_a_tab);
825 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
826 meta, ecbdata->label_path[1], reset, name_b_tab);
827 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
828 }
829
830 if (diff_suppress_blank_empty
831 && len == 2 && line[0] == ' ' && line[1] == '\n') {
832 line[0] = '\n';
833 len = 1;
834 }
835
836 if (line[0] == '@') {
837 if (ecbdata->diff_words)
838 diff_words_flush(ecbdata);
839 len = sane_truncate_line(ecbdata, line, len);
840 find_lno(line, ecbdata);
841 emit_hunk_header(ecbdata, line, len);
842 if (line[len-1] != '\n')
843 putc('\n', ecbdata->file);
844 return;
845 }
846
847 if (len < 1) {
848 emit_line(ecbdata->file, reset, reset, line, len);
849 return;
850 }
851
852 if (ecbdata->diff_words) {
853 if (line[0] == '-') {
854 diff_words_append(line, len,
855 &ecbdata->diff_words->minus);
856 return;
857 } else if (line[0] == '+') {
858 diff_words_append(line, len,
859 &ecbdata->diff_words->plus);
860 return;
861 }
862 diff_words_flush(ecbdata);
863 line++;
864 len--;
865 emit_line(ecbdata->file, plain, reset, line, len);
866 return;
867 }
868
869 if (line[0] != '+') {
870 const char *color =
871 diff_get_color(ecbdata->color_diff,
872 line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
873 ecbdata->lno_in_preimage++;
874 if (line[0] == ' ')
875 ecbdata->lno_in_postimage++;
876 emit_line(ecbdata->file, color, reset, line, len);
877 } else {
878 ecbdata->lno_in_postimage++;
879 emit_add_line(reset, ecbdata, line + 1, len - 1);
880 }
881 }
882
883 static char *pprint_rename(const char *a, const char *b)
884 {
885 const char *old = a;
886 const char *new = b;
887 struct strbuf name = STRBUF_INIT;
888 int pfx_length, sfx_length;
889 int len_a = strlen(a);
890 int len_b = strlen(b);
891 int a_midlen, b_midlen;
892 int qlen_a = quote_c_style(a, NULL, NULL, 0);
893 int qlen_b = quote_c_style(b, NULL, NULL, 0);
894
895 if (qlen_a || qlen_b) {
896 quote_c_style(a, &name, NULL, 0);
897 strbuf_addstr(&name, " => ");
898 quote_c_style(b, &name, NULL, 0);
899 return strbuf_detach(&name, NULL);
900 }
901
902 /* Find common prefix */
903 pfx_length = 0;
904 while (*old && *new && *old == *new) {
905 if (*old == '/')
906 pfx_length = old - a + 1;
907 old++;
908 new++;
909 }
910
911 /* Find common suffix */
912 old = a + len_a;
913 new = b + len_b;
914 sfx_length = 0;
915 while (a <= old && b <= new && *old == *new) {
916 if (*old == '/')
917 sfx_length = len_a - (old - a);
918 old--;
919 new--;
920 }
921
922 /*
923 * pfx{mid-a => mid-b}sfx
924 * {pfx-a => pfx-b}sfx
925 * pfx{sfx-a => sfx-b}
926 * name-a => name-b
927 */
928 a_midlen = len_a - pfx_length - sfx_length;
929 b_midlen = len_b - pfx_length - sfx_length;
930 if (a_midlen < 0)
931 a_midlen = 0;
932 if (b_midlen < 0)
933 b_midlen = 0;
934
935 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
936 if (pfx_length + sfx_length) {
937 strbuf_add(&name, a, pfx_length);
938 strbuf_addch(&name, '{');
939 }
940 strbuf_add(&name, a + pfx_length, a_midlen);
941 strbuf_addstr(&name, " => ");
942 strbuf_add(&name, b + pfx_length, b_midlen);
943 if (pfx_length + sfx_length) {
944 strbuf_addch(&name, '}');
945 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
946 }
947 return strbuf_detach(&name, NULL);
948 }
949
950 struct diffstat_t {
951 int nr;
952 int alloc;
953 struct diffstat_file {
954 char *from_name;
955 char *name;
956 char *print_name;
957 unsigned is_unmerged:1;
958 unsigned is_binary:1;
959 unsigned is_renamed:1;
960 unsigned int added, deleted;
961 } **files;
962 };
963
964 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
965 const char *name_a,
966 const char *name_b)
967 {
968 struct diffstat_file *x;
969 x = xcalloc(sizeof (*x), 1);
970 if (diffstat->nr == diffstat->alloc) {
971 diffstat->alloc = alloc_nr(diffstat->alloc);
972 diffstat->files = xrealloc(diffstat->files,
973 diffstat->alloc * sizeof(x));
974 }
975 diffstat->files[diffstat->nr++] = x;
976 if (name_b) {
977 x->from_name = xstrdup(name_a);
978 x->name = xstrdup(name_b);
979 x->is_renamed = 1;
980 }
981 else {
982 x->from_name = NULL;
983 x->name = xstrdup(name_a);
984 }
985 return x;
986 }
987
988 static void diffstat_consume(void *priv, char *line, unsigned long len)
989 {
990 struct diffstat_t *diffstat = priv;
991 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
992
993 if (line[0] == '+')
994 x->added++;
995 else if (line[0] == '-')
996 x->deleted++;
997 }
998
999 const char mime_boundary_leader[] = "------------";
1000
1001 static int scale_linear(int it, int width, int max_change)
1002 {
1003 /*
1004 * make sure that at least one '-' is printed if there were deletions,
1005 * and likewise for '+'.
1006 */
1007 if (max_change < 2)
1008 return it;
1009 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1010 }
1011
1012 static void show_name(FILE *file,
1013 const char *prefix, const char *name, int len)
1014 {
1015 fprintf(file, " %s%-*s |", prefix, len, name);
1016 }
1017
1018 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1019 {
1020 if (cnt <= 0)
1021 return;
1022 fprintf(file, "%s", set);
1023 while (cnt--)
1024 putc(ch, file);
1025 fprintf(file, "%s", reset);
1026 }
1027
1028 static void fill_print_name(struct diffstat_file *file)
1029 {
1030 char *pname;
1031
1032 if (file->print_name)
1033 return;
1034
1035 if (!file->is_renamed) {
1036 struct strbuf buf = STRBUF_INIT;
1037 if (quote_c_style(file->name, &buf, NULL, 0)) {
1038 pname = strbuf_detach(&buf, NULL);
1039 } else {
1040 pname = file->name;
1041 strbuf_release(&buf);
1042 }
1043 } else {
1044 pname = pprint_rename(file->from_name, file->name);
1045 }
1046 file->print_name = pname;
1047 }
1048
1049 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1050 {
1051 int i, len, add, del, adds = 0, dels = 0;
1052 int max_change = 0, max_len = 0;
1053 int total_files = data->nr;
1054 int width, name_width;
1055 const char *reset, *set, *add_c, *del_c;
1056
1057 if (data->nr == 0)
1058 return;
1059
1060 width = options->stat_width ? options->stat_width : 80;
1061 name_width = options->stat_name_width ? options->stat_name_width : 50;
1062
1063 /* Sanity: give at least 5 columns to the graph,
1064 * but leave at least 10 columns for the name.
1065 */
1066 if (width < 25)
1067 width = 25;
1068 if (name_width < 10)
1069 name_width = 10;
1070 else if (width < name_width + 15)
1071 name_width = width - 15;
1072
1073 /* Find the longest filename and max number of changes */
1074 reset = diff_get_color_opt(options, DIFF_RESET);
1075 set = diff_get_color_opt(options, DIFF_PLAIN);
1076 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1077 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1078
1079 for (i = 0; i < data->nr; i++) {
1080 struct diffstat_file *file = data->files[i];
1081 int change = file->added + file->deleted;
1082 fill_print_name(file);
1083 len = strlen(file->print_name);
1084 if (max_len < len)
1085 max_len = len;
1086
1087 if (file->is_binary || file->is_unmerged)
1088 continue;
1089 if (max_change < change)
1090 max_change = change;
1091 }
1092
1093 /* Compute the width of the graph part;
1094 * 10 is for one blank at the beginning of the line plus
1095 * " | count " between the name and the graph.
1096 *
1097 * From here on, name_width is the width of the name area,
1098 * and width is the width of the graph area.
1099 */
1100 name_width = (name_width < max_len) ? name_width : max_len;
1101 if (width < (name_width + 10) + max_change)
1102 width = width - (name_width + 10);
1103 else
1104 width = max_change;
1105
1106 for (i = 0; i < data->nr; i++) {
1107 const char *prefix = "";
1108 char *name = data->files[i]->print_name;
1109 int added = data->files[i]->added;
1110 int deleted = data->files[i]->deleted;
1111 int name_len;
1112
1113 /*
1114 * "scale" the filename
1115 */
1116 len = name_width;
1117 name_len = strlen(name);
1118 if (name_width < name_len) {
1119 char *slash;
1120 prefix = "...";
1121 len -= 3;
1122 name += name_len - len;
1123 slash = strchr(name, '/');
1124 if (slash)
1125 name = slash;
1126 }
1127
1128 if (data->files[i]->is_binary) {
1129 show_name(options->file, prefix, name, len);
1130 fprintf(options->file, " Bin ");
1131 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
1132 fprintf(options->file, " -> ");
1133 fprintf(options->file, "%s%d%s", add_c, added, reset);
1134 fprintf(options->file, " bytes");
1135 fprintf(options->file, "\n");
1136 continue;
1137 }
1138 else if (data->files[i]->is_unmerged) {
1139 show_name(options->file, prefix, name, len);
1140 fprintf(options->file, " Unmerged\n");
1141 continue;
1142 }
1143 else if (!data->files[i]->is_renamed &&
1144 (added + deleted == 0)) {
1145 total_files--;
1146 continue;
1147 }
1148
1149 /*
1150 * scale the add/delete
1151 */
1152 add = added;
1153 del = deleted;
1154 adds += add;
1155 dels += del;
1156
1157 if (width <= max_change) {
1158 add = scale_linear(add, width, max_change);
1159 del = scale_linear(del, width, max_change);
1160 }
1161 show_name(options->file, prefix, name, len);
1162 fprintf(options->file, "%5d%s", added + deleted,
1163 added + deleted ? " " : "");
1164 show_graph(options->file, '+', add, add_c, reset);
1165 show_graph(options->file, '-', del, del_c, reset);
1166 fprintf(options->file, "\n");
1167 }
1168 fprintf(options->file,
1169 " %d files changed, %d insertions(+), %d deletions(-)\n",
1170 total_files, adds, dels);
1171 }
1172
1173 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1174 {
1175 int i, adds = 0, dels = 0, total_files = data->nr;
1176
1177 if (data->nr == 0)
1178 return;
1179
1180 for (i = 0; i < data->nr; i++) {
1181 if (!data->files[i]->is_binary &&
1182 !data->files[i]->is_unmerged) {
1183 int added = data->files[i]->added;
1184 int deleted= data->files[i]->deleted;
1185 if (!data->files[i]->is_renamed &&
1186 (added + deleted == 0)) {
1187 total_files--;
1188 } else {
1189 adds += added;
1190 dels += deleted;
1191 }
1192 }
1193 }
1194 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1195 total_files, adds, dels);
1196 }
1197
1198 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1199 {
1200 int i;
1201
1202 if (data->nr == 0)
1203 return;
1204
1205 for (i = 0; i < data->nr; i++) {
1206 struct diffstat_file *file = data->files[i];
1207
1208 if (file->is_binary)
1209 fprintf(options->file, "-\t-\t");
1210 else
1211 fprintf(options->file,
1212 "%d\t%d\t", file->added, file->deleted);
1213 if (options->line_termination) {
1214 fill_print_name(file);
1215 if (!file->is_renamed)
1216 write_name_quoted(file->name, options->file,
1217 options->line_termination);
1218 else {
1219 fputs(file->print_name, options->file);
1220 putc(options->line_termination, options->file);
1221 }
1222 } else {
1223 if (file->is_renamed) {
1224 putc('\0', options->file);
1225 write_name_quoted(file->from_name, options->file, '\0');
1226 }
1227 write_name_quoted(file->name, options->file, '\0');
1228 }
1229 }
1230 }
1231
1232 struct dirstat_file {
1233 const char *name;
1234 unsigned long changed;
1235 };
1236
1237 struct dirstat_dir {
1238 struct dirstat_file *files;
1239 int alloc, nr, percent, cumulative;
1240 };
1241
1242 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1243 {
1244 unsigned long this_dir = 0;
1245 unsigned int sources = 0;
1246
1247 while (dir->nr) {
1248 struct dirstat_file *f = dir->files;
1249 int namelen = strlen(f->name);
1250 unsigned long this;
1251 char *slash;
1252
1253 if (namelen < baselen)
1254 break;
1255 if (memcmp(f->name, base, baselen))
1256 break;
1257 slash = strchr(f->name + baselen, '/');
1258 if (slash) {
1259 int newbaselen = slash + 1 - f->name;
1260 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1261 sources++;
1262 } else {
1263 this = f->changed;
1264 dir->files++;
1265 dir->nr--;
1266 sources += 2;
1267 }
1268 this_dir += this;
1269 }
1270
1271 /*
1272 * We don't report dirstat's for
1273 * - the top level
1274 * - or cases where everything came from a single directory
1275 * under this directory (sources == 1).
1276 */
1277 if (baselen && sources != 1) {
1278 int permille = this_dir * 1000 / changed;
1279 if (permille) {
1280 int percent = permille / 10;
1281 if (percent >= dir->percent) {
1282 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1283 if (!dir->cumulative)
1284 return 0;
1285 }
1286 }
1287 }
1288 return this_dir;
1289 }
1290
1291 static int dirstat_compare(const void *_a, const void *_b)
1292 {
1293 const struct dirstat_file *a = _a;
1294 const struct dirstat_file *b = _b;
1295 return strcmp(a->name, b->name);
1296 }
1297
1298 static void show_dirstat(struct diff_options *options)
1299 {
1300 int i;
1301 unsigned long changed;
1302 struct dirstat_dir dir;
1303 struct diff_queue_struct *q = &diff_queued_diff;
1304
1305 dir.files = NULL;
1306 dir.alloc = 0;
1307 dir.nr = 0;
1308 dir.percent = options->dirstat_percent;
1309 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1310
1311 changed = 0;
1312 for (i = 0; i < q->nr; i++) {
1313 struct diff_filepair *p = q->queue[i];
1314 const char *name;
1315 unsigned long copied, added, damage;
1316
1317 name = p->one->path ? p->one->path : p->two->path;
1318
1319 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1320 diff_populate_filespec(p->one, 0);
1321 diff_populate_filespec(p->two, 0);
1322 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1323 &copied, &added);
1324 diff_free_filespec_data(p->one);
1325 diff_free_filespec_data(p->two);
1326 } else if (DIFF_FILE_VALID(p->one)) {
1327 diff_populate_filespec(p->one, 1);
1328 copied = added = 0;
1329 diff_free_filespec_data(p->one);
1330 } else if (DIFF_FILE_VALID(p->two)) {
1331 diff_populate_filespec(p->two, 1);
1332 copied = 0;
1333 added = p->two->size;
1334 diff_free_filespec_data(p->two);
1335 } else
1336 continue;
1337
1338 /*
1339 * Original minus copied is the removed material,
1340 * added is the new material. They are both damages
1341 * made to the preimage. In --dirstat-by-file mode, count
1342 * damaged files, not damaged lines. This is done by
1343 * counting only a single damaged line per file.
1344 */
1345 damage = (p->one->size - copied) + added;
1346 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1347 damage = 1;
1348
1349 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1350 dir.files[dir.nr].name = name;
1351 dir.files[dir.nr].changed = damage;
1352 changed += damage;
1353 dir.nr++;
1354 }
1355
1356 /* This can happen even with many files, if everything was renames */
1357 if (!changed)
1358 return;
1359
1360 /* Show all directories with more than x% of the changes */
1361 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1362 gather_dirstat(options->file, &dir, changed, "", 0);
1363 }
1364
1365 static void free_diffstat_info(struct diffstat_t *diffstat)
1366 {
1367 int i;
1368 for (i = 0; i < diffstat->nr; i++) {
1369 struct diffstat_file *f = diffstat->files[i];
1370 if (f->name != f->print_name)
1371 free(f->print_name);
1372 free(f->name);
1373 free(f->from_name);
1374 free(f);
1375 }
1376 free(diffstat->files);
1377 }
1378
1379 struct checkdiff_t {
1380 const char *filename;
1381 int lineno;
1382 struct diff_options *o;
1383 unsigned ws_rule;
1384 unsigned status;
1385 };
1386
1387 static int is_conflict_marker(const char *line, unsigned long len)
1388 {
1389 char firstchar;
1390 int cnt;
1391
1392 if (len < 8)
1393 return 0;
1394 firstchar = line[0];
1395 switch (firstchar) {
1396 case '=': case '>': case '<':
1397 break;
1398 default:
1399 return 0;
1400 }
1401 for (cnt = 1; cnt < 7; cnt++)
1402 if (line[cnt] != firstchar)
1403 return 0;
1404 /* line[0] thru line[6] are same as firstchar */
1405 if (firstchar == '=') {
1406 /* divider between ours and theirs? */
1407 if (len != 8 || line[7] != '\n')
1408 return 0;
1409 } else if (len < 8 || !isspace(line[7])) {
1410 /* not divider before ours nor after theirs */
1411 return 0;
1412 }
1413 return 1;
1414 }
1415
1416 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1417 {
1418 struct checkdiff_t *data = priv;
1419 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1420 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1421 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1422 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1423 char *err;
1424
1425 if (line[0] == '+') {
1426 unsigned bad;
1427 data->lineno++;
1428 if (is_conflict_marker(line + 1, len - 1)) {
1429 data->status |= 1;
1430 fprintf(data->o->file,
1431 "%s:%d: leftover conflict marker\n",
1432 data->filename, data->lineno);
1433 }
1434 bad = ws_check(line + 1, len - 1, data->ws_rule);
1435 if (!bad)
1436 return;
1437 data->status |= bad;
1438 err = whitespace_error_string(bad);
1439 fprintf(data->o->file, "%s:%d: %s.\n",
1440 data->filename, data->lineno, err);
1441 free(err);
1442 emit_line(data->o->file, set, reset, line, 1);
1443 ws_check_emit(line + 1, len - 1, data->ws_rule,
1444 data->o->file, set, reset, ws);
1445 } else if (line[0] == ' ') {
1446 data->lineno++;
1447 } else if (line[0] == '@') {
1448 char *plus = strchr(line, '+');
1449 if (plus)
1450 data->lineno = strtol(plus, NULL, 10) - 1;
1451 else
1452 die("invalid diff");
1453 }
1454 }
1455
1456 static unsigned char *deflate_it(char *data,
1457 unsigned long size,
1458 unsigned long *result_size)
1459 {
1460 int bound;
1461 unsigned char *deflated;
1462 z_stream stream;
1463
1464 memset(&stream, 0, sizeof(stream));
1465 deflateInit(&stream, zlib_compression_level);
1466 bound = deflateBound(&stream, size);
1467 deflated = xmalloc(bound);
1468 stream.next_out = deflated;
1469 stream.avail_out = bound;
1470
1471 stream.next_in = (unsigned char *)data;
1472 stream.avail_in = size;
1473 while (deflate(&stream, Z_FINISH) == Z_OK)
1474 ; /* nothing */
1475 deflateEnd(&stream);
1476 *result_size = stream.total_out;
1477 return deflated;
1478 }
1479
1480 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1481 {
1482 void *cp;
1483 void *delta;
1484 void *deflated;
1485 void *data;
1486 unsigned long orig_size;
1487 unsigned long delta_size;
1488 unsigned long deflate_size;
1489 unsigned long data_size;
1490
1491 /* We could do deflated delta, or we could do just deflated two,
1492 * whichever is smaller.
1493 */
1494 delta = NULL;
1495 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1496 if (one->size && two->size) {
1497 delta = diff_delta(one->ptr, one->size,
1498 two->ptr, two->size,
1499 &delta_size, deflate_size);
1500 if (delta) {
1501 void *to_free = delta;
1502 orig_size = delta_size;
1503 delta = deflate_it(delta, delta_size, &delta_size);
1504 free(to_free);
1505 }
1506 }
1507
1508 if (delta && delta_size < deflate_size) {
1509 fprintf(file, "delta %lu\n", orig_size);
1510 free(deflated);
1511 data = delta;
1512 data_size = delta_size;
1513 }
1514 else {
1515 fprintf(file, "literal %lu\n", two->size);
1516 free(delta);
1517 data = deflated;
1518 data_size = deflate_size;
1519 }
1520
1521 /* emit data encoded in base85 */
1522 cp = data;
1523 while (data_size) {
1524 int bytes = (52 < data_size) ? 52 : data_size;
1525 char line[70];
1526 data_size -= bytes;
1527 if (bytes <= 26)
1528 line[0] = bytes + 'A' - 1;
1529 else
1530 line[0] = bytes - 26 + 'a' - 1;
1531 encode_85(line + 1, cp, bytes);
1532 cp = (char *) cp + bytes;
1533 fputs(line, file);
1534 fputc('\n', file);
1535 }
1536 fprintf(file, "\n");
1537 free(data);
1538 }
1539
1540 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1541 {
1542 fprintf(file, "GIT binary patch\n");
1543 emit_binary_diff_body(file, one, two);
1544 emit_binary_diff_body(file, two, one);
1545 }
1546
1547 static void diff_filespec_load_driver(struct diff_filespec *one)
1548 {
1549 if (!one->driver)
1550 one->driver = userdiff_find_by_path(one->path);
1551 if (!one->driver)
1552 one->driver = userdiff_find_by_name("default");
1553 }
1554
1555 int diff_filespec_is_binary(struct diff_filespec *one)
1556 {
1557 if (one->is_binary == -1) {
1558 diff_filespec_load_driver(one);
1559 if (one->driver->binary != -1)
1560 one->is_binary = one->driver->binary;
1561 else {
1562 if (!one->data && DIFF_FILE_VALID(one))
1563 diff_populate_filespec(one, 0);
1564 if (one->data)
1565 one->is_binary = buffer_is_binary(one->data,
1566 one->size);
1567 if (one->is_binary == -1)
1568 one->is_binary = 0;
1569 }
1570 }
1571 return one->is_binary;
1572 }
1573
1574 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1575 {
1576 diff_filespec_load_driver(one);
1577 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1578 }
1579
1580 static const char *userdiff_word_regex(struct diff_filespec *one)
1581 {
1582 diff_filespec_load_driver(one);
1583 return one->driver->word_regex;
1584 }
1585
1586 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1587 {
1588 if (!options->a_prefix)
1589 options->a_prefix = a;
1590 if (!options->b_prefix)
1591 options->b_prefix = b;
1592 }
1593
1594 static const char *get_textconv(struct diff_filespec *one)
1595 {
1596 if (!DIFF_FILE_VALID(one))
1597 return NULL;
1598 if (!S_ISREG(one->mode))
1599 return NULL;
1600 diff_filespec_load_driver(one);
1601 return one->driver->textconv;
1602 }
1603
1604 static void builtin_diff(const char *name_a,
1605 const char *name_b,
1606 struct diff_filespec *one,
1607 struct diff_filespec *two,
1608 const char *xfrm_msg,
1609 struct diff_options *o,
1610 int complete_rewrite)
1611 {
1612 mmfile_t mf1, mf2;
1613 const char *lbl[2];
1614 char *a_one, *b_two;
1615 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1616 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1617 const char *a_prefix, *b_prefix;
1618 const char *textconv_one = NULL, *textconv_two = NULL;
1619 struct strbuf header = STRBUF_INIT;
1620
1621 if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1622 (!one->mode || S_ISGITLINK(one->mode)) &&
1623 (!two->mode || S_ISGITLINK(two->mode))) {
1624 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1625 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1626 show_submodule_summary(o->file, one ? one->path : two->path,
1627 one->sha1, two->sha1, two->dirty_submodule,
1628 del, add, reset);
1629 return;
1630 }
1631
1632 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1633 textconv_one = get_textconv(one);
1634 textconv_two = get_textconv(two);
1635 }
1636
1637 diff_set_mnemonic_prefix(o, "a/", "b/");
1638 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1639 a_prefix = o->b_prefix;
1640 b_prefix = o->a_prefix;
1641 } else {
1642 a_prefix = o->a_prefix;
1643 b_prefix = o->b_prefix;
1644 }
1645
1646 /* Never use a non-valid filename anywhere if at all possible */
1647 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1648 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1649
1650 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1651 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1652 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1653 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1654 strbuf_addf(&header, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1655 if (lbl[0][0] == '/') {
1656 /* /dev/null */
1657 strbuf_addf(&header, "%snew file mode %06o%s\n", set, two->mode, reset);
1658 if (xfrm_msg && xfrm_msg[0])
1659 strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1660 }
1661 else if (lbl[1][0] == '/') {
1662 strbuf_addf(&header, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1663 if (xfrm_msg && xfrm_msg[0])
1664 strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1665 }
1666 else {
1667 if (one->mode != two->mode) {
1668 strbuf_addf(&header, "%sold mode %06o%s\n", set, one->mode, reset);
1669 strbuf_addf(&header, "%snew mode %06o%s\n", set, two->mode, reset);
1670 }
1671 if (xfrm_msg && xfrm_msg[0])
1672 strbuf_addf(&header, "%s%s%s\n", set, xfrm_msg, reset);
1673
1674 /*
1675 * we do not run diff between different kind
1676 * of objects.
1677 */
1678 if ((one->mode ^ two->mode) & S_IFMT)
1679 goto free_ab_and_return;
1680 if (complete_rewrite &&
1681 (textconv_one || !diff_filespec_is_binary(one)) &&
1682 (textconv_two || !diff_filespec_is_binary(two))) {
1683 fprintf(o->file, "%s", header.buf);
1684 strbuf_reset(&header);
1685 emit_rewrite_diff(name_a, name_b, one, two,
1686 textconv_one, textconv_two, o);
1687 o->found_changes = 1;
1688 goto free_ab_and_return;
1689 }
1690 }
1691
1692 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1693 die("unable to read files to diff");
1694
1695 if (!DIFF_OPT_TST(o, TEXT) &&
1696 ( (diff_filespec_is_binary(one) && !textconv_one) ||
1697 (diff_filespec_is_binary(two) && !textconv_two) )) {
1698 /* Quite common confusing case */
1699 if (mf1.size == mf2.size &&
1700 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1701 goto free_ab_and_return;
1702 fprintf(o->file, "%s", header.buf);
1703 strbuf_reset(&header);
1704 if (DIFF_OPT_TST(o, BINARY))
1705 emit_binary_diff(o->file, &mf1, &mf2);
1706 else
1707 fprintf(o->file, "Binary files %s and %s differ\n",
1708 lbl[0], lbl[1]);
1709 o->found_changes = 1;
1710 }
1711 else {
1712 /* Crazy xdl interfaces.. */
1713 const char *diffopts = getenv("GIT_DIFF_OPTS");
1714 xpparam_t xpp;
1715 xdemitconf_t xecfg;
1716 xdemitcb_t ecb;
1717 struct emit_callback ecbdata;
1718 const struct userdiff_funcname *pe;
1719
1720 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS)) {
1721 fprintf(o->file, "%s", header.buf);
1722 strbuf_reset(&header);
1723 }
1724
1725 if (textconv_one) {
1726 size_t size;
1727 mf1.ptr = run_textconv(textconv_one, one, &size);
1728 if (!mf1.ptr)
1729 die("unable to read files to diff");
1730 mf1.size = size;
1731 }
1732 if (textconv_two) {
1733 size_t size;
1734 mf2.ptr = run_textconv(textconv_two, two, &size);
1735 if (!mf2.ptr)
1736 die("unable to read files to diff");
1737 mf2.size = size;
1738 }
1739
1740 pe = diff_funcname_pattern(one);
1741 if (!pe)
1742 pe = diff_funcname_pattern(two);
1743
1744 memset(&xpp, 0, sizeof(xpp));
1745 memset(&xecfg, 0, sizeof(xecfg));
1746 memset(&ecbdata, 0, sizeof(ecbdata));
1747 ecbdata.label_path = lbl;
1748 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1749 ecbdata.found_changesp = &o->found_changes;
1750 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1751 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1752 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1753 ecbdata.file = o->file;
1754 ecbdata.header = header.len ? &header : NULL;
1755 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1756 xecfg.ctxlen = o->context;
1757 xecfg.interhunkctxlen = o->interhunkcontext;
1758 xecfg.flags = XDL_EMIT_FUNCNAMES;
1759 if (pe)
1760 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1761 if (!diffopts)
1762 ;
1763 else if (!prefixcmp(diffopts, "--unified="))
1764 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1765 else if (!prefixcmp(diffopts, "-u"))
1766 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1767 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1768 ecbdata.diff_words =
1769 xcalloc(1, sizeof(struct diff_words_data));
1770 ecbdata.diff_words->file = o->file;
1771 if (!o->word_regex)
1772 o->word_regex = userdiff_word_regex(one);
1773 if (!o->word_regex)
1774 o->word_regex = userdiff_word_regex(two);
1775 if (!o->word_regex)
1776 o->word_regex = diff_word_regex_cfg;
1777 if (o->word_regex) {
1778 ecbdata.diff_words->word_regex = (regex_t *)
1779 xmalloc(sizeof(regex_t));
1780 if (regcomp(ecbdata.diff_words->word_regex,
1781 o->word_regex,
1782 REG_EXTENDED | REG_NEWLINE))
1783 die ("Invalid regular expression: %s",
1784 o->word_regex);
1785 }
1786 }
1787 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1788 &xpp, &xecfg, &ecb);
1789 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1790 free_diff_words_data(&ecbdata);
1791 if (textconv_one)
1792 free(mf1.ptr);
1793 if (textconv_two)
1794 free(mf2.ptr);
1795 xdiff_clear_find_func(&xecfg);
1796 }
1797
1798 free_ab_and_return:
1799 strbuf_release(&header);
1800 diff_free_filespec_data(one);
1801 diff_free_filespec_data(two);
1802 free(a_one);
1803 free(b_two);
1804 return;
1805 }
1806
1807 static void builtin_diffstat(const char *name_a, const char *name_b,
1808 struct diff_filespec *one,
1809 struct diff_filespec *two,
1810 struct diffstat_t *diffstat,
1811 struct diff_options *o,
1812 int complete_rewrite)
1813 {
1814 mmfile_t mf1, mf2;
1815 struct diffstat_file *data;
1816
1817 data = diffstat_add(diffstat, name_a, name_b);
1818
1819 if (!one || !two) {
1820 data->is_unmerged = 1;
1821 return;
1822 }
1823
1824 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1825 data->is_binary = 1;
1826 data->added = diff_filespec_size(two);
1827 data->deleted = diff_filespec_size(one);
1828 }
1829
1830 else if (complete_rewrite) {
1831 diff_populate_filespec(one, 0);
1832 diff_populate_filespec(two, 0);
1833 data->deleted = count_lines(one->data, one->size);
1834 data->added = count_lines(two->data, two->size);
1835 }
1836
1837 else {
1838 /* Crazy xdl interfaces.. */
1839 xpparam_t xpp;
1840 xdemitconf_t xecfg;
1841 xdemitcb_t ecb;
1842
1843 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1844 die("unable to read files to diff");
1845
1846 memset(&xpp, 0, sizeof(xpp));
1847 memset(&xecfg, 0, sizeof(xecfg));
1848 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1849 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1850 &xpp, &xecfg, &ecb);
1851 }
1852
1853 diff_free_filespec_data(one);
1854 diff_free_filespec_data(two);
1855 }
1856
1857 static void builtin_checkdiff(const char *name_a, const char *name_b,
1858 const char *attr_path,
1859 struct diff_filespec *one,
1860 struct diff_filespec *two,
1861 struct diff_options *o)
1862 {
1863 mmfile_t mf1, mf2;
1864 struct checkdiff_t data;
1865
1866 if (!two)
1867 return;
1868
1869 memset(&data, 0, sizeof(data));
1870 data.filename = name_b ? name_b : name_a;
1871 data.lineno = 0;
1872 data.o = o;
1873 data.ws_rule = whitespace_rule(attr_path);
1874
1875 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1876 die("unable to read files to diff");
1877
1878 /*
1879 * All the other codepaths check both sides, but not checking
1880 * the "old" side here is deliberate. We are checking the newly
1881 * introduced changes, and as long as the "new" side is text, we
1882 * can and should check what it introduces.
1883 */
1884 if (diff_filespec_is_binary(two))
1885 goto free_and_return;
1886 else {
1887 /* Crazy xdl interfaces.. */
1888 xpparam_t xpp;
1889 xdemitconf_t xecfg;
1890 xdemitcb_t ecb;
1891
1892 memset(&xpp, 0, sizeof(xpp));
1893 memset(&xecfg, 0, sizeof(xecfg));
1894 xecfg.ctxlen = 1; /* at least one context line */
1895 xpp.flags = XDF_NEED_MINIMAL;
1896 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1897 &xpp, &xecfg, &ecb);
1898
1899 if (data.ws_rule & WS_BLANK_AT_EOF) {
1900 struct emit_callback ecbdata;
1901 int blank_at_eof;
1902
1903 ecbdata.ws_rule = data.ws_rule;
1904 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1905 blank_at_eof = ecbdata.blank_at_eof_in_preimage;
1906
1907 if (blank_at_eof) {
1908 static char *err;
1909 if (!err)
1910 err = whitespace_error_string(WS_BLANK_AT_EOF);
1911 fprintf(o->file, "%s:%d: %s.\n",
1912 data.filename, blank_at_eof, err);
1913 data.status = 1; /* report errors */
1914 }
1915 }
1916 }
1917 free_and_return:
1918 diff_free_filespec_data(one);
1919 diff_free_filespec_data(two);
1920 if (data.status)
1921 DIFF_OPT_SET(o, CHECK_FAILED);
1922 }
1923
1924 struct diff_filespec *alloc_filespec(const char *path)
1925 {
1926 int namelen = strlen(path);
1927 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1928
1929 memset(spec, 0, sizeof(*spec));
1930 spec->path = (char *)(spec + 1);
1931 memcpy(spec->path, path, namelen+1);
1932 spec->count = 1;
1933 spec->is_binary = -1;
1934 return spec;
1935 }
1936
1937 void free_filespec(struct diff_filespec *spec)
1938 {
1939 if (!--spec->count) {
1940 diff_free_filespec_data(spec);
1941 free(spec);
1942 }
1943 }
1944
1945 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1946 unsigned short mode)
1947 {
1948 if (mode) {
1949 spec->mode = canon_mode(mode);
1950 hashcpy(spec->sha1, sha1);
1951 spec->sha1_valid = !is_null_sha1(sha1);
1952 }
1953 }
1954
1955 /*
1956 * Given a name and sha1 pair, if the index tells us the file in
1957 * the work tree has that object contents, return true, so that
1958 * prepare_temp_file() does not have to inflate and extract.
1959 */
1960 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1961 {
1962 struct cache_entry *ce;
1963 struct stat st;
1964 int pos, len;
1965
1966 /*
1967 * We do not read the cache ourselves here, because the
1968 * benchmark with my previous version that always reads cache
1969 * shows that it makes things worse for diff-tree comparing
1970 * two linux-2.6 kernel trees in an already checked out work
1971 * tree. This is because most diff-tree comparisons deal with
1972 * only a small number of files, while reading the cache is
1973 * expensive for a large project, and its cost outweighs the
1974 * savings we get by not inflating the object to a temporary
1975 * file. Practically, this code only helps when we are used
1976 * by diff-cache --cached, which does read the cache before
1977 * calling us.
1978 */
1979 if (!active_cache)
1980 return 0;
1981
1982 /* We want to avoid the working directory if our caller
1983 * doesn't need the data in a normal file, this system
1984 * is rather slow with its stat/open/mmap/close syscalls,
1985 * and the object is contained in a pack file. The pack
1986 * is probably already open and will be faster to obtain
1987 * the data through than the working directory. Loose
1988 * objects however would tend to be slower as they need
1989 * to be individually opened and inflated.
1990 */
1991 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1992 return 0;
1993
1994 len = strlen(name);
1995 pos = cache_name_pos(name, len);
1996 if (pos < 0)
1997 return 0;
1998 ce = active_cache[pos];
1999
2000 /*
2001 * This is not the sha1 we are looking for, or
2002 * unreusable because it is not a regular file.
2003 */
2004 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2005 return 0;
2006
2007 /*
2008 * If ce is marked as "assume unchanged", there is no
2009 * guarantee that work tree matches what we are looking for.
2010 */
2011 if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2012 return 0;
2013
2014 /*
2015 * If ce matches the file in the work tree, we can reuse it.
2016 */
2017 if (ce_uptodate(ce) ||
2018 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2019 return 1;
2020
2021 return 0;
2022 }
2023
2024 static int populate_from_stdin(struct diff_filespec *s)
2025 {
2026 struct strbuf buf = STRBUF_INIT;
2027 size_t size = 0;
2028
2029 if (strbuf_read(&buf, 0, 0) < 0)
2030 return error("error while reading from stdin %s",
2031 strerror(errno));
2032
2033 s->should_munmap = 0;
2034 s->data = strbuf_detach(&buf, &size);
2035 s->size = size;
2036 s->should_free = 1;
2037 return 0;
2038 }
2039
2040 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2041 {
2042 int len;
2043 char *data = xmalloc(100), *dirty = "";
2044
2045 /* Are we looking at the work tree? */
2046 if (!s->sha1_valid && s->dirty_submodule)
2047 dirty = "-dirty";
2048
2049 len = snprintf(data, 100,
2050 "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2051 s->data = data;
2052 s->size = len;
2053 s->should_free = 1;
2054 if (size_only) {
2055 s->data = NULL;
2056 free(data);
2057 }
2058 return 0;
2059 }
2060
2061 /*
2062 * While doing rename detection and pickaxe operation, we may need to
2063 * grab the data for the blob (or file) for our own in-core comparison.
2064 * diff_filespec has data and size fields for this purpose.
2065 */
2066 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2067 {
2068 int err = 0;
2069 if (!DIFF_FILE_VALID(s))
2070 die("internal error: asking to populate invalid file.");
2071 if (S_ISDIR(s->mode))
2072 return -1;
2073
2074 if (s->data)
2075 return 0;
2076
2077 if (size_only && 0 < s->size)
2078 return 0;
2079
2080 if (S_ISGITLINK(s->mode))
2081 return diff_populate_gitlink(s, size_only);
2082
2083 if (!s->sha1_valid ||
2084 reuse_worktree_file(s->path, s->sha1, 0)) {
2085 struct strbuf buf = STRBUF_INIT;
2086 struct stat st;
2087 int fd;
2088
2089 if (!strcmp(s->path, "-"))
2090 return populate_from_stdin(s);
2091
2092 if (lstat(s->path, &st) < 0) {
2093 if (errno == ENOENT) {
2094 err_empty:
2095 err = -1;
2096 empty:
2097 s->data = (char *)"";
2098 s->size = 0;
2099 return err;
2100 }
2101 }
2102 s->size = xsize_t(st.st_size);
2103 if (!s->size)
2104 goto empty;
2105 if (S_ISLNK(st.st_mode)) {
2106 struct strbuf sb = STRBUF_INIT;
2107
2108 if (strbuf_readlink(&sb, s->path, s->size))
2109 goto err_empty;
2110 s->size = sb.len;
2111 s->data = strbuf_detach(&sb, NULL);
2112 s->should_free = 1;
2113 return 0;
2114 }
2115 if (size_only)
2116 return 0;
2117 fd = open(s->path, O_RDONLY);
2118 if (fd < 0)
2119 goto err_empty;
2120 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2121 close(fd);
2122 s->should_munmap = 1;
2123
2124 /*
2125 * Convert from working tree format to canonical git format
2126 */
2127 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2128 size_t size = 0;
2129 munmap(s->data, s->size);
2130 s->should_munmap = 0;
2131 s->data = strbuf_detach(&buf, &size);
2132 s->size = size;
2133 s->should_free = 1;
2134 }
2135 }
2136 else {
2137 enum object_type type;
2138 if (size_only)
2139 type = sha1_object_info(s->sha1, &s->size);
2140 else {
2141 s->data = read_sha1_file(s->sha1, &type, &s->size);
2142 s->should_free = 1;
2143 }
2144 }
2145 return 0;
2146 }
2147
2148 void diff_free_filespec_blob(struct diff_filespec *s)
2149 {
2150 if (s->should_free)
2151 free(s->data);
2152 else if (s->should_munmap)
2153 munmap(s->data, s->size);
2154
2155 if (s->should_free || s->should_munmap) {
2156 s->should_free = s->should_munmap = 0;
2157 s->data = NULL;
2158 }
2159 }
2160
2161 void diff_free_filespec_data(struct diff_filespec *s)
2162 {
2163 diff_free_filespec_blob(s);
2164 free(s->cnt_data);
2165 s->cnt_data = NULL;
2166 }
2167
2168 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2169 void *blob,
2170 unsigned long size,
2171 const unsigned char *sha1,
2172 int mode)
2173 {
2174 int fd;
2175 struct strbuf buf = STRBUF_INIT;
2176 struct strbuf template = STRBUF_INIT;
2177 char *path_dup = xstrdup(path);
2178 const char *base = basename(path_dup);
2179
2180 /* Generate "XXXXXX_basename.ext" */
2181 strbuf_addstr(&template, "XXXXXX_");
2182 strbuf_addstr(&template, base);
2183
2184 fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2185 strlen(base) + 1);
2186 if (fd < 0)
2187 die_errno("unable to create temp-file");
2188 if (convert_to_working_tree(path,
2189 (const char *)blob, (size_t)size, &buf)) {
2190 blob = buf.buf;
2191 size = buf.len;
2192 }
2193 if (write_in_full(fd, blob, size) != size)
2194 die_errno("unable to write temp-file");
2195 close(fd);
2196 temp->name = temp->tmp_path;
2197 strcpy(temp->hex, sha1_to_hex(sha1));
2198 temp->hex[40] = 0;
2199 sprintf(temp->mode, "%06o", mode);
2200 strbuf_release(&buf);
2201 strbuf_release(&template);
2202 free(path_dup);
2203 }
2204
2205 static struct diff_tempfile *prepare_temp_file(const char *name,
2206 struct diff_filespec *one)
2207 {
2208 struct diff_tempfile *temp = claim_diff_tempfile();
2209
2210 if (!DIFF_FILE_VALID(one)) {
2211 not_a_valid_file:
2212 /* A '-' entry produces this for file-2, and
2213 * a '+' entry produces this for file-1.
2214 */
2215 temp->name = "/dev/null";
2216 strcpy(temp->hex, ".");
2217 strcpy(temp->mode, ".");
2218 return temp;
2219 }
2220
2221 if (!remove_tempfile_installed) {
2222 atexit(remove_tempfile);
2223 sigchain_push_common(remove_tempfile_on_signal);
2224 remove_tempfile_installed = 1;
2225 }
2226
2227 if (!one->sha1_valid ||
2228 reuse_worktree_file(name, one->sha1, 1)) {
2229 struct stat st;
2230 if (lstat(name, &st) < 0) {
2231 if (errno == ENOENT)
2232 goto not_a_valid_file;
2233 die_errno("stat(%s)", name);
2234 }
2235 if (S_ISLNK(st.st_mode)) {
2236 struct strbuf sb = STRBUF_INIT;
2237 if (strbuf_readlink(&sb, name, st.st_size) < 0)
2238 die_errno("readlink(%s)", name);
2239 prep_temp_blob(name, temp, sb.buf, sb.len,
2240 (one->sha1_valid ?
2241 one->sha1 : null_sha1),
2242 (one->sha1_valid ?
2243 one->mode : S_IFLNK));
2244 strbuf_release(&sb);
2245 }
2246 else {
2247 /* we can borrow from the file in the work tree */
2248 temp->name = name;
2249 if (!one->sha1_valid)
2250 strcpy(temp->hex, sha1_to_hex(null_sha1));
2251 else
2252 strcpy(temp->hex, sha1_to_hex(one->sha1));
2253 /* Even though we may sometimes borrow the
2254 * contents from the work tree, we always want
2255 * one->mode. mode is trustworthy even when
2256 * !(one->sha1_valid), as long as
2257 * DIFF_FILE_VALID(one).
2258 */
2259 sprintf(temp->mode, "%06o", one->mode);
2260 }
2261 return temp;
2262 }
2263 else {
2264 if (diff_populate_filespec(one, 0))
2265 die("cannot read data blob for %s", one->path);
2266 prep_temp_blob(name, temp, one->data, one->size,
2267 one->sha1, one->mode);
2268 }
2269 return temp;
2270 }
2271
2272 /* An external diff command takes:
2273 *
2274 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2275 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2276 *
2277 */
2278 static void run_external_diff(const char *pgm,
2279 const char *name,
2280 const char *other,
2281 struct diff_filespec *one,
2282 struct diff_filespec *two,
2283 const char *xfrm_msg,
2284 int complete_rewrite)
2285 {
2286 const char *spawn_arg[10];
2287 int retval;
2288 const char **arg = &spawn_arg[0];
2289
2290 if (one && two) {
2291 struct diff_tempfile *temp_one, *temp_two;
2292 const char *othername = (other ? other : name);
2293 temp_one = prepare_temp_file(name, one);
2294 temp_two = prepare_temp_file(othername, two);
2295 *arg++ = pgm;
2296 *arg++ = name;
2297 *arg++ = temp_one->name;
2298 *arg++ = temp_one->hex;
2299 *arg++ = temp_one->mode;
2300 *arg++ = temp_two->name;
2301 *arg++ = temp_two->hex;
2302 *arg++ = temp_two->mode;
2303 if (other) {
2304 *arg++ = other;
2305 *arg++ = xfrm_msg;
2306 }
2307 } else {
2308 *arg++ = pgm;
2309 *arg++ = name;
2310 }
2311 *arg = NULL;
2312 fflush(NULL);
2313 retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2314 remove_tempfile();
2315 if (retval) {
2316 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2317 exit(1);
2318 }
2319 }
2320
2321 static int similarity_index(struct diff_filepair *p)
2322 {
2323 return p->score * 100 / MAX_SCORE;
2324 }
2325
2326 static void fill_metainfo(struct strbuf *msg,
2327 const char *name,
2328 const char *other,
2329 struct diff_filespec *one,
2330 struct diff_filespec *two,
2331 struct diff_options *o,
2332 struct diff_filepair *p)
2333 {
2334 strbuf_init(msg, PATH_MAX * 2 + 300);
2335 switch (p->status) {
2336 case DIFF_STATUS_COPIED:
2337 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2338 strbuf_addstr(msg, "\ncopy from ");
2339 quote_c_style(name, msg, NULL, 0);
2340 strbuf_addstr(msg, "\ncopy to ");
2341 quote_c_style(other, msg, NULL, 0);
2342 strbuf_addch(msg, '\n');
2343 break;
2344 case DIFF_STATUS_RENAMED:
2345 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2346 strbuf_addstr(msg, "\nrename from ");
2347 quote_c_style(name, msg, NULL, 0);
2348 strbuf_addstr(msg, "\nrename to ");
2349 quote_c_style(other, msg, NULL, 0);
2350 strbuf_addch(msg, '\n');
2351 break;
2352 case DIFF_STATUS_MODIFIED:
2353 if (p->score) {
2354 strbuf_addf(msg, "dissimilarity index %d%%\n",
2355 similarity_index(p));
2356 break;
2357 }
2358 /* fallthru */
2359 default:
2360 /* nothing */
2361 ;
2362 }
2363 if (one && two && hashcmp(one->sha1, two->sha1)) {
2364 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2365
2366 if (DIFF_OPT_TST(o, BINARY)) {
2367 mmfile_t mf;
2368 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2369 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2370 abbrev = 40;
2371 }
2372 strbuf_addf(msg, "index %.*s..%.*s",
2373 abbrev, sha1_to_hex(one->sha1),
2374 abbrev, sha1_to_hex(two->sha1));
2375 if (one->mode == two->mode)
2376 strbuf_addf(msg, " %06o", one->mode);
2377 strbuf_addch(msg, '\n');
2378 }
2379 if (msg->len)
2380 strbuf_setlen(msg, msg->len - 1);
2381 }
2382
2383 static void run_diff_cmd(const char *pgm,
2384 const char *name,
2385 const char *other,
2386 const char *attr_path,
2387 struct diff_filespec *one,
2388 struct diff_filespec *two,
2389 struct strbuf *msg,
2390 struct diff_options *o,
2391 struct diff_filepair *p)
2392 {
2393 const char *xfrm_msg = NULL;
2394 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2395
2396 if (msg) {
2397 fill_metainfo(msg, name, other, one, two, o, p);
2398 xfrm_msg = msg->len ? msg->buf : NULL;
2399 }
2400
2401 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2402 pgm = NULL;
2403 else {
2404 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2405 if (drv && drv->external)
2406 pgm = drv->external;
2407 }
2408
2409 if (pgm) {
2410 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2411 complete_rewrite);
2412 return;
2413 }
2414 if (one && two)
2415 builtin_diff(name, other ? other : name,
2416 one, two, xfrm_msg, o, complete_rewrite);
2417 else
2418 fprintf(o->file, "* Unmerged path %s\n", name);
2419 }
2420
2421 static void diff_fill_sha1_info(struct diff_filespec *one)
2422 {
2423 if (DIFF_FILE_VALID(one)) {
2424 if (!one->sha1_valid) {
2425 struct stat st;
2426 if (!strcmp(one->path, "-")) {
2427 hashcpy(one->sha1, null_sha1);
2428 return;
2429 }
2430 if (lstat(one->path, &st) < 0)
2431 die_errno("stat '%s'", one->path);
2432 if (index_path(one->sha1, one->path, &st, 0))
2433 die("cannot hash %s", one->path);
2434 }
2435 }
2436 else
2437 hashclr(one->sha1);
2438 }
2439
2440 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2441 {
2442 /* Strip the prefix but do not molest /dev/null and absolute paths */
2443 if (*namep && **namep != '/')
2444 *namep += prefix_length;
2445 if (*otherp && **otherp != '/')
2446 *otherp += prefix_length;
2447 }
2448
2449 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2450 {
2451 const char *pgm = external_diff();
2452 struct strbuf msg;
2453 struct diff_filespec *one = p->one;
2454 struct diff_filespec *two = p->two;
2455 const char *name;
2456 const char *other;
2457 const char *attr_path;
2458
2459 name = p->one->path;
2460 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2461 attr_path = name;
2462 if (o->prefix_length)
2463 strip_prefix(o->prefix_length, &name, &other);
2464
2465 if (DIFF_PAIR_UNMERGED(p)) {
2466 run_diff_cmd(pgm, name, NULL, attr_path,
2467 NULL, NULL, NULL, o, p);
2468 return;
2469 }
2470
2471 diff_fill_sha1_info(one);
2472 diff_fill_sha1_info(two);
2473
2474 if (!pgm &&
2475 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2476 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2477 /*
2478 * a filepair that changes between file and symlink
2479 * needs to be split into deletion and creation.
2480 */
2481 struct diff_filespec *null = alloc_filespec(two->path);
2482 run_diff_cmd(NULL, name, other, attr_path,
2483 one, null, &msg, o, p);
2484 free(null);
2485 strbuf_release(&msg);
2486
2487 null = alloc_filespec(one->path);
2488 run_diff_cmd(NULL, name, other, attr_path,
2489 null, two, &msg, o, p);
2490 free(null);
2491 }
2492 else
2493 run_diff_cmd(pgm, name, other, attr_path,
2494 one, two, &msg, o, p);
2495
2496 strbuf_release(&msg);
2497 }
2498
2499 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2500 struct diffstat_t *diffstat)
2501 {
2502 const char *name;
2503 const char *other;
2504 int complete_rewrite = 0;
2505
2506 if (DIFF_PAIR_UNMERGED(p)) {
2507 /* unmerged */
2508 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2509 return;
2510 }
2511
2512 name = p->one->path;
2513 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2514
2515 if (o->prefix_length)
2516 strip_prefix(o->prefix_length, &name, &other);
2517
2518 diff_fill_sha1_info(p->one);
2519 diff_fill_sha1_info(p->two);
2520
2521 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2522 complete_rewrite = 1;
2523 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2524 }
2525
2526 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2527 {
2528 const char *name;
2529 const char *other;
2530 const char *attr_path;
2531
2532 if (DIFF_PAIR_UNMERGED(p)) {
2533 /* unmerged */
2534 return;
2535 }
2536
2537 name = p->one->path;
2538 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2539 attr_path = other ? other : name;
2540
2541 if (o->prefix_length)
2542 strip_prefix(o->prefix_length, &name, &other);
2543
2544 diff_fill_sha1_info(p->one);
2545 diff_fill_sha1_info(p->two);
2546
2547 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2548 }
2549
2550 void diff_setup(struct diff_options *options)
2551 {
2552 memset(options, 0, sizeof(*options));
2553
2554 options->file = stdout;
2555
2556 options->line_termination = '\n';
2557 options->break_opt = -1;
2558 options->rename_limit = -1;
2559 options->dirstat_percent = 3;
2560 options->context = 3;
2561
2562 options->change = diff_change;
2563 options->add_remove = diff_addremove;
2564 if (diff_use_color_default > 0)
2565 DIFF_OPT_SET(options, COLOR_DIFF);
2566 options->detect_rename = diff_detect_rename_default;
2567
2568 if (!diff_mnemonic_prefix) {
2569 options->a_prefix = "a/";
2570 options->b_prefix = "b/";
2571 }
2572 }
2573
2574 int diff_setup_done(struct diff_options *options)
2575 {
2576 int count = 0;
2577
2578 if (options->output_format & DIFF_FORMAT_NAME)
2579 count++;
2580 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2581 count++;
2582 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2583 count++;
2584 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2585 count++;
2586 if (count > 1)
2587 die("--name-only, --name-status, --check and -s are mutually exclusive");
2588
2589 /*
2590 * Most of the time we can say "there are changes"
2591 * only by checking if there are changed paths, but
2592 * --ignore-whitespace* options force us to look
2593 * inside contents.
2594 */
2595
2596 if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2597 DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2598 DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2599 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2600 else
2601 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2602
2603 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2604 options->detect_rename = DIFF_DETECT_COPY;
2605
2606 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2607 options->prefix = NULL;
2608 if (options->prefix)
2609 options->prefix_length = strlen(options->prefix);
2610 else
2611 options->prefix_length = 0;
2612
2613 if (options->output_format & (DIFF_FORMAT_NAME |
2614 DIFF_FORMAT_NAME_STATUS |
2615 DIFF_FORMAT_CHECKDIFF |
2616 DIFF_FORMAT_NO_OUTPUT))
2617 options->output_format &= ~(DIFF_FORMAT_RAW |
2618 DIFF_FORMAT_NUMSTAT |
2619 DIFF_FORMAT_DIFFSTAT |
2620 DIFF_FORMAT_SHORTSTAT |
2621 DIFF_FORMAT_DIRSTAT |
2622 DIFF_FORMAT_SUMMARY |
2623 DIFF_FORMAT_PATCH);
2624
2625 /*
2626 * These cases always need recursive; we do not drop caller-supplied
2627 * recursive bits for other formats here.
2628 */
2629 if (options->output_format & (DIFF_FORMAT_PATCH |
2630 DIFF_FORMAT_NUMSTAT |
2631 DIFF_FORMAT_DIFFSTAT |
2632 DIFF_FORMAT_SHORTSTAT |
2633 DIFF_FORMAT_DIRSTAT |
2634 DIFF_FORMAT_SUMMARY |
2635 DIFF_FORMAT_CHECKDIFF))
2636 DIFF_OPT_SET(options, RECURSIVE);
2637 /*
2638 * Also pickaxe would not work very well if you do not say recursive
2639 */
2640 if (options->pickaxe)
2641 DIFF_OPT_SET(options, RECURSIVE);
2642
2643 if (options->detect_rename && options->rename_limit < 0)
2644 options->rename_limit = diff_rename_limit_default;
2645 if (options->setup & DIFF_SETUP_USE_CACHE) {
2646 if (!active_cache)
2647 /* read-cache does not die even when it fails
2648 * so it is safe for us to do this here. Also
2649 * it does not smudge active_cache or active_nr
2650 * when it fails, so we do not have to worry about
2651 * cleaning it up ourselves either.
2652 */
2653 read_cache();
2654 }
2655 if (options->abbrev <= 0 || 40 < options->abbrev)
2656 options->abbrev = 40; /* full */
2657
2658 /*
2659 * It does not make sense to show the first hit we happened
2660 * to have found. It does not make sense not to return with
2661 * exit code in such a case either.
2662 */
2663 if (DIFF_OPT_TST(options, QUICK)) {
2664 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2665 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2666 }
2667
2668 return 0;
2669 }
2670
2671 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2672 {
2673 char c, *eq;
2674 int len;
2675
2676 if (*arg != '-')
2677 return 0;
2678 c = *++arg;
2679 if (!c)
2680 return 0;
2681 if (c == arg_short) {
2682 c = *++arg;
2683 if (!c)
2684 return 1;
2685 if (val && isdigit(c)) {
2686 char *end;
2687 int n = strtoul(arg, &end, 10);
2688 if (*end)
2689 return 0;
2690 *val = n;
2691 return 1;
2692 }
2693 return 0;
2694 }
2695 if (c != '-')
2696 return 0;
2697 arg++;
2698 eq = strchr(arg, '=');
2699 if (eq)
2700 len = eq - arg;
2701 else
2702 len = strlen(arg);
2703 if (!len || strncmp(arg, arg_long, len))
2704 return 0;
2705 if (eq) {
2706 int n;
2707 char *end;
2708 if (!isdigit(*++eq))
2709 return 0;
2710 n = strtoul(eq, &end, 10);
2711 if (*end)
2712 return 0;
2713 *val = n;
2714 }
2715 return 1;
2716 }
2717
2718 static int diff_scoreopt_parse(const char *opt);
2719
2720 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2721 {
2722 const char *arg = av[0];
2723
2724 /* Output format options */
2725 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2726 options->output_format |= DIFF_FORMAT_PATCH;
2727 else if (opt_arg(arg, 'U', "unified", &options->context))
2728 options->output_format |= DIFF_FORMAT_PATCH;
2729 else if (!strcmp(arg, "--raw"))
2730 options->output_format |= DIFF_FORMAT_RAW;
2731 else if (!strcmp(arg, "--patch-with-raw"))
2732 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2733 else if (!strcmp(arg, "--numstat"))
2734 options->output_format |= DIFF_FORMAT_NUMSTAT;
2735 else if (!strcmp(arg, "--shortstat"))
2736 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2737 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2738 options->output_format |= DIFF_FORMAT_DIRSTAT;
2739 else if (!strcmp(arg, "--cumulative")) {
2740 options->output_format |= DIFF_FORMAT_DIRSTAT;
2741 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2742 } else if (opt_arg(arg, 0, "dirstat-by-file",
2743 &options->dirstat_percent)) {
2744 options->output_format |= DIFF_FORMAT_DIRSTAT;
2745 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2746 }
2747 else if (!strcmp(arg, "--check"))
2748 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2749 else if (!strcmp(arg, "--summary"))
2750 options->output_format |= DIFF_FORMAT_SUMMARY;
2751 else if (!strcmp(arg, "--patch-with-stat"))
2752 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2753 else if (!strcmp(arg, "--name-only"))
2754 options->output_format |= DIFF_FORMAT_NAME;
2755 else if (!strcmp(arg, "--name-status"))
2756 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2757 else if (!strcmp(arg, "-s"))
2758 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2759 else if (!prefixcmp(arg, "--stat")) {
2760 char *end;
2761 int width = options->stat_width;
2762 int name_width = options->stat_name_width;
2763 arg += 6;
2764 end = (char *)arg;
2765
2766 switch (*arg) {
2767 case '-':
2768 if (!prefixcmp(arg, "-width="))
2769 width = strtoul(arg + 7, &end, 10);
2770 else if (!prefixcmp(arg, "-name-width="))
2771 name_width = strtoul(arg + 12, &end, 10);
2772 break;
2773 case '=':
2774 width = strtoul(arg+1, &end, 10);
2775 if (*end == ',')
2776 name_width = strtoul(end+1, &end, 10);
2777 }
2778
2779 /* Important! This checks all the error cases! */
2780 if (*end)
2781 return 0;
2782 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2783 options->stat_name_width = name_width;
2784 options->stat_width = width;
2785 }
2786
2787 /* renames options */
2788 else if (!prefixcmp(arg, "-B")) {
2789 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2790 return -1;
2791 }
2792 else if (!prefixcmp(arg, "-M")) {
2793 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2794 return -1;
2795 options->detect_rename = DIFF_DETECT_RENAME;
2796 }
2797 else if (!prefixcmp(arg, "-C")) {
2798 if (options->detect_rename == DIFF_DETECT_COPY)
2799 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2800 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2801 return -1;
2802 options->detect_rename = DIFF_DETECT_COPY;
2803 }
2804 else if (!strcmp(arg, "--no-renames"))
2805 options->detect_rename = 0;
2806 else if (!strcmp(arg, "--relative"))
2807 DIFF_OPT_SET(options, RELATIVE_NAME);
2808 else if (!prefixcmp(arg, "--relative=")) {
2809 DIFF_OPT_SET(options, RELATIVE_NAME);
2810 options->prefix = arg + 11;
2811 }
2812
2813 /* xdiff options */
2814 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2815 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
2816 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2817 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
2818 else if (!strcmp(arg, "--ignore-space-at-eol"))
2819 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
2820 else if (!strcmp(arg, "--patience"))
2821 DIFF_XDL_SET(options, PATIENCE_DIFF);
2822
2823 /* flags options */
2824 else if (!strcmp(arg, "--binary")) {
2825 options->output_format |= DIFF_FORMAT_PATCH;
2826 DIFF_OPT_SET(options, BINARY);
2827 }
2828 else if (!strcmp(arg, "--full-index"))
2829 DIFF_OPT_SET(options, FULL_INDEX);
2830 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2831 DIFF_OPT_SET(options, TEXT);
2832 else if (!strcmp(arg, "-R"))
2833 DIFF_OPT_SET(options, REVERSE_DIFF);
2834 else if (!strcmp(arg, "--find-copies-harder"))
2835 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2836 else if (!strcmp(arg, "--follow"))
2837 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2838 else if (!strcmp(arg, "--color"))
2839 DIFF_OPT_SET(options, COLOR_DIFF);
2840 else if (!strcmp(arg, "--no-color"))
2841 DIFF_OPT_CLR(options, COLOR_DIFF);
2842 else if (!strcmp(arg, "--color-words")) {
2843 DIFF_OPT_SET(options, COLOR_DIFF);
2844 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2845 }
2846 else if (!prefixcmp(arg, "--color-words=")) {
2847 DIFF_OPT_SET(options, COLOR_DIFF);
2848 DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
2849 options->word_regex = arg + 14;
2850 }
2851 else if (!strcmp(arg, "--exit-code"))
2852 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2853 else if (!strcmp(arg, "--quiet"))
2854 DIFF_OPT_SET(options, QUICK);
2855 else if (!strcmp(arg, "--ext-diff"))
2856 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2857 else if (!strcmp(arg, "--no-ext-diff"))
2858 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2859 else if (!strcmp(arg, "--textconv"))
2860 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2861 else if (!strcmp(arg, "--no-textconv"))
2862 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2863 else if (!strcmp(arg, "--ignore-submodules"))
2864 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2865 else if (!strcmp(arg, "--submodule"))
2866 DIFF_OPT_SET(options, SUBMODULE_LOG);
2867 else if (!prefixcmp(arg, "--submodule=")) {
2868 if (!strcmp(arg + 12, "log"))
2869 DIFF_OPT_SET(options, SUBMODULE_LOG);
2870 }
2871
2872 /* misc options */
2873 else if (!strcmp(arg, "-z"))
2874 options->line_termination = 0;
2875 else if (!prefixcmp(arg, "-l"))
2876 options->rename_limit = strtoul(arg+2, NULL, 10);
2877 else if (!prefixcmp(arg, "-S"))
2878 options->pickaxe = arg + 2;
2879 else if (!strcmp(arg, "--pickaxe-all"))
2880 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2881 else if (!strcmp(arg, "--pickaxe-regex"))
2882 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2883 else if (!prefixcmp(arg, "-O"))
2884 options->orderfile = arg + 2;
2885 else if (!prefixcmp(arg, "--diff-filter="))
2886 options->filter = arg + 14;
2887 else if (!strcmp(arg, "--abbrev"))
2888 options->abbrev = DEFAULT_ABBREV;
2889 else if (!prefixcmp(arg, "--abbrev=")) {
2890 options->abbrev = strtoul(arg + 9, NULL, 10);
2891 if (options->abbrev < MINIMUM_ABBREV)
2892 options->abbrev = MINIMUM_ABBREV;
2893 else if (40 < options->abbrev)
2894 options->abbrev = 40;
2895 }
2896 else if (!prefixcmp(arg, "--src-prefix="))
2897 options->a_prefix = arg + 13;
2898 else if (!prefixcmp(arg, "--dst-prefix="))
2899 options->b_prefix = arg + 13;
2900 else if (!strcmp(arg, "--no-prefix"))
2901 options->a_prefix = options->b_prefix = "";
2902 else if (opt_arg(arg, '\0', "inter-hunk-context",
2903 &options->interhunkcontext))
2904 ;
2905 else if (!prefixcmp(arg, "--output=")) {
2906 options->file = fopen(arg + strlen("--output="), "w");
2907 options->close_file = 1;
2908 } else
2909 return 0;
2910 return 1;
2911 }
2912
2913 static int parse_num(const char **cp_p)
2914 {
2915 unsigned long num, scale;
2916 int ch, dot;
2917 const char *cp = *cp_p;
2918
2919 num = 0;
2920 scale = 1;
2921 dot = 0;
2922 for (;;) {
2923 ch = *cp;
2924 if ( !dot && ch == '.' ) {
2925 scale = 1;
2926 dot = 1;
2927 } else if ( ch == '%' ) {
2928 scale = dot ? scale*100 : 100;
2929 cp++; /* % is always at the end */
2930 break;
2931 } else if ( ch >= '0' && ch <= '9' ) {
2932 if ( scale < 100000 ) {
2933 scale *= 10;
2934 num = (num*10) + (ch-'0');
2935 }
2936 } else {
2937 break;
2938 }
2939 cp++;
2940 }
2941 *cp_p = cp;
2942
2943 /* user says num divided by scale and we say internally that
2944 * is MAX_SCORE * num / scale.
2945 */
2946 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2947 }
2948
2949 static int diff_scoreopt_parse(const char *opt)
2950 {
2951 int opt1, opt2, cmd;
2952
2953 if (*opt++ != '-')
2954 return -1;
2955 cmd = *opt++;
2956 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2957 return -1; /* that is not a -M, -C nor -B option */
2958
2959 opt1 = parse_num(&opt);
2960 if (cmd != 'B')
2961 opt2 = 0;
2962 else {
2963 if (*opt == 0)
2964 opt2 = 0;
2965 else if (*opt != '/')
2966 return -1; /* we expect -B80/99 or -B80 */
2967 else {
2968 opt++;
2969 opt2 = parse_num(&opt);
2970 }
2971 }
2972 if (*opt != 0)
2973 return -1;
2974 return opt1 | (opt2 << 16);
2975 }
2976
2977 struct diff_queue_struct diff_queued_diff;
2978
2979 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2980 {
2981 if (queue->alloc <= queue->nr) {
2982 queue->alloc = alloc_nr(queue->alloc);
2983 queue->queue = xrealloc(queue->queue,
2984 sizeof(dp) * queue->alloc);
2985 }
2986 queue->queue[queue->nr++] = dp;
2987 }
2988
2989 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2990 struct diff_filespec *one,
2991 struct diff_filespec *two)
2992 {
2993 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2994 dp->one = one;
2995 dp->two = two;
2996 if (queue)
2997 diff_q(queue, dp);
2998 return dp;
2999 }
3000
3001 void diff_free_filepair(struct diff_filepair *p)
3002 {
3003 free_filespec(p->one);
3004 free_filespec(p->two);
3005 free(p);
3006 }
3007
3008 /* This is different from find_unique_abbrev() in that
3009 * it stuffs the result with dots for alignment.
3010 */
3011 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3012 {
3013 int abblen;
3014 const char *abbrev;
3015 if (len == 40)
3016 return sha1_to_hex(sha1);
3017
3018 abbrev = find_unique_abbrev(sha1, len);
3019 abblen = strlen(abbrev);
3020 if (abblen < 37) {
3021 static char hex[41];
3022 if (len < abblen && abblen <= len + 2)
3023 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3024 else
3025 sprintf(hex, "%s...", abbrev);
3026 return hex;
3027 }
3028 return sha1_to_hex(sha1);
3029 }
3030
3031 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3032 {
3033 int line_termination = opt->line_termination;
3034 int inter_name_termination = line_termination ? '\t' : '\0';
3035
3036 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3037 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3038 diff_unique_abbrev(p->one->sha1, opt->abbrev));
3039 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3040 }
3041 if (p->score) {
3042 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3043 inter_name_termination);
3044 } else {
3045 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3046 }
3047
3048 if (p->status == DIFF_STATUS_COPIED ||
3049 p->status == DIFF_STATUS_RENAMED) {
3050 const char *name_a, *name_b;
3051 name_a = p->one->path;
3052 name_b = p->two->path;
3053 strip_prefix(opt->prefix_length, &name_a, &name_b);
3054 write_name_quoted(name_a, opt->file, inter_name_termination);
3055 write_name_quoted(name_b, opt->file, line_termination);
3056 } else {
3057 const char *name_a, *name_b;
3058 name_a = p->one->mode ? p->one->path : p->two->path;
3059 name_b = NULL;
3060 strip_prefix(opt->prefix_length, &name_a, &name_b);
3061 write_name_quoted(name_a, opt->file, line_termination);
3062 }
3063 }
3064
3065 int diff_unmodified_pair(struct diff_filepair *p)
3066 {
3067 /* This function is written stricter than necessary to support
3068 * the currently implemented transformers, but the idea is to
3069 * let transformers to produce diff_filepairs any way they want,
3070 * and filter and clean them up here before producing the output.
3071 */
3072 struct diff_filespec *one = p->one, *two = p->two;
3073
3074 if (DIFF_PAIR_UNMERGED(p))
3075 return 0; /* unmerged is interesting */
3076
3077 /* deletion, addition, mode or type change
3078 * and rename are all interesting.
3079 */
3080 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3081 DIFF_PAIR_MODE_CHANGED(p) ||
3082 strcmp(one->path, two->path))
3083 return 0;
3084
3085 /* both are valid and point at the same path. that is, we are
3086 * dealing with a change.
3087 */
3088 if (one->sha1_valid && two->sha1_valid &&
3089 !hashcmp(one->sha1, two->sha1))
3090 return 1; /* no change */
3091 if (!one->sha1_valid && !two->sha1_valid)
3092 return 1; /* both look at the same file on the filesystem. */
3093 return 0;
3094 }
3095
3096 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3097 {
3098 if (diff_unmodified_pair(p))
3099 return;
3100
3101 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3102 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3103 return; /* no tree diffs in patch format */
3104
3105 run_diff(p, o);
3106 }
3107
3108 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3109 struct diffstat_t *diffstat)
3110 {
3111 if (diff_unmodified_pair(p))
3112 return;
3113
3114 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3115 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3116 return; /* no tree diffs in patch format */
3117
3118 run_diffstat(p, o, diffstat);
3119 }
3120
3121 static void diff_flush_checkdiff(struct diff_filepair *p,
3122 struct diff_options *o)
3123 {
3124 if (diff_unmodified_pair(p))
3125 return;
3126
3127 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3128 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3129 return; /* no tree diffs in patch format */
3130
3131 run_checkdiff(p, o);
3132 }
3133
3134 int diff_queue_is_empty(void)
3135 {
3136 struct diff_queue_struct *q = &diff_queued_diff;
3137 int i;
3138 for (i = 0; i < q->nr; i++)
3139 if (!diff_unmodified_pair(q->queue[i]))
3140 return 0;
3141 return 1;
3142 }
3143
3144 #if DIFF_DEBUG
3145 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3146 {
3147 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3148 x, one ? one : "",
3149 s->path,
3150 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3151 s->mode,
3152 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3153 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3154 x, one ? one : "",
3155 s->size, s->xfrm_flags);
3156 }
3157
3158 void diff_debug_filepair(const struct diff_filepair *p, int i)
3159 {
3160 diff_debug_filespec(p->one, i, "one");
3161 diff_debug_filespec(p->two, i, "two");
3162 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3163 p->score, p->status ? p->status : '?',
3164 p->one->rename_used, p->broken_pair);
3165 }
3166
3167 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3168 {
3169 int i;
3170 if (msg)
3171 fprintf(stderr, "%s\n", msg);
3172 fprintf(stderr, "q->nr = %d\n", q->nr);
3173 for (i = 0; i < q->nr; i++) {
3174 struct diff_filepair *p = q->queue[i];
3175 diff_debug_filepair(p, i);
3176 }
3177 }
3178 #endif
3179
3180 static void diff_resolve_rename_copy(void)
3181 {
3182 int i;
3183 struct diff_filepair *p;
3184 struct diff_queue_struct *q = &diff_queued_diff;
3185
3186 diff_debug_queue("resolve-rename-copy", q);
3187
3188 for (i = 0; i < q->nr; i++) {
3189 p = q->queue[i];
3190 p->status = 0; /* undecided */
3191 if (DIFF_PAIR_UNMERGED(p))
3192 p->status = DIFF_STATUS_UNMERGED;
3193 else if (!DIFF_FILE_VALID(p->one))
3194 p->status = DIFF_STATUS_ADDED;
3195 else if (!DIFF_FILE_VALID(p->two))
3196 p->status = DIFF_STATUS_DELETED;
3197 else if (DIFF_PAIR_TYPE_CHANGED(p))
3198 p->status = DIFF_STATUS_TYPE_CHANGED;
3199
3200 /* from this point on, we are dealing with a pair
3201 * whose both sides are valid and of the same type, i.e.
3202 * either in-place edit or rename/copy edit.
3203 */
3204 else if (DIFF_PAIR_RENAME(p)) {
3205 /*
3206 * A rename might have re-connected a broken
3207 * pair up, causing the pathnames to be the
3208 * same again. If so, that's not a rename at
3209 * all, just a modification..
3210 *
3211 * Otherwise, see if this source was used for
3212 * multiple renames, in which case we decrement
3213 * the count, and call it a copy.
3214 */
3215 if (!strcmp(p->one->path, p->two->path))
3216 p->status = DIFF_STATUS_MODIFIED;
3217 else if (--p->one->rename_used > 0)
3218 p->status = DIFF_STATUS_COPIED;
3219 else
3220 p->status = DIFF_STATUS_RENAMED;
3221 }
3222 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3223 p->one->mode != p->two->mode ||
3224 is_null_sha1(p->one->sha1))
3225 p->status = DIFF_STATUS_MODIFIED;
3226 else {
3227 /* This is a "no-change" entry and should not
3228 * happen anymore, but prepare for broken callers.
3229 */
3230 error("feeding unmodified %s to diffcore",
3231 p->one->path);
3232 p->status = DIFF_STATUS_UNKNOWN;
3233 }
3234 }
3235 diff_debug_queue("resolve-rename-copy done", q);
3236 }
3237
3238 static int check_pair_status(struct diff_filepair *p)
3239 {
3240 switch (p->status) {
3241 case DIFF_STATUS_UNKNOWN:
3242 return 0;
3243 case 0:
3244 die("internal error in diff-resolve-rename-copy");
3245 default:
3246 return 1;
3247 }
3248 }
3249
3250 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3251 {
3252 int fmt = opt->output_format;
3253
3254 if (fmt & DIFF_FORMAT_CHECKDIFF)
3255 diff_flush_checkdiff(p, opt);
3256 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3257 diff_flush_raw(p, opt);
3258 else if (fmt & DIFF_FORMAT_NAME) {
3259 const char *name_a, *name_b;
3260 name_a = p->two->path;
3261 name_b = NULL;
3262 strip_prefix(opt->prefix_length, &name_a, &name_b);
3263 write_name_quoted(name_a, opt->file, opt->line_termination);
3264 }
3265 }
3266
3267 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3268 {
3269 if (fs->mode)
3270 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3271 else
3272 fprintf(file, " %s ", newdelete);
3273 write_name_quoted(fs->path, file, '\n');
3274 }
3275
3276
3277 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
3278 {
3279 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3280 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
3281 show_name ? ' ' : '\n');
3282 if (show_name) {
3283 write_name_quoted(p->two->path, file, '\n');
3284 }
3285 }
3286 }
3287
3288 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
3289 {
3290 char *names = pprint_rename(p->one->path, p->two->path);
3291
3292 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3293 free(names);
3294 show_mode_change(file, p, 0);
3295 }
3296
3297 static void diff_summary(FILE *file, struct diff_filepair *p)
3298 {
3299 switch(p->status) {
3300 case DIFF_STATUS_DELETED:
3301 show_file_mode_name(file, "delete", p->one);
3302 break;
3303 case DIFF_STATUS_ADDED:
3304 show_file_mode_name(file, "create", p->two);
3305 break;
3306 case DIFF_STATUS_COPIED:
3307 show_rename_copy(file, "copy", p);
3308 break;
3309 case DIFF_STATUS_RENAMED:
3310 show_rename_copy(file, "rename", p);
3311 break;
3312 default:
3313 if (p->score) {
3314 fputs(" rewrite ", file);
3315 write_name_quoted(p->two->path, file, ' ');
3316 fprintf(file, "(%d%%)\n", similarity_index(p));
3317 }
3318 show_mode_change(file, p, !p->score);
3319 break;
3320 }
3321 }
3322
3323 struct patch_id_t {
3324 git_SHA_CTX *ctx;
3325 int patchlen;
3326 };
3327
3328 static int remove_space(char *line, int len)
3329 {
3330 int i;
3331 char *dst = line;
3332 unsigned char c;
3333
3334 for (i = 0; i < len; i++)
3335 if (!isspace((c = line[i])))
3336 *dst++ = c;
3337
3338 return dst - line;
3339 }
3340
3341 static void patch_id_consume(void *priv, char *line, unsigned long len)
3342 {
3343 struct patch_id_t *data = priv;
3344 int new_len;
3345
3346 /* Ignore line numbers when computing the SHA1 of the patch */
3347 if (!prefixcmp(line, "@@ -"))
3348 return;
3349
3350 new_len = remove_space(line, len);
3351
3352 git_SHA1_Update(data->ctx, line, new_len);
3353 data->patchlen += new_len;
3354 }
3355
3356 /* returns 0 upon success, and writes result into sha1 */
3357 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3358 {
3359 struct diff_queue_struct *q = &diff_queued_diff;
3360 int i;
3361 git_SHA_CTX ctx;
3362 struct patch_id_t data;
3363 char buffer[PATH_MAX * 4 + 20];
3364
3365 git_SHA1_Init(&ctx);
3366 memset(&data, 0, sizeof(struct patch_id_t));
3367 data.ctx = &ctx;
3368
3369 for (i = 0; i < q->nr; i++) {
3370 xpparam_t xpp;
3371 xdemitconf_t xecfg;
3372 xdemitcb_t ecb;
3373 mmfile_t mf1, mf2;
3374 struct diff_filepair *p = q->queue[i];
3375 int len1, len2;
3376
3377 memset(&xpp, 0, sizeof(xpp));
3378 memset(&xecfg, 0, sizeof(xecfg));
3379 if (p->status == 0)
3380 return error("internal diff status error");
3381 if (p->status == DIFF_STATUS_UNKNOWN)
3382 continue;
3383 if (diff_unmodified_pair(p))
3384 continue;
3385 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3386 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3387 continue;
3388 if (DIFF_PAIR_UNMERGED(p))
3389 continue;
3390
3391 diff_fill_sha1_info(p->one);
3392 diff_fill_sha1_info(p->two);
3393 if (fill_mmfile(&mf1, p->one) < 0 ||
3394 fill_mmfile(&mf2, p->two) < 0)
3395 return error("unable to read files to diff");
3396
3397 len1 = remove_space(p->one->path, strlen(p->one->path));
3398 len2 = remove_space(p->two->path, strlen(p->two->path));
3399 if (p->one->mode == 0)
3400 len1 = snprintf(buffer, sizeof(buffer),
3401 "diff--gita/%.*sb/%.*s"
3402 "newfilemode%06o"
3403 "---/dev/null"
3404 "+++b/%.*s",
3405 len1, p->one->path,
3406 len2, p->two->path,
3407 p->two->mode,
3408 len2, p->two->path);
3409 else if (p->two->mode == 0)
3410 len1 = snprintf(buffer, sizeof(buffer),
3411 "diff--gita/%.*sb/%.*s"
3412 "deletedfilemode%06o"
3413 "---a/%.*s"
3414 "+++/dev/null",
3415 len1, p->one->path,
3416 len2, p->two->path,
3417 p->one->mode,
3418 len1, p->one->path);
3419 else
3420 len1 = snprintf(buffer, sizeof(buffer),
3421 "diff--gita/%.*sb/%.*s"
3422 "---a/%.*s"
3423 "+++b/%.*s",
3424 len1, p->one->path,
3425 len2, p->two->path,
3426 len1, p->one->path,
3427 len2, p->two->path);
3428 git_SHA1_Update(&ctx, buffer, len1);
3429
3430 xpp.flags = XDF_NEED_MINIMAL;
3431 xecfg.ctxlen = 3;
3432 xecfg.flags = XDL_EMIT_FUNCNAMES;
3433 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3434 &xpp, &xecfg, &ecb);
3435 }
3436
3437 git_SHA1_Final(sha1, &ctx);
3438 return 0;
3439 }
3440
3441 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3442 {
3443 struct diff_queue_struct *q = &diff_queued_diff;
3444 int i;
3445 int result = diff_get_patch_id(options, sha1);
3446
3447 for (i = 0; i < q->nr; i++)
3448 diff_free_filepair(q->queue[i]);
3449
3450 free(q->queue);
3451 q->queue = NULL;
3452 q->nr = q->alloc = 0;
3453
3454 return result;
3455 }
3456
3457 static int is_summary_empty(const struct diff_queue_struct *q)
3458 {
3459 int i;
3460
3461 for (i = 0; i < q->nr; i++) {
3462 const struct diff_filepair *p = q->queue[i];
3463
3464 switch (p->status) {
3465 case DIFF_STATUS_DELETED:
3466 case DIFF_STATUS_ADDED:
3467 case DIFF_STATUS_COPIED:
3468 case DIFF_STATUS_RENAMED:
3469 return 0;
3470 default:
3471 if (p->score)
3472 return 0;
3473 if (p->one->mode && p->two->mode &&
3474 p->one->mode != p->two->mode)
3475 return 0;
3476 break;
3477 }
3478 }
3479 return 1;
3480 }
3481
3482 void diff_flush(struct diff_options *options)
3483 {
3484 struct diff_queue_struct *q = &diff_queued_diff;
3485 int i, output_format = options->output_format;
3486 int separator = 0;
3487
3488 /*
3489 * Order: raw, stat, summary, patch
3490 * or: name/name-status/checkdiff (other bits clear)
3491 */
3492 if (!q->nr)
3493 goto free_queue;
3494
3495 if (output_format & (DIFF_FORMAT_RAW |
3496 DIFF_FORMAT_NAME |
3497 DIFF_FORMAT_NAME_STATUS |
3498 DIFF_FORMAT_CHECKDIFF)) {
3499 for (i = 0; i < q->nr; i++) {
3500 struct diff_filepair *p = q->queue[i];
3501 if (check_pair_status(p))
3502 flush_one_pair(p, options);
3503 }
3504 separator++;
3505 }
3506
3507 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3508 struct diffstat_t diffstat;
3509
3510 memset(&diffstat, 0, sizeof(struct diffstat_t));
3511 for (i = 0; i < q->nr; i++) {
3512 struct diff_filepair *p = q->queue[i];
3513 if (check_pair_status(p))
3514 diff_flush_stat(p, options, &diffstat);
3515 }
3516 if (output_format & DIFF_FORMAT_NUMSTAT)
3517 show_numstat(&diffstat, options);
3518 if (output_format & DIFF_FORMAT_DIFFSTAT)
3519 show_stats(&diffstat, options);
3520 if (output_format & DIFF_FORMAT_SHORTSTAT)
3521 show_shortstats(&diffstat, options);
3522 free_diffstat_info(&diffstat);
3523 separator++;
3524 }
3525 if (output_format & DIFF_FORMAT_DIRSTAT)
3526 show_dirstat(options);
3527
3528 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3529 for (i = 0; i < q->nr; i++)
3530 diff_summary(options->file, q->queue[i]);
3531 separator++;
3532 }
3533
3534 if (output_format & DIFF_FORMAT_PATCH) {
3535 if (separator) {
3536 putc(options->line_termination, options->file);
3537 if (options->stat_sep) {
3538 /* attach patch instead of inline */
3539 fputs(options->stat_sep, options->file);
3540 }
3541 }
3542
3543 for (i = 0; i < q->nr; i++) {
3544 struct diff_filepair *p = q->queue[i];
3545 if (check_pair_status(p))
3546 diff_flush_patch(p, options);
3547 }
3548 }
3549
3550 if (output_format & DIFF_FORMAT_CALLBACK)
3551 options->format_callback(q, options, options->format_callback_data);
3552
3553 for (i = 0; i < q->nr; i++)
3554 diff_free_filepair(q->queue[i]);
3555 free_queue:
3556 free(q->queue);
3557 q->queue = NULL;
3558 q->nr = q->alloc = 0;
3559 if (options->close_file)
3560 fclose(options->file);
3561
3562 /*
3563 * Report the content-level differences with HAS_CHANGES;
3564 * diff_addremove/diff_change does not set the bit when
3565 * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3566 */
3567 if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3568 if (options->found_changes)
3569 DIFF_OPT_SET(options, HAS_CHANGES);
3570 else
3571 DIFF_OPT_CLR(options, HAS_CHANGES);
3572 }
3573 }
3574
3575 static void diffcore_apply_filter(const char *filter)
3576 {
3577 int i;
3578 struct diff_queue_struct *q = &diff_queued_diff;
3579 struct diff_queue_struct outq;
3580 outq.queue = NULL;
3581 outq.nr = outq.alloc = 0;
3582
3583 if (!filter)
3584 return;
3585
3586 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3587 int found;
3588 for (i = found = 0; !found && i < q->nr; i++) {
3589 struct diff_filepair *p = q->queue[i];
3590 if (((p->status == DIFF_STATUS_MODIFIED) &&
3591 ((p->score &&
3592 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3593 (!p->score &&
3594 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3595 ((p->status != DIFF_STATUS_MODIFIED) &&
3596 strchr(filter, p->status)))
3597 found++;
3598 }
3599 if (found)
3600 return;
3601
3602 /* otherwise we will clear the whole queue
3603 * by copying the empty outq at the end of this
3604 * function, but first clear the current entries
3605 * in the queue.
3606 */
3607 for (i = 0; i < q->nr; i++)
3608 diff_free_filepair(q->queue[i]);
3609 }
3610 else {
3611 /* Only the matching ones */
3612 for (i = 0; i < q->nr; i++) {
3613 struct diff_filepair *p = q->queue[i];
3614
3615 if (((p->status == DIFF_STATUS_MODIFIED) &&
3616 ((p->score &&
3617 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3618 (!p->score &&
3619 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3620 ((p->status != DIFF_STATUS_MODIFIED) &&
3621 strchr(filter, p->status)))
3622 diff_q(&outq, p);
3623 else
3624 diff_free_filepair(p);
3625 }
3626 }
3627 free(q->queue);
3628 *q = outq;
3629 }
3630
3631 /* Check whether two filespecs with the same mode and size are identical */
3632 static int diff_filespec_is_identical(struct diff_filespec *one,
3633 struct diff_filespec *two)
3634 {
3635 if (S_ISGITLINK(one->mode))
3636 return 0;
3637 if (diff_populate_filespec(one, 0))
3638 return 0;
3639 if (diff_populate_filespec(two, 0))
3640 return 0;
3641 return !memcmp(one->data, two->data, one->size);
3642 }
3643
3644 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3645 {
3646 int i;
3647 struct diff_queue_struct *q = &diff_queued_diff;
3648 struct diff_queue_struct outq;
3649 outq.queue = NULL;
3650 outq.nr = outq.alloc = 0;
3651
3652 for (i = 0; i < q->nr; i++) {
3653 struct diff_filepair *p = q->queue[i];
3654
3655 /*
3656 * 1. Entries that come from stat info dirtyness
3657 * always have both sides (iow, not create/delete),
3658 * one side of the object name is unknown, with
3659 * the same mode and size. Keep the ones that
3660 * do not match these criteria. They have real
3661 * differences.
3662 *
3663 * 2. At this point, the file is known to be modified,
3664 * with the same mode and size, and the object
3665 * name of one side is unknown. Need to inspect
3666 * the identical contents.
3667 */
3668 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3669 !DIFF_FILE_VALID(p->two) ||
3670 (p->one->sha1_valid && p->two->sha1_valid) ||
3671 (p->one->mode != p->two->mode) ||
3672 diff_populate_filespec(p->one, 1) ||
3673 diff_populate_filespec(p->two, 1) ||
3674 (p->one->size != p->two->size) ||
3675 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3676 diff_q(&outq, p);
3677 else {
3678 /*
3679 * The caller can subtract 1 from skip_stat_unmatch
3680 * to determine how many paths were dirty only
3681 * due to stat info mismatch.
3682 */
3683 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3684 diffopt->skip_stat_unmatch++;
3685 diff_free_filepair(p);
3686 }
3687 }
3688 free(q->queue);
3689 *q = outq;
3690 }
3691
3692 static int diffnamecmp(const void *a_, const void *b_)
3693 {
3694 const struct diff_filepair *a = *((const struct diff_filepair **)a_);
3695 const struct diff_filepair *b = *((const struct diff_filepair **)b_);
3696 const char *name_a, *name_b;
3697
3698 name_a = a->one ? a->one->path : a->two->path;
3699 name_b = b->one ? b->one->path : b->two->path;
3700 return strcmp(name_a, name_b);
3701 }
3702
3703 void diffcore_fix_diff_index(struct diff_options *options)
3704 {
3705 struct diff_queue_struct *q = &diff_queued_diff;
3706 qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
3707 }
3708
3709 void diffcore_std(struct diff_options *options)
3710 {
3711 if (options->skip_stat_unmatch)
3712 diffcore_skip_stat_unmatch(options);
3713 if (options->break_opt != -1)
3714 diffcore_break(options->break_opt);
3715 if (options->detect_rename)
3716 diffcore_rename(options);
3717 if (options->break_opt != -1)
3718 diffcore_merge_broken();
3719 if (options->pickaxe)
3720 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3721 if (options->orderfile)
3722 diffcore_order(options->orderfile);
3723 diff_resolve_rename_copy();
3724 diffcore_apply_filter(options->filter);
3725
3726 if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3727 DIFF_OPT_SET(options, HAS_CHANGES);
3728 else
3729 DIFF_OPT_CLR(options, HAS_CHANGES);
3730 }
3731
3732 int diff_result_code(struct diff_options *opt, int status)
3733 {
3734 int result = 0;
3735 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3736 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3737 return status;
3738 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3739 DIFF_OPT_TST(opt, HAS_CHANGES))
3740 result |= 01;
3741 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3742 DIFF_OPT_TST(opt, CHECK_FAILED))
3743 result |= 02;
3744 return result;
3745 }
3746
3747 void diff_addremove(struct diff_options *options,
3748 int addremove, unsigned mode,
3749 const unsigned char *sha1,
3750 const char *concatpath, unsigned dirty_submodule)
3751 {
3752 struct diff_filespec *one, *two;
3753
3754 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3755 return;
3756
3757 /* This may look odd, but it is a preparation for
3758 * feeding "there are unchanged files which should
3759 * not produce diffs, but when you are doing copy
3760 * detection you would need them, so here they are"
3761 * entries to the diff-core. They will be prefixed
3762 * with something like '=' or '*' (I haven't decided
3763 * which but should not make any difference).
3764 * Feeding the same new and old to diff_change()
3765 * also has the same effect.
3766 * Before the final output happens, they are pruned after
3767 * merged into rename/copy pairs as appropriate.
3768 */
3769 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3770 addremove = (addremove == '+' ? '-' :
3771 addremove == '-' ? '+' : addremove);
3772
3773 if (options->prefix &&
3774 strncmp(concatpath, options->prefix, options->prefix_length))
3775 return;
3776
3777 one = alloc_filespec(concatpath);
3778 two = alloc_filespec(concatpath);
3779
3780 if (addremove != '+')
3781 fill_filespec(one, sha1, mode);
3782 if (addremove != '-') {
3783 fill_filespec(two, sha1, mode);
3784 two->dirty_submodule = dirty_submodule;
3785 }
3786
3787 diff_queue(&diff_queued_diff, one, two);
3788 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3789 DIFF_OPT_SET(options, HAS_CHANGES);
3790 }
3791
3792 void diff_change(struct diff_options *options,
3793 unsigned old_mode, unsigned new_mode,
3794 const unsigned char *old_sha1,
3795 const unsigned char *new_sha1,
3796 const char *concatpath,
3797 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
3798 {
3799 struct diff_filespec *one, *two;
3800
3801 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3802 && S_ISGITLINK(new_mode))
3803 return;
3804
3805 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3806 unsigned tmp;
3807 const unsigned char *tmp_c;
3808 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3809 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3810 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
3811 new_dirty_submodule = tmp;
3812 }
3813
3814 if (options->prefix &&
3815 strncmp(concatpath, options->prefix, options->prefix_length))
3816 return;
3817
3818 one = alloc_filespec(concatpath);
3819 two = alloc_filespec(concatpath);
3820 fill_filespec(one, old_sha1, old_mode);
3821 fill_filespec(two, new_sha1, new_mode);
3822 one->dirty_submodule = old_dirty_submodule;
3823 two->dirty_submodule = new_dirty_submodule;
3824
3825 diff_queue(&diff_queued_diff, one, two);
3826 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
3827 DIFF_OPT_SET(options, HAS_CHANGES);
3828 }
3829
3830 void diff_unmerge(struct diff_options *options,
3831 const char *path,
3832 unsigned mode, const unsigned char *sha1)
3833 {
3834 struct diff_filespec *one, *two;
3835
3836 if (options->prefix &&
3837 strncmp(path, options->prefix, options->prefix_length))
3838 return;
3839
3840 one = alloc_filespec(path);
3841 two = alloc_filespec(path);
3842 fill_filespec(one, sha1, mode);
3843 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3844 }
3845
3846 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3847 size_t *outsize)
3848 {
3849 struct diff_tempfile *temp;
3850 const char *argv[3];
3851 const char **arg = argv;
3852 struct child_process child;
3853 struct strbuf buf = STRBUF_INIT;
3854
3855 temp = prepare_temp_file(spec->path, spec);
3856 *arg++ = pgm;
3857 *arg++ = temp->name;
3858 *arg = NULL;
3859
3860 memset(&child, 0, sizeof(child));
3861 child.use_shell = 1;
3862 child.argv = argv;
3863 child.out = -1;
3864 if (start_command(&child) != 0 ||
3865 strbuf_read(&buf, child.out, 0) < 0 ||
3866 finish_command(&child) != 0) {
3867 close(child.out);
3868 strbuf_release(&buf);
3869 remove_tempfile();
3870 error("error running textconv command '%s'", pgm);
3871 return NULL;
3872 }
3873 close(child.out);
3874 remove_tempfile();
3875
3876 return strbuf_detach(&buf, outsize);
3877 }
The Best VCS Around