?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'bg/fix-t7003'
[git/dscho.git] / diff.c
blob9a5c77c13ff3406a85e8a94f309a639529f2d88b
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 #include "ll-merge.h"
18
19 #ifdef NO_FAST_WORKING_DIRECTORY
20 #define FAST_WORKING_DIRECTORY 0
21 #else
22 #define FAST_WORKING_DIRECTORY 1
23 #endif
24
25 static int diff_detect_rename_default;
26 static int diff_rename_limit_default = 200;
27 static int diff_suppress_blank_empty;
28 int diff_use_color_default = -1;
29 static const char *diff_word_regex_cfg;
30 static const char *external_diff_cmd_cfg;
31 int diff_auto_refresh_index = 1;
32 static int diff_mnemonic_prefix;
33 static int diff_no_prefix;
34 static struct diff_options default_diff_options;
35
36 static char diff_colors[][COLOR_MAXLEN] = {
37 GIT_COLOR_RESET,
38 GIT_COLOR_NORMAL, /* PLAIN */
39 GIT_COLOR_BOLD, /* METAINFO */
40 GIT_COLOR_CYAN, /* FRAGINFO */
41 GIT_COLOR_RED, /* OLD */
42 GIT_COLOR_GREEN, /* NEW */
43 GIT_COLOR_YELLOW, /* COMMIT */
44 GIT_COLOR_BG_RED, /* WHITESPACE */
45 GIT_COLOR_NORMAL, /* FUNCINFO */
46 };
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.noprefix")) {
103 diff_no_prefix = git_config_bool(var, value);
104 return 0;
105 }
106 if (!strcmp(var, "diff.external"))
107 return git_config_string(&external_diff_cmd_cfg, var, value);
108 if (!strcmp(var, "diff.wordregex"))
109 return git_config_string(&diff_word_regex_cfg, var, value);
110
111 if (!strcmp(var, "diff.ignoresubmodules"))
112 handle_ignore_submodules_arg(&default_diff_options, value);
113
114 return git_diff_basic_config(var, value, cb);
115 }
116
117 int git_diff_basic_config(const char *var, const char *value, void *cb)
118 {
119 if (!strcmp(var, "diff.renamelimit")) {
120 diff_rename_limit_default = git_config_int(var, value);
121 return 0;
122 }
123
124 switch (userdiff_config(var, value)) {
125 case 0: break;
126 case -1: return -1;
127 default: return 0;
128 }
129
130 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
131 int slot = parse_diff_color_slot(var, 11);
132 if (slot < 0)
133 return 0;
134 if (!value)
135 return config_error_nonbool(var);
136 color_parse(value, var, diff_colors[slot]);
137 return 0;
138 }
139
140 /* like GNU diff's --suppress-blank-empty option */
141 if (!strcmp(var, "diff.suppressblankempty") ||
142 /* for backwards compatibility */
143 !strcmp(var, "diff.suppress-blank-empty")) {
144 diff_suppress_blank_empty = git_config_bool(var, value);
145 return 0;
146 }
147
148 if (!prefixcmp(var, "submodule."))
149 return parse_submodule_config_option(var, value);
150
151 return git_color_default_config(var, value, cb);
152 }
153
154 static char *quote_two(const char *one, const char *two)
155 {
156 int need_one = quote_c_style(one, NULL, NULL, 1);
157 int need_two = quote_c_style(two, NULL, NULL, 1);
158 struct strbuf res = STRBUF_INIT;
159
160 if (need_one + need_two) {
161 strbuf_addch(&res, '"');
162 quote_c_style(one, &res, NULL, 1);
163 quote_c_style(two, &res, NULL, 1);
164 strbuf_addch(&res, '"');
165 } else {
166 strbuf_addstr(&res, one);
167 strbuf_addstr(&res, two);
168 }
169 return strbuf_detach(&res, NULL);
170 }
171
172 static const char *external_diff(void)
173 {
174 static const char *external_diff_cmd = NULL;
175 static int done_preparing = 0;
176
177 if (done_preparing)
178 return external_diff_cmd;
179 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
180 if (!external_diff_cmd)
181 external_diff_cmd = external_diff_cmd_cfg;
182 done_preparing = 1;
183 return external_diff_cmd;
184 }
185
186 static struct diff_tempfile {
187 const char *name; /* filename external diff should read from */
188 char hex[41];
189 char mode[10];
190 char tmp_path[PATH_MAX];
191 } diff_temp[2];
192
193 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
194
195 struct emit_callback {
196 int color_diff;
197 unsigned ws_rule;
198 int blank_at_eof_in_preimage;
199 int blank_at_eof_in_postimage;
200 int lno_in_preimage;
201 int lno_in_postimage;
202 sane_truncate_fn truncate;
203 const char **label_path;
204 struct diff_words_data *diff_words;
205 struct diff_options *opt;
206 int *found_changesp;
207 struct strbuf *header;
208 };
209
210 static int count_lines(const char *data, int size)
211 {
212 int count, ch, completely_empty = 1, nl_just_seen = 0;
213 count = 0;
214 while (0 < size--) {
215 ch = *data++;
216 if (ch == '\n') {
217 count++;
218 nl_just_seen = 1;
219 completely_empty = 0;
220 }
221 else {
222 nl_just_seen = 0;
223 completely_empty = 0;
224 }
225 }
226 if (completely_empty)
227 return 0;
228 if (!nl_just_seen)
229 count++; /* no trailing newline */
230 return count;
231 }
232
233 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
234 {
235 if (!DIFF_FILE_VALID(one)) {
236 mf->ptr = (char *)""; /* does not matter */
237 mf->size = 0;
238 return 0;
239 }
240 else if (diff_populate_filespec(one, 0))
241 return -1;
242
243 mf->ptr = one->data;
244 mf->size = one->size;
245 return 0;
246 }
247
248 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
249 {
250 char *ptr = mf->ptr;
251 long size = mf->size;
252 int cnt = 0;
253
254 if (!size)
255 return cnt;
256 ptr += size - 1; /* pointing at the very end */
257 if (*ptr != '\n')
258 ; /* incomplete line */
259 else
260 ptr--; /* skip the last LF */
261 while (mf->ptr < ptr) {
262 char *prev_eol;
263 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
264 if (*prev_eol == '\n')
265 break;
266 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
267 break;
268 cnt++;
269 ptr = prev_eol - 1;
270 }
271 return cnt;
272 }
273
274 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
275 struct emit_callback *ecbdata)
276 {
277 int l1, l2, at;
278 unsigned ws_rule = ecbdata->ws_rule;
279 l1 = count_trailing_blank(mf1, ws_rule);
280 l2 = count_trailing_blank(mf2, ws_rule);
281 if (l2 <= l1) {
282 ecbdata->blank_at_eof_in_preimage = 0;
283 ecbdata->blank_at_eof_in_postimage = 0;
284 return;
285 }
286 at = count_lines(mf1->ptr, mf1->size);
287 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
288
289 at = count_lines(mf2->ptr, mf2->size);
290 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
291 }
292
293 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
294 int first, const char *line, int len)
295 {
296 int has_trailing_newline, has_trailing_carriage_return;
297 int nofirst;
298 FILE *file = o->file;
299
300 if (o->output_prefix) {
301 struct strbuf *msg = NULL;
302 msg = o->output_prefix(o, o->output_prefix_data);
303 assert(msg);
304 fwrite(msg->buf, msg->len, 1, file);
305 }
306
307 if (len == 0) {
308 has_trailing_newline = (first == '\n');
309 has_trailing_carriage_return = (!has_trailing_newline &&
310 (first == '\r'));
311 nofirst = has_trailing_newline || has_trailing_carriage_return;
312 } else {
313 has_trailing_newline = (len > 0 && line[len-1] == '\n');
314 if (has_trailing_newline)
315 len--;
316 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
317 if (has_trailing_carriage_return)
318 len--;
319 nofirst = 0;
320 }
321
322 if (len || !nofirst) {
323 fputs(set, file);
324 if (!nofirst)
325 fputc(first, file);
326 fwrite(line, len, 1, file);
327 fputs(reset, file);
328 }
329 if (has_trailing_carriage_return)
330 fputc('\r', file);
331 if (has_trailing_newline)
332 fputc('\n', file);
333 }
334
335 static void emit_line(struct diff_options *o, const char *set, const char *reset,
336 const char *line, int len)
337 {
338 emit_line_0(o, set, reset, line[0], line+1, len-1);
339 }
340
341 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
342 {
343 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
344 ecbdata->blank_at_eof_in_preimage &&
345 ecbdata->blank_at_eof_in_postimage &&
346 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
347 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
348 return 0;
349 return ws_blank_line(line, len, ecbdata->ws_rule);
350 }
351
352 static void emit_add_line(const char *reset,
353 struct emit_callback *ecbdata,
354 const char *line, int len)
355 {
356 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
357 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
358
359 if (!*ws)
360 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
361 else if (new_blank_line_at_eof(ecbdata, line, len))
362 /* Blank line at EOF - paint '+' as well */
363 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
364 else {
365 /* Emit just the prefix, then the rest. */
366 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
367 ws_check_emit(line, len, ecbdata->ws_rule,
368 ecbdata->opt->file, set, reset, ws);
369 }
370 }
371
372 static void emit_hunk_header(struct emit_callback *ecbdata,
373 const char *line, int len)
374 {
375 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
376 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
377 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
378 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
379 static const char atat[2] = { '@', '@' };
380 const char *cp, *ep;
381 struct strbuf msgbuf = STRBUF_INIT;
382 int org_len = len;
383 int i = 1;
384
385 /*
386 * As a hunk header must begin with "@@ -<old>, +<new> @@",
387 * it always is at least 10 bytes long.
388 */
389 if (len < 10 ||
390 memcmp(line, atat, 2) ||
391 !(ep = memmem(line + 2, len - 2, atat, 2))) {
392 emit_line(ecbdata->opt, plain, reset, line, len);
393 return;
394 }
395 ep += 2; /* skip over @@ */
396
397 /* The hunk header in fraginfo color */
398 strbuf_add(&msgbuf, frag, strlen(frag));
399 strbuf_add(&msgbuf, line, ep - line);
400 strbuf_add(&msgbuf, reset, strlen(reset));
401
402 /*
403 * trailing "\r\n"
404 */
405 for ( ; i < 3; i++)
406 if (line[len - i] == '\r' || line[len - i] == '\n')
407 len--;
408
409 /* blank before the func header */
410 for (cp = ep; ep - line < len; ep++)
411 if (*ep != ' ' && *ep != '\t')
412 break;
413 if (ep != cp) {
414 strbuf_add(&msgbuf, plain, strlen(plain));
415 strbuf_add(&msgbuf, cp, ep - cp);
416 strbuf_add(&msgbuf, reset, strlen(reset));
417 }
418
419 if (ep < line + len) {
420 strbuf_add(&msgbuf, func, strlen(func));
421 strbuf_add(&msgbuf, ep, line + len - ep);
422 strbuf_add(&msgbuf, reset, strlen(reset));
423 }
424
425 strbuf_add(&msgbuf, line + len, org_len - len);
426 emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
427 strbuf_release(&msgbuf);
428 }
429
430 static struct diff_tempfile *claim_diff_tempfile(void) {
431 int i;
432 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
433 if (!diff_temp[i].name)
434 return diff_temp + i;
435 die("BUG: diff is failing to clean up its tempfiles");
436 }
437
438 static int remove_tempfile_installed;
439
440 static void remove_tempfile(void)
441 {
442 int i;
443 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
444 if (diff_temp[i].name == diff_temp[i].tmp_path)
445 unlink_or_warn(diff_temp[i].name);
446 diff_temp[i].name = NULL;
447 }
448 }
449
450 static void remove_tempfile_on_signal(int signo)
451 {
452 remove_tempfile();
453 sigchain_pop(signo);
454 raise(signo);
455 }
456
457 static void print_line_count(FILE *file, int count)
458 {
459 switch (count) {
460 case 0:
461 fprintf(file, "0,0");
462 break;
463 case 1:
464 fprintf(file, "1");
465 break;
466 default:
467 fprintf(file, "1,%d", count);
468 break;
469 }
470 }
471
472 static void emit_rewrite_lines(struct emit_callback *ecb,
473 int prefix, const char *data, int size)
474 {
475 const char *endp = NULL;
476 static const char *nneof = " No newline at end of file\n";
477 const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
478 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
479
480 while (0 < size) {
481 int len;
482
483 endp = memchr(data, '\n', size);
484 len = endp ? (endp - data + 1) : size;
485 if (prefix != '+') {
486 ecb->lno_in_preimage++;
487 emit_line_0(ecb->opt, old, reset, '-',
488 data, len);
489 } else {
490 ecb->lno_in_postimage++;
491 emit_add_line(reset, ecb, data, len);
492 }
493 size -= len;
494 data += len;
495 }
496 if (!endp) {
497 const char *plain = diff_get_color(ecb->color_diff,
498 DIFF_PLAIN);
499 emit_line_0(ecb->opt, plain, reset, '\\',
500 nneof, strlen(nneof));
501 }
502 }
503
504 static void emit_rewrite_diff(const char *name_a,
505 const char *name_b,
506 struct diff_filespec *one,
507 struct diff_filespec *two,
508 struct userdiff_driver *textconv_one,
509 struct userdiff_driver *textconv_two,
510 struct diff_options *o)
511 {
512 int lc_a, lc_b;
513 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
514 const char *name_a_tab, *name_b_tab;
515 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
516 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
517 const char *reset = diff_get_color(color_diff, DIFF_RESET);
518 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
519 const char *a_prefix, *b_prefix;
520 char *data_one, *data_two;
521 size_t size_one, size_two;
522 struct emit_callback ecbdata;
523 char *line_prefix = "";
524 struct strbuf *msgbuf;
525
526 if (o && o->output_prefix) {
527 msgbuf = o->output_prefix(o, o->output_prefix_data);
528 line_prefix = msgbuf->buf;
529 }
530
531 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
532 a_prefix = o->b_prefix;
533 b_prefix = o->a_prefix;
534 } else {
535 a_prefix = o->a_prefix;
536 b_prefix = o->b_prefix;
537 }
538
539 name_a += (*name_a == '/');
540 name_b += (*name_b == '/');
541 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
542 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
543
544 strbuf_reset(&a_name);
545 strbuf_reset(&b_name);
546 quote_two_c_style(&a_name, a_prefix, name_a, 0);
547 quote_two_c_style(&b_name, b_prefix, name_b, 0);
548
549 size_one = fill_textconv(textconv_one, one, &data_one);
550 size_two = fill_textconv(textconv_two, two, &data_two);
551
552 memset(&ecbdata, 0, sizeof(ecbdata));
553 ecbdata.color_diff = color_diff;
554 ecbdata.found_changesp = &o->found_changes;
555 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
556 ecbdata.opt = o;
557 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
558 mmfile_t mf1, mf2;
559 mf1.ptr = (char *)data_one;
560 mf2.ptr = (char *)data_two;
561 mf1.size = size_one;
562 mf2.size = size_two;
563 check_blank_at_eof(&mf1, &mf2, &ecbdata);
564 }
565 ecbdata.lno_in_preimage = 1;
566 ecbdata.lno_in_postimage = 1;
567
568 lc_a = count_lines(data_one, size_one);
569 lc_b = count_lines(data_two, size_two);
570 fprintf(o->file,
571 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
572 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
573 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
574 line_prefix, fraginfo);
575 print_line_count(o->file, lc_a);
576 fprintf(o->file, " +");
577 print_line_count(o->file, lc_b);
578 fprintf(o->file, " @@%s\n", reset);
579 if (lc_a)
580 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
581 if (lc_b)
582 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
583 if (textconv_one)
584 free((char *)data_one);
585 if (textconv_two)
586 free((char *)data_two);
587 }
588
589 struct diff_words_buffer {
590 mmfile_t text;
591 long alloc;
592 struct diff_words_orig {
593 const char *begin, *end;
594 } *orig;
595 int orig_nr, orig_alloc;
596 };
597
598 static void diff_words_append(char *line, unsigned long len,
599 struct diff_words_buffer *buffer)
600 {
601 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
602 line++;
603 len--;
604 memcpy(buffer->text.ptr + buffer->text.size, line, len);
605 buffer->text.size += len;
606 buffer->text.ptr[buffer->text.size] = '\0';
607 }
608
609 struct diff_words_style_elem
610 {
611 const char *prefix;
612 const char *suffix;
613 const char *color; /* NULL; filled in by the setup code if
614 * color is enabled */
615 };
616
617 struct diff_words_style
618 {
619 enum diff_words_type type;
620 struct diff_words_style_elem new, old, ctx;
621 const char *newline;
622 };
623
624 struct diff_words_style diff_words_styles[] = {
625 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
626 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
627 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
628 };
629
630 struct diff_words_data {
631 struct diff_words_buffer minus, plus;
632 const char *current_plus;
633 int last_minus;
634 struct diff_options *opt;
635 regex_t *word_regex;
636 enum diff_words_type type;
637 struct diff_words_style *style;
638 };
639
640 static int fn_out_diff_words_write_helper(FILE *fp,
641 struct diff_words_style_elem *st_el,
642 const char *newline,
643 size_t count, const char *buf,
644 const char *line_prefix)
645 {
646 int print = 0;
647
648 while (count) {
649 char *p = memchr(buf, '\n', count);
650 if (print)
651 fputs(line_prefix, fp);
652 if (p != buf) {
653 if (st_el->color && fputs(st_el->color, fp) < 0)
654 return -1;
655 if (fputs(st_el->prefix, fp) < 0 ||
656 fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
657 fputs(st_el->suffix, fp) < 0)
658 return -1;
659 if (st_el->color && *st_el->color
660 && fputs(GIT_COLOR_RESET, fp) < 0)
661 return -1;
662 }
663 if (!p)
664 return 0;
665 if (fputs(newline, fp) < 0)
666 return -1;
667 count -= p + 1 - buf;
668 buf = p + 1;
669 print = 1;
670 }
671 return 0;
672 }
673
674 /*
675 * '--color-words' algorithm can be described as:
676 *
677 * 1. collect a the minus/plus lines of a diff hunk, divided into
678 * minus-lines and plus-lines;
679 *
680 * 2. break both minus-lines and plus-lines into words and
681 * place them into two mmfile_t with one word for each line;
682 *
683 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
684 *
685 * And for the common parts of the both file, we output the plus side text.
686 * diff_words->current_plus is used to trace the current position of the plus file
687 * which printed. diff_words->last_minus is used to trace the last minus word
688 * printed.
689 *
690 * For '--graph' to work with '--color-words', we need to output the graph prefix
691 * on each line of color words output. Generally, there are two conditions on
692 * which we should output the prefix.
693 *
694 * 1. diff_words->last_minus == 0 &&
695 * diff_words->current_plus == diff_words->plus.text.ptr
696 *
697 * that is: the plus text must start as a new line, and if there is no minus
698 * word printed, a graph prefix must be printed.
699 *
700 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
701 * *(diff_words->current_plus - 1) == '\n'
702 *
703 * that is: a graph prefix must be printed following a '\n'
704 */
705 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
706 {
707 if ((diff_words->last_minus == 0 &&
708 diff_words->current_plus == diff_words->plus.text.ptr) ||
709 (diff_words->current_plus > diff_words->plus.text.ptr &&
710 *(diff_words->current_plus - 1) == '\n')) {
711 return 1;
712 } else {
713 return 0;
714 }
715 }
716
717 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
718 {
719 struct diff_words_data *diff_words = priv;
720 struct diff_words_style *style = diff_words->style;
721 int minus_first, minus_len, plus_first, plus_len;
722 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
723 struct diff_options *opt = diff_words->opt;
724 struct strbuf *msgbuf;
725 char *line_prefix = "";
726
727 if (line[0] != '@' || parse_hunk_header(line, len,
728 &minus_first, &minus_len, &plus_first, &plus_len))
729 return;
730
731 assert(opt);
732 if (opt->output_prefix) {
733 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
734 line_prefix = msgbuf->buf;
735 }
736
737 /* POSIX requires that first be decremented by one if len == 0... */
738 if (minus_len) {
739 minus_begin = diff_words->minus.orig[minus_first].begin;
740 minus_end =
741 diff_words->minus.orig[minus_first + minus_len - 1].end;
742 } else
743 minus_begin = minus_end =
744 diff_words->minus.orig[minus_first].end;
745
746 if (plus_len) {
747 plus_begin = diff_words->plus.orig[plus_first].begin;
748 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
749 } else
750 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
751
752 if (color_words_output_graph_prefix(diff_words)) {
753 fputs(line_prefix, diff_words->opt->file);
754 }
755 if (diff_words->current_plus != plus_begin) {
756 fn_out_diff_words_write_helper(diff_words->opt->file,
757 &style->ctx, style->newline,
758 plus_begin - diff_words->current_plus,
759 diff_words->current_plus, line_prefix);
760 if (*(plus_begin - 1) == '\n')
761 fputs(line_prefix, diff_words->opt->file);
762 }
763 if (minus_begin != minus_end) {
764 fn_out_diff_words_write_helper(diff_words->opt->file,
765 &style->old, style->newline,
766 minus_end - minus_begin, minus_begin,
767 line_prefix);
768 }
769 if (plus_begin != plus_end) {
770 fn_out_diff_words_write_helper(diff_words->opt->file,
771 &style->new, style->newline,
772 plus_end - plus_begin, plus_begin,
773 line_prefix);
774 }
775
776 diff_words->current_plus = plus_end;
777 diff_words->last_minus = minus_first;
778 }
779
780 /* This function starts looking at *begin, and returns 0 iff a word was found. */
781 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
782 int *begin, int *end)
783 {
784 if (word_regex && *begin < buffer->size) {
785 regmatch_t match[1];
786 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
787 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
788 '\n', match[0].rm_eo - match[0].rm_so);
789 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
790 *begin += match[0].rm_so;
791 return *begin >= *end;
792 }
793 return -1;
794 }
795
796 /* find the next word */
797 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
798 (*begin)++;
799 if (*begin >= buffer->size)
800 return -1;
801
802 /* find the end of the word */
803 *end = *begin + 1;
804 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
805 (*end)++;
806
807 return 0;
808 }
809
810 /*
811 * This function splits the words in buffer->text, stores the list with
812 * newline separator into out, and saves the offsets of the original words
813 * in buffer->orig.
814 */
815 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
816 regex_t *word_regex)
817 {
818 int i, j;
819 long alloc = 0;
820
821 out->size = 0;
822 out->ptr = NULL;
823
824 /* fake an empty "0th" word */
825 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
826 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
827 buffer->orig_nr = 1;
828
829 for (i = 0; i < buffer->text.size; i++) {
830 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
831 return;
832
833 /* store original boundaries */
834 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
835 buffer->orig_alloc);
836 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
837 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
838 buffer->orig_nr++;
839
840 /* store one word */
841 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
842 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
843 out->ptr[out->size + j - i] = '\n';
844 out->size += j - i + 1;
845
846 i = j - 1;
847 }
848 }
849
850 /* this executes the word diff on the accumulated buffers */
851 static void diff_words_show(struct diff_words_data *diff_words)
852 {
853 xpparam_t xpp;
854 xdemitconf_t xecfg;
855 mmfile_t minus, plus;
856 struct diff_words_style *style = diff_words->style;
857
858 struct diff_options *opt = diff_words->opt;
859 struct strbuf *msgbuf;
860 char *line_prefix = "";
861
862 assert(opt);
863 if (opt->output_prefix) {
864 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
865 line_prefix = msgbuf->buf;
866 }
867
868 /* special case: only removal */
869 if (!diff_words->plus.text.size) {
870 fputs(line_prefix, diff_words->opt->file);
871 fn_out_diff_words_write_helper(diff_words->opt->file,
872 &style->old, style->newline,
873 diff_words->minus.text.size,
874 diff_words->minus.text.ptr, line_prefix);
875 diff_words->minus.text.size = 0;
876 return;
877 }
878
879 diff_words->current_plus = diff_words->plus.text.ptr;
880 diff_words->last_minus = 0;
881
882 memset(&xpp, 0, sizeof(xpp));
883 memset(&xecfg, 0, sizeof(xecfg));
884 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
885 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
886 xpp.flags = 0;
887 /* as only the hunk header will be parsed, we need a 0-context */
888 xecfg.ctxlen = 0;
889 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
890 &xpp, &xecfg);
891 free(minus.ptr);
892 free(plus.ptr);
893 if (diff_words->current_plus != diff_words->plus.text.ptr +
894 diff_words->plus.text.size) {
895 if (color_words_output_graph_prefix(diff_words))
896 fputs(line_prefix, diff_words->opt->file);
897 fn_out_diff_words_write_helper(diff_words->opt->file,
898 &style->ctx, style->newline,
899 diff_words->plus.text.ptr + diff_words->plus.text.size
900 - diff_words->current_plus, diff_words->current_plus,
901 line_prefix);
902 }
903 diff_words->minus.text.size = diff_words->plus.text.size = 0;
904 }
905
906 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
907 static void diff_words_flush(struct emit_callback *ecbdata)
908 {
909 if (ecbdata->diff_words->minus.text.size ||
910 ecbdata->diff_words->plus.text.size)
911 diff_words_show(ecbdata->diff_words);
912 }
913
914 static void free_diff_words_data(struct emit_callback *ecbdata)
915 {
916 if (ecbdata->diff_words) {
917 diff_words_flush(ecbdata);
918 free (ecbdata->diff_words->minus.text.ptr);
919 free (ecbdata->diff_words->minus.orig);
920 free (ecbdata->diff_words->plus.text.ptr);
921 free (ecbdata->diff_words->plus.orig);
922 if (ecbdata->diff_words->word_regex) {
923 regfree(ecbdata->diff_words->word_regex);
924 free(ecbdata->diff_words->word_regex);
925 }
926 free(ecbdata->diff_words);
927 ecbdata->diff_words = NULL;
928 }
929 }
930
931 const char *diff_get_color(int diff_use_color, enum color_diff ix)
932 {
933 if (diff_use_color)
934 return diff_colors[ix];
935 return "";
936 }
937
938 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
939 {
940 const char *cp;
941 unsigned long allot;
942 size_t l = len;
943
944 if (ecb->truncate)
945 return ecb->truncate(line, len);
946 cp = line;
947 allot = l;
948 while (0 < l) {
949 (void) utf8_width(&cp, &l);
950 if (!cp)
951 break; /* truncated in the middle? */
952 }
953 return allot - l;
954 }
955
956 static void find_lno(const char *line, struct emit_callback *ecbdata)
957 {
958 const char *p;
959 ecbdata->lno_in_preimage = 0;
960 ecbdata->lno_in_postimage = 0;
961 p = strchr(line, '-');
962 if (!p)
963 return; /* cannot happen */
964 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
965 p = strchr(p, '+');
966 if (!p)
967 return; /* cannot happen */
968 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
969 }
970
971 static void fn_out_consume(void *priv, char *line, unsigned long len)
972 {
973 struct emit_callback *ecbdata = priv;
974 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
975 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
976 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
977 struct diff_options *o = ecbdata->opt;
978 char *line_prefix = "";
979 struct strbuf *msgbuf;
980
981 if (o && o->output_prefix) {
982 msgbuf = o->output_prefix(o, o->output_prefix_data);
983 line_prefix = msgbuf->buf;
984 }
985
986 if (ecbdata->header) {
987 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
988 strbuf_reset(ecbdata->header);
989 ecbdata->header = NULL;
990 }
991 *(ecbdata->found_changesp) = 1;
992
993 if (ecbdata->label_path[0]) {
994 const char *name_a_tab, *name_b_tab;
995
996 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
997 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
998
999 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1000 line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1001 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1002 line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1003 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1004 }
1005
1006 if (diff_suppress_blank_empty
1007 && len == 2 && line[0] == ' ' && line[1] == '\n') {
1008 line[0] = '\n';
1009 len = 1;
1010 }
1011
1012 if (line[0] == '@') {
1013 if (ecbdata->diff_words)
1014 diff_words_flush(ecbdata);
1015 len = sane_truncate_line(ecbdata, line, len);
1016 find_lno(line, ecbdata);
1017 emit_hunk_header(ecbdata, line, len);
1018 if (line[len-1] != '\n')
1019 putc('\n', ecbdata->opt->file);
1020 return;
1021 }
1022
1023 if (len < 1) {
1024 emit_line(ecbdata->opt, reset, reset, line, len);
1025 if (ecbdata->diff_words
1026 && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1027 fputs("~\n", ecbdata->opt->file);
1028 return;
1029 }
1030
1031 if (ecbdata->diff_words) {
1032 if (line[0] == '-') {
1033 diff_words_append(line, len,
1034 &ecbdata->diff_words->minus);
1035 return;
1036 } else if (line[0] == '+') {
1037 diff_words_append(line, len,
1038 &ecbdata->diff_words->plus);
1039 return;
1040 }
1041 diff_words_flush(ecbdata);
1042 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1043 emit_line(ecbdata->opt, plain, reset, line, len);
1044 fputs("~\n", ecbdata->opt->file);
1045 } else {
1046 /* don't print the prefix character */
1047 emit_line(ecbdata->opt, plain, reset, line+1, len-1);
1048 }
1049 return;
1050 }
1051
1052 if (line[0] != '+') {
1053 const char *color =
1054 diff_get_color(ecbdata->color_diff,
1055 line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1056 ecbdata->lno_in_preimage++;
1057 if (line[0] == ' ')
1058 ecbdata->lno_in_postimage++;
1059 emit_line(ecbdata->opt, color, reset, line, len);
1060 } else {
1061 ecbdata->lno_in_postimage++;
1062 emit_add_line(reset, ecbdata, line + 1, len - 1);
1063 }
1064 }
1065
1066 static char *pprint_rename(const char *a, const char *b)
1067 {
1068 const char *old = a;
1069 const char *new = b;
1070 struct strbuf name = STRBUF_INIT;
1071 int pfx_length, sfx_length;
1072 int len_a = strlen(a);
1073 int len_b = strlen(b);
1074 int a_midlen, b_midlen;
1075 int qlen_a = quote_c_style(a, NULL, NULL, 0);
1076 int qlen_b = quote_c_style(b, NULL, NULL, 0);
1077
1078 if (qlen_a || qlen_b) {
1079 quote_c_style(a, &name, NULL, 0);
1080 strbuf_addstr(&name, " => ");
1081 quote_c_style(b, &name, NULL, 0);
1082 return strbuf_detach(&name, NULL);
1083 }
1084
1085 /* Find common prefix */
1086 pfx_length = 0;
1087 while (*old && *new && *old == *new) {
1088 if (*old == '/')
1089 pfx_length = old - a + 1;
1090 old++;
1091 new++;
1092 }
1093
1094 /* Find common suffix */
1095 old = a + len_a;
1096 new = b + len_b;
1097 sfx_length = 0;
1098 while (a <= old && b <= new && *old == *new) {
1099 if (*old == '/')
1100 sfx_length = len_a - (old - a);
1101 old--;
1102 new--;
1103 }
1104
1105 /*
1106 * pfx{mid-a => mid-b}sfx
1107 * {pfx-a => pfx-b}sfx
1108 * pfx{sfx-a => sfx-b}
1109 * name-a => name-b
1110 */
1111 a_midlen = len_a - pfx_length - sfx_length;
1112 b_midlen = len_b - pfx_length - sfx_length;
1113 if (a_midlen < 0)
1114 a_midlen = 0;
1115 if (b_midlen < 0)
1116 b_midlen = 0;
1117
1118 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1119 if (pfx_length + sfx_length) {
1120 strbuf_add(&name, a, pfx_length);
1121 strbuf_addch(&name, '{');
1122 }
1123 strbuf_add(&name, a + pfx_length, a_midlen);
1124 strbuf_addstr(&name, " => ");
1125 strbuf_add(&name, b + pfx_length, b_midlen);
1126 if (pfx_length + sfx_length) {
1127 strbuf_addch(&name, '}');
1128 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1129 }
1130 return strbuf_detach(&name, NULL);
1131 }
1132
1133 struct diffstat_t {
1134 int nr;
1135 int alloc;
1136 struct diffstat_file {
1137 char *from_name;
1138 char *name;
1139 char *print_name;
1140 unsigned is_unmerged:1;
1141 unsigned is_binary:1;
1142 unsigned is_renamed:1;
1143 uintmax_t added, deleted;
1144 } **files;
1145 };
1146
1147 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1148 const char *name_a,
1149 const char *name_b)
1150 {
1151 struct diffstat_file *x;
1152 x = xcalloc(sizeof (*x), 1);
1153 if (diffstat->nr == diffstat->alloc) {
1154 diffstat->alloc = alloc_nr(diffstat->alloc);
1155 diffstat->files = xrealloc(diffstat->files,
1156 diffstat->alloc * sizeof(x));
1157 }
1158 diffstat->files[diffstat->nr++] = x;
1159 if (name_b) {
1160 x->from_name = xstrdup(name_a);
1161 x->name = xstrdup(name_b);
1162 x->is_renamed = 1;
1163 }
1164 else {
1165 x->from_name = NULL;
1166 x->name = xstrdup(name_a);
1167 }
1168 return x;
1169 }
1170
1171 static void diffstat_consume(void *priv, char *line, unsigned long len)
1172 {
1173 struct diffstat_t *diffstat = priv;
1174 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1175
1176 if (line[0] == '+')
1177 x->added++;
1178 else if (line[0] == '-')
1179 x->deleted++;
1180 }
1181
1182 const char mime_boundary_leader[] = "------------";
1183
1184 static int scale_linear(int it, int width, int max_change)
1185 {
1186 /*
1187 * make sure that at least one '-' is printed if there were deletions,
1188 * and likewise for '+'.
1189 */
1190 if (max_change < 2)
1191 return it;
1192 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1193 }
1194
1195 static void show_name(FILE *file,
1196 const char *prefix, const char *name, int len)
1197 {
1198 fprintf(file, " %s%-*s |", prefix, len, name);
1199 }
1200
1201 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1202 {
1203 if (cnt <= 0)
1204 return;
1205 fprintf(file, "%s", set);
1206 while (cnt--)
1207 putc(ch, file);
1208 fprintf(file, "%s", reset);
1209 }
1210
1211 static void fill_print_name(struct diffstat_file *file)
1212 {
1213 char *pname;
1214
1215 if (file->print_name)
1216 return;
1217
1218 if (!file->is_renamed) {
1219 struct strbuf buf = STRBUF_INIT;
1220 if (quote_c_style(file->name, &buf, NULL, 0)) {
1221 pname = strbuf_detach(&buf, NULL);
1222 } else {
1223 pname = file->name;
1224 strbuf_release(&buf);
1225 }
1226 } else {
1227 pname = pprint_rename(file->from_name, file->name);
1228 }
1229 file->print_name = pname;
1230 }
1231
1232 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1233 {
1234 int i, len, add, del, adds = 0, dels = 0;
1235 uintmax_t max_change = 0, max_len = 0;
1236 int total_files = data->nr;
1237 int width, name_width;
1238 const char *reset, *set, *add_c, *del_c;
1239 const char *line_prefix = "";
1240 struct strbuf *msg = NULL;
1241
1242 if (data->nr == 0)
1243 return;
1244
1245 if (options->output_prefix) {
1246 msg = options->output_prefix(options, options->output_prefix_data);
1247 line_prefix = msg->buf;
1248 }
1249
1250 width = options->stat_width ? options->stat_width : 80;
1251 name_width = options->stat_name_width ? options->stat_name_width : 50;
1252
1253 /* Sanity: give at least 5 columns to the graph,
1254 * but leave at least 10 columns for the name.
1255 */
1256 if (width < 25)
1257 width = 25;
1258 if (name_width < 10)
1259 name_width = 10;
1260 else if (width < name_width + 15)
1261 name_width = width - 15;
1262
1263 /* Find the longest filename and max number of changes */
1264 reset = diff_get_color_opt(options, DIFF_RESET);
1265 set = diff_get_color_opt(options, DIFF_PLAIN);
1266 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1267 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1268
1269 for (i = 0; i < data->nr; i++) {
1270 struct diffstat_file *file = data->files[i];
1271 uintmax_t change = file->added + file->deleted;
1272 fill_print_name(file);
1273 len = strlen(file->print_name);
1274 if (max_len < len)
1275 max_len = len;
1276
1277 if (file->is_binary || file->is_unmerged)
1278 continue;
1279 if (max_change < change)
1280 max_change = change;
1281 }
1282
1283 /* Compute the width of the graph part;
1284 * 10 is for one blank at the beginning of the line plus
1285 * " | count " between the name and the graph.
1286 *
1287 * From here on, name_width is the width of the name area,
1288 * and width is the width of the graph area.
1289 */
1290 name_width = (name_width < max_len) ? name_width : max_len;
1291 if (width < (name_width + 10) + max_change)
1292 width = width - (name_width + 10);
1293 else
1294 width = max_change;
1295
1296 for (i = 0; i < data->nr; i++) {
1297 const char *prefix = "";
1298 char *name = data->files[i]->print_name;
1299 uintmax_t added = data->files[i]->added;
1300 uintmax_t deleted = data->files[i]->deleted;
1301 int name_len;
1302
1303 /*
1304 * "scale" the filename
1305 */
1306 len = name_width;
1307 name_len = strlen(name);
1308 if (name_width < name_len) {
1309 char *slash;
1310 prefix = "...";
1311 len -= 3;
1312 name += name_len - len;
1313 slash = strchr(name, '/');
1314 if (slash)
1315 name = slash;
1316 }
1317
1318 if (data->files[i]->is_binary) {
1319 fprintf(options->file, "%s", line_prefix);
1320 show_name(options->file, prefix, name, len);
1321 fprintf(options->file, " Bin ");
1322 fprintf(options->file, "%s%"PRIuMAX"%s",
1323 del_c, deleted, reset);
1324 fprintf(options->file, " -> ");
1325 fprintf(options->file, "%s%"PRIuMAX"%s",
1326 add_c, added, reset);
1327 fprintf(options->file, " bytes");
1328 fprintf(options->file, "\n");
1329 continue;
1330 }
1331 else if (data->files[i]->is_unmerged) {
1332 fprintf(options->file, "%s", line_prefix);
1333 show_name(options->file, prefix, name, len);
1334 fprintf(options->file, " Unmerged\n");
1335 continue;
1336 }
1337 else if (!data->files[i]->is_renamed &&
1338 (added + deleted == 0)) {
1339 total_files--;
1340 continue;
1341 }
1342
1343 /*
1344 * scale the add/delete
1345 */
1346 add = added;
1347 del = deleted;
1348 adds += add;
1349 dels += del;
1350
1351 if (width <= max_change) {
1352 add = scale_linear(add, width, max_change);
1353 del = scale_linear(del, width, max_change);
1354 }
1355 fprintf(options->file, "%s", line_prefix);
1356 show_name(options->file, prefix, name, len);
1357 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1358 added + deleted ? " " : "");
1359 show_graph(options->file, '+', add, add_c, reset);
1360 show_graph(options->file, '-', del, del_c, reset);
1361 fprintf(options->file, "\n");
1362 }
1363 fprintf(options->file, "%s", line_prefix);
1364 fprintf(options->file,
1365 " %d files changed, %d insertions(+), %d deletions(-)\n",
1366 total_files, adds, dels);
1367 }
1368
1369 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1370 {
1371 int i, adds = 0, dels = 0, total_files = data->nr;
1372
1373 if (data->nr == 0)
1374 return;
1375
1376 for (i = 0; i < data->nr; i++) {
1377 if (!data->files[i]->is_binary &&
1378 !data->files[i]->is_unmerged) {
1379 int added = data->files[i]->added;
1380 int deleted= data->files[i]->deleted;
1381 if (!data->files[i]->is_renamed &&
1382 (added + deleted == 0)) {
1383 total_files--;
1384 } else {
1385 adds += added;
1386 dels += deleted;
1387 }
1388 }
1389 }
1390 if (options->output_prefix) {
1391 struct strbuf *msg = NULL;
1392 msg = options->output_prefix(options,
1393 options->output_prefix_data);
1394 fprintf(options->file, "%s", msg->buf);
1395 }
1396 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1397 total_files, adds, dels);
1398 }
1399
1400 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1401 {
1402 int i;
1403
1404 if (data->nr == 0)
1405 return;
1406
1407 for (i = 0; i < data->nr; i++) {
1408 struct diffstat_file *file = data->files[i];
1409
1410 if (options->output_prefix) {
1411 struct strbuf *msg = NULL;
1412 msg = options->output_prefix(options,
1413 options->output_prefix_data);
1414 fprintf(options->file, "%s", msg->buf);
1415 }
1416
1417 if (file->is_binary)
1418 fprintf(options->file, "-\t-\t");
1419 else
1420 fprintf(options->file,
1421 "%"PRIuMAX"\t%"PRIuMAX"\t",
1422 file->added, file->deleted);
1423 if (options->line_termination) {
1424 fill_print_name(file);
1425 if (!file->is_renamed)
1426 write_name_quoted(file->name, options->file,
1427 options->line_termination);
1428 else {
1429 fputs(file->print_name, options->file);
1430 putc(options->line_termination, options->file);
1431 }
1432 } else {
1433 if (file->is_renamed) {
1434 putc('\0', options->file);
1435 write_name_quoted(file->from_name, options->file, '\0');
1436 }
1437 write_name_quoted(file->name, options->file, '\0');
1438 }
1439 }
1440 }
1441
1442 struct dirstat_file {
1443 const char *name;
1444 unsigned long changed;
1445 };
1446
1447 struct dirstat_dir {
1448 struct dirstat_file *files;
1449 int alloc, nr, percent, cumulative;
1450 };
1451
1452 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1453 unsigned long changed, const char *base, int baselen)
1454 {
1455 unsigned long this_dir = 0;
1456 unsigned int sources = 0;
1457 const char *line_prefix = "";
1458 struct strbuf *msg = NULL;
1459
1460 if (opt->output_prefix) {
1461 msg = opt->output_prefix(opt, opt->output_prefix_data);
1462 line_prefix = msg->buf;
1463 }
1464
1465 while (dir->nr) {
1466 struct dirstat_file *f = dir->files;
1467 int namelen = strlen(f->name);
1468 unsigned long this;
1469 char *slash;
1470
1471 if (namelen < baselen)
1472 break;
1473 if (memcmp(f->name, base, baselen))
1474 break;
1475 slash = strchr(f->name + baselen, '/');
1476 if (slash) {
1477 int newbaselen = slash + 1 - f->name;
1478 this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1479 sources++;
1480 } else {
1481 this = f->changed;
1482 dir->files++;
1483 dir->nr--;
1484 sources += 2;
1485 }
1486 this_dir += this;
1487 }
1488
1489 /*
1490 * We don't report dirstat's for
1491 * - the top level
1492 * - or cases where everything came from a single directory
1493 * under this directory (sources == 1).
1494 */
1495 if (baselen && sources != 1) {
1496 int permille = this_dir * 1000 / changed;
1497 if (permille) {
1498 int percent = permille / 10;
1499 if (percent >= dir->percent) {
1500 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1501 percent, permille % 10, baselen, base);
1502 if (!dir->cumulative)
1503 return 0;
1504 }
1505 }
1506 }
1507 return this_dir;
1508 }
1509
1510 static int dirstat_compare(const void *_a, const void *_b)
1511 {
1512 const struct dirstat_file *a = _a;
1513 const struct dirstat_file *b = _b;
1514 return strcmp(a->name, b->name);
1515 }
1516
1517 static void show_dirstat(struct diff_options *options)
1518 {
1519 int i;
1520 unsigned long changed;
1521 struct dirstat_dir dir;
1522 struct diff_queue_struct *q = &diff_queued_diff;
1523
1524 dir.files = NULL;
1525 dir.alloc = 0;
1526 dir.nr = 0;
1527 dir.percent = options->dirstat_percent;
1528 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1529
1530 changed = 0;
1531 for (i = 0; i < q->nr; i++) {
1532 struct diff_filepair *p = q->queue[i];
1533 const char *name;
1534 unsigned long copied, added, damage;
1535
1536 name = p->one->path ? p->one->path : p->two->path;
1537
1538 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1539 diff_populate_filespec(p->one, 0);
1540 diff_populate_filespec(p->two, 0);
1541 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1542 &copied, &added);
1543 diff_free_filespec_data(p->one);
1544 diff_free_filespec_data(p->two);
1545 } else if (DIFF_FILE_VALID(p->one)) {
1546 diff_populate_filespec(p->one, 1);
1547 copied = added = 0;
1548 diff_free_filespec_data(p->one);
1549 } else if (DIFF_FILE_VALID(p->two)) {
1550 diff_populate_filespec(p->two, 1);
1551 copied = 0;
1552 added = p->two->size;
1553 diff_free_filespec_data(p->two);
1554 } else
1555 continue;
1556
1557 /*
1558 * Original minus copied is the removed material,
1559 * added is the new material. They are both damages
1560 * made to the preimage. In --dirstat-by-file mode, count
1561 * damaged files, not damaged lines. This is done by
1562 * counting only a single damaged line per file.
1563 */
1564 damage = (p->one->size - copied) + added;
1565 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1566 damage = 1;
1567
1568 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1569 dir.files[dir.nr].name = name;
1570 dir.files[dir.nr].changed = damage;
1571 changed += damage;
1572 dir.nr++;
1573 }
1574
1575 /* This can happen even with many files, if everything was renames */
1576 if (!changed)
1577 return;
1578
1579 /* Show all directories with more than x% of the changes */
1580 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1581 gather_dirstat(options, &dir, changed, "", 0);
1582 }
1583
1584 static void free_diffstat_info(struct diffstat_t *diffstat)
1585 {
1586 int i;
1587 for (i = 0; i < diffstat->nr; i++) {
1588 struct diffstat_file *f = diffstat->files[i];
1589 if (f->name != f->print_name)
1590 free(f->print_name);
1591 free(f->name);
1592 free(f->from_name);
1593 free(f);
1594 }
1595 free(diffstat->files);
1596 }
1597
1598 struct checkdiff_t {
1599 const char *filename;
1600 int lineno;
1601 int conflict_marker_size;
1602 struct diff_options *o;
1603 unsigned ws_rule;
1604 unsigned status;
1605 };
1606
1607 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1608 {
1609 char firstchar;
1610 int cnt;
1611
1612 if (len < marker_size + 1)
1613 return 0;
1614 firstchar = line[0];
1615 switch (firstchar) {
1616 case '=': case '>': case '<': case '|':
1617 break;
1618 default:
1619 return 0;
1620 }
1621 for (cnt = 1; cnt < marker_size; cnt++)
1622 if (line[cnt] != firstchar)
1623 return 0;
1624 /* line[1] thru line[marker_size-1] are same as firstchar */
1625 if (len < marker_size + 1 || !isspace(line[marker_size]))
1626 return 0;
1627 return 1;
1628 }
1629
1630 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1631 {
1632 struct checkdiff_t *data = priv;
1633 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1634 int marker_size = data->conflict_marker_size;
1635 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1636 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1637 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1638 char *err;
1639 char *line_prefix = "";
1640 struct strbuf *msgbuf;
1641
1642 assert(data->o);
1643 if (data->o->output_prefix) {
1644 msgbuf = data->o->output_prefix(data->o,
1645 data->o->output_prefix_data);
1646 line_prefix = msgbuf->buf;
1647 }
1648
1649 if (line[0] == '+') {
1650 unsigned bad;
1651 data->lineno++;
1652 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1653 data->status |= 1;
1654 fprintf(data->o->file,
1655 "%s%s:%d: leftover conflict marker\n",
1656 line_prefix, data->filename, data->lineno);
1657 }
1658 bad = ws_check(line + 1, len - 1, data->ws_rule);
1659 if (!bad)
1660 return;
1661 data->status |= bad;
1662 err = whitespace_error_string(bad);
1663 fprintf(data->o->file, "%s%s:%d: %s.\n",
1664 line_prefix, data->filename, data->lineno, err);
1665 free(err);
1666 emit_line(data->o, set, reset, line, 1);
1667 ws_check_emit(line + 1, len - 1, data->ws_rule,
1668 data->o->file, set, reset, ws);
1669 } else if (line[0] == ' ') {
1670 data->lineno++;
1671 } else if (line[0] == '@') {
1672 char *plus = strchr(line, '+');
1673 if (plus)
1674 data->lineno = strtol(plus, NULL, 10) - 1;
1675 else
1676 die("invalid diff");
1677 }
1678 }
1679
1680 static unsigned char *deflate_it(char *data,
1681 unsigned long size,
1682 unsigned long *result_size)
1683 {
1684 int bound;
1685 unsigned char *deflated;
1686 z_stream stream;
1687
1688 memset(&stream, 0, sizeof(stream));
1689 deflateInit(&stream, zlib_compression_level);
1690 bound = deflateBound(&stream, size);
1691 deflated = xmalloc(bound);
1692 stream.next_out = deflated;
1693 stream.avail_out = bound;
1694
1695 stream.next_in = (unsigned char *)data;
1696 stream.avail_in = size;
1697 while (deflate(&stream, Z_FINISH) == Z_OK)
1698 ; /* nothing */
1699 deflateEnd(&stream);
1700 *result_size = stream.total_out;
1701 return deflated;
1702 }
1703
1704 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1705 {
1706 void *cp;
1707 void *delta;
1708 void *deflated;
1709 void *data;
1710 unsigned long orig_size;
1711 unsigned long delta_size;
1712 unsigned long deflate_size;
1713 unsigned long data_size;
1714
1715 /* We could do deflated delta, or we could do just deflated two,
1716 * whichever is smaller.
1717 */
1718 delta = NULL;
1719 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1720 if (one->size && two->size) {
1721 delta = diff_delta(one->ptr, one->size,
1722 two->ptr, two->size,
1723 &delta_size, deflate_size);
1724 if (delta) {
1725 void *to_free = delta;
1726 orig_size = delta_size;
1727 delta = deflate_it(delta, delta_size, &delta_size);
1728 free(to_free);
1729 }
1730 }
1731
1732 if (delta && delta_size < deflate_size) {
1733 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
1734 free(deflated);
1735 data = delta;
1736 data_size = delta_size;
1737 }
1738 else {
1739 fprintf(file, "%sliteral %lu\n", prefix, two->size);
1740 free(delta);
1741 data = deflated;
1742 data_size = deflate_size;
1743 }
1744
1745 /* emit data encoded in base85 */
1746 cp = data;
1747 while (data_size) {
1748 int bytes = (52 < data_size) ? 52 : data_size;
1749 char line[70];
1750 data_size -= bytes;
1751 if (bytes <= 26)
1752 line[0] = bytes + 'A' - 1;
1753 else
1754 line[0] = bytes - 26 + 'a' - 1;
1755 encode_85(line + 1, cp, bytes);
1756 cp = (char *) cp + bytes;
1757 fprintf(file, "%s", prefix);
1758 fputs(line, file);
1759 fputc('\n', file);
1760 }
1761 fprintf(file, "%s\n", prefix);
1762 free(data);
1763 }
1764
1765 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1766 {
1767 fprintf(file, "%sGIT binary patch\n", prefix);
1768 emit_binary_diff_body(file, one, two, prefix);
1769 emit_binary_diff_body(file, two, one, prefix);
1770 }
1771
1772 static void diff_filespec_load_driver(struct diff_filespec *one)
1773 {
1774 if (!one->driver)
1775 one->driver = userdiff_find_by_path(one->path);
1776 if (!one->driver)
1777 one->driver = userdiff_find_by_name("default");
1778 }
1779
1780 int diff_filespec_is_binary(struct diff_filespec *one)
1781 {
1782 if (one->is_binary == -1) {
1783 diff_filespec_load_driver(one);
1784 if (one->driver->binary != -1)
1785 one->is_binary = one->driver->binary;
1786 else {
1787 if (!one->data && DIFF_FILE_VALID(one))
1788 diff_populate_filespec(one, 0);
1789 if (one->data)
1790 one->is_binary = buffer_is_binary(one->data,
1791 one->size);
1792 if (one->is_binary == -1)
1793 one->is_binary = 0;
1794 }
1795 }
1796 return one->is_binary;
1797 }
1798
1799 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1800 {
1801 diff_filespec_load_driver(one);
1802 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1803 }
1804
1805 static const char *userdiff_word_regex(struct diff_filespec *one)
1806 {
1807 diff_filespec_load_driver(one);
1808 return one->driver->word_regex;
1809 }
1810
1811 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1812 {
1813 if (!options->a_prefix)
1814 options->a_prefix = a;
1815 if (!options->b_prefix)
1816 options->b_prefix = b;
1817 }
1818
1819 struct userdiff_driver *get_textconv(struct diff_filespec *one)
1820 {
1821 if (!DIFF_FILE_VALID(one))
1822 return NULL;
1823 if (!S_ISREG(one->mode))
1824 return NULL;
1825 diff_filespec_load_driver(one);
1826 if (!one->driver->textconv)
1827 return NULL;
1828
1829 if (one->driver->textconv_want_cache && !one->driver->textconv_cache) {
1830 struct notes_cache *c = xmalloc(sizeof(*c));
1831 struct strbuf name = STRBUF_INIT;
1832
1833 strbuf_addf(&name, "textconv/%s", one->driver->name);
1834 notes_cache_init(c, name.buf, one->driver->textconv);
1835 one->driver->textconv_cache = c;
1836 }
1837
1838 return one->driver;
1839 }
1840
1841 static void builtin_diff(const char *name_a,
1842 const char *name_b,
1843 struct diff_filespec *one,
1844 struct diff_filespec *two,
1845 const char *xfrm_msg,
1846 int must_show_header,
1847 struct diff_options *o,
1848 int complete_rewrite)
1849 {
1850 mmfile_t mf1, mf2;
1851 const char *lbl[2];
1852 char *a_one, *b_two;
1853 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1854 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1855 const char *a_prefix, *b_prefix;
1856 struct userdiff_driver *textconv_one = NULL;
1857 struct userdiff_driver *textconv_two = NULL;
1858 struct strbuf header = STRBUF_INIT;
1859 struct strbuf *msgbuf;
1860 char *line_prefix = "";
1861
1862 if (o->output_prefix) {
1863 msgbuf = o->output_prefix(o, o->output_prefix_data);
1864 line_prefix = msgbuf->buf;
1865 }
1866
1867 if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1868 (!one->mode || S_ISGITLINK(one->mode)) &&
1869 (!two->mode || S_ISGITLINK(two->mode))) {
1870 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1871 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1872 show_submodule_summary(o->file, one ? one->path : two->path,
1873 one->sha1, two->sha1, two->dirty_submodule,
1874 del, add, reset);
1875 return;
1876 }
1877
1878 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1879 textconv_one = get_textconv(one);
1880 textconv_two = get_textconv(two);
1881 }
1882
1883 diff_set_mnemonic_prefix(o, "a/", "b/");
1884 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1885 a_prefix = o->b_prefix;
1886 b_prefix = o->a_prefix;
1887 } else {
1888 a_prefix = o->a_prefix;
1889 b_prefix = o->b_prefix;
1890 }
1891
1892 /* Never use a non-valid filename anywhere if at all possible */
1893 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1894 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1895
1896 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1897 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1898 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1899 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1900 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
1901 if (lbl[0][0] == '/') {
1902 /* /dev/null */
1903 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
1904 if (xfrm_msg)
1905 strbuf_addstr(&header, xfrm_msg);
1906 must_show_header = 1;
1907 }
1908 else if (lbl[1][0] == '/') {
1909 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
1910 if (xfrm_msg)
1911 strbuf_addstr(&header, xfrm_msg);
1912 must_show_header = 1;
1913 }
1914 else {
1915 if (one->mode != two->mode) {
1916 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
1917 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
1918 must_show_header = 1;
1919 }
1920 if (xfrm_msg)
1921 strbuf_addstr(&header, xfrm_msg);
1922
1923 /*
1924 * we do not run diff between different kind
1925 * of objects.
1926 */
1927 if ((one->mode ^ two->mode) & S_IFMT)
1928 goto free_ab_and_return;
1929 if (complete_rewrite &&
1930 (textconv_one || !diff_filespec_is_binary(one)) &&
1931 (textconv_two || !diff_filespec_is_binary(two))) {
1932 fprintf(o->file, "%s", header.buf);
1933 strbuf_reset(&header);
1934 emit_rewrite_diff(name_a, name_b, one, two,
1935 textconv_one, textconv_two, o);
1936 o->found_changes = 1;
1937 goto free_ab_and_return;
1938 }
1939 }
1940
1941 if (!DIFF_OPT_TST(o, TEXT) &&
1942 ( (!textconv_one && diff_filespec_is_binary(one)) ||
1943 (!textconv_two && diff_filespec_is_binary(two)) )) {
1944 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1945 die("unable to read files to diff");
1946 /* Quite common confusing case */
1947 if (mf1.size == mf2.size &&
1948 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
1949 if (must_show_header)
1950 fprintf(o->file, "%s", header.buf);
1951 goto free_ab_and_return;
1952 }
1953 fprintf(o->file, "%s", header.buf);
1954 strbuf_reset(&header);
1955 if (DIFF_OPT_TST(o, BINARY))
1956 emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
1957 else
1958 fprintf(o->file, "%sBinary files %s and %s differ\n",
1959 line_prefix, lbl[0], lbl[1]);
1960 o->found_changes = 1;
1961 }
1962 else {
1963 /* Crazy xdl interfaces.. */
1964 const char *diffopts = getenv("GIT_DIFF_OPTS");
1965 xpparam_t xpp;
1966 xdemitconf_t xecfg;
1967 struct emit_callback ecbdata;
1968 const struct userdiff_funcname *pe;
1969
1970 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || must_show_header) {
1971 fprintf(o->file, "%s", header.buf);
1972 strbuf_reset(&header);
1973 }
1974
1975 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
1976 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
1977
1978 pe = diff_funcname_pattern(one);
1979 if (!pe)
1980 pe = diff_funcname_pattern(two);
1981
1982 memset(&xpp, 0, sizeof(xpp));
1983 memset(&xecfg, 0, sizeof(xecfg));
1984 memset(&ecbdata, 0, sizeof(ecbdata));
1985 ecbdata.label_path = lbl;
1986 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1987 ecbdata.found_changesp = &o->found_changes;
1988 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1989 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1990 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1991 ecbdata.opt = o;
1992 ecbdata.header = header.len ? &header : NULL;
1993 xpp.flags = o->xdl_opts;
1994 xecfg.ctxlen = o->context;
1995 xecfg.interhunkctxlen = o->interhunkcontext;
1996 xecfg.flags = XDL_EMIT_FUNCNAMES;
1997 if (pe)
1998 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1999 if (!diffopts)
2000 ;
2001 else if (!prefixcmp(diffopts, "--unified="))
2002 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2003 else if (!prefixcmp(diffopts, "-u"))
2004 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2005 if (o->word_diff) {
2006 int i;
2007
2008 ecbdata.diff_words =
2009 xcalloc(1, sizeof(struct diff_words_data));
2010 ecbdata.diff_words->type = o->word_diff;
2011 ecbdata.diff_words->opt = o;
2012 if (!o->word_regex)
2013 o->word_regex = userdiff_word_regex(one);
2014 if (!o->word_regex)
2015 o->word_regex = userdiff_word_regex(two);
2016 if (!o->word_regex)
2017 o->word_regex = diff_word_regex_cfg;
2018 if (o->word_regex) {
2019 ecbdata.diff_words->word_regex = (regex_t *)
2020 xmalloc(sizeof(regex_t));
2021 if (regcomp(ecbdata.diff_words->word_regex,
2022 o->word_regex,
2023 REG_EXTENDED | REG_NEWLINE))
2024 die ("Invalid regular expression: %s",
2025 o->word_regex);
2026 }
2027 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2028 if (o->word_diff == diff_words_styles[i].type) {
2029 ecbdata.diff_words->style =
2030 &diff_words_styles[i];
2031 break;
2032 }
2033 }
2034 if (DIFF_OPT_TST(o, COLOR_DIFF)) {
2035 struct diff_words_style *st = ecbdata.diff_words->style;
2036 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2037 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2038 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
2039 }
2040 }
2041 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2042 &xpp, &xecfg);
2043 if (o->word_diff)
2044 free_diff_words_data(&ecbdata);
2045 if (textconv_one)
2046 free(mf1.ptr);
2047 if (textconv_two)
2048 free(mf2.ptr);
2049 xdiff_clear_find_func(&xecfg);
2050 }
2051
2052 free_ab_and_return:
2053 strbuf_release(&header);
2054 diff_free_filespec_data(one);
2055 diff_free_filespec_data(two);
2056 free(a_one);
2057 free(b_two);
2058 return;
2059 }
2060
2061 static void builtin_diffstat(const char *name_a, const char *name_b,
2062 struct diff_filespec *one,
2063 struct diff_filespec *two,
2064 struct diffstat_t *diffstat,
2065 struct diff_options *o,
2066 int complete_rewrite)
2067 {
2068 mmfile_t mf1, mf2;
2069 struct diffstat_file *data;
2070
2071 data = diffstat_add(diffstat, name_a, name_b);
2072
2073 if (!one || !two) {
2074 data->is_unmerged = 1;
2075 return;
2076 }
2077 if (complete_rewrite) {
2078 diff_populate_filespec(one, 0);
2079 diff_populate_filespec(two, 0);
2080 data->deleted = count_lines(one->data, one->size);
2081 data->added = count_lines(two->data, two->size);
2082 goto free_and_return;
2083 }
2084 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2085 die("unable to read files to diff");
2086
2087 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2088 data->is_binary = 1;
2089 data->added = mf2.size;
2090 data->deleted = mf1.size;
2091 } else {
2092 /* Crazy xdl interfaces.. */
2093 xpparam_t xpp;
2094 xdemitconf_t xecfg;
2095
2096 memset(&xpp, 0, sizeof(xpp));
2097 memset(&xecfg, 0, sizeof(xecfg));
2098 xpp.flags = o->xdl_opts;
2099 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2100 &xpp, &xecfg);
2101 }
2102
2103 free_and_return:
2104 diff_free_filespec_data(one);
2105 diff_free_filespec_data(two);
2106 }
2107
2108 static void builtin_checkdiff(const char *name_a, const char *name_b,
2109 const char *attr_path,
2110 struct diff_filespec *one,
2111 struct diff_filespec *two,
2112 struct diff_options *o)
2113 {
2114 mmfile_t mf1, mf2;
2115 struct checkdiff_t data;
2116
2117 if (!two)
2118 return;
2119
2120 memset(&data, 0, sizeof(data));
2121 data.filename = name_b ? name_b : name_a;
2122 data.lineno = 0;
2123 data.o = o;
2124 data.ws_rule = whitespace_rule(attr_path);
2125 data.conflict_marker_size = ll_merge_marker_size(attr_path);
2126
2127 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2128 die("unable to read files to diff");
2129
2130 /*
2131 * All the other codepaths check both sides, but not checking
2132 * the "old" side here is deliberate. We are checking the newly
2133 * introduced changes, and as long as the "new" side is text, we
2134 * can and should check what it introduces.
2135 */
2136 if (diff_filespec_is_binary(two))
2137 goto free_and_return;
2138 else {
2139 /* Crazy xdl interfaces.. */
2140 xpparam_t xpp;
2141 xdemitconf_t xecfg;
2142
2143 memset(&xpp, 0, sizeof(xpp));
2144 memset(&xecfg, 0, sizeof(xecfg));
2145 xecfg.ctxlen = 1; /* at least one context line */
2146 xpp.flags = 0;
2147 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2148 &xpp, &xecfg);
2149
2150 if (data.ws_rule & WS_BLANK_AT_EOF) {
2151 struct emit_callback ecbdata;
2152 int blank_at_eof;
2153
2154 ecbdata.ws_rule = data.ws_rule;
2155 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2156 blank_at_eof = ecbdata.blank_at_eof_in_preimage;
2157
2158 if (blank_at_eof) {
2159 static char *err;
2160 if (!err)
2161 err = whitespace_error_string(WS_BLANK_AT_EOF);
2162 fprintf(o->file, "%s:%d: %s.\n",
2163 data.filename, blank_at_eof, err);
2164 data.status = 1; /* report errors */
2165 }
2166 }
2167 }
2168 free_and_return:
2169 diff_free_filespec_data(one);
2170 diff_free_filespec_data(two);
2171 if (data.status)
2172 DIFF_OPT_SET(o, CHECK_FAILED);
2173 }
2174
2175 struct diff_filespec *alloc_filespec(const char *path)
2176 {
2177 int namelen = strlen(path);
2178 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2179
2180 memset(spec, 0, sizeof(*spec));
2181 spec->path = (char *)(spec + 1);
2182 memcpy(spec->path, path, namelen+1);
2183 spec->count = 1;
2184 spec->is_binary = -1;
2185 return spec;
2186 }
2187
2188 void free_filespec(struct diff_filespec *spec)
2189 {
2190 if (!--spec->count) {
2191 diff_free_filespec_data(spec);
2192 free(spec);
2193 }
2194 }
2195
2196 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2197 unsigned short mode)
2198 {
2199 if (mode) {
2200 spec->mode = canon_mode(mode);
2201 hashcpy(spec->sha1, sha1);
2202 spec->sha1_valid = !is_null_sha1(sha1);
2203 }
2204 }
2205
2206 /*
2207 * Given a name and sha1 pair, if the index tells us the file in
2208 * the work tree has that object contents, return true, so that
2209 * prepare_temp_file() does not have to inflate and extract.
2210 */
2211 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2212 {
2213 struct cache_entry *ce;
2214 struct stat st;
2215 int pos, len;
2216
2217 /*
2218 * We do not read the cache ourselves here, because the
2219 * benchmark with my previous version that always reads cache
2220 * shows that it makes things worse for diff-tree comparing
2221 * two linux-2.6 kernel trees in an already checked out work
2222 * tree. This is because most diff-tree comparisons deal with
2223 * only a small number of files, while reading the cache is
2224 * expensive for a large project, and its cost outweighs the
2225 * savings we get by not inflating the object to a temporary
2226 * file. Practically, this code only helps when we are used
2227 * by diff-cache --cached, which does read the cache before
2228 * calling us.
2229 */
2230 if (!active_cache)
2231 return 0;
2232
2233 /* We want to avoid the working directory if our caller
2234 * doesn't need the data in a normal file, this system
2235 * is rather slow with its stat/open/mmap/close syscalls,
2236 * and the object is contained in a pack file. The pack
2237 * is probably already open and will be faster to obtain
2238 * the data through than the working directory. Loose
2239 * objects however would tend to be slower as they need
2240 * to be individually opened and inflated.
2241 */
2242 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2243 return 0;
2244
2245 len = strlen(name);
2246 pos = cache_name_pos(name, len);
2247 if (pos < 0)
2248 return 0;
2249 ce = active_cache[pos];
2250
2251 /*
2252 * This is not the sha1 we are looking for, or
2253 * unreusable because it is not a regular file.
2254 */
2255 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2256 return 0;
2257
2258 /*
2259 * If ce is marked as "assume unchanged", there is no
2260 * guarantee that work tree matches what we are looking for.
2261 */
2262 if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2263 return 0;
2264
2265 /*
2266 * If ce matches the file in the work tree, we can reuse it.
2267 */
2268 if (ce_uptodate(ce) ||
2269 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2270 return 1;
2271
2272 return 0;
2273 }
2274
2275 static int populate_from_stdin(struct diff_filespec *s)
2276 {
2277 struct strbuf buf = STRBUF_INIT;
2278 size_t size = 0;
2279
2280 if (strbuf_read(&buf, 0, 0) < 0)
2281 return error("error while reading from stdin %s",
2282 strerror(errno));
2283
2284 s->should_munmap = 0;
2285 s->data = strbuf_detach(&buf, &size);
2286 s->size = size;
2287 s->should_free = 1;
2288 return 0;
2289 }
2290
2291 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2292 {
2293 int len;
2294 char *data = xmalloc(100), *dirty = "";
2295
2296 /* Are we looking at the work tree? */
2297 if (s->dirty_submodule)
2298 dirty = "-dirty";
2299
2300 len = snprintf(data, 100,
2301 "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2302 s->data = data;
2303 s->size = len;
2304 s->should_free = 1;
2305 if (size_only) {
2306 s->data = NULL;
2307 free(data);
2308 }
2309 return 0;
2310 }
2311
2312 /*
2313 * While doing rename detection and pickaxe operation, we may need to
2314 * grab the data for the blob (or file) for our own in-core comparison.
2315 * diff_filespec has data and size fields for this purpose.
2316 */
2317 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2318 {
2319 int err = 0;
2320 if (!DIFF_FILE_VALID(s))
2321 die("internal error: asking to populate invalid file.");
2322 if (S_ISDIR(s->mode))
2323 return -1;
2324
2325 if (s->data)
2326 return 0;
2327
2328 if (size_only && 0 < s->size)
2329 return 0;
2330
2331 if (S_ISGITLINK(s->mode))
2332 return diff_populate_gitlink(s, size_only);
2333
2334 if (!s->sha1_valid ||
2335 reuse_worktree_file(s->path, s->sha1, 0)) {
2336 struct strbuf buf = STRBUF_INIT;
2337 struct stat st;
2338 int fd;
2339
2340 if (!strcmp(s->path, "-"))
2341 return populate_from_stdin(s);
2342
2343 if (lstat(s->path, &st) < 0) {
2344 if (errno == ENOENT) {
2345 err_empty:
2346 err = -1;
2347 empty:
2348 s->data = (char *)"";
2349 s->size = 0;
2350 return err;
2351 }
2352 }
2353 s->size = xsize_t(st.st_size);
2354 if (!s->size)
2355 goto empty;
2356 if (S_ISLNK(st.st_mode)) {
2357 struct strbuf sb = STRBUF_INIT;
2358
2359 if (strbuf_readlink(&sb, s->path, s->size))
2360 goto err_empty;
2361 s->size = sb.len;
2362 s->data = strbuf_detach(&sb, NULL);
2363 s->should_free = 1;
2364 return 0;
2365 }
2366 if (size_only)
2367 return 0;
2368 fd = open(s->path, O_RDONLY);
2369 if (fd < 0)
2370 goto err_empty;
2371 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2372 close(fd);
2373 s->should_munmap = 1;
2374
2375 /*
2376 * Convert from working tree format to canonical git format
2377 */
2378 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2379 size_t size = 0;
2380 munmap(s->data, s->size);
2381 s->should_munmap = 0;
2382 s->data = strbuf_detach(&buf, &size);
2383 s->size = size;
2384 s->should_free = 1;
2385 }
2386 }
2387 else {
2388 enum object_type type;
2389 if (size_only)
2390 type = sha1_object_info(s->sha1, &s->size);
2391 else {
2392 s->data = read_sha1_file(s->sha1, &type, &s->size);
2393 s->should_free = 1;
2394 }
2395 }
2396 return 0;
2397 }
2398
2399 void diff_free_filespec_blob(struct diff_filespec *s)
2400 {
2401 if (s->should_free)
2402 free(s->data);
2403 else if (s->should_munmap)
2404 munmap(s->data, s->size);
2405
2406 if (s->should_free || s->should_munmap) {
2407 s->should_free = s->should_munmap = 0;
2408 s->data = NULL;
2409 }
2410 }
2411
2412 void diff_free_filespec_data(struct diff_filespec *s)
2413 {
2414 diff_free_filespec_blob(s);
2415 free(s->cnt_data);
2416 s->cnt_data = NULL;
2417 }
2418
2419 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2420 void *blob,
2421 unsigned long size,
2422 const unsigned char *sha1,
2423 int mode)
2424 {
2425 int fd;
2426 struct strbuf buf = STRBUF_INIT;
2427 struct strbuf template = STRBUF_INIT;
2428 char *path_dup = xstrdup(path);
2429 const char *base = basename(path_dup);
2430
2431 /* Generate "XXXXXX_basename.ext" */
2432 strbuf_addstr(&template, "XXXXXX_");
2433 strbuf_addstr(&template, base);
2434
2435 fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2436 strlen(base) + 1);
2437 if (fd < 0)
2438 die_errno("unable to create temp-file");
2439 if (convert_to_working_tree(path,
2440 (const char *)blob, (size_t)size, &buf)) {
2441 blob = buf.buf;
2442 size = buf.len;
2443 }
2444 if (write_in_full(fd, blob, size) != size)
2445 die_errno("unable to write temp-file");
2446 close(fd);
2447 temp->name = temp->tmp_path;
2448 strcpy(temp->hex, sha1_to_hex(sha1));
2449 temp->hex[40] = 0;
2450 sprintf(temp->mode, "%06o", mode);
2451 strbuf_release(&buf);
2452 strbuf_release(&template);
2453 free(path_dup);
2454 }
2455
2456 static struct diff_tempfile *prepare_temp_file(const char *name,
2457 struct diff_filespec *one)
2458 {
2459 struct diff_tempfile *temp = claim_diff_tempfile();
2460
2461 if (!DIFF_FILE_VALID(one)) {
2462 not_a_valid_file:
2463 /* A '-' entry produces this for file-2, and
2464 * a '+' entry produces this for file-1.
2465 */
2466 temp->name = "/dev/null";
2467 strcpy(temp->hex, ".");
2468 strcpy(temp->mode, ".");
2469 return temp;
2470 }
2471
2472 if (!remove_tempfile_installed) {
2473 atexit(remove_tempfile);
2474 sigchain_push_common(remove_tempfile_on_signal);
2475 remove_tempfile_installed = 1;
2476 }
2477
2478 if (!one->sha1_valid ||
2479 reuse_worktree_file(name, one->sha1, 1)) {
2480 struct stat st;
2481 if (lstat(name, &st) < 0) {
2482 if (errno == ENOENT)
2483 goto not_a_valid_file;
2484 die_errno("stat(%s)", name);
2485 }
2486 if (S_ISLNK(st.st_mode)) {
2487 struct strbuf sb = STRBUF_INIT;
2488 if (strbuf_readlink(&sb, name, st.st_size) < 0)
2489 die_errno("readlink(%s)", name);
2490 prep_temp_blob(name, temp, sb.buf, sb.len,
2491 (one->sha1_valid ?
2492 one->sha1 : null_sha1),
2493 (one->sha1_valid ?
2494 one->mode : S_IFLNK));
2495 strbuf_release(&sb);
2496 }
2497 else {
2498 /* we can borrow from the file in the work tree */
2499 temp->name = name;
2500 if (!one->sha1_valid)
2501 strcpy(temp->hex, sha1_to_hex(null_sha1));
2502 else
2503 strcpy(temp->hex, sha1_to_hex(one->sha1));
2504 /* Even though we may sometimes borrow the
2505 * contents from the work tree, we always want
2506 * one->mode. mode is trustworthy even when
2507 * !(one->sha1_valid), as long as
2508 * DIFF_FILE_VALID(one).
2509 */
2510 sprintf(temp->mode, "%06o", one->mode);
2511 }
2512 return temp;
2513 }
2514 else {
2515 if (diff_populate_filespec(one, 0))
2516 die("cannot read data blob for %s", one->path);
2517 prep_temp_blob(name, temp, one->data, one->size,
2518 one->sha1, one->mode);
2519 }
2520 return temp;
2521 }
2522
2523 /* An external diff command takes:
2524 *
2525 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2526 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2527 *
2528 */
2529 static void run_external_diff(const char *pgm,
2530 const char *name,
2531 const char *other,
2532 struct diff_filespec *one,
2533 struct diff_filespec *two,
2534 const char *xfrm_msg,
2535 int complete_rewrite)
2536 {
2537 const char *spawn_arg[10];
2538 int retval;
2539 const char **arg = &spawn_arg[0];
2540
2541 if (one && two) {
2542 struct diff_tempfile *temp_one, *temp_two;
2543 const char *othername = (other ? other : name);
2544 temp_one = prepare_temp_file(name, one);
2545 temp_two = prepare_temp_file(othername, two);
2546 *arg++ = pgm;
2547 *arg++ = name;
2548 *arg++ = temp_one->name;
2549 *arg++ = temp_one->hex;
2550 *arg++ = temp_one->mode;
2551 *arg++ = temp_two->name;
2552 *arg++ = temp_two->hex;
2553 *arg++ = temp_two->mode;
2554 if (other) {
2555 *arg++ = other;
2556 *arg++ = xfrm_msg;
2557 }
2558 } else {
2559 *arg++ = pgm;
2560 *arg++ = name;
2561 }
2562 *arg = NULL;
2563 fflush(NULL);
2564 retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2565 remove_tempfile();
2566 if (retval) {
2567 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2568 exit(1);
2569 }
2570 }
2571
2572 static int similarity_index(struct diff_filepair *p)
2573 {
2574 return p->score * 100 / MAX_SCORE;
2575 }
2576
2577 static void fill_metainfo(struct strbuf *msg,
2578 const char *name,
2579 const char *other,
2580 struct diff_filespec *one,
2581 struct diff_filespec *two,
2582 struct diff_options *o,
2583 struct diff_filepair *p,
2584 int *must_show_header,
2585 int use_color)
2586 {
2587 const char *set = diff_get_color(use_color, DIFF_METAINFO);
2588 const char *reset = diff_get_color(use_color, DIFF_RESET);
2589 struct strbuf *msgbuf;
2590 char *line_prefix = "";
2591
2592 *must_show_header = 1;
2593 if (o->output_prefix) {
2594 msgbuf = o->output_prefix(o, o->output_prefix_data);
2595 line_prefix = msgbuf->buf;
2596 }
2597 strbuf_init(msg, PATH_MAX * 2 + 300);
2598 switch (p->status) {
2599 case DIFF_STATUS_COPIED:
2600 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2601 line_prefix, set, similarity_index(p));
2602 strbuf_addf(msg, "%s\n%s%scopy from ",
2603 reset, line_prefix, set);
2604 quote_c_style(name, msg, NULL, 0);
2605 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2606 quote_c_style(other, msg, NULL, 0);
2607 strbuf_addf(msg, "%s\n", reset);
2608 break;
2609 case DIFF_STATUS_RENAMED:
2610 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2611 line_prefix, set, similarity_index(p));
2612 strbuf_addf(msg, "%s\n%s%srename from ",
2613 reset, line_prefix, set);
2614 quote_c_style(name, msg, NULL, 0);
2615 strbuf_addf(msg, "%s\n%s%srename to ",
2616 reset, line_prefix, set);
2617 quote_c_style(other, msg, NULL, 0);
2618 strbuf_addf(msg, "%s\n", reset);
2619 break;
2620 case DIFF_STATUS_MODIFIED:
2621 if (p->score) {
2622 strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2623 line_prefix,
2624 set, similarity_index(p), reset);
2625 break;
2626 }
2627 /* fallthru */
2628 default:
2629 *must_show_header = 0;
2630 }
2631 if (one && two && hashcmp(one->sha1, two->sha1)) {
2632 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2633
2634 if (DIFF_OPT_TST(o, BINARY)) {
2635 mmfile_t mf;
2636 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2637 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2638 abbrev = 40;
2639 }
2640 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2641 find_unique_abbrev(one->sha1, abbrev));
2642 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2643 if (one->mode == two->mode)
2644 strbuf_addf(msg, " %06o", one->mode);
2645 strbuf_addf(msg, "%s\n", reset);
2646 }
2647 }
2648
2649 static void run_diff_cmd(const char *pgm,
2650 const char *name,
2651 const char *other,
2652 const char *attr_path,
2653 struct diff_filespec *one,
2654 struct diff_filespec *two,
2655 struct strbuf *msg,
2656 struct diff_options *o,
2657 struct diff_filepair *p)
2658 {
2659 const char *xfrm_msg = NULL;
2660 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2661 int must_show_header = 0;
2662
2663 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2664 pgm = NULL;
2665 else {
2666 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2667 if (drv && drv->external)
2668 pgm = drv->external;
2669 }
2670
2671 if (msg) {
2672 /*
2673 * don't use colors when the header is intended for an
2674 * external diff driver
2675 */
2676 fill_metainfo(msg, name, other, one, two, o, p,
2677 &must_show_header,
2678 DIFF_OPT_TST(o, COLOR_DIFF) && !pgm);
2679 xfrm_msg = msg->len ? msg->buf : NULL;
2680 }
2681
2682 if (pgm) {
2683 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2684 complete_rewrite);
2685 return;
2686 }
2687 if (one && two)
2688 builtin_diff(name, other ? other : name,
2689 one, two, xfrm_msg, must_show_header,
2690 o, complete_rewrite);
2691 else
2692 fprintf(o->file, "* Unmerged path %s\n", name);
2693 }
2694
2695 static void diff_fill_sha1_info(struct diff_filespec *one)
2696 {
2697 if (DIFF_FILE_VALID(one)) {
2698 if (!one->sha1_valid) {
2699 struct stat st;
2700 if (!strcmp(one->path, "-")) {
2701 hashcpy(one->sha1, null_sha1);
2702 return;
2703 }
2704 if (lstat(one->path, &st) < 0)
2705 die_errno("stat '%s'", one->path);
2706 if (index_path(one->sha1, one->path, &st, 0))
2707 die("cannot hash %s", one->path);
2708 }
2709 }
2710 else
2711 hashclr(one->sha1);
2712 }
2713
2714 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2715 {
2716 /* Strip the prefix but do not molest /dev/null and absolute paths */
2717 if (*namep && **namep != '/') {
2718 *namep += prefix_length;
2719 if (**namep == '/')
2720 ++*namep;
2721 }
2722 if (*otherp && **otherp != '/') {
2723 *otherp += prefix_length;
2724 if (**otherp == '/')
2725 ++*otherp;
2726 }
2727 }
2728
2729 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2730 {
2731 const char *pgm = external_diff();
2732 struct strbuf msg;
2733 struct diff_filespec *one = p->one;
2734 struct diff_filespec *two = p->two;
2735 const char *name;
2736 const char *other;
2737 const char *attr_path;
2738
2739 name = p->one->path;
2740 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2741 attr_path = name;
2742 if (o->prefix_length)
2743 strip_prefix(o->prefix_length, &name, &other);
2744
2745 if (DIFF_PAIR_UNMERGED(p)) {
2746 run_diff_cmd(pgm, name, NULL, attr_path,
2747 NULL, NULL, NULL, o, p);
2748 return;
2749 }
2750
2751 diff_fill_sha1_info(one);
2752 diff_fill_sha1_info(two);
2753
2754 if (!pgm &&
2755 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2756 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2757 /*
2758 * a filepair that changes between file and symlink
2759 * needs to be split into deletion and creation.
2760 */
2761 struct diff_filespec *null = alloc_filespec(two->path);
2762 run_diff_cmd(NULL, name, other, attr_path,
2763 one, null, &msg, o, p);
2764 free(null);
2765 strbuf_release(&msg);
2766
2767 null = alloc_filespec(one->path);
2768 run_diff_cmd(NULL, name, other, attr_path,
2769 null, two, &msg, o, p);
2770 free(null);
2771 }
2772 else
2773 run_diff_cmd(pgm, name, other, attr_path,
2774 one, two, &msg, o, p);
2775
2776 strbuf_release(&msg);
2777 }
2778
2779 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2780 struct diffstat_t *diffstat)
2781 {
2782 const char *name;
2783 const char *other;
2784 int complete_rewrite = 0;
2785
2786 if (DIFF_PAIR_UNMERGED(p)) {
2787 /* unmerged */
2788 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2789 return;
2790 }
2791
2792 name = p->one->path;
2793 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2794
2795 if (o->prefix_length)
2796 strip_prefix(o->prefix_length, &name, &other);
2797
2798 diff_fill_sha1_info(p->one);
2799 diff_fill_sha1_info(p->two);
2800
2801 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2802 complete_rewrite = 1;
2803 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2804 }
2805
2806 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2807 {
2808 const char *name;
2809 const char *other;
2810 const char *attr_path;
2811
2812 if (DIFF_PAIR_UNMERGED(p)) {
2813 /* unmerged */
2814 return;
2815 }
2816
2817 name = p->one->path;
2818 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2819 attr_path = other ? other : name;
2820
2821 if (o->prefix_length)
2822 strip_prefix(o->prefix_length, &name, &other);
2823
2824 diff_fill_sha1_info(p->one);
2825 diff_fill_sha1_info(p->two);
2826
2827 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2828 }
2829
2830 void diff_setup(struct diff_options *options)
2831 {
2832 memcpy(options, &default_diff_options, sizeof(*options));
2833
2834 options->file = stdout;
2835
2836 options->line_termination = '\n';
2837 options->break_opt = -1;
2838 options->rename_limit = -1;
2839 options->dirstat_percent = 3;
2840 options->context = 3;
2841
2842 options->change = diff_change;
2843 options->add_remove = diff_addremove;
2844 if (diff_use_color_default > 0)
2845 DIFF_OPT_SET(options, COLOR_DIFF);
2846 options->detect_rename = diff_detect_rename_default;
2847
2848 if (diff_no_prefix) {
2849 options->a_prefix = options->b_prefix = "";
2850 } else if (!diff_mnemonic_prefix) {
2851 options->a_prefix = "a/";
2852 options->b_prefix = "b/";
2853 }
2854 }
2855
2856 int diff_setup_done(struct diff_options *options)
2857 {
2858 int count = 0;
2859
2860 if (options->output_format & DIFF_FORMAT_NAME)
2861 count++;
2862 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2863 count++;
2864 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2865 count++;
2866 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2867 count++;
2868 if (count > 1)
2869 die("--name-only, --name-status, --check and -s are mutually exclusive");
2870
2871 /*
2872 * Most of the time we can say "there are changes"
2873 * only by checking if there are changed paths, but
2874 * --ignore-whitespace* options force us to look
2875 * inside contents.
2876 */
2877
2878 if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2879 DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2880 DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2881 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2882 else
2883 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2884
2885 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2886 options->detect_rename = DIFF_DETECT_COPY;
2887
2888 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2889 options->prefix = NULL;
2890 if (options->prefix)
2891 options->prefix_length = strlen(options->prefix);
2892 else
2893 options->prefix_length = 0;
2894
2895 if (options->output_format & (DIFF_FORMAT_NAME |
2896 DIFF_FORMAT_NAME_STATUS |
2897 DIFF_FORMAT_CHECKDIFF |
2898 DIFF_FORMAT_NO_OUTPUT))
2899 options->output_format &= ~(DIFF_FORMAT_RAW |
2900 DIFF_FORMAT_NUMSTAT |
2901 DIFF_FORMAT_DIFFSTAT |
2902 DIFF_FORMAT_SHORTSTAT |
2903 DIFF_FORMAT_DIRSTAT |
2904 DIFF_FORMAT_SUMMARY |
2905 DIFF_FORMAT_PATCH);
2906
2907 /*
2908 * These cases always need recursive; we do not drop caller-supplied
2909 * recursive bits for other formats here.
2910 */
2911 if (options->output_format & (DIFF_FORMAT_PATCH |
2912 DIFF_FORMAT_NUMSTAT |
2913 DIFF_FORMAT_DIFFSTAT |
2914 DIFF_FORMAT_SHORTSTAT |
2915 DIFF_FORMAT_DIRSTAT |
2916 DIFF_FORMAT_SUMMARY |
2917 DIFF_FORMAT_CHECKDIFF))
2918 DIFF_OPT_SET(options, RECURSIVE);
2919 /*
2920 * Also pickaxe would not work very well if you do not say recursive
2921 */
2922 if (options->pickaxe)
2923 DIFF_OPT_SET(options, RECURSIVE);
2924 /*
2925 * When patches are generated, submodules diffed against the work tree
2926 * must be checked for dirtiness too so it can be shown in the output
2927 */
2928 if (options->output_format & DIFF_FORMAT_PATCH)
2929 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2930
2931 if (options->detect_rename && options->rename_limit < 0)
2932 options->rename_limit = diff_rename_limit_default;
2933 if (options->setup & DIFF_SETUP_USE_CACHE) {
2934 if (!active_cache)
2935 /* read-cache does not die even when it fails
2936 * so it is safe for us to do this here. Also
2937 * it does not smudge active_cache or active_nr
2938 * when it fails, so we do not have to worry about
2939 * cleaning it up ourselves either.
2940 */
2941 read_cache();
2942 }
2943 if (options->abbrev <= 0 || 40 < options->abbrev)
2944 options->abbrev = 40; /* full */
2945
2946 /*
2947 * It does not make sense to show the first hit we happened
2948 * to have found. It does not make sense not to return with
2949 * exit code in such a case either.
2950 */
2951 if (DIFF_OPT_TST(options, QUICK)) {
2952 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2953 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2954 }
2955
2956 return 0;
2957 }
2958
2959 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2960 {
2961 char c, *eq;
2962 int len;
2963
2964 if (*arg != '-')
2965 return 0;
2966 c = *++arg;
2967 if (!c)
2968 return 0;
2969 if (c == arg_short) {
2970 c = *++arg;
2971 if (!c)
2972 return 1;
2973 if (val && isdigit(c)) {
2974 char *end;
2975 int n = strtoul(arg, &end, 10);
2976 if (*end)
2977 return 0;
2978 *val = n;
2979 return 1;
2980 }
2981 return 0;
2982 }
2983 if (c != '-')
2984 return 0;
2985 arg++;
2986 eq = strchr(arg, '=');
2987 if (eq)
2988 len = eq - arg;
2989 else
2990 len = strlen(arg);
2991 if (!len || strncmp(arg, arg_long, len))
2992 return 0;
2993 if (eq) {
2994 int n;
2995 char *end;
2996 if (!isdigit(*++eq))
2997 return 0;
2998 n = strtoul(eq, &end, 10);
2999 if (*end)
3000 return 0;
3001 *val = n;
3002 }
3003 return 1;
3004 }
3005
3006 static int diff_scoreopt_parse(const char *opt);
3007
3008 static inline int short_opt(char opt, const char **argv,
3009 const char **optarg)
3010 {
3011 const char *arg = argv[0];
3012 if (arg[0] != '-' || arg[1] != opt)
3013 return 0;
3014 if (arg[2] != '\0') {
3015 *optarg = arg + 2;
3016 return 1;
3017 }
3018 if (!argv[1])
3019 die("Option '%c' requires a value", opt);
3020 *optarg = argv[1];
3021 return 2;
3022 }
3023
3024 int parse_long_opt(const char *opt, const char **argv,
3025 const char **optarg)
3026 {
3027 const char *arg = argv[0];
3028 if (arg[0] != '-' || arg[1] != '-')
3029 return 0;
3030 arg += strlen("--");
3031 if (prefixcmp(arg, opt))
3032 return 0;
3033 arg += strlen(opt);
3034 if (*arg == '=') { /* sticked form: --option=value */
3035 *optarg = arg + 1;
3036 return 1;
3037 }
3038 if (*arg != '\0')
3039 return 0;
3040 /* separate form: --option value */
3041 if (!argv[1])
3042 die("Option '--%s' requires a value", opt);
3043 *optarg = argv[1];
3044 return 2;
3045 }
3046
3047 static int stat_opt(struct diff_options *options, const char **av)
3048 {
3049 const char *arg = av[0];
3050 char *end;
3051 int width = options->stat_width;
3052 int name_width = options->stat_name_width;
3053 int argcount = 1;
3054
3055 arg += strlen("--stat");
3056 end = (char *)arg;
3057
3058 switch (*arg) {
3059 case '-':
3060 if (!prefixcmp(arg, "-width")) {
3061 arg += strlen("-width");
3062 if (*arg == '=')
3063 width = strtoul(arg + 1, &end, 10);
3064 else if (!*arg && !av[1])
3065 die("Option '--stat-width' requires a value");
3066 else if (!*arg) {
3067 width = strtoul(av[1], &end, 10);
3068 argcount = 2;
3069 }
3070 } else if (!prefixcmp(arg, "-name-width")) {
3071 arg += strlen("-name-width");
3072 if (*arg == '=')
3073 name_width = strtoul(arg + 1, &end, 10);
3074 else if (!*arg && !av[1])
3075 die("Option '--stat-name-width' requires a value");
3076 else if (!*arg) {
3077 name_width = strtoul(av[1], &end, 10);
3078 argcount = 2;
3079 }
3080 }
3081 break;
3082 case '=':
3083 width = strtoul(arg+1, &end, 10);
3084 if (*end == ',')
3085 name_width = strtoul(end+1, &end, 10);
3086 }
3087
3088 /* Important! This checks all the error cases! */
3089 if (*end)
3090 return 0;
3091 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3092 options->stat_name_width = name_width;
3093 options->stat_width = width;
3094 return argcount;
3095 }
3096
3097 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3098 {
3099 const char *arg = av[0];
3100 const char *optarg;
3101 int argcount;
3102
3103 /* Output format options */
3104 if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3105 options->output_format |= DIFF_FORMAT_PATCH;
3106 else if (opt_arg(arg, 'U', "unified", &options->context))
3107 options->output_format |= DIFF_FORMAT_PATCH;
3108 else if (!strcmp(arg, "--raw"))
3109 options->output_format |= DIFF_FORMAT_RAW;
3110 else if (!strcmp(arg, "--patch-with-raw"))
3111 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3112 else if (!strcmp(arg, "--numstat"))
3113 options->output_format |= DIFF_FORMAT_NUMSTAT;
3114 else if (!strcmp(arg, "--shortstat"))
3115 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3116 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
3117 options->output_format |= DIFF_FORMAT_DIRSTAT;
3118 else if (!strcmp(arg, "--cumulative")) {
3119 options->output_format |= DIFF_FORMAT_DIRSTAT;
3120 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
3121 } else if (opt_arg(arg, 0, "dirstat-by-file",
3122 &options->dirstat_percent)) {
3123 options->output_format |= DIFF_FORMAT_DIRSTAT;
3124 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
3125 }
3126 else if (!strcmp(arg, "--check"))
3127 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3128 else if (!strcmp(arg, "--summary"))
3129 options->output_format |= DIFF_FORMAT_SUMMARY;
3130 else if (!strcmp(arg, "--patch-with-stat"))
3131 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3132 else if (!strcmp(arg, "--name-only"))
3133 options->output_format |= DIFF_FORMAT_NAME;
3134 else if (!strcmp(arg, "--name-status"))
3135 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3136 else if (!strcmp(arg, "-s"))
3137 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3138 else if (!prefixcmp(arg, "--stat"))
3139 /* --stat, --stat-width, or --stat-name-width */
3140 return stat_opt(options, av);
3141
3142 /* renames options */
3143 else if (!prefixcmp(arg, "-B")) {
3144 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3145 return -1;
3146 }
3147 else if (!prefixcmp(arg, "-M")) {
3148 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3149 return -1;
3150 options->detect_rename = DIFF_DETECT_RENAME;
3151 }
3152 else if (!prefixcmp(arg, "-C")) {
3153 if (options->detect_rename == DIFF_DETECT_COPY)
3154 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3155 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3156 return -1;
3157 options->detect_rename = DIFF_DETECT_COPY;
3158 }
3159 else if (!strcmp(arg, "--no-renames"))
3160 options->detect_rename = 0;
3161 else if (!strcmp(arg, "--relative"))
3162 DIFF_OPT_SET(options, RELATIVE_NAME);
3163 else if (!prefixcmp(arg, "--relative=")) {
3164 DIFF_OPT_SET(options, RELATIVE_NAME);
3165 options->prefix = arg + 11;
3166 }
3167
3168 /* xdiff options */
3169 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3170 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3171 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3172 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3173 else if (!strcmp(arg, "--ignore-space-at-eol"))
3174 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3175 else if (!strcmp(arg, "--patience"))
3176 DIFF_XDL_SET(options, PATIENCE_DIFF);
3177
3178 /* flags options */
3179 else if (!strcmp(arg, "--binary")) {
3180 options->output_format |= DIFF_FORMAT_PATCH;
3181 DIFF_OPT_SET(options, BINARY);
3182 }
3183 else if (!strcmp(arg, "--full-index"))
3184 DIFF_OPT_SET(options, FULL_INDEX);
3185 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3186 DIFF_OPT_SET(options, TEXT);
3187 else if (!strcmp(arg, "-R"))
3188 DIFF_OPT_SET(options, REVERSE_DIFF);
3189 else if (!strcmp(arg, "--find-copies-harder"))
3190 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3191 else if (!strcmp(arg, "--follow"))
3192 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3193 else if (!strcmp(arg, "--color"))
3194 DIFF_OPT_SET(options, COLOR_DIFF);
3195 else if (!prefixcmp(arg, "--color=")) {
3196 int value = git_config_colorbool(NULL, arg+8, -1);
3197 if (value == 0)
3198 DIFF_OPT_CLR(options, COLOR_DIFF);
3199 else if (value > 0)
3200 DIFF_OPT_SET(options, COLOR_DIFF);
3201 else
3202 return error("option `color' expects \"always\", \"auto\", or \"never\"");
3203 }
3204 else if (!strcmp(arg, "--no-color"))
3205 DIFF_OPT_CLR(options, COLOR_DIFF);
3206 else if (!strcmp(arg, "--color-words")) {
3207 DIFF_OPT_SET(options, COLOR_DIFF);
3208 options->word_diff = DIFF_WORDS_COLOR;
3209 }
3210 else if (!prefixcmp(arg, "--color-words=")) {
3211 DIFF_OPT_SET(options, COLOR_DIFF);
3212 options->word_diff = DIFF_WORDS_COLOR;
3213 options->word_regex = arg + 14;
3214 }
3215 else if (!strcmp(arg, "--word-diff")) {
3216 if (options->word_diff == DIFF_WORDS_NONE)
3217 options->word_diff = DIFF_WORDS_PLAIN;
3218 }
3219 else if (!prefixcmp(arg, "--word-diff=")) {
3220 const char *type = arg + 12;
3221 if (!strcmp(type, "plain"))
3222 options->word_diff = DIFF_WORDS_PLAIN;
3223 else if (!strcmp(type, "color")) {
3224 DIFF_OPT_SET(options, COLOR_DIFF);
3225 options->word_diff = DIFF_WORDS_COLOR;
3226 }
3227 else if (!strcmp(type, "porcelain"))
3228 options->word_diff = DIFF_WORDS_PORCELAIN;
3229 else if (!strcmp(type, "none"))
3230 options->word_diff = DIFF_WORDS_NONE;
3231 else
3232 die("bad --word-diff argument: %s", type);
3233 }
3234 else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3235 if (options->word_diff == DIFF_WORDS_NONE)
3236 options->word_diff = DIFF_WORDS_PLAIN;
3237 options->word_regex = optarg;
3238 return argcount;
3239 }
3240 else if (!strcmp(arg, "--exit-code"))
3241 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3242 else if (!strcmp(arg, "--quiet"))
3243 DIFF_OPT_SET(options, QUICK);
3244 else if (!strcmp(arg, "--ext-diff"))
3245 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3246 else if (!strcmp(arg, "--no-ext-diff"))
3247 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3248 else if (!strcmp(arg, "--textconv"))
3249 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3250 else if (!strcmp(arg, "--no-textconv"))
3251 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3252 else if (!strcmp(arg, "--ignore-submodules")) {
3253 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3254 handle_ignore_submodules_arg(options, "all");
3255 } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3256 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3257 handle_ignore_submodules_arg(options, arg + 20);
3258 } else if (!strcmp(arg, "--submodule"))
3259 DIFF_OPT_SET(options, SUBMODULE_LOG);
3260 else if (!prefixcmp(arg, "--submodule=")) {
3261 if (!strcmp(arg + 12, "log"))
3262 DIFF_OPT_SET(options, SUBMODULE_LOG);
3263 }
3264
3265 /* misc options */
3266 else if (!strcmp(arg, "-z"))
3267 options->line_termination = 0;
3268 else if ((argcount = short_opt('l', av, &optarg))) {
3269 options->rename_limit = strtoul(optarg, NULL, 10);
3270 return argcount;
3271 }
3272 else if ((argcount = short_opt('S', av, &optarg))) {
3273 options->pickaxe = optarg;
3274 return argcount;
3275 }
3276 else if (!strcmp(arg, "--pickaxe-all"))
3277 options->pickaxe_opts = DIFF_PICKAXE_ALL;
3278 else if (!strcmp(arg, "--pickaxe-regex"))
3279 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
3280 else if ((argcount = short_opt('O', av, &optarg))) {
3281 options->orderfile = optarg;
3282 return argcount;
3283 }
3284 else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3285 options->filter = optarg;
3286 return argcount;
3287 }
3288 else if (!strcmp(arg, "--abbrev"))
3289 options->abbrev = DEFAULT_ABBREV;
3290 else if (!prefixcmp(arg, "--abbrev=")) {
3291 options->abbrev = strtoul(arg + 9, NULL, 10);
3292 if (options->abbrev < MINIMUM_ABBREV)
3293 options->abbrev = MINIMUM_ABBREV;
3294 else if (40 < options->abbrev)
3295 options->abbrev = 40;
3296 }
3297 else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3298 options->a_prefix = optarg;
3299 return argcount;
3300 }
3301 else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3302 options->b_prefix = optarg;
3303 return argcount;
3304 }
3305 else if (!strcmp(arg, "--no-prefix"))
3306 options->a_prefix = options->b_prefix = "";
3307 else if (opt_arg(arg, '\0', "inter-hunk-context",
3308 &options->interhunkcontext))
3309 ;
3310 else if ((argcount = parse_long_opt("output", av, &optarg))) {
3311 options->file = fopen(optarg, "w");
3312 if (!options->file)
3313 die_errno("Could not open '%s'", arg + strlen("--output="));
3314 options->close_file = 1;
3315 return argcount;
3316 } else
3317 return 0;
3318 return 1;
3319 }
3320
3321 static int parse_num(const char **cp_p)
3322 {
3323 unsigned long num, scale;
3324 int ch, dot;
3325 const char *cp = *cp_p;
3326
3327 num = 0;
3328 scale = 1;
3329 dot = 0;
3330 for (;;) {
3331 ch = *cp;
3332 if ( !dot && ch == '.' ) {
3333 scale = 1;
3334 dot = 1;
3335 } else if ( ch == '%' ) {
3336 scale = dot ? scale*100 : 100;
3337 cp++; /* % is always at the end */
3338 break;
3339 } else if ( ch >= '0' && ch <= '9' ) {
3340 if ( scale < 100000 ) {
3341 scale *= 10;
3342 num = (num*10) + (ch-'0');
3343 }
3344 } else {
3345 break;
3346 }
3347 cp++;
3348 }
3349 *cp_p = cp;
3350
3351 /* user says num divided by scale and we say internally that
3352 * is MAX_SCORE * num / scale.
3353 */
3354 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3355 }
3356
3357 static int diff_scoreopt_parse(const char *opt)
3358 {
3359 int opt1, opt2, cmd;
3360
3361 if (*opt++ != '-')
3362 return -1;
3363 cmd = *opt++;
3364 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3365 return -1; /* that is not a -M, -C nor -B option */
3366
3367 opt1 = parse_num(&opt);
3368 if (cmd != 'B')
3369 opt2 = 0;
3370 else {
3371 if (*opt == 0)
3372 opt2 = 0;
3373 else if (*opt != '/')
3374 return -1; /* we expect -B80/99 or -B80 */
3375 else {
3376 opt++;
3377 opt2 = parse_num(&opt);
3378 }
3379 }
3380 if (*opt != 0)
3381 return -1;
3382 return opt1 | (opt2 << 16);
3383 }
3384
3385 struct diff_queue_struct diff_queued_diff;
3386
3387 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3388 {
3389 if (queue->alloc <= queue->nr) {
3390 queue->alloc = alloc_nr(queue->alloc);
3391 queue->queue = xrealloc(queue->queue,
3392 sizeof(dp) * queue->alloc);
3393 }
3394 queue->queue[queue->nr++] = dp;
3395 }
3396
3397 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3398 struct diff_filespec *one,
3399 struct diff_filespec *two)
3400 {
3401 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3402 dp->one = one;
3403 dp->two = two;
3404 if (queue)
3405 diff_q(queue, dp);
3406 return dp;
3407 }
3408
3409 void diff_free_filepair(struct diff_filepair *p)
3410 {
3411 free_filespec(p->one);
3412 free_filespec(p->two);
3413 free(p);
3414 }
3415
3416 /* This is different from find_unique_abbrev() in that
3417 * it stuffs the result with dots for alignment.
3418 */
3419 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3420 {
3421 int abblen;
3422 const char *abbrev;
3423 if (len == 40)
3424 return sha1_to_hex(sha1);
3425
3426 abbrev = find_unique_abbrev(sha1, len);
3427 abblen = strlen(abbrev);
3428 if (abblen < 37) {
3429 static char hex[41];
3430 if (len < abblen && abblen <= len + 2)
3431 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3432 else
3433 sprintf(hex, "%s...", abbrev);
3434 return hex;
3435 }
3436 return sha1_to_hex(sha1);
3437 }
3438
3439 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3440 {
3441 int line_termination = opt->line_termination;
3442 int inter_name_termination = line_termination ? '\t' : '\0';
3443 if (opt->output_prefix) {
3444 struct strbuf *msg = NULL;
3445 msg = opt->output_prefix(opt, opt->output_prefix_data);
3446 fprintf(opt->file, "%s", msg->buf);
3447 }
3448
3449 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3450 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3451 diff_unique_abbrev(p->one->sha1, opt->abbrev));
3452 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3453 }
3454 if (p->score) {
3455 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3456 inter_name_termination);
3457 } else {
3458 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3459 }
3460
3461 if (p->status == DIFF_STATUS_COPIED ||
3462 p->status == DIFF_STATUS_RENAMED) {
3463 const char *name_a, *name_b;
3464 name_a = p->one->path;
3465 name_b = p->two->path;
3466 strip_prefix(opt->prefix_length, &name_a, &name_b);
3467 write_name_quoted(name_a, opt->file, inter_name_termination);
3468 write_name_quoted(name_b, opt->file, line_termination);
3469 } else {
3470 const char *name_a, *name_b;
3471 name_a = p->one->mode ? p->one->path : p->two->path;
3472 name_b = NULL;
3473 strip_prefix(opt->prefix_length, &name_a, &name_b);
3474 write_name_quoted(name_a, opt->file, line_termination);
3475 }
3476 }
3477
3478 int diff_unmodified_pair(struct diff_filepair *p)
3479 {
3480 /* This function is written stricter than necessary to support
3481 * the currently implemented transformers, but the idea is to
3482 * let transformers to produce diff_filepairs any way they want,
3483 * and filter and clean them up here before producing the output.
3484 */
3485 struct diff_filespec *one = p->one, *two = p->two;
3486
3487 if (DIFF_PAIR_UNMERGED(p))
3488 return 0; /* unmerged is interesting */
3489
3490 /* deletion, addition, mode or type change
3491 * and rename are all interesting.
3492 */
3493 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3494 DIFF_PAIR_MODE_CHANGED(p) ||
3495 strcmp(one->path, two->path))
3496 return 0;
3497
3498 /* both are valid and point at the same path. that is, we are
3499 * dealing with a change.
3500 */
3501 if (one->sha1_valid && two->sha1_valid &&
3502 !hashcmp(one->sha1, two->sha1) &&
3503 !one->dirty_submodule && !two->dirty_submodule)
3504 return 1; /* no change */
3505 if (!one->sha1_valid && !two->sha1_valid)
3506 return 1; /* both look at the same file on the filesystem. */
3507 return 0;
3508 }
3509
3510 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3511 {
3512 if (diff_unmodified_pair(p))
3513 return;
3514
3515 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3516 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3517 return; /* no tree diffs in patch format */
3518
3519 run_diff(p, o);
3520 }
3521
3522 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3523 struct diffstat_t *diffstat)
3524 {
3525 if (diff_unmodified_pair(p))
3526 return;
3527
3528 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3529 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3530 return; /* no tree diffs in patch format */
3531
3532 run_diffstat(p, o, diffstat);
3533 }
3534
3535 static void diff_flush_checkdiff(struct diff_filepair *p,
3536 struct diff_options *o)
3537 {
3538 if (diff_unmodified_pair(p))
3539 return;
3540
3541 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3542 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3543 return; /* no tree diffs in patch format */
3544
3545 run_checkdiff(p, o);
3546 }
3547
3548 int diff_queue_is_empty(void)
3549 {
3550 struct diff_queue_struct *q = &diff_queued_diff;
3551 int i;
3552 for (i = 0; i < q->nr; i++)
3553 if (!diff_unmodified_pair(q->queue[i]))
3554 return 0;
3555 return 1;
3556 }
3557
3558 #if DIFF_DEBUG
3559 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3560 {
3561 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3562 x, one ? one : "",
3563 s->path,
3564 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3565 s->mode,
3566 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3567 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3568 x, one ? one : "",
3569 s->size, s->xfrm_flags);
3570 }
3571
3572 void diff_debug_filepair(const struct diff_filepair *p, int i)
3573 {
3574 diff_debug_filespec(p->one, i, "one");
3575 diff_debug_filespec(p->two, i, "two");
3576 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3577 p->score, p->status ? p->status : '?',
3578 p->one->rename_used, p->broken_pair);
3579 }
3580
3581 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3582 {
3583 int i;
3584 if (msg)
3585 fprintf(stderr, "%s\n", msg);
3586 fprintf(stderr, "q->nr = %d\n", q->nr);
3587 for (i = 0; i < q->nr; i++) {
3588 struct diff_filepair *p = q->queue[i];
3589 diff_debug_filepair(p, i);
3590 }
3591 }
3592 #endif
3593
3594 static void diff_resolve_rename_copy(void)
3595 {
3596 int i;
3597 struct diff_filepair *p;
3598 struct diff_queue_struct *q = &diff_queued_diff;
3599
3600 diff_debug_queue("resolve-rename-copy", q);
3601
3602 for (i = 0; i < q->nr; i++) {
3603 p = q->queue[i];
3604 p->status = 0; /* undecided */
3605 if (DIFF_PAIR_UNMERGED(p))
3606 p->status = DIFF_STATUS_UNMERGED;
3607 else if (!DIFF_FILE_VALID(p->one))
3608 p->status = DIFF_STATUS_ADDED;
3609 else if (!DIFF_FILE_VALID(p->two))
3610 p->status = DIFF_STATUS_DELETED;
3611 else if (DIFF_PAIR_TYPE_CHANGED(p))
3612 p->status = DIFF_STATUS_TYPE_CHANGED;
3613
3614 /* from this point on, we are dealing with a pair
3615 * whose both sides are valid and of the same type, i.e.
3616 * either in-place edit or rename/copy edit.
3617 */
3618 else if (DIFF_PAIR_RENAME(p)) {
3619 /*
3620 * A rename might have re-connected a broken
3621 * pair up, causing the pathnames to be the
3622 * same again. If so, that's not a rename at
3623 * all, just a modification..
3624 *
3625 * Otherwise, see if this source was used for
3626 * multiple renames, in which case we decrement
3627 * the count, and call it a copy.
3628 */
3629 if (!strcmp(p->one->path, p->two->path))
3630 p->status = DIFF_STATUS_MODIFIED;
3631 else if (--p->one->rename_used > 0)
3632 p->status = DIFF_STATUS_COPIED;
3633 else
3634 p->status = DIFF_STATUS_RENAMED;
3635 }
3636 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3637 p->one->mode != p->two->mode ||
3638 p->one->dirty_submodule ||
3639 p->two->dirty_submodule ||
3640 is_null_sha1(p->one->sha1))
3641 p->status = DIFF_STATUS_MODIFIED;
3642 else {
3643 /* This is a "no-change" entry and should not
3644 * happen anymore, but prepare for broken callers.
3645 */
3646 error("feeding unmodified %s to diffcore",
3647 p->one->path);
3648 p->status = DIFF_STATUS_UNKNOWN;
3649 }
3650 }
3651 diff_debug_queue("resolve-rename-copy done", q);
3652 }
3653
3654 static int check_pair_status(struct diff_filepair *p)
3655 {
3656 switch (p->status) {
3657 case DIFF_STATUS_UNKNOWN:
3658 return 0;
3659 case 0:
3660 die("internal error in diff-resolve-rename-copy");
3661 default:
3662 return 1;
3663 }
3664 }
3665
3666 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3667 {
3668 int fmt = opt->output_format;
3669
3670 if (fmt & DIFF_FORMAT_CHECKDIFF)
3671 diff_flush_checkdiff(p, opt);
3672 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3673 diff_flush_raw(p, opt);
3674 else if (fmt & DIFF_FORMAT_NAME) {
3675 const char *name_a, *name_b;
3676 name_a = p->two->path;
3677 name_b = NULL;
3678 strip_prefix(opt->prefix_length, &name_a, &name_b);
3679 write_name_quoted(name_a, opt->file, opt->line_termination);
3680 }
3681 }
3682
3683 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3684 {
3685 if (fs->mode)
3686 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3687 else
3688 fprintf(file, " %s ", newdelete);
3689 write_name_quoted(fs->path, file, '\n');
3690 }
3691
3692
3693 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
3694 const char *line_prefix)
3695 {
3696 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3697 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
3698 p->two->mode, show_name ? ' ' : '\n');
3699 if (show_name) {
3700 write_name_quoted(p->two->path, file, '\n');
3701 }
3702 }
3703 }
3704
3705 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
3706 const char *line_prefix)
3707 {
3708 char *names = pprint_rename(p->one->path, p->two->path);
3709
3710 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3711 free(names);
3712 show_mode_change(file, p, 0, line_prefix);
3713 }
3714
3715 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
3716 {
3717 FILE *file = opt->file;
3718 char *line_prefix = "";
3719
3720 if (opt->output_prefix) {
3721 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
3722 line_prefix = buf->buf;
3723 }
3724
3725 switch(p->status) {
3726 case DIFF_STATUS_DELETED:
3727 fputs(line_prefix, file);
3728 show_file_mode_name(file, "delete", p->one);
3729 break;
3730 case DIFF_STATUS_ADDED:
3731 fputs(line_prefix, file);
3732 show_file_mode_name(file, "create", p->two);
3733 break;
3734 case DIFF_STATUS_COPIED:
3735 fputs(line_prefix, file);
3736 show_rename_copy(file, "copy", p, line_prefix);
3737 break;
3738 case DIFF_STATUS_RENAMED:
3739 fputs(line_prefix, file);
3740 show_rename_copy(file, "rename", p, line_prefix);
3741 break;
3742 default:
3743 if (p->score) {
3744 fprintf(file, "%s rewrite ", line_prefix);
3745 write_name_quoted(p->two->path, file, ' ');
3746 fprintf(file, "(%d%%)\n", similarity_index(p));
3747 }
3748 show_mode_change(file, p, !p->score, line_prefix);
3749 break;
3750 }
3751 }
3752
3753 struct patch_id_t {
3754 git_SHA_CTX *ctx;
3755 int patchlen;
3756 };
3757
3758 static int remove_space(char *line, int len)
3759 {
3760 int i;
3761 char *dst = line;
3762 unsigned char c;
3763
3764 for (i = 0; i < len; i++)
3765 if (!isspace((c = line[i])))
3766 *dst++ = c;
3767
3768 return dst - line;
3769 }
3770
3771 static void patch_id_consume(void *priv, char *line, unsigned long len)
3772 {
3773 struct patch_id_t *data = priv;
3774 int new_len;
3775
3776 /* Ignore line numbers when computing the SHA1 of the patch */
3777 if (!prefixcmp(line, "@@ -"))
3778 return;
3779
3780 new_len = remove_space(line, len);
3781
3782 git_SHA1_Update(data->ctx, line, new_len);
3783 data->patchlen += new_len;
3784 }
3785
3786 /* returns 0 upon success, and writes result into sha1 */
3787 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3788 {
3789 struct diff_queue_struct *q = &diff_queued_diff;
3790 int i;
3791 git_SHA_CTX ctx;
3792 struct patch_id_t data;
3793 char buffer[PATH_MAX * 4 + 20];
3794
3795 git_SHA1_Init(&ctx);
3796 memset(&data, 0, sizeof(struct patch_id_t));
3797 data.ctx = &ctx;
3798
3799 for (i = 0; i < q->nr; i++) {
3800 xpparam_t xpp;
3801 xdemitconf_t xecfg;
3802 mmfile_t mf1, mf2;
3803 struct diff_filepair *p = q->queue[i];
3804 int len1, len2;
3805
3806 memset(&xpp, 0, sizeof(xpp));
3807 memset(&xecfg, 0, sizeof(xecfg));
3808 if (p->status == 0)
3809 return error("internal diff status error");
3810 if (p->status == DIFF_STATUS_UNKNOWN)
3811 continue;
3812 if (diff_unmodified_pair(p))
3813 continue;
3814 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3815 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3816 continue;
3817 if (DIFF_PAIR_UNMERGED(p))
3818 continue;
3819
3820 diff_fill_sha1_info(p->one);
3821 diff_fill_sha1_info(p->two);
3822 if (fill_mmfile(&mf1, p->one) < 0 ||
3823 fill_mmfile(&mf2, p->two) < 0)
3824 return error("unable to read files to diff");
3825
3826 len1 = remove_space(p->one->path, strlen(p->one->path));
3827 len2 = remove_space(p->two->path, strlen(p->two->path));
3828 if (p->one->mode == 0)
3829 len1 = snprintf(buffer, sizeof(buffer),
3830 "diff--gita/%.*sb/%.*s"
3831 "newfilemode%06o"
3832 "---/dev/null"
3833 "+++b/%.*s",
3834 len1, p->one->path,
3835 len2, p->two->path,
3836 p->two->mode,
3837 len2, p->two->path);
3838 else if (p->two->mode == 0)
3839 len1 = snprintf(buffer, sizeof(buffer),
3840 "diff--gita/%.*sb/%.*s"
3841 "deletedfilemode%06o"
3842 "---a/%.*s"
3843 "+++/dev/null",
3844 len1, p->one->path,
3845 len2, p->two->path,
3846 p->one->mode,
3847 len1, p->one->path);
3848 else
3849 len1 = snprintf(buffer, sizeof(buffer),
3850 "diff--gita/%.*sb/%.*s"
3851 "---a/%.*s"
3852 "+++b/%.*s",
3853 len1, p->one->path,
3854 len2, p->two->path,
3855 len1, p->one->path,
3856 len2, p->two->path);
3857 git_SHA1_Update(&ctx, buffer, len1);
3858
3859 if (diff_filespec_is_binary(p->one) ||
3860 diff_filespec_is_binary(p->two)) {
3861 git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
3862 git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
3863 continue;
3864 }
3865
3866 xpp.flags = 0;
3867 xecfg.ctxlen = 3;
3868 xecfg.flags = XDL_EMIT_FUNCNAMES;
3869 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3870 &xpp, &xecfg);
3871 }
3872
3873 git_SHA1_Final(sha1, &ctx);
3874 return 0;
3875 }
3876
3877 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3878 {
3879 struct diff_queue_struct *q = &diff_queued_diff;
3880 int i;
3881 int result = diff_get_patch_id(options, sha1);
3882
3883 for (i = 0; i < q->nr; i++)
3884 diff_free_filepair(q->queue[i]);
3885
3886 free(q->queue);
3887 DIFF_QUEUE_CLEAR(q);
3888
3889 return result;
3890 }
3891
3892 static int is_summary_empty(const struct diff_queue_struct *q)
3893 {
3894 int i;
3895
3896 for (i = 0; i < q->nr; i++) {
3897 const struct diff_filepair *p = q->queue[i];
3898
3899 switch (p->status) {
3900 case DIFF_STATUS_DELETED:
3901 case DIFF_STATUS_ADDED:
3902 case DIFF_STATUS_COPIED:
3903 case DIFF_STATUS_RENAMED:
3904 return 0;
3905 default:
3906 if (p->score)
3907 return 0;
3908 if (p->one->mode && p->two->mode &&
3909 p->one->mode != p->two->mode)
3910 return 0;
3911 break;
3912 }
3913 }
3914 return 1;
3915 }
3916
3917 void diff_flush(struct diff_options *options)
3918 {
3919 struct diff_queue_struct *q = &diff_queued_diff;
3920 int i, output_format = options->output_format;
3921 int separator = 0;
3922
3923 /*
3924 * Order: raw, stat, summary, patch
3925 * or: name/name-status/checkdiff (other bits clear)
3926 */
3927 if (!q->nr)
3928 goto free_queue;
3929
3930 if (output_format & (DIFF_FORMAT_RAW |
3931 DIFF_FORMAT_NAME |
3932 DIFF_FORMAT_NAME_STATUS |
3933 DIFF_FORMAT_CHECKDIFF)) {
3934 for (i = 0; i < q->nr; i++) {
3935 struct diff_filepair *p = q->queue[i];
3936 if (check_pair_status(p))
3937 flush_one_pair(p, options);
3938 }
3939 separator++;
3940 }
3941
3942 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3943 struct diffstat_t diffstat;
3944
3945 memset(&diffstat, 0, sizeof(struct diffstat_t));
3946 for (i = 0; i < q->nr; i++) {
3947 struct diff_filepair *p = q->queue[i];
3948 if (check_pair_status(p))
3949 diff_flush_stat(p, options, &diffstat);
3950 }
3951 if (output_format & DIFF_FORMAT_NUMSTAT)
3952 show_numstat(&diffstat, options);
3953 if (output_format & DIFF_FORMAT_DIFFSTAT)
3954 show_stats(&diffstat, options);
3955 if (output_format & DIFF_FORMAT_SHORTSTAT)
3956 show_shortstats(&diffstat, options);
3957 free_diffstat_info(&diffstat);
3958 separator++;
3959 }
3960 if (output_format & DIFF_FORMAT_DIRSTAT)
3961 show_dirstat(options);
3962
3963 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3964 for (i = 0; i < q->nr; i++) {
3965 diff_summary(options, q->queue[i]);
3966 }
3967 separator++;
3968 }
3969
3970 if (output_format & DIFF_FORMAT_NO_OUTPUT &&
3971 DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
3972 DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3973 /*
3974 * run diff_flush_patch for the exit status. setting
3975 * options->file to /dev/null should be safe, becaue we
3976 * aren't supposed to produce any output anyway.
3977 */
3978 if (options->close_file)
3979 fclose(options->file);
3980 options->file = fopen("/dev/null", "w");
3981 if (!options->file)
3982 die_errno("Could not open /dev/null");
3983 options->close_file = 1;
3984 for (i = 0; i < q->nr; i++) {
3985 struct diff_filepair *p = q->queue[i];
3986 if (check_pair_status(p))
3987 diff_flush_patch(p, options);
3988 if (options->found_changes)
3989 break;
3990 }
3991 }
3992
3993 if (output_format & DIFF_FORMAT_PATCH) {
3994 if (separator) {
3995 putc(options->line_termination, options->file);
3996 if (options->stat_sep) {
3997 /* attach patch instead of inline */
3998 fputs(options->stat_sep, options->file);
3999 }
4000 }
4001
4002 for (i = 0; i < q->nr; i++) {
4003 struct diff_filepair *p = q->queue[i];
4004 if (check_pair_status(p))
4005 diff_flush_patch(p, options);
4006 }
4007 }
4008
4009 if (output_format & DIFF_FORMAT_CALLBACK)
4010 options->format_callback(q, options, options->format_callback_data);
4011
4012 for (i = 0; i < q->nr; i++)
4013 diff_free_filepair(q->queue[i]);
4014 free_queue:
4015 free(q->queue);
4016 DIFF_QUEUE_CLEAR(q);
4017 if (options->close_file)
4018 fclose(options->file);
4019
4020 /*
4021 * Report the content-level differences with HAS_CHANGES;
4022 * diff_addremove/diff_change does not set the bit when
4023 * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4024 */
4025 if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4026 if (options->found_changes)
4027 DIFF_OPT_SET(options, HAS_CHANGES);
4028 else
4029 DIFF_OPT_CLR(options, HAS_CHANGES);
4030 }
4031 }
4032
4033 static void diffcore_apply_filter(const char *filter)
4034 {
4035 int i;
4036 struct diff_queue_struct *q = &diff_queued_diff;
4037 struct diff_queue_struct outq;
4038 DIFF_QUEUE_CLEAR(&outq);
4039
4040 if (!filter)
4041 return;
4042
4043 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4044 int found;
4045 for (i = found = 0; !found && i < q->nr; i++) {
4046 struct diff_filepair *p = q->queue[i];
4047 if (((p->status == DIFF_STATUS_MODIFIED) &&
4048 ((p->score &&
4049 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4050 (!p->score &&
4051 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4052 ((p->status != DIFF_STATUS_MODIFIED) &&
4053 strchr(filter, p->status)))
4054 found++;
4055 }
4056 if (found)
4057 return;
4058
4059 /* otherwise we will clear the whole queue
4060 * by copying the empty outq at the end of this
4061 * function, but first clear the current entries
4062 * in the queue.
4063 */
4064 for (i = 0; i < q->nr; i++)
4065 diff_free_filepair(q->queue[i]);
4066 }
4067 else {
4068 /* Only the matching ones */
4069 for (i = 0; i < q->nr; i++) {
4070 struct diff_filepair *p = q->queue[i];
4071
4072 if (((p->status == DIFF_STATUS_MODIFIED) &&
4073 ((p->score &&
4074 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4075 (!p->score &&
4076 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4077 ((p->status != DIFF_STATUS_MODIFIED) &&
4078 strchr(filter, p->status)))
4079 diff_q(&outq, p);
4080 else
4081 diff_free_filepair(p);
4082 }
4083 }
4084 free(q->queue);
4085 *q = outq;
4086 }
4087
4088 /* Check whether two filespecs with the same mode and size are identical */
4089 static int diff_filespec_is_identical(struct diff_filespec *one,
4090 struct diff_filespec *two)
4091 {
4092 if (S_ISGITLINK(one->mode))
4093 return 0;
4094 if (diff_populate_filespec(one, 0))
4095 return 0;
4096 if (diff_populate_filespec(two, 0))
4097 return 0;
4098 return !memcmp(one->data, two->data, one->size);
4099 }
4100
4101 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4102 {
4103 int i;
4104 struct diff_queue_struct *q = &diff_queued_diff;
4105 struct diff_queue_struct outq;
4106 DIFF_QUEUE_CLEAR(&outq);
4107
4108 for (i = 0; i < q->nr; i++) {
4109 struct diff_filepair *p = q->queue[i];
4110
4111 /*
4112 * 1. Entries that come from stat info dirtiness
4113 * always have both sides (iow, not create/delete),
4114 * one side of the object name is unknown, with
4115 * the same mode and size. Keep the ones that
4116 * do not match these criteria. They have real
4117 * differences.
4118 *
4119 * 2. At this point, the file is known to be modified,
4120 * with the same mode and size, and the object
4121 * name of one side is unknown. Need to inspect
4122 * the identical contents.
4123 */
4124 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4125 !DIFF_FILE_VALID(p->two) ||
4126 (p->one->sha1_valid && p->two->sha1_valid) ||
4127 (p->one->mode != p->two->mode) ||
4128 diff_populate_filespec(p->one, 1) ||
4129 diff_populate_filespec(p->two, 1) ||
4130 (p->one->size != p->two->size) ||
4131 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4132 diff_q(&outq, p);
4133 else {
4134 /*
4135 * The caller can subtract 1 from skip_stat_unmatch
4136 * to determine how many paths were dirty only
4137 * due to stat info mismatch.
4138 */
4139 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4140 diffopt->skip_stat_unmatch++;
4141 diff_free_filepair(p);
4142 }
4143 }
4144 free(q->queue);
4145 *q = outq;
4146 }
4147
4148 static int diffnamecmp(const void *a_, const void *b_)
4149 {
4150 const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4151 const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4152 const char *name_a, *name_b;
4153
4154 name_a = a->one ? a->one->path : a->two->path;
4155 name_b = b->one ? b->one->path : b->two->path;
4156 return strcmp(name_a, name_b);
4157 }
4158
4159 void diffcore_fix_diff_index(struct diff_options *options)
4160 {
4161 struct diff_queue_struct *q = &diff_queued_diff;
4162 qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4163 }
4164
4165 void diffcore_std(struct diff_options *options)
4166 {
4167 if (options->skip_stat_unmatch)
4168 diffcore_skip_stat_unmatch(options);
4169 if (!options->found_follow) {
4170 /* See try_to_follow_renames() in tree-diff.c */
4171 if (options->break_opt != -1)
4172 diffcore_break(options->break_opt);
4173 if (options->detect_rename)
4174 diffcore_rename(options);
4175 if (options->break_opt != -1)
4176 diffcore_merge_broken();
4177 }
4178 if (options->pickaxe)
4179 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
4180 if (options->orderfile)
4181 diffcore_order(options->orderfile);
4182 if (!options->found_follow)
4183 /* See try_to_follow_renames() in tree-diff.c */
4184 diff_resolve_rename_copy();
4185 diffcore_apply_filter(options->filter);
4186
4187 if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4188 DIFF_OPT_SET(options, HAS_CHANGES);
4189 else
4190 DIFF_OPT_CLR(options, HAS_CHANGES);
4191
4192 options->found_follow = 0;
4193 }
4194
4195 int diff_result_code(struct diff_options *opt, int status)
4196 {
4197 int result = 0;
4198 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4199 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4200 return status;
4201 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4202 DIFF_OPT_TST(opt, HAS_CHANGES))
4203 result |= 01;
4204 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4205 DIFF_OPT_TST(opt, CHECK_FAILED))
4206 result |= 02;
4207 return result;
4208 }
4209
4210 /*
4211 * Shall changes to this submodule be ignored?
4212 *
4213 * Submodule changes can be configured to be ignored separately for each path,
4214 * but that configuration can be overridden from the command line.
4215 */
4216 static int is_submodule_ignored(const char *path, struct diff_options *options)
4217 {
4218 int ignored = 0;
4219 unsigned orig_flags = options->flags;
4220 if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4221 set_diffopt_flags_from_submodule_config(options, path);
4222 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4223 ignored = 1;
4224 options->flags = orig_flags;
4225 return ignored;
4226 }
4227
4228 void diff_addremove(struct diff_options *options,
4229 int addremove, unsigned mode,
4230 const unsigned char *sha1,
4231 const char *concatpath, unsigned dirty_submodule)
4232 {
4233 struct diff_filespec *one, *two;
4234
4235 if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4236 return;
4237
4238 /* This may look odd, but it is a preparation for
4239 * feeding "there are unchanged files which should
4240 * not produce diffs, but when you are doing copy
4241 * detection you would need them, so here they are"
4242 * entries to the diff-core. They will be prefixed
4243 * with something like '=' or '*' (I haven't decided
4244 * which but should not make any difference).
4245 * Feeding the same new and old to diff_change()
4246 * also has the same effect.
4247 * Before the final output happens, they are pruned after
4248 * merged into rename/copy pairs as appropriate.
4249 */
4250 if (DIFF_OPT_TST(options, REVERSE_DIFF))
4251 addremove = (addremove == '+' ? '-' :
4252 addremove == '-' ? '+' : addremove);
4253
4254 if (options->prefix &&
4255 strncmp(concatpath, options->prefix, options->prefix_length))
4256 return;
4257
4258 one = alloc_filespec(concatpath);
4259 two = alloc_filespec(concatpath);
4260
4261 if (addremove != '+')
4262 fill_filespec(one, sha1, mode);
4263 if (addremove != '-') {
4264 fill_filespec(two, sha1, mode);
4265 two->dirty_submodule = dirty_submodule;
4266 }
4267
4268 diff_queue(&diff_queued_diff, one, two);
4269 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4270 DIFF_OPT_SET(options, HAS_CHANGES);
4271 }
4272
4273 void diff_change(struct diff_options *options,
4274 unsigned old_mode, unsigned new_mode,
4275 const unsigned char *old_sha1,
4276 const unsigned char *new_sha1,
4277 const char *concatpath,
4278 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4279 {
4280 struct diff_filespec *one, *two;
4281
4282 if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4283 is_submodule_ignored(concatpath, options))
4284 return;
4285
4286 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4287 unsigned tmp;
4288 const unsigned char *tmp_c;
4289 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4290 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4291 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4292 new_dirty_submodule = tmp;
4293 }
4294
4295 if (options->prefix &&
4296 strncmp(concatpath, options->prefix, options->prefix_length))
4297 return;
4298
4299 one = alloc_filespec(concatpath);
4300 two = alloc_filespec(concatpath);
4301 fill_filespec(one, old_sha1, old_mode);
4302 fill_filespec(two, new_sha1, new_mode);
4303 one->dirty_submodule = old_dirty_submodule;
4304 two->dirty_submodule = new_dirty_submodule;
4305
4306 diff_queue(&diff_queued_diff, one, two);
4307 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4308 DIFF_OPT_SET(options, HAS_CHANGES);
4309 }
4310
4311 void diff_unmerge(struct diff_options *options,
4312 const char *path,
4313 unsigned mode, const unsigned char *sha1)
4314 {
4315 struct diff_filespec *one, *two;
4316
4317 if (options->prefix &&
4318 strncmp(path, options->prefix, options->prefix_length))
4319 return;
4320
4321 one = alloc_filespec(path);
4322 two = alloc_filespec(path);
4323 fill_filespec(one, sha1, mode);
4324 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
4325 }
4326
4327 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4328 size_t *outsize)
4329 {
4330 struct diff_tempfile *temp;
4331 const char *argv[3];
4332 const char **arg = argv;
4333 struct child_process child;
4334 struct strbuf buf = STRBUF_INIT;
4335 int err = 0;
4336
4337 temp = prepare_temp_file(spec->path, spec);
4338 *arg++ = pgm;
4339 *arg++ = temp->name;
4340 *arg = NULL;
4341
4342 memset(&child, 0, sizeof(child));
4343 child.use_shell = 1;
4344 child.argv = argv;
4345 child.out = -1;
4346 if (start_command(&child)) {
4347 remove_tempfile();
4348 return NULL;
4349 }
4350
4351 if (strbuf_read(&buf, child.out, 0) < 0)
4352 err = error("error reading from textconv command '%s'", pgm);
4353 close(child.out);
4354
4355 if (finish_command(&child) || err) {
4356 strbuf_release(&buf);
4357 remove_tempfile();
4358 return NULL;
4359 }
4360 remove_tempfile();
4361
4362 return strbuf_detach(&buf, outsize);
4363 }
4364
4365 size_t fill_textconv(struct userdiff_driver *driver,
4366 struct diff_filespec *df,
4367 char **outbuf)
4368 {
4369 size_t size;
4370
4371 if (!driver || !driver->textconv) {
4372 if (!DIFF_FILE_VALID(df)) {
4373 *outbuf = "";
4374 return 0;
4375 }
4376 if (diff_populate_filespec(df, 0))
4377 die("unable to read files to diff");
4378 *outbuf = df->data;
4379 return df->size;
4380 }
4381
4382 if (driver->textconv_cache) {
4383 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4384 &size);
4385 if (*outbuf)
4386 return size;
4387 }
4388
4389 *outbuf = run_textconv(driver->textconv, df, &size);
4390 if (!*outbuf)
4391 die("unable to read files to diff");
4392
4393 if (driver->textconv_cache) {
4394 /* ignore errors, as we might be in a readonly repository */
4395 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4396 size);
4397 /*
4398 * we could save up changes and flush them all at the end,
4399 * but we would need an extra call after all diffing is done.
4400 * Since generating a cache entry is the slow path anyway,
4401 * this extra overhead probably isn't a big deal.
4402 */
4403 notes_cache_write(driver->textconv_cache);
4404 }
4405
4406 return size;
4407 }
Dscho's miscellaneous git branches