?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'maint-1.7.1' into maint-1.7.2
[git/jrn.git] / diff.c
blob19b5bf63ed37aa49c00e63f7aec621ce601b6daa
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
35 static char diff_colors[][COLOR_MAXLEN] = {
36 GIT_COLOR_RESET,
37 GIT_COLOR_NORMAL, /* PLAIN */
38 GIT_COLOR_BOLD, /* METAINFO */
39 GIT_COLOR_CYAN, /* FRAGINFO */
40 GIT_COLOR_RED, /* OLD */
41 GIT_COLOR_GREEN, /* NEW */
42 GIT_COLOR_YELLOW, /* COMMIT */
43 GIT_COLOR_BG_RED, /* WHITESPACE */
44 GIT_COLOR_NORMAL, /* FUNCINFO */
45 };
46
47 static int parse_diff_color_slot(const char *var, int ofs)
48 {
49 if (!strcasecmp(var+ofs, "plain"))
50 return DIFF_PLAIN;
51 if (!strcasecmp(var+ofs, "meta"))
52 return DIFF_METAINFO;
53 if (!strcasecmp(var+ofs, "frag"))
54 return DIFF_FRAGINFO;
55 if (!strcasecmp(var+ofs, "old"))
56 return DIFF_FILE_OLD;
57 if (!strcasecmp(var+ofs, "new"))
58 return DIFF_FILE_NEW;
59 if (!strcasecmp(var+ofs, "commit"))
60 return DIFF_COMMIT;
61 if (!strcasecmp(var+ofs, "whitespace"))
62 return DIFF_WHITESPACE;
63 if (!strcasecmp(var+ofs, "func"))
64 return DIFF_FUNCINFO;
65 return -1;
66 }
67
68 static int git_config_rename(const char *var, const char *value)
69 {
70 if (!value)
71 return DIFF_DETECT_RENAME;
72 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
73 return DIFF_DETECT_COPY;
74 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
75 }
76
77 /*
78 * These are to give UI layer defaults.
79 * The core-level commands such as git-diff-files should
80 * never be affected by the setting of diff.renames
81 * the user happens to have in the configuration file.
82 */
83 int git_diff_ui_config(const char *var, const char *value, void *cb)
84 {
85 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
86 diff_use_color_default = git_config_colorbool(var, value, -1);
87 return 0;
88 }
89 if (!strcmp(var, "diff.renames")) {
90 diff_detect_rename_default = git_config_rename(var, value);
91 return 0;
92 }
93 if (!strcmp(var, "diff.autorefreshindex")) {
94 diff_auto_refresh_index = git_config_bool(var, value);
95 return 0;
96 }
97 if (!strcmp(var, "diff.mnemonicprefix")) {
98 diff_mnemonic_prefix = git_config_bool(var, value);
99 return 0;
100 }
101 if (!strcmp(var, "diff.noprefix")) {
102 diff_no_prefix = git_config_bool(var, value);
103 return 0;
104 }
105 if (!strcmp(var, "diff.external"))
106 return git_config_string(&external_diff_cmd_cfg, var, value);
107 if (!strcmp(var, "diff.wordregex"))
108 return git_config_string(&diff_word_regex_cfg, var, value);
109
110 return git_diff_basic_config(var, value, cb);
111 }
112
113 int git_diff_basic_config(const char *var, const char *value, void *cb)
114 {
115 if (!strcmp(var, "diff.renamelimit")) {
116 diff_rename_limit_default = git_config_int(var, value);
117 return 0;
118 }
119
120 switch (userdiff_config(var, value)) {
121 case 0: break;
122 case -1: return -1;
123 default: return 0;
124 }
125
126 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
127 int slot = parse_diff_color_slot(var, 11);
128 if (slot < 0)
129 return 0;
130 if (!value)
131 return config_error_nonbool(var);
132 color_parse(value, var, diff_colors[slot]);
133 return 0;
134 }
135
136 /* like GNU diff's --suppress-blank-empty option */
137 if (!strcmp(var, "diff.suppressblankempty") ||
138 /* for backwards compatibility */
139 !strcmp(var, "diff.suppress-blank-empty")) {
140 diff_suppress_blank_empty = git_config_bool(var, value);
141 return 0;
142 }
143
144 return git_color_default_config(var, value, cb);
145 }
146
147 static char *quote_two(const char *one, const char *two)
148 {
149 int need_one = quote_c_style(one, NULL, NULL, 1);
150 int need_two = quote_c_style(two, NULL, NULL, 1);
151 struct strbuf res = STRBUF_INIT;
152
153 if (need_one + need_two) {
154 strbuf_addch(&res, '"');
155 quote_c_style(one, &res, NULL, 1);
156 quote_c_style(two, &res, NULL, 1);
157 strbuf_addch(&res, '"');
158 } else {
159 strbuf_addstr(&res, one);
160 strbuf_addstr(&res, two);
161 }
162 return strbuf_detach(&res, NULL);
163 }
164
165 static const char *external_diff(void)
166 {
167 static const char *external_diff_cmd = NULL;
168 static int done_preparing = 0;
169
170 if (done_preparing)
171 return external_diff_cmd;
172 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
173 if (!external_diff_cmd)
174 external_diff_cmd = external_diff_cmd_cfg;
175 done_preparing = 1;
176 return external_diff_cmd;
177 }
178
179 static struct diff_tempfile {
180 const char *name; /* filename external diff should read from */
181 char hex[41];
182 char mode[10];
183 char tmp_path[PATH_MAX];
184 } diff_temp[2];
185
186 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
187
188 struct emit_callback {
189 int color_diff;
190 unsigned ws_rule;
191 int blank_at_eof_in_preimage;
192 int blank_at_eof_in_postimage;
193 int lno_in_preimage;
194 int lno_in_postimage;
195 sane_truncate_fn truncate;
196 const char **label_path;
197 struct diff_words_data *diff_words;
198 struct diff_options *opt;
199 int *found_changesp;
200 struct strbuf *header;
201 };
202
203 static int count_lines(const char *data, int size)
204 {
205 int count, ch, completely_empty = 1, nl_just_seen = 0;
206 count = 0;
207 while (0 < size--) {
208 ch = *data++;
209 if (ch == '\n') {
210 count++;
211 nl_just_seen = 1;
212 completely_empty = 0;
213 }
214 else {
215 nl_just_seen = 0;
216 completely_empty = 0;
217 }
218 }
219 if (completely_empty)
220 return 0;
221 if (!nl_just_seen)
222 count++; /* no trailing newline */
223 return count;
224 }
225
226 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
227 {
228 if (!DIFF_FILE_VALID(one)) {
229 mf->ptr = (char *)""; /* does not matter */
230 mf->size = 0;
231 return 0;
232 }
233 else if (diff_populate_filespec(one, 0))
234 return -1;
235
236 mf->ptr = one->data;
237 mf->size = one->size;
238 return 0;
239 }
240
241 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
242 {
243 char *ptr = mf->ptr;
244 long size = mf->size;
245 int cnt = 0;
246
247 if (!size)
248 return cnt;
249 ptr += size - 1; /* pointing at the very end */
250 if (*ptr != '\n')
251 ; /* incomplete line */
252 else
253 ptr--; /* skip the last LF */
254 while (mf->ptr < ptr) {
255 char *prev_eol;
256 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
257 if (*prev_eol == '\n')
258 break;
259 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
260 break;
261 cnt++;
262 ptr = prev_eol - 1;
263 }
264 return cnt;
265 }
266
267 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
268 struct emit_callback *ecbdata)
269 {
270 int l1, l2, at;
271 unsigned ws_rule = ecbdata->ws_rule;
272 l1 = count_trailing_blank(mf1, ws_rule);
273 l2 = count_trailing_blank(mf2, ws_rule);
274 if (l2 <= l1) {
275 ecbdata->blank_at_eof_in_preimage = 0;
276 ecbdata->blank_at_eof_in_postimage = 0;
277 return;
278 }
279 at = count_lines(mf1->ptr, mf1->size);
280 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
281
282 at = count_lines(mf2->ptr, mf2->size);
283 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
284 }
285
286 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
287 int first, const char *line, int len)
288 {
289 int has_trailing_newline, has_trailing_carriage_return;
290 int nofirst;
291 FILE *file = o->file;
292
293 if (o->output_prefix) {
294 struct strbuf *msg = NULL;
295 msg = o->output_prefix(o, o->output_prefix_data);
296 assert(msg);
297 fwrite(msg->buf, msg->len, 1, file);
298 }
299
300 if (len == 0) {
301 has_trailing_newline = (first == '\n');
302 has_trailing_carriage_return = (!has_trailing_newline &&
303 (first == '\r'));
304 nofirst = has_trailing_newline || has_trailing_carriage_return;
305 } else {
306 has_trailing_newline = (len > 0 && line[len-1] == '\n');
307 if (has_trailing_newline)
308 len--;
309 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
310 if (has_trailing_carriage_return)
311 len--;
312 nofirst = 0;
313 }
314
315 if (len || !nofirst) {
316 fputs(set, file);
317 if (!nofirst)
318 fputc(first, file);
319 fwrite(line, len, 1, file);
320 fputs(reset, file);
321 }
322 if (has_trailing_carriage_return)
323 fputc('\r', file);
324 if (has_trailing_newline)
325 fputc('\n', file);
326 }
327
328 static void emit_line(struct diff_options *o, const char *set, const char *reset,
329 const char *line, int len)
330 {
331 emit_line_0(o, set, reset, line[0], line+1, len-1);
332 }
333
334 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
335 {
336 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
337 ecbdata->blank_at_eof_in_preimage &&
338 ecbdata->blank_at_eof_in_postimage &&
339 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
340 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
341 return 0;
342 return ws_blank_line(line, len, ecbdata->ws_rule);
343 }
344
345 static void emit_add_line(const char *reset,
346 struct emit_callback *ecbdata,
347 const char *line, int len)
348 {
349 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
350 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
351
352 if (!*ws)
353 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
354 else if (new_blank_line_at_eof(ecbdata, line, len))
355 /* Blank line at EOF - paint '+' as well */
356 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
357 else {
358 /* Emit just the prefix, then the rest. */
359 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
360 ws_check_emit(line, len, ecbdata->ws_rule,
361 ecbdata->opt->file, set, reset, ws);
362 }
363 }
364
365 static void emit_hunk_header(struct emit_callback *ecbdata,
366 const char *line, int len)
367 {
368 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
369 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
370 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
371 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
372 static const char atat[2] = { '@', '@' };
373 const char *cp, *ep;
374 struct strbuf msgbuf = STRBUF_INIT;
375 int org_len = len;
376 int i = 1;
377
378 /*
379 * As a hunk header must begin with "@@ -<old>, +<new> @@",
380 * it always is at least 10 bytes long.
381 */
382 if (len < 10 ||
383 memcmp(line, atat, 2) ||
384 !(ep = memmem(line + 2, len - 2, atat, 2))) {
385 emit_line(ecbdata->opt, plain, reset, line, len);
386 return;
387 }
388 ep += 2; /* skip over @@ */
389
390 /* The hunk header in fraginfo color */
391 strbuf_add(&msgbuf, frag, strlen(frag));
392 strbuf_add(&msgbuf, line, ep - line);
393 strbuf_add(&msgbuf, reset, strlen(reset));
394
395 /*
396 * trailing "\r\n"
397 */
398 for ( ; i < 3; i++)
399 if (line[len - i] == '\r' || line[len - i] == '\n')
400 len--;
401
402 /* blank before the func header */
403 for (cp = ep; ep - line < len; ep++)
404 if (*ep != ' ' && *ep != '\t')
405 break;
406 if (ep != cp) {
407 strbuf_add(&msgbuf, plain, strlen(plain));
408 strbuf_add(&msgbuf, cp, ep - cp);
409 strbuf_add(&msgbuf, reset, strlen(reset));
410 }
411
412 if (ep < line + len) {
413 strbuf_add(&msgbuf, func, strlen(func));
414 strbuf_add(&msgbuf, ep, line + len - ep);
415 strbuf_add(&msgbuf, reset, strlen(reset));
416 }
417
418 strbuf_add(&msgbuf, line + len, org_len - len);
419 emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
420 strbuf_release(&msgbuf);
421 }
422
423 static struct diff_tempfile *claim_diff_tempfile(void) {
424 int i;
425 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
426 if (!diff_temp[i].name)
427 return diff_temp + i;
428 die("BUG: diff is failing to clean up its tempfiles");
429 }
430
431 static int remove_tempfile_installed;
432
433 static void remove_tempfile(void)
434 {
435 int i;
436 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
437 if (diff_temp[i].name == diff_temp[i].tmp_path)
438 unlink_or_warn(diff_temp[i].name);
439 diff_temp[i].name = NULL;
440 }
441 }
442
443 static void remove_tempfile_on_signal(int signo)
444 {
445 remove_tempfile();
446 sigchain_pop(signo);
447 raise(signo);
448 }
449
450 static void print_line_count(FILE *file, int count)
451 {
452 switch (count) {
453 case 0:
454 fprintf(file, "0,0");
455 break;
456 case 1:
457 fprintf(file, "1");
458 break;
459 default:
460 fprintf(file, "1,%d", count);
461 break;
462 }
463 }
464
465 static void emit_rewrite_lines(struct emit_callback *ecb,
466 int prefix, const char *data, int size)
467 {
468 const char *endp = NULL;
469 static const char *nneof = " No newline at end of file\n";
470 const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
471 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
472
473 while (0 < size) {
474 int len;
475
476 endp = memchr(data, '\n', size);
477 len = endp ? (endp - data + 1) : size;
478 if (prefix != '+') {
479 ecb->lno_in_preimage++;
480 emit_line_0(ecb->opt, old, reset, '-',
481 data, len);
482 } else {
483 ecb->lno_in_postimage++;
484 emit_add_line(reset, ecb, data, len);
485 }
486 size -= len;
487 data += len;
488 }
489 if (!endp) {
490 const char *plain = diff_get_color(ecb->color_diff,
491 DIFF_PLAIN);
492 emit_line_0(ecb->opt, plain, reset, '\\',
493 nneof, strlen(nneof));
494 }
495 }
496
497 static void emit_rewrite_diff(const char *name_a,
498 const char *name_b,
499 struct diff_filespec *one,
500 struct diff_filespec *two,
501 struct userdiff_driver *textconv_one,
502 struct userdiff_driver *textconv_two,
503 struct diff_options *o)
504 {
505 int lc_a, lc_b;
506 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
507 const char *name_a_tab, *name_b_tab;
508 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
509 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
510 const char *reset = diff_get_color(color_diff, DIFF_RESET);
511 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
512 const char *a_prefix, *b_prefix;
513 char *data_one, *data_two;
514 size_t size_one, size_two;
515 struct emit_callback ecbdata;
516 char *line_prefix = "";
517 struct strbuf *msgbuf;
518
519 if (o && o->output_prefix) {
520 msgbuf = o->output_prefix(o, o->output_prefix_data);
521 line_prefix = msgbuf->buf;
522 }
523
524 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
525 a_prefix = o->b_prefix;
526 b_prefix = o->a_prefix;
527 } else {
528 a_prefix = o->a_prefix;
529 b_prefix = o->b_prefix;
530 }
531
532 name_a += (*name_a == '/');
533 name_b += (*name_b == '/');
534 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
535 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
536
537 strbuf_reset(&a_name);
538 strbuf_reset(&b_name);
539 quote_two_c_style(&a_name, a_prefix, name_a, 0);
540 quote_two_c_style(&b_name, b_prefix, name_b, 0);
541
542 size_one = fill_textconv(textconv_one, one, &data_one);
543 size_two = fill_textconv(textconv_two, two, &data_two);
544
545 memset(&ecbdata, 0, sizeof(ecbdata));
546 ecbdata.color_diff = color_diff;
547 ecbdata.found_changesp = &o->found_changes;
548 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
549 ecbdata.opt = o;
550 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
551 mmfile_t mf1, mf2;
552 mf1.ptr = (char *)data_one;
553 mf2.ptr = (char *)data_two;
554 mf1.size = size_one;
555 mf2.size = size_two;
556 check_blank_at_eof(&mf1, &mf2, &ecbdata);
557 }
558 ecbdata.lno_in_preimage = 1;
559 ecbdata.lno_in_postimage = 1;
560
561 lc_a = count_lines(data_one, size_one);
562 lc_b = count_lines(data_two, size_two);
563 fprintf(o->file,
564 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
565 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
566 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
567 line_prefix, fraginfo);
568 print_line_count(o->file, lc_a);
569 fprintf(o->file, " +");
570 print_line_count(o->file, lc_b);
571 fprintf(o->file, " @@%s\n", reset);
572 if (lc_a)
573 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
574 if (lc_b)
575 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
576 if (textconv_one)
577 free((char *)data_one);
578 if (textconv_two)
579 free((char *)data_two);
580 }
581
582 struct diff_words_buffer {
583 mmfile_t text;
584 long alloc;
585 struct diff_words_orig {
586 const char *begin, *end;
587 } *orig;
588 int orig_nr, orig_alloc;
589 };
590
591 static void diff_words_append(char *line, unsigned long len,
592 struct diff_words_buffer *buffer)
593 {
594 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
595 line++;
596 len--;
597 memcpy(buffer->text.ptr + buffer->text.size, line, len);
598 buffer->text.size += len;
599 buffer->text.ptr[buffer->text.size] = '\0';
600 }
601
602 struct diff_words_style_elem
603 {
604 const char *prefix;
605 const char *suffix;
606 const char *color; /* NULL; filled in by the setup code if
607 * color is enabled */
608 };
609
610 struct diff_words_style
611 {
612 enum diff_words_type type;
613 struct diff_words_style_elem new, old, ctx;
614 const char *newline;
615 };
616
617 struct diff_words_style diff_words_styles[] = {
618 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
619 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
620 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
621 };
622
623 struct diff_words_data {
624 struct diff_words_buffer minus, plus;
625 const char *current_plus;
626 int last_minus;
627 struct diff_options *opt;
628 regex_t *word_regex;
629 enum diff_words_type type;
630 struct diff_words_style *style;
631 };
632
633 static int fn_out_diff_words_write_helper(FILE *fp,
634 struct diff_words_style_elem *st_el,
635 const char *newline,
636 size_t count, const char *buf,
637 const char *line_prefix)
638 {
639 int print = 0;
640
641 while (count) {
642 char *p = memchr(buf, '\n', count);
643 if (print)
644 fputs(line_prefix, fp);
645 if (p != buf) {
646 if (st_el->color && fputs(st_el->color, fp) < 0)
647 return -1;
648 if (fputs(st_el->prefix, fp) < 0 ||
649 fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
650 fputs(st_el->suffix, fp) < 0)
651 return -1;
652 if (st_el->color && *st_el->color
653 && fputs(GIT_COLOR_RESET, fp) < 0)
654 return -1;
655 }
656 if (!p)
657 return 0;
658 if (fputs(newline, fp) < 0)
659 return -1;
660 count -= p + 1 - buf;
661 buf = p + 1;
662 print = 1;
663 }
664 return 0;
665 }
666
667 /*
668 * '--color-words' algorithm can be described as:
669 *
670 * 1. collect a the minus/plus lines of a diff hunk, divided into
671 * minus-lines and plus-lines;
672 *
673 * 2. break both minus-lines and plus-lines into words and
674 * place them into two mmfile_t with one word for each line;
675 *
676 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
677 *
678 * And for the common parts of the both file, we output the plus side text.
679 * diff_words->current_plus is used to trace the current position of the plus file
680 * which printed. diff_words->last_minus is used to trace the last minus word
681 * printed.
682 *
683 * For '--graph' to work with '--color-words', we need to output the graph prefix
684 * on each line of color words output. Generally, there are two conditions on
685 * which we should output the prefix.
686 *
687 * 1. diff_words->last_minus == 0 &&
688 * diff_words->current_plus == diff_words->plus.text.ptr
689 *
690 * that is: the plus text must start as a new line, and if there is no minus
691 * word printed, a graph prefix must be printed.
692 *
693 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
694 * *(diff_words->current_plus - 1) == '\n'
695 *
696 * that is: a graph prefix must be printed following a '\n'
697 */
698 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
699 {
700 if ((diff_words->last_minus == 0 &&
701 diff_words->current_plus == diff_words->plus.text.ptr) ||
702 (diff_words->current_plus > diff_words->plus.text.ptr &&
703 *(diff_words->current_plus - 1) == '\n')) {
704 return 1;
705 } else {
706 return 0;
707 }
708 }
709
710 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
711 {
712 struct diff_words_data *diff_words = priv;
713 struct diff_words_style *style = diff_words->style;
714 int minus_first, minus_len, plus_first, plus_len;
715 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
716 struct diff_options *opt = diff_words->opt;
717 struct strbuf *msgbuf;
718 char *line_prefix = "";
719
720 if (line[0] != '@' || parse_hunk_header(line, len,
721 &minus_first, &minus_len, &plus_first, &plus_len))
722 return;
723
724 assert(opt);
725 if (opt->output_prefix) {
726 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
727 line_prefix = msgbuf->buf;
728 }
729
730 /* POSIX requires that first be decremented by one if len == 0... */
731 if (minus_len) {
732 minus_begin = diff_words->minus.orig[minus_first].begin;
733 minus_end =
734 diff_words->minus.orig[minus_first + minus_len - 1].end;
735 } else
736 minus_begin = minus_end =
737 diff_words->minus.orig[minus_first].end;
738
739 if (plus_len) {
740 plus_begin = diff_words->plus.orig[plus_first].begin;
741 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
742 } else
743 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
744
745 if (color_words_output_graph_prefix(diff_words)) {
746 fputs(line_prefix, diff_words->opt->file);
747 }
748 if (diff_words->current_plus != plus_begin) {
749 fn_out_diff_words_write_helper(diff_words->opt->file,
750 &style->ctx, style->newline,
751 plus_begin - diff_words->current_plus,
752 diff_words->current_plus, line_prefix);
753 if (*(plus_begin - 1) == '\n')
754 fputs(line_prefix, diff_words->opt->file);
755 }
756 if (minus_begin != minus_end) {
757 fn_out_diff_words_write_helper(diff_words->opt->file,
758 &style->old, style->newline,
759 minus_end - minus_begin, minus_begin,
760 line_prefix);
761 }
762 if (plus_begin != plus_end) {
763 fn_out_diff_words_write_helper(diff_words->opt->file,
764 &style->new, style->newline,
765 plus_end - plus_begin, plus_begin,
766 line_prefix);
767 }
768
769 diff_words->current_plus = plus_end;
770 diff_words->last_minus = minus_first;
771 }
772
773 /* This function starts looking at *begin, and returns 0 iff a word was found. */
774 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
775 int *begin, int *end)
776 {
777 if (word_regex && *begin < buffer->size) {
778 regmatch_t match[1];
779 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
780 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
781 '\n', match[0].rm_eo - match[0].rm_so);
782 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
783 *begin += match[0].rm_so;
784 return *begin >= *end;
785 }
786 return -1;
787 }
788
789 /* find the next word */
790 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
791 (*begin)++;
792 if (*begin >= buffer->size)
793 return -1;
794
795 /* find the end of the word */
796 *end = *begin + 1;
797 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
798 (*end)++;
799
800 return 0;
801 }
802
803 /*
804 * This function splits the words in buffer->text, stores the list with
805 * newline separator into out, and saves the offsets of the original words
806 * in buffer->orig.
807 */
808 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
809 regex_t *word_regex)
810 {
811 int i, j;
812 long alloc = 0;
813
814 out->size = 0;
815 out->ptr = NULL;
816
817 /* fake an empty "0th" word */
818 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
819 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
820 buffer->orig_nr = 1;
821
822 for (i = 0; i < buffer->text.size; i++) {
823 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
824 return;
825
826 /* store original boundaries */
827 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
828 buffer->orig_alloc);
829 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
830 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
831 buffer->orig_nr++;
832
833 /* store one word */
834 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
835 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
836 out->ptr[out->size + j - i] = '\n';
837 out->size += j - i + 1;
838
839 i = j - 1;
840 }
841 }
842
843 /* this executes the word diff on the accumulated buffers */
844 static void diff_words_show(struct diff_words_data *diff_words)
845 {
846 xpparam_t xpp;
847 xdemitconf_t xecfg;
848 mmfile_t minus, plus;
849 struct diff_words_style *style = diff_words->style;
850
851 struct diff_options *opt = diff_words->opt;
852 struct strbuf *msgbuf;
853 char *line_prefix = "";
854
855 assert(opt);
856 if (opt->output_prefix) {
857 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
858 line_prefix = msgbuf->buf;
859 }
860
861 /* special case: only removal */
862 if (!diff_words->plus.text.size) {
863 fputs(line_prefix, diff_words->opt->file);
864 fn_out_diff_words_write_helper(diff_words->opt->file,
865 &style->old, style->newline,
866 diff_words->minus.text.size,
867 diff_words->minus.text.ptr, line_prefix);
868 diff_words->minus.text.size = 0;
869 return;
870 }
871
872 diff_words->current_plus = diff_words->plus.text.ptr;
873 diff_words->last_minus = 0;
874
875 memset(&xpp, 0, sizeof(xpp));
876 memset(&xecfg, 0, sizeof(xecfg));
877 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
878 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
879 xpp.flags = 0;
880 /* as only the hunk header will be parsed, we need a 0-context */
881 xecfg.ctxlen = 0;
882 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
883 &xpp, &xecfg);
884 free(minus.ptr);
885 free(plus.ptr);
886 if (diff_words->current_plus != diff_words->plus.text.ptr +
887 diff_words->plus.text.size) {
888 if (color_words_output_graph_prefix(diff_words))
889 fputs(line_prefix, diff_words->opt->file);
890 fn_out_diff_words_write_helper(diff_words->opt->file,
891 &style->ctx, style->newline,
892 diff_words->plus.text.ptr + diff_words->plus.text.size
893 - diff_words->current_plus, diff_words->current_plus,
894 line_prefix);
895 }
896 diff_words->minus.text.size = diff_words->plus.text.size = 0;
897 }
898
899 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
900 static void diff_words_flush(struct emit_callback *ecbdata)
901 {
902 if (ecbdata->diff_words->minus.text.size ||
903 ecbdata->diff_words->plus.text.size)
904 diff_words_show(ecbdata->diff_words);
905 }
906
907 static void free_diff_words_data(struct emit_callback *ecbdata)
908 {
909 if (ecbdata->diff_words) {
910 diff_words_flush(ecbdata);
911 free (ecbdata->diff_words->minus.text.ptr);
912 free (ecbdata->diff_words->minus.orig);
913 free (ecbdata->diff_words->plus.text.ptr);
914 free (ecbdata->diff_words->plus.orig);
915 if (ecbdata->diff_words->word_regex) {
916 regfree(ecbdata->diff_words->word_regex);
917 free(ecbdata->diff_words->word_regex);
918 }
919 free(ecbdata->diff_words);
920 ecbdata->diff_words = NULL;
921 }
922 }
923
924 const char *diff_get_color(int diff_use_color, enum color_diff ix)
925 {
926 if (diff_use_color)
927 return diff_colors[ix];
928 return "";
929 }
930
931 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
932 {
933 const char *cp;
934 unsigned long allot;
935 size_t l = len;
936
937 if (ecb->truncate)
938 return ecb->truncate(line, len);
939 cp = line;
940 allot = l;
941 while (0 < l) {
942 (void) utf8_width(&cp, &l);
943 if (!cp)
944 break; /* truncated in the middle? */
945 }
946 return allot - l;
947 }
948
949 static void find_lno(const char *line, struct emit_callback *ecbdata)
950 {
951 const char *p;
952 ecbdata->lno_in_preimage = 0;
953 ecbdata->lno_in_postimage = 0;
954 p = strchr(line, '-');
955 if (!p)
956 return; /* cannot happen */
957 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
958 p = strchr(p, '+');
959 if (!p)
960 return; /* cannot happen */
961 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
962 }
963
964 static void fn_out_consume(void *priv, char *line, unsigned long len)
965 {
966 struct emit_callback *ecbdata = priv;
967 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
968 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
969 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
970 struct diff_options *o = ecbdata->opt;
971 char *line_prefix = "";
972 struct strbuf *msgbuf;
973
974 if (o && o->output_prefix) {
975 msgbuf = o->output_prefix(o, o->output_prefix_data);
976 line_prefix = msgbuf->buf;
977 }
978
979 if (ecbdata->header) {
980 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
981 strbuf_reset(ecbdata->header);
982 ecbdata->header = NULL;
983 }
984 *(ecbdata->found_changesp) = 1;
985
986 if (ecbdata->label_path[0]) {
987 const char *name_a_tab, *name_b_tab;
988
989 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
990 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
991
992 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
993 line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
994 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
995 line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
996 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
997 }
998
999 if (diff_suppress_blank_empty
1000 && len == 2 && line[0] == ' ' && line[1] == '\n') {
1001 line[0] = '\n';
1002 len = 1;
1003 }
1004
1005 if (line[0] == '@') {
1006 if (ecbdata->diff_words)
1007 diff_words_flush(ecbdata);
1008 len = sane_truncate_line(ecbdata, line, len);
1009 find_lno(line, ecbdata);
1010 emit_hunk_header(ecbdata, line, len);
1011 if (line[len-1] != '\n')
1012 putc('\n', ecbdata->opt->file);
1013 return;
1014 }
1015
1016 if (len < 1) {
1017 emit_line(ecbdata->opt, reset, reset, line, len);
1018 if (ecbdata->diff_words
1019 && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1020 fputs("~\n", ecbdata->opt->file);
1021 return;
1022 }
1023
1024 if (ecbdata->diff_words) {
1025 if (line[0] == '-') {
1026 diff_words_append(line, len,
1027 &ecbdata->diff_words->minus);
1028 return;
1029 } else if (line[0] == '+') {
1030 diff_words_append(line, len,
1031 &ecbdata->diff_words->plus);
1032 return;
1033 }
1034 diff_words_flush(ecbdata);
1035 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1036 emit_line(ecbdata->opt, plain, reset, line, len);
1037 fputs("~\n", ecbdata->opt->file);
1038 } else {
1039 /* don't print the prefix character */
1040 emit_line(ecbdata->opt, plain, reset, line+1, len-1);
1041 }
1042 return;
1043 }
1044
1045 if (line[0] != '+') {
1046 const char *color =
1047 diff_get_color(ecbdata->color_diff,
1048 line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1049 ecbdata->lno_in_preimage++;
1050 if (line[0] == ' ')
1051 ecbdata->lno_in_postimage++;
1052 emit_line(ecbdata->opt, color, reset, line, len);
1053 } else {
1054 ecbdata->lno_in_postimage++;
1055 emit_add_line(reset, ecbdata, line + 1, len - 1);
1056 }
1057 }
1058
1059 static char *pprint_rename(const char *a, const char *b)
1060 {
1061 const char *old = a;
1062 const char *new = b;
1063 struct strbuf name = STRBUF_INIT;
1064 int pfx_length, sfx_length;
1065 int len_a = strlen(a);
1066 int len_b = strlen(b);
1067 int a_midlen, b_midlen;
1068 int qlen_a = quote_c_style(a, NULL, NULL, 0);
1069 int qlen_b = quote_c_style(b, NULL, NULL, 0);
1070
1071 if (qlen_a || qlen_b) {
1072 quote_c_style(a, &name, NULL, 0);
1073 strbuf_addstr(&name, " => ");
1074 quote_c_style(b, &name, NULL, 0);
1075 return strbuf_detach(&name, NULL);
1076 }
1077
1078 /* Find common prefix */
1079 pfx_length = 0;
1080 while (*old && *new && *old == *new) {
1081 if (*old == '/')
1082 pfx_length = old - a + 1;
1083 old++;
1084 new++;
1085 }
1086
1087 /* Find common suffix */
1088 old = a + len_a;
1089 new = b + len_b;
1090 sfx_length = 0;
1091 while (a <= old && b <= new && *old == *new) {
1092 if (*old == '/')
1093 sfx_length = len_a - (old - a);
1094 old--;
1095 new--;
1096 }
1097
1098 /*
1099 * pfx{mid-a => mid-b}sfx
1100 * {pfx-a => pfx-b}sfx
1101 * pfx{sfx-a => sfx-b}
1102 * name-a => name-b
1103 */
1104 a_midlen = len_a - pfx_length - sfx_length;
1105 b_midlen = len_b - pfx_length - sfx_length;
1106 if (a_midlen < 0)
1107 a_midlen = 0;
1108 if (b_midlen < 0)
1109 b_midlen = 0;
1110
1111 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1112 if (pfx_length + sfx_length) {
1113 strbuf_add(&name, a, pfx_length);
1114 strbuf_addch(&name, '{');
1115 }
1116 strbuf_add(&name, a + pfx_length, a_midlen);
1117 strbuf_addstr(&name, " => ");
1118 strbuf_add(&name, b + pfx_length, b_midlen);
1119 if (pfx_length + sfx_length) {
1120 strbuf_addch(&name, '}');
1121 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1122 }
1123 return strbuf_detach(&name, NULL);
1124 }
1125
1126 struct diffstat_t {
1127 int nr;
1128 int alloc;
1129 struct diffstat_file {
1130 char *from_name;
1131 char *name;
1132 char *print_name;
1133 unsigned is_unmerged:1;
1134 unsigned is_binary:1;
1135 unsigned is_renamed:1;
1136 uintmax_t added, deleted;
1137 } **files;
1138 };
1139
1140 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1141 const char *name_a,
1142 const char *name_b)
1143 {
1144 struct diffstat_file *x;
1145 x = xcalloc(sizeof (*x), 1);
1146 if (diffstat->nr == diffstat->alloc) {
1147 diffstat->alloc = alloc_nr(diffstat->alloc);
1148 diffstat->files = xrealloc(diffstat->files,
1149 diffstat->alloc * sizeof(x));
1150 }
1151 diffstat->files[diffstat->nr++] = x;
1152 if (name_b) {
1153 x->from_name = xstrdup(name_a);
1154 x->name = xstrdup(name_b);
1155 x->is_renamed = 1;
1156 }
1157 else {
1158 x->from_name = NULL;
1159 x->name = xstrdup(name_a);
1160 }
1161 return x;
1162 }
1163
1164 static void diffstat_consume(void *priv, char *line, unsigned long len)
1165 {
1166 struct diffstat_t *diffstat = priv;
1167 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1168
1169 if (line[0] == '+')
1170 x->added++;
1171 else if (line[0] == '-')
1172 x->deleted++;
1173 }
1174
1175 const char mime_boundary_leader[] = "------------";
1176
1177 static int scale_linear(int it, int width, int max_change)
1178 {
1179 /*
1180 * make sure that at least one '-' is printed if there were deletions,
1181 * and likewise for '+'.
1182 */
1183 if (max_change < 2)
1184 return it;
1185 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1186 }
1187
1188 static void show_name(FILE *file,
1189 const char *prefix, const char *name, int len)
1190 {
1191 fprintf(file, " %s%-*s |", prefix, len, name);
1192 }
1193
1194 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1195 {
1196 if (cnt <= 0)
1197 return;
1198 fprintf(file, "%s", set);
1199 while (cnt--)
1200 putc(ch, file);
1201 fprintf(file, "%s", reset);
1202 }
1203
1204 static void fill_print_name(struct diffstat_file *file)
1205 {
1206 char *pname;
1207
1208 if (file->print_name)
1209 return;
1210
1211 if (!file->is_renamed) {
1212 struct strbuf buf = STRBUF_INIT;
1213 if (quote_c_style(file->name, &buf, NULL, 0)) {
1214 pname = strbuf_detach(&buf, NULL);
1215 } else {
1216 pname = file->name;
1217 strbuf_release(&buf);
1218 }
1219 } else {
1220 pname = pprint_rename(file->from_name, file->name);
1221 }
1222 file->print_name = pname;
1223 }
1224
1225 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1226 {
1227 int i, len, add, del, adds = 0, dels = 0;
1228 uintmax_t max_change = 0, max_len = 0;
1229 int total_files = data->nr;
1230 int width, name_width;
1231 const char *reset, *set, *add_c, *del_c;
1232 const char *line_prefix = "";
1233 struct strbuf *msg = NULL;
1234
1235 if (data->nr == 0)
1236 return;
1237
1238 if (options->output_prefix) {
1239 msg = options->output_prefix(options, options->output_prefix_data);
1240 line_prefix = msg->buf;
1241 }
1242
1243 width = options->stat_width ? options->stat_width : 80;
1244 name_width = options->stat_name_width ? options->stat_name_width : 50;
1245
1246 /* Sanity: give at least 5 columns to the graph,
1247 * but leave at least 10 columns for the name.
1248 */
1249 if (width < 25)
1250 width = 25;
1251 if (name_width < 10)
1252 name_width = 10;
1253 else if (width < name_width + 15)
1254 name_width = width - 15;
1255
1256 /* Find the longest filename and max number of changes */
1257 reset = diff_get_color_opt(options, DIFF_RESET);
1258 set = diff_get_color_opt(options, DIFF_PLAIN);
1259 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1260 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1261
1262 for (i = 0; i < data->nr; i++) {
1263 struct diffstat_file *file = data->files[i];
1264 uintmax_t change = file->added + file->deleted;
1265 fill_print_name(file);
1266 len = strlen(file->print_name);
1267 if (max_len < len)
1268 max_len = len;
1269
1270 if (file->is_binary || file->is_unmerged)
1271 continue;
1272 if (max_change < change)
1273 max_change = change;
1274 }
1275
1276 /* Compute the width of the graph part;
1277 * 10 is for one blank at the beginning of the line plus
1278 * " | count " between the name and the graph.
1279 *
1280 * From here on, name_width is the width of the name area,
1281 * and width is the width of the graph area.
1282 */
1283 name_width = (name_width < max_len) ? name_width : max_len;
1284 if (width < (name_width + 10) + max_change)
1285 width = width - (name_width + 10);
1286 else
1287 width = max_change;
1288
1289 for (i = 0; i < data->nr; i++) {
1290 const char *prefix = "";
1291 char *name = data->files[i]->print_name;
1292 uintmax_t added = data->files[i]->added;
1293 uintmax_t deleted = data->files[i]->deleted;
1294 int name_len;
1295
1296 /*
1297 * "scale" the filename
1298 */
1299 len = name_width;
1300 name_len = strlen(name);
1301 if (name_width < name_len) {
1302 char *slash;
1303 prefix = "...";
1304 len -= 3;
1305 name += name_len - len;
1306 slash = strchr(name, '/');
1307 if (slash)
1308 name = slash;
1309 }
1310
1311 if (data->files[i]->is_binary) {
1312 fprintf(options->file, "%s", line_prefix);
1313 show_name(options->file, prefix, name, len);
1314 fprintf(options->file, " Bin ");
1315 fprintf(options->file, "%s%"PRIuMAX"%s",
1316 del_c, deleted, reset);
1317 fprintf(options->file, " -> ");
1318 fprintf(options->file, "%s%"PRIuMAX"%s",
1319 add_c, added, reset);
1320 fprintf(options->file, " bytes");
1321 fprintf(options->file, "\n");
1322 continue;
1323 }
1324 else if (data->files[i]->is_unmerged) {
1325 fprintf(options->file, "%s", line_prefix);
1326 show_name(options->file, prefix, name, len);
1327 fprintf(options->file, " Unmerged\n");
1328 continue;
1329 }
1330 else if (!data->files[i]->is_renamed &&
1331 (added + deleted == 0)) {
1332 total_files--;
1333 continue;
1334 }
1335
1336 /*
1337 * scale the add/delete
1338 */
1339 add = added;
1340 del = deleted;
1341 adds += add;
1342 dels += del;
1343
1344 if (width <= max_change) {
1345 add = scale_linear(add, width, max_change);
1346 del = scale_linear(del, width, max_change);
1347 }
1348 fprintf(options->file, "%s", line_prefix);
1349 show_name(options->file, prefix, name, len);
1350 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1351 added + deleted ? " " : "");
1352 show_graph(options->file, '+', add, add_c, reset);
1353 show_graph(options->file, '-', del, del_c, reset);
1354 fprintf(options->file, "\n");
1355 }
1356 fprintf(options->file, "%s", line_prefix);
1357 fprintf(options->file,
1358 " %d files changed, %d insertions(+), %d deletions(-)\n",
1359 total_files, adds, dels);
1360 }
1361
1362 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1363 {
1364 int i, adds = 0, dels = 0, total_files = data->nr;
1365
1366 if (data->nr == 0)
1367 return;
1368
1369 for (i = 0; i < data->nr; i++) {
1370 if (!data->files[i]->is_binary &&
1371 !data->files[i]->is_unmerged) {
1372 int added = data->files[i]->added;
1373 int deleted= data->files[i]->deleted;
1374 if (!data->files[i]->is_renamed &&
1375 (added + deleted == 0)) {
1376 total_files--;
1377 } else {
1378 adds += added;
1379 dels += deleted;
1380 }
1381 }
1382 }
1383 if (options->output_prefix) {
1384 struct strbuf *msg = NULL;
1385 msg = options->output_prefix(options,
1386 options->output_prefix_data);
1387 fprintf(options->file, "%s", msg->buf);
1388 }
1389 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1390 total_files, adds, dels);
1391 }
1392
1393 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1394 {
1395 int i;
1396
1397 if (data->nr == 0)
1398 return;
1399
1400 for (i = 0; i < data->nr; i++) {
1401 struct diffstat_file *file = data->files[i];
1402
1403 if (options->output_prefix) {
1404 struct strbuf *msg = NULL;
1405 msg = options->output_prefix(options,
1406 options->output_prefix_data);
1407 fprintf(options->file, "%s", msg->buf);
1408 }
1409
1410 if (file->is_binary)
1411 fprintf(options->file, "-\t-\t");
1412 else
1413 fprintf(options->file,
1414 "%"PRIuMAX"\t%"PRIuMAX"\t",
1415 file->added, file->deleted);
1416 if (options->line_termination) {
1417 fill_print_name(file);
1418 if (!file->is_renamed)
1419 write_name_quoted(file->name, options->file,
1420 options->line_termination);
1421 else {
1422 fputs(file->print_name, options->file);
1423 putc(options->line_termination, options->file);
1424 }
1425 } else {
1426 if (file->is_renamed) {
1427 putc('\0', options->file);
1428 write_name_quoted(file->from_name, options->file, '\0');
1429 }
1430 write_name_quoted(file->name, options->file, '\0');
1431 }
1432 }
1433 }
1434
1435 struct dirstat_file {
1436 const char *name;
1437 unsigned long changed;
1438 };
1439
1440 struct dirstat_dir {
1441 struct dirstat_file *files;
1442 int alloc, nr, percent, cumulative;
1443 };
1444
1445 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1446 unsigned long changed, const char *base, int baselen)
1447 {
1448 unsigned long this_dir = 0;
1449 unsigned int sources = 0;
1450 const char *line_prefix = "";
1451 struct strbuf *msg = NULL;
1452
1453 if (opt->output_prefix) {
1454 msg = opt->output_prefix(opt, opt->output_prefix_data);
1455 line_prefix = msg->buf;
1456 }
1457
1458 while (dir->nr) {
1459 struct dirstat_file *f = dir->files;
1460 int namelen = strlen(f->name);
1461 unsigned long this;
1462 char *slash;
1463
1464 if (namelen < baselen)
1465 break;
1466 if (memcmp(f->name, base, baselen))
1467 break;
1468 slash = strchr(f->name + baselen, '/');
1469 if (slash) {
1470 int newbaselen = slash + 1 - f->name;
1471 this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1472 sources++;
1473 } else {
1474 this = f->changed;
1475 dir->files++;
1476 dir->nr--;
1477 sources += 2;
1478 }
1479 this_dir += this;
1480 }
1481
1482 /*
1483 * We don't report dirstat's for
1484 * - the top level
1485 * - or cases where everything came from a single directory
1486 * under this directory (sources == 1).
1487 */
1488 if (baselen && sources != 1) {
1489 int permille = this_dir * 1000 / changed;
1490 if (permille) {
1491 int percent = permille / 10;
1492 if (percent >= dir->percent) {
1493 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1494 percent, permille % 10, baselen, base);
1495 if (!dir->cumulative)
1496 return 0;
1497 }
1498 }
1499 }
1500 return this_dir;
1501 }
1502
1503 static int dirstat_compare(const void *_a, const void *_b)
1504 {
1505 const struct dirstat_file *a = _a;
1506 const struct dirstat_file *b = _b;
1507 return strcmp(a->name, b->name);
1508 }
1509
1510 static void show_dirstat(struct diff_options *options)
1511 {
1512 int i;
1513 unsigned long changed;
1514 struct dirstat_dir dir;
1515 struct diff_queue_struct *q = &diff_queued_diff;
1516
1517 dir.files = NULL;
1518 dir.alloc = 0;
1519 dir.nr = 0;
1520 dir.percent = options->dirstat_percent;
1521 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1522
1523 changed = 0;
1524 for (i = 0; i < q->nr; i++) {
1525 struct diff_filepair *p = q->queue[i];
1526 const char *name;
1527 unsigned long copied, added, damage;
1528
1529 name = p->one->path ? p->one->path : p->two->path;
1530
1531 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1532 diff_populate_filespec(p->one, 0);
1533 diff_populate_filespec(p->two, 0);
1534 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1535 &copied, &added);
1536 diff_free_filespec_data(p->one);
1537 diff_free_filespec_data(p->two);
1538 } else if (DIFF_FILE_VALID(p->one)) {
1539 diff_populate_filespec(p->one, 1);
1540 copied = added = 0;
1541 diff_free_filespec_data(p->one);
1542 } else if (DIFF_FILE_VALID(p->two)) {
1543 diff_populate_filespec(p->two, 1);
1544 copied = 0;
1545 added = p->two->size;
1546 diff_free_filespec_data(p->two);
1547 } else
1548 continue;
1549
1550 /*
1551 * Original minus copied is the removed material,
1552 * added is the new material. They are both damages
1553 * made to the preimage. In --dirstat-by-file mode, count
1554 * damaged files, not damaged lines. This is done by
1555 * counting only a single damaged line per file.
1556 */
1557 damage = (p->one->size - copied) + added;
1558 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1559 damage = 1;
1560
1561 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1562 dir.files[dir.nr].name = name;
1563 dir.files[dir.nr].changed = damage;
1564 changed += damage;
1565 dir.nr++;
1566 }
1567
1568 /* This can happen even with many files, if everything was renames */
1569 if (!changed)
1570 return;
1571
1572 /* Show all directories with more than x% of the changes */
1573 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1574 gather_dirstat(options, &dir, changed, "", 0);
1575 }
1576
1577 static void free_diffstat_info(struct diffstat_t *diffstat)
1578 {
1579 int i;
1580 for (i = 0; i < diffstat->nr; i++) {
1581 struct diffstat_file *f = diffstat->files[i];
1582 if (f->name != f->print_name)
1583 free(f->print_name);
1584 free(f->name);
1585 free(f->from_name);
1586 free(f);
1587 }
1588 free(diffstat->files);
1589 }
1590
1591 struct checkdiff_t {
1592 const char *filename;
1593 int lineno;
1594 int conflict_marker_size;
1595 struct diff_options *o;
1596 unsigned ws_rule;
1597 unsigned status;
1598 };
1599
1600 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1601 {
1602 char firstchar;
1603 int cnt;
1604
1605 if (len < marker_size + 1)
1606 return 0;
1607 firstchar = line[0];
1608 switch (firstchar) {
1609 case '=': case '>': case '<': case '|':
1610 break;
1611 default:
1612 return 0;
1613 }
1614 for (cnt = 1; cnt < marker_size; cnt++)
1615 if (line[cnt] != firstchar)
1616 return 0;
1617 /* line[1] thru line[marker_size-1] are same as firstchar */
1618 if (len < marker_size + 1 || !isspace(line[marker_size]))
1619 return 0;
1620 return 1;
1621 }
1622
1623 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1624 {
1625 struct checkdiff_t *data = priv;
1626 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1627 int marker_size = data->conflict_marker_size;
1628 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1629 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1630 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1631 char *err;
1632 char *line_prefix = "";
1633 struct strbuf *msgbuf;
1634
1635 assert(data->o);
1636 if (data->o->output_prefix) {
1637 msgbuf = data->o->output_prefix(data->o,
1638 data->o->output_prefix_data);
1639 line_prefix = msgbuf->buf;
1640 }
1641
1642 if (line[0] == '+') {
1643 unsigned bad;
1644 data->lineno++;
1645 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1646 data->status |= 1;
1647 fprintf(data->o->file,
1648 "%s%s:%d: leftover conflict marker\n",
1649 line_prefix, data->filename, data->lineno);
1650 }
1651 bad = ws_check(line + 1, len - 1, data->ws_rule);
1652 if (!bad)
1653 return;
1654 data->status |= bad;
1655 err = whitespace_error_string(bad);
1656 fprintf(data->o->file, "%s%s:%d: %s.\n",
1657 line_prefix, data->filename, data->lineno, err);
1658 free(err);
1659 emit_line(data->o, set, reset, line, 1);
1660 ws_check_emit(line + 1, len - 1, data->ws_rule,
1661 data->o->file, set, reset, ws);
1662 } else if (line[0] == ' ') {
1663 data->lineno++;
1664 } else if (line[0] == '@') {
1665 char *plus = strchr(line, '+');
1666 if (plus)
1667 data->lineno = strtol(plus, NULL, 10) - 1;
1668 else
1669 die("invalid diff");
1670 }
1671 }
1672
1673 static unsigned char *deflate_it(char *data,
1674 unsigned long size,
1675 unsigned long *result_size)
1676 {
1677 int bound;
1678 unsigned char *deflated;
1679 z_stream stream;
1680
1681 memset(&stream, 0, sizeof(stream));
1682 deflateInit(&stream, zlib_compression_level);
1683 bound = deflateBound(&stream, size);
1684 deflated = xmalloc(bound);
1685 stream.next_out = deflated;
1686 stream.avail_out = bound;
1687
1688 stream.next_in = (unsigned char *)data;
1689 stream.avail_in = size;
1690 while (deflate(&stream, Z_FINISH) == Z_OK)
1691 ; /* nothing */
1692 deflateEnd(&stream);
1693 *result_size = stream.total_out;
1694 return deflated;
1695 }
1696
1697 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1698 {
1699 void *cp;
1700 void *delta;
1701 void *deflated;
1702 void *data;
1703 unsigned long orig_size;
1704 unsigned long delta_size;
1705 unsigned long deflate_size;
1706 unsigned long data_size;
1707
1708 /* We could do deflated delta, or we could do just deflated two,
1709 * whichever is smaller.
1710 */
1711 delta = NULL;
1712 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1713 if (one->size && two->size) {
1714 delta = diff_delta(one->ptr, one->size,
1715 two->ptr, two->size,
1716 &delta_size, deflate_size);
1717 if (delta) {
1718 void *to_free = delta;
1719 orig_size = delta_size;
1720 delta = deflate_it(delta, delta_size, &delta_size);
1721 free(to_free);
1722 }
1723 }
1724
1725 if (delta && delta_size < deflate_size) {
1726 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
1727 free(deflated);
1728 data = delta;
1729 data_size = delta_size;
1730 }
1731 else {
1732 fprintf(file, "%sliteral %lu\n", prefix, two->size);
1733 free(delta);
1734 data = deflated;
1735 data_size = deflate_size;
1736 }
1737
1738 /* emit data encoded in base85 */
1739 cp = data;
1740 while (data_size) {
1741 int bytes = (52 < data_size) ? 52 : data_size;
1742 char line[70];
1743 data_size -= bytes;
1744 if (bytes <= 26)
1745 line[0] = bytes + 'A' - 1;
1746 else
1747 line[0] = bytes - 26 + 'a' - 1;
1748 encode_85(line + 1, cp, bytes);
1749 cp = (char *) cp + bytes;
1750 fprintf(file, "%s", prefix);
1751 fputs(line, file);
1752 fputc('\n', file);
1753 }
1754 fprintf(file, "%s\n", prefix);
1755 free(data);
1756 }
1757
1758 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1759 {
1760 fprintf(file, "%sGIT binary patch\n", prefix);
1761 emit_binary_diff_body(file, one, two, prefix);
1762 emit_binary_diff_body(file, two, one, prefix);
1763 }
1764
1765 static void diff_filespec_load_driver(struct diff_filespec *one)
1766 {
1767 if (!one->driver)
1768 one->driver = userdiff_find_by_path(one->path);
1769 if (!one->driver)
1770 one->driver = userdiff_find_by_name("default");
1771 }
1772
1773 int diff_filespec_is_binary(struct diff_filespec *one)
1774 {
1775 if (one->is_binary == -1) {
1776 diff_filespec_load_driver(one);
1777 if (one->driver->binary != -1)
1778 one->is_binary = one->driver->binary;
1779 else {
1780 if (!one->data && DIFF_FILE_VALID(one))
1781 diff_populate_filespec(one, 0);
1782 if (one->data)
1783 one->is_binary = buffer_is_binary(one->data,
1784 one->size);
1785 if (one->is_binary == -1)
1786 one->is_binary = 0;
1787 }
1788 }
1789 return one->is_binary;
1790 }
1791
1792 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1793 {
1794 diff_filespec_load_driver(one);
1795 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1796 }
1797
1798 static const char *userdiff_word_regex(struct diff_filespec *one)
1799 {
1800 diff_filespec_load_driver(one);
1801 return one->driver->word_regex;
1802 }
1803
1804 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1805 {
1806 if (!options->a_prefix)
1807 options->a_prefix = a;
1808 if (!options->b_prefix)
1809 options->b_prefix = b;
1810 }
1811
1812 struct userdiff_driver *get_textconv(struct diff_filespec *one)
1813 {
1814 if (!DIFF_FILE_VALID(one))
1815 return NULL;
1816 if (!S_ISREG(one->mode))
1817 return NULL;
1818 diff_filespec_load_driver(one);
1819 if (!one->driver->textconv)
1820 return NULL;
1821
1822 if (one->driver->textconv_want_cache && !one->driver->textconv_cache) {
1823 struct notes_cache *c = xmalloc(sizeof(*c));
1824 struct strbuf name = STRBUF_INIT;
1825
1826 strbuf_addf(&name, "textconv/%s", one->driver->name);
1827 notes_cache_init(c, name.buf, one->driver->textconv);
1828 one->driver->textconv_cache = c;
1829 }
1830
1831 return one->driver;
1832 }
1833
1834 static void builtin_diff(const char *name_a,
1835 const char *name_b,
1836 struct diff_filespec *one,
1837 struct diff_filespec *two,
1838 const char *xfrm_msg,
1839 int must_show_header,
1840 struct diff_options *o,
1841 int complete_rewrite)
1842 {
1843 mmfile_t mf1, mf2;
1844 const char *lbl[2];
1845 char *a_one, *b_two;
1846 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1847 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1848 const char *a_prefix, *b_prefix;
1849 struct userdiff_driver *textconv_one = NULL;
1850 struct userdiff_driver *textconv_two = NULL;
1851 struct strbuf header = STRBUF_INIT;
1852 struct strbuf *msgbuf;
1853 char *line_prefix = "";
1854
1855 if (o->output_prefix) {
1856 msgbuf = o->output_prefix(o, o->output_prefix_data);
1857 line_prefix = msgbuf->buf;
1858 }
1859
1860 if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1861 (!one->mode || S_ISGITLINK(one->mode)) &&
1862 (!two->mode || S_ISGITLINK(two->mode))) {
1863 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1864 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1865 show_submodule_summary(o->file, one ? one->path : two->path,
1866 one->sha1, two->sha1, two->dirty_submodule,
1867 del, add, reset);
1868 return;
1869 }
1870
1871 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1872 textconv_one = get_textconv(one);
1873 textconv_two = get_textconv(two);
1874 }
1875
1876 diff_set_mnemonic_prefix(o, "a/", "b/");
1877 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1878 a_prefix = o->b_prefix;
1879 b_prefix = o->a_prefix;
1880 } else {
1881 a_prefix = o->a_prefix;
1882 b_prefix = o->b_prefix;
1883 }
1884
1885 /* Never use a non-valid filename anywhere if at all possible */
1886 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1887 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1888
1889 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1890 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1891 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1892 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1893 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
1894 if (lbl[0][0] == '/') {
1895 /* /dev/null */
1896 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
1897 if (xfrm_msg)
1898 strbuf_addstr(&header, xfrm_msg);
1899 must_show_header = 1;
1900 }
1901 else if (lbl[1][0] == '/') {
1902 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
1903 if (xfrm_msg)
1904 strbuf_addstr(&header, xfrm_msg);
1905 must_show_header = 1;
1906 }
1907 else {
1908 if (one->mode != two->mode) {
1909 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
1910 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
1911 must_show_header = 1;
1912 }
1913 if (xfrm_msg)
1914 strbuf_addstr(&header, xfrm_msg);
1915
1916 /*
1917 * we do not run diff between different kind
1918 * of objects.
1919 */
1920 if ((one->mode ^ two->mode) & S_IFMT)
1921 goto free_ab_and_return;
1922 if (complete_rewrite &&
1923 (textconv_one || !diff_filespec_is_binary(one)) &&
1924 (textconv_two || !diff_filespec_is_binary(two))) {
1925 fprintf(o->file, "%s", header.buf);
1926 strbuf_reset(&header);
1927 emit_rewrite_diff(name_a, name_b, one, two,
1928 textconv_one, textconv_two, o);
1929 o->found_changes = 1;
1930 goto free_ab_and_return;
1931 }
1932 }
1933
1934 if (!DIFF_OPT_TST(o, TEXT) &&
1935 ( (!textconv_one && diff_filespec_is_binary(one)) ||
1936 (!textconv_two && diff_filespec_is_binary(two)) )) {
1937 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1938 die("unable to read files to diff");
1939 /* Quite common confusing case */
1940 if (mf1.size == mf2.size &&
1941 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
1942 if (must_show_header)
1943 fprintf(o->file, "%s", header.buf);
1944 goto free_ab_and_return;
1945 }
1946 fprintf(o->file, "%s", header.buf);
1947 strbuf_reset(&header);
1948 if (DIFF_OPT_TST(o, BINARY))
1949 emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
1950 else
1951 fprintf(o->file, "%sBinary files %s and %s differ\n",
1952 line_prefix, lbl[0], lbl[1]);
1953 o->found_changes = 1;
1954 }
1955 else {
1956 /* Crazy xdl interfaces.. */
1957 const char *diffopts = getenv("GIT_DIFF_OPTS");
1958 xpparam_t xpp;
1959 xdemitconf_t xecfg;
1960 struct emit_callback ecbdata;
1961 const struct userdiff_funcname *pe;
1962
1963 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || must_show_header) {
1964 fprintf(o->file, "%s", header.buf);
1965 strbuf_reset(&header);
1966 }
1967
1968 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
1969 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
1970
1971 pe = diff_funcname_pattern(one);
1972 if (!pe)
1973 pe = diff_funcname_pattern(two);
1974
1975 memset(&xpp, 0, sizeof(xpp));
1976 memset(&xecfg, 0, sizeof(xecfg));
1977 memset(&ecbdata, 0, sizeof(ecbdata));
1978 ecbdata.label_path = lbl;
1979 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1980 ecbdata.found_changesp = &o->found_changes;
1981 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1982 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1983 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1984 ecbdata.opt = o;
1985 ecbdata.header = header.len ? &header : NULL;
1986 xpp.flags = o->xdl_opts;
1987 xecfg.ctxlen = o->context;
1988 xecfg.interhunkctxlen = o->interhunkcontext;
1989 xecfg.flags = XDL_EMIT_FUNCNAMES;
1990 if (pe)
1991 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1992 if (!diffopts)
1993 ;
1994 else if (!prefixcmp(diffopts, "--unified="))
1995 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1996 else if (!prefixcmp(diffopts, "-u"))
1997 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1998 if (o->word_diff) {
1999 int i;
2000
2001 ecbdata.diff_words =
2002 xcalloc(1, sizeof(struct diff_words_data));
2003 ecbdata.diff_words->type = o->word_diff;
2004 ecbdata.diff_words->opt = o;
2005 if (!o->word_regex)
2006 o->word_regex = userdiff_word_regex(one);
2007 if (!o->word_regex)
2008 o->word_regex = userdiff_word_regex(two);
2009 if (!o->word_regex)
2010 o->word_regex = diff_word_regex_cfg;
2011 if (o->word_regex) {
2012 ecbdata.diff_words->word_regex = (regex_t *)
2013 xmalloc(sizeof(regex_t));
2014 if (regcomp(ecbdata.diff_words->word_regex,
2015 o->word_regex,
2016 REG_EXTENDED | REG_NEWLINE))
2017 die ("Invalid regular expression: %s",
2018 o->word_regex);
2019 }
2020 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2021 if (o->word_diff == diff_words_styles[i].type) {
2022 ecbdata.diff_words->style =
2023 &diff_words_styles[i];
2024 break;
2025 }
2026 }
2027 if (DIFF_OPT_TST(o, COLOR_DIFF)) {
2028 struct diff_words_style *st = ecbdata.diff_words->style;
2029 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2030 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2031 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
2032 }
2033 }
2034 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2035 &xpp, &xecfg);
2036 if (o->word_diff)
2037 free_diff_words_data(&ecbdata);
2038 if (textconv_one)
2039 free(mf1.ptr);
2040 if (textconv_two)
2041 free(mf2.ptr);
2042 xdiff_clear_find_func(&xecfg);
2043 }
2044
2045 free_ab_and_return:
2046 strbuf_release(&header);
2047 diff_free_filespec_data(one);
2048 diff_free_filespec_data(two);
2049 free(a_one);
2050 free(b_two);
2051 return;
2052 }
2053
2054 static void builtin_diffstat(const char *name_a, const char *name_b,
2055 struct diff_filespec *one,
2056 struct diff_filespec *two,
2057 struct diffstat_t *diffstat,
2058 struct diff_options *o,
2059 int complete_rewrite)
2060 {
2061 mmfile_t mf1, mf2;
2062 struct diffstat_file *data;
2063
2064 data = diffstat_add(diffstat, name_a, name_b);
2065
2066 if (!one || !two) {
2067 data->is_unmerged = 1;
2068 return;
2069 }
2070 if (complete_rewrite) {
2071 diff_populate_filespec(one, 0);
2072 diff_populate_filespec(two, 0);
2073 data->deleted = count_lines(one->data, one->size);
2074 data->added = count_lines(two->data, two->size);
2075 goto free_and_return;
2076 }
2077 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2078 die("unable to read files to diff");
2079
2080 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2081 data->is_binary = 1;
2082 data->added = mf2.size;
2083 data->deleted = mf1.size;
2084 } else {
2085 /* Crazy xdl interfaces.. */
2086 xpparam_t xpp;
2087 xdemitconf_t xecfg;
2088
2089 memset(&xpp, 0, sizeof(xpp));
2090 memset(&xecfg, 0, sizeof(xecfg));
2091 xpp.flags = o->xdl_opts;
2092 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2093 &xpp, &xecfg);
2094 }
2095
2096 free_and_return:
2097 diff_free_filespec_data(one);
2098 diff_free_filespec_data(two);
2099 }
2100
2101 static void builtin_checkdiff(const char *name_a, const char *name_b,
2102 const char *attr_path,
2103 struct diff_filespec *one,
2104 struct diff_filespec *two,
2105 struct diff_options *o)
2106 {
2107 mmfile_t mf1, mf2;
2108 struct checkdiff_t data;
2109
2110 if (!two)
2111 return;
2112
2113 memset(&data, 0, sizeof(data));
2114 data.filename = name_b ? name_b : name_a;
2115 data.lineno = 0;
2116 data.o = o;
2117 data.ws_rule = whitespace_rule(attr_path);
2118 data.conflict_marker_size = ll_merge_marker_size(attr_path);
2119
2120 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2121 die("unable to read files to diff");
2122
2123 /*
2124 * All the other codepaths check both sides, but not checking
2125 * the "old" side here is deliberate. We are checking the newly
2126 * introduced changes, and as long as the "new" side is text, we
2127 * can and should check what it introduces.
2128 */
2129 if (diff_filespec_is_binary(two))
2130 goto free_and_return;
2131 else {
2132 /* Crazy xdl interfaces.. */
2133 xpparam_t xpp;
2134 xdemitconf_t xecfg;
2135
2136 memset(&xpp, 0, sizeof(xpp));
2137 memset(&xecfg, 0, sizeof(xecfg));
2138 xecfg.ctxlen = 1; /* at least one context line */
2139 xpp.flags = 0;
2140 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2141 &xpp, &xecfg);
2142
2143 if (data.ws_rule & WS_BLANK_AT_EOF) {
2144 struct emit_callback ecbdata;
2145 int blank_at_eof;
2146
2147 ecbdata.ws_rule = data.ws_rule;
2148 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2149 blank_at_eof = ecbdata.blank_at_eof_in_preimage;
2150
2151 if (blank_at_eof) {
2152 static char *err;
2153 if (!err)
2154 err = whitespace_error_string(WS_BLANK_AT_EOF);
2155 fprintf(o->file, "%s:%d: %s.\n",
2156 data.filename, blank_at_eof, err);
2157 data.status = 1; /* report errors */
2158 }
2159 }
2160 }
2161 free_and_return:
2162 diff_free_filespec_data(one);
2163 diff_free_filespec_data(two);
2164 if (data.status)
2165 DIFF_OPT_SET(o, CHECK_FAILED);
2166 }
2167
2168 struct diff_filespec *alloc_filespec(const char *path)
2169 {
2170 int namelen = strlen(path);
2171 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2172
2173 memset(spec, 0, sizeof(*spec));
2174 spec->path = (char *)(spec + 1);
2175 memcpy(spec->path, path, namelen+1);
2176 spec->count = 1;
2177 spec->is_binary = -1;
2178 return spec;
2179 }
2180
2181 void free_filespec(struct diff_filespec *spec)
2182 {
2183 if (!--spec->count) {
2184 diff_free_filespec_data(spec);
2185 free(spec);
2186 }
2187 }
2188
2189 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2190 unsigned short mode)
2191 {
2192 if (mode) {
2193 spec->mode = canon_mode(mode);
2194 hashcpy(spec->sha1, sha1);
2195 spec->sha1_valid = !is_null_sha1(sha1);
2196 }
2197 }
2198
2199 /*
2200 * Given a name and sha1 pair, if the index tells us the file in
2201 * the work tree has that object contents, return true, so that
2202 * prepare_temp_file() does not have to inflate and extract.
2203 */
2204 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2205 {
2206 struct cache_entry *ce;
2207 struct stat st;
2208 int pos, len;
2209
2210 /*
2211 * We do not read the cache ourselves here, because the
2212 * benchmark with my previous version that always reads cache
2213 * shows that it makes things worse for diff-tree comparing
2214 * two linux-2.6 kernel trees in an already checked out work
2215 * tree. This is because most diff-tree comparisons deal with
2216 * only a small number of files, while reading the cache is
2217 * expensive for a large project, and its cost outweighs the
2218 * savings we get by not inflating the object to a temporary
2219 * file. Practically, this code only helps when we are used
2220 * by diff-cache --cached, which does read the cache before
2221 * calling us.
2222 */
2223 if (!active_cache)
2224 return 0;
2225
2226 /* We want to avoid the working directory if our caller
2227 * doesn't need the data in a normal file, this system
2228 * is rather slow with its stat/open/mmap/close syscalls,
2229 * and the object is contained in a pack file. The pack
2230 * is probably already open and will be faster to obtain
2231 * the data through than the working directory. Loose
2232 * objects however would tend to be slower as they need
2233 * to be individually opened and inflated.
2234 */
2235 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2236 return 0;
2237
2238 len = strlen(name);
2239 pos = cache_name_pos(name, len);
2240 if (pos < 0)
2241 return 0;
2242 ce = active_cache[pos];
2243
2244 /*
2245 * This is not the sha1 we are looking for, or
2246 * unreusable because it is not a regular file.
2247 */
2248 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2249 return 0;
2250
2251 /*
2252 * If ce is marked as "assume unchanged", there is no
2253 * guarantee that work tree matches what we are looking for.
2254 */
2255 if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2256 return 0;
2257
2258 /*
2259 * If ce matches the file in the work tree, we can reuse it.
2260 */
2261 if (ce_uptodate(ce) ||
2262 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2263 return 1;
2264
2265 return 0;
2266 }
2267
2268 static int populate_from_stdin(struct diff_filespec *s)
2269 {
2270 struct strbuf buf = STRBUF_INIT;
2271 size_t size = 0;
2272
2273 if (strbuf_read(&buf, 0, 0) < 0)
2274 return error("error while reading from stdin %s",
2275 strerror(errno));
2276
2277 s->should_munmap = 0;
2278 s->data = strbuf_detach(&buf, &size);
2279 s->size = size;
2280 s->should_free = 1;
2281 return 0;
2282 }
2283
2284 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2285 {
2286 int len;
2287 char *data = xmalloc(100), *dirty = "";
2288
2289 /* Are we looking at the work tree? */
2290 if (s->dirty_submodule)
2291 dirty = "-dirty";
2292
2293 len = snprintf(data, 100,
2294 "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2295 s->data = data;
2296 s->size = len;
2297 s->should_free = 1;
2298 if (size_only) {
2299 s->data = NULL;
2300 free(data);
2301 }
2302 return 0;
2303 }
2304
2305 /*
2306 * While doing rename detection and pickaxe operation, we may need to
2307 * grab the data for the blob (or file) for our own in-core comparison.
2308 * diff_filespec has data and size fields for this purpose.
2309 */
2310 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2311 {
2312 int err = 0;
2313 if (!DIFF_FILE_VALID(s))
2314 die("internal error: asking to populate invalid file.");
2315 if (S_ISDIR(s->mode))
2316 return -1;
2317
2318 if (s->data)
2319 return 0;
2320
2321 if (size_only && 0 < s->size)
2322 return 0;
2323
2324 if (S_ISGITLINK(s->mode))
2325 return diff_populate_gitlink(s, size_only);
2326
2327 if (!s->sha1_valid ||
2328 reuse_worktree_file(s->path, s->sha1, 0)) {
2329 struct strbuf buf = STRBUF_INIT;
2330 struct stat st;
2331 int fd;
2332
2333 if (!strcmp(s->path, "-"))
2334 return populate_from_stdin(s);
2335
2336 if (lstat(s->path, &st) < 0) {
2337 if (errno == ENOENT) {
2338 err_empty:
2339 err = -1;
2340 empty:
2341 s->data = (char *)"";
2342 s->size = 0;
2343 return err;
2344 }
2345 }
2346 s->size = xsize_t(st.st_size);
2347 if (!s->size)
2348 goto empty;
2349 if (S_ISLNK(st.st_mode)) {
2350 struct strbuf sb = STRBUF_INIT;
2351
2352 if (strbuf_readlink(&sb, s->path, s->size))
2353 goto err_empty;
2354 s->size = sb.len;
2355 s->data = strbuf_detach(&sb, NULL);
2356 s->should_free = 1;
2357 return 0;
2358 }
2359 if (size_only)
2360 return 0;
2361 fd = open(s->path, O_RDONLY);
2362 if (fd < 0)
2363 goto err_empty;
2364 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2365 close(fd);
2366 s->should_munmap = 1;
2367
2368 /*
2369 * Convert from working tree format to canonical git format
2370 */
2371 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2372 size_t size = 0;
2373 munmap(s->data, s->size);
2374 s->should_munmap = 0;
2375 s->data = strbuf_detach(&buf, &size);
2376 s->size = size;
2377 s->should_free = 1;
2378 }
2379 }
2380 else {
2381 enum object_type type;
2382 if (size_only)
2383 type = sha1_object_info(s->sha1, &s->size);
2384 else {
2385 s->data = read_sha1_file(s->sha1, &type, &s->size);
2386 s->should_free = 1;
2387 }
2388 }
2389 return 0;
2390 }
2391
2392 void diff_free_filespec_blob(struct diff_filespec *s)
2393 {
2394 if (s->should_free)
2395 free(s->data);
2396 else if (s->should_munmap)
2397 munmap(s->data, s->size);
2398
2399 if (s->should_free || s->should_munmap) {
2400 s->should_free = s->should_munmap = 0;
2401 s->data = NULL;
2402 }
2403 }
2404
2405 void diff_free_filespec_data(struct diff_filespec *s)
2406 {
2407 diff_free_filespec_blob(s);
2408 free(s->cnt_data);
2409 s->cnt_data = NULL;
2410 }
2411
2412 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2413 void *blob,
2414 unsigned long size,
2415 const unsigned char *sha1,
2416 int mode)
2417 {
2418 int fd;
2419 struct strbuf buf = STRBUF_INIT;
2420 struct strbuf template = STRBUF_INIT;
2421 char *path_dup = xstrdup(path);
2422 const char *base = basename(path_dup);
2423
2424 /* Generate "XXXXXX_basename.ext" */
2425 strbuf_addstr(&template, "XXXXXX_");
2426 strbuf_addstr(&template, base);
2427
2428 fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2429 strlen(base) + 1);
2430 if (fd < 0)
2431 die_errno("unable to create temp-file");
2432 if (convert_to_working_tree(path,
2433 (const char *)blob, (size_t)size, &buf)) {
2434 blob = buf.buf;
2435 size = buf.len;
2436 }
2437 if (write_in_full(fd, blob, size) != size)
2438 die_errno("unable to write temp-file");
2439 close(fd);
2440 temp->name = temp->tmp_path;
2441 strcpy(temp->hex, sha1_to_hex(sha1));
2442 temp->hex[40] = 0;
2443 sprintf(temp->mode, "%06o", mode);
2444 strbuf_release(&buf);
2445 strbuf_release(&template);
2446 free(path_dup);
2447 }
2448
2449 static struct diff_tempfile *prepare_temp_file(const char *name,
2450 struct diff_filespec *one)
2451 {
2452 struct diff_tempfile *temp = claim_diff_tempfile();
2453
2454 if (!DIFF_FILE_VALID(one)) {
2455 not_a_valid_file:
2456 /* A '-' entry produces this for file-2, and
2457 * a '+' entry produces this for file-1.
2458 */
2459 temp->name = "/dev/null";
2460 strcpy(temp->hex, ".");
2461 strcpy(temp->mode, ".");
2462 return temp;
2463 }
2464
2465 if (!remove_tempfile_installed) {
2466 atexit(remove_tempfile);
2467 sigchain_push_common(remove_tempfile_on_signal);
2468 remove_tempfile_installed = 1;
2469 }
2470
2471 if (!one->sha1_valid ||
2472 reuse_worktree_file(name, one->sha1, 1)) {
2473 struct stat st;
2474 if (lstat(name, &st) < 0) {
2475 if (errno == ENOENT)
2476 goto not_a_valid_file;
2477 die_errno("stat(%s)", name);
2478 }
2479 if (S_ISLNK(st.st_mode)) {
2480 struct strbuf sb = STRBUF_INIT;
2481 if (strbuf_readlink(&sb, name, st.st_size) < 0)
2482 die_errno("readlink(%s)", name);
2483 prep_temp_blob(name, temp, sb.buf, sb.len,
2484 (one->sha1_valid ?
2485 one->sha1 : null_sha1),
2486 (one->sha1_valid ?
2487 one->mode : S_IFLNK));
2488 strbuf_release(&sb);
2489 }
2490 else {
2491 /* we can borrow from the file in the work tree */
2492 temp->name = name;
2493 if (!one->sha1_valid)
2494 strcpy(temp->hex, sha1_to_hex(null_sha1));
2495 else
2496 strcpy(temp->hex, sha1_to_hex(one->sha1));
2497 /* Even though we may sometimes borrow the
2498 * contents from the work tree, we always want
2499 * one->mode. mode is trustworthy even when
2500 * !(one->sha1_valid), as long as
2501 * DIFF_FILE_VALID(one).
2502 */
2503 sprintf(temp->mode, "%06o", one->mode);
2504 }
2505 return temp;
2506 }
2507 else {
2508 if (diff_populate_filespec(one, 0))
2509 die("cannot read data blob for %s", one->path);
2510 prep_temp_blob(name, temp, one->data, one->size,
2511 one->sha1, one->mode);
2512 }
2513 return temp;
2514 }
2515
2516 /* An external diff command takes:
2517 *
2518 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2519 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2520 *
2521 */
2522 static void run_external_diff(const char *pgm,
2523 const char *name,
2524 const char *other,
2525 struct diff_filespec *one,
2526 struct diff_filespec *two,
2527 const char *xfrm_msg,
2528 int complete_rewrite)
2529 {
2530 const char *spawn_arg[10];
2531 int retval;
2532 const char **arg = &spawn_arg[0];
2533
2534 if (one && two) {
2535 struct diff_tempfile *temp_one, *temp_two;
2536 const char *othername = (other ? other : name);
2537 temp_one = prepare_temp_file(name, one);
2538 temp_two = prepare_temp_file(othername, two);
2539 *arg++ = pgm;
2540 *arg++ = name;
2541 *arg++ = temp_one->name;
2542 *arg++ = temp_one->hex;
2543 *arg++ = temp_one->mode;
2544 *arg++ = temp_two->name;
2545 *arg++ = temp_two->hex;
2546 *arg++ = temp_two->mode;
2547 if (other) {
2548 *arg++ = other;
2549 *arg++ = xfrm_msg;
2550 }
2551 } else {
2552 *arg++ = pgm;
2553 *arg++ = name;
2554 }
2555 *arg = NULL;
2556 fflush(NULL);
2557 retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2558 remove_tempfile();
2559 if (retval) {
2560 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2561 exit(1);
2562 }
2563 }
2564
2565 static int similarity_index(struct diff_filepair *p)
2566 {
2567 return p->score * 100 / MAX_SCORE;
2568 }
2569
2570 static void fill_metainfo(struct strbuf *msg,
2571 const char *name,
2572 const char *other,
2573 struct diff_filespec *one,
2574 struct diff_filespec *two,
2575 struct diff_options *o,
2576 struct diff_filepair *p,
2577 int *must_show_header,
2578 int use_color)
2579 {
2580 const char *set = diff_get_color(use_color, DIFF_METAINFO);
2581 const char *reset = diff_get_color(use_color, DIFF_RESET);
2582 struct strbuf *msgbuf;
2583 char *line_prefix = "";
2584
2585 *must_show_header = 1;
2586 if (o->output_prefix) {
2587 msgbuf = o->output_prefix(o, o->output_prefix_data);
2588 line_prefix = msgbuf->buf;
2589 }
2590 strbuf_init(msg, PATH_MAX * 2 + 300);
2591 switch (p->status) {
2592 case DIFF_STATUS_COPIED:
2593 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2594 line_prefix, set, similarity_index(p));
2595 strbuf_addf(msg, "%s\n%s%scopy from ",
2596 reset, line_prefix, set);
2597 quote_c_style(name, msg, NULL, 0);
2598 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2599 quote_c_style(other, msg, NULL, 0);
2600 strbuf_addf(msg, "%s\n", reset);
2601 break;
2602 case DIFF_STATUS_RENAMED:
2603 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2604 line_prefix, set, similarity_index(p));
2605 strbuf_addf(msg, "%s\n%s%srename from ",
2606 reset, line_prefix, set);
2607 quote_c_style(name, msg, NULL, 0);
2608 strbuf_addf(msg, "%s\n%s%srename to ",
2609 reset, line_prefix, set);
2610 quote_c_style(other, msg, NULL, 0);
2611 strbuf_addf(msg, "%s\n", reset);
2612 break;
2613 case DIFF_STATUS_MODIFIED:
2614 if (p->score) {
2615 strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2616 line_prefix,
2617 set, similarity_index(p), reset);
2618 break;
2619 }
2620 /* fallthru */
2621 default:
2622 *must_show_header = 0;
2623 }
2624 if (one && two && hashcmp(one->sha1, two->sha1)) {
2625 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2626
2627 if (DIFF_OPT_TST(o, BINARY)) {
2628 mmfile_t mf;
2629 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2630 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2631 abbrev = 40;
2632 }
2633 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2634 find_unique_abbrev(one->sha1, abbrev));
2635 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2636 if (one->mode == two->mode)
2637 strbuf_addf(msg, " %06o", one->mode);
2638 strbuf_addf(msg, "%s\n", reset);
2639 }
2640 }
2641
2642 static void run_diff_cmd(const char *pgm,
2643 const char *name,
2644 const char *other,
2645 const char *attr_path,
2646 struct diff_filespec *one,
2647 struct diff_filespec *two,
2648 struct strbuf *msg,
2649 struct diff_options *o,
2650 struct diff_filepair *p)
2651 {
2652 const char *xfrm_msg = NULL;
2653 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2654 int must_show_header = 0;
2655
2656 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2657 pgm = NULL;
2658 else {
2659 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2660 if (drv && drv->external)
2661 pgm = drv->external;
2662 }
2663
2664 if (msg) {
2665 /*
2666 * don't use colors when the header is intended for an
2667 * external diff driver
2668 */
2669 fill_metainfo(msg, name, other, one, two, o, p,
2670 &must_show_header,
2671 DIFF_OPT_TST(o, COLOR_DIFF) && !pgm);
2672 xfrm_msg = msg->len ? msg->buf : NULL;
2673 }
2674
2675 if (pgm) {
2676 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2677 complete_rewrite);
2678 return;
2679 }
2680 if (one && two)
2681 builtin_diff(name, other ? other : name,
2682 one, two, xfrm_msg, must_show_header,
2683 o, complete_rewrite);
2684 else
2685 fprintf(o->file, "* Unmerged path %s\n", name);
2686 }
2687
2688 static void diff_fill_sha1_info(struct diff_filespec *one)
2689 {
2690 if (DIFF_FILE_VALID(one)) {
2691 if (!one->sha1_valid) {
2692 struct stat st;
2693 if (!strcmp(one->path, "-")) {
2694 hashcpy(one->sha1, null_sha1);
2695 return;
2696 }
2697 if (lstat(one->path, &st) < 0)
2698 die_errno("stat '%s'", one->path);
2699 if (index_path(one->sha1, one->path, &st, 0))
2700 die("cannot hash %s", one->path);
2701 }
2702 }
2703 else
2704 hashclr(one->sha1);
2705 }
2706
2707 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2708 {
2709 /* Strip the prefix but do not molest /dev/null and absolute paths */
2710 if (*namep && **namep != '/') {
2711 *namep += prefix_length;
2712 if (**namep == '/')
2713 ++*namep;
2714 }
2715 if (*otherp && **otherp != '/') {
2716 *otherp += prefix_length;
2717 if (**otherp == '/')
2718 ++*otherp;
2719 }
2720 }
2721
2722 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2723 {
2724 const char *pgm = external_diff();
2725 struct strbuf msg;
2726 struct diff_filespec *one = p->one;
2727 struct diff_filespec *two = p->two;
2728 const char *name;
2729 const char *other;
2730 const char *attr_path;
2731
2732 name = p->one->path;
2733 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2734 attr_path = name;
2735 if (o->prefix_length)
2736 strip_prefix(o->prefix_length, &name, &other);
2737
2738 if (DIFF_PAIR_UNMERGED(p)) {
2739 run_diff_cmd(pgm, name, NULL, attr_path,
2740 NULL, NULL, NULL, o, p);
2741 return;
2742 }
2743
2744 diff_fill_sha1_info(one);
2745 diff_fill_sha1_info(two);
2746
2747 if (!pgm &&
2748 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2749 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2750 /*
2751 * a filepair that changes between file and symlink
2752 * needs to be split into deletion and creation.
2753 */
2754 struct diff_filespec *null = alloc_filespec(two->path);
2755 run_diff_cmd(NULL, name, other, attr_path,
2756 one, null, &msg, o, p);
2757 free(null);
2758 strbuf_release(&msg);
2759
2760 null = alloc_filespec(one->path);
2761 run_diff_cmd(NULL, name, other, attr_path,
2762 null, two, &msg, o, p);
2763 free(null);
2764 }
2765 else
2766 run_diff_cmd(pgm, name, other, attr_path,
2767 one, two, &msg, o, p);
2768
2769 strbuf_release(&msg);
2770 }
2771
2772 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2773 struct diffstat_t *diffstat)
2774 {
2775 const char *name;
2776 const char *other;
2777 int complete_rewrite = 0;
2778
2779 if (DIFF_PAIR_UNMERGED(p)) {
2780 /* unmerged */
2781 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2782 return;
2783 }
2784
2785 name = p->one->path;
2786 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2787
2788 if (o->prefix_length)
2789 strip_prefix(o->prefix_length, &name, &other);
2790
2791 diff_fill_sha1_info(p->one);
2792 diff_fill_sha1_info(p->two);
2793
2794 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2795 complete_rewrite = 1;
2796 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2797 }
2798
2799 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2800 {
2801 const char *name;
2802 const char *other;
2803 const char *attr_path;
2804
2805 if (DIFF_PAIR_UNMERGED(p)) {
2806 /* unmerged */
2807 return;
2808 }
2809
2810 name = p->one->path;
2811 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2812 attr_path = other ? other : name;
2813
2814 if (o->prefix_length)
2815 strip_prefix(o->prefix_length, &name, &other);
2816
2817 diff_fill_sha1_info(p->one);
2818 diff_fill_sha1_info(p->two);
2819
2820 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2821 }
2822
2823 void diff_setup(struct diff_options *options)
2824 {
2825 memset(options, 0, sizeof(*options));
2826
2827 options->file = stdout;
2828
2829 options->line_termination = '\n';
2830 options->break_opt = -1;
2831 options->rename_limit = -1;
2832 options->dirstat_percent = 3;
2833 options->context = 3;
2834
2835 options->change = diff_change;
2836 options->add_remove = diff_addremove;
2837 if (diff_use_color_default > 0)
2838 DIFF_OPT_SET(options, COLOR_DIFF);
2839 options->detect_rename = diff_detect_rename_default;
2840
2841 if (diff_no_prefix) {
2842 options->a_prefix = options->b_prefix = "";
2843 } else if (!diff_mnemonic_prefix) {
2844 options->a_prefix = "a/";
2845 options->b_prefix = "b/";
2846 }
2847 }
2848
2849 int diff_setup_done(struct diff_options *options)
2850 {
2851 int count = 0;
2852
2853 if (options->output_format & DIFF_FORMAT_NAME)
2854 count++;
2855 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2856 count++;
2857 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2858 count++;
2859 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2860 count++;
2861 if (count > 1)
2862 die("--name-only, --name-status, --check and -s are mutually exclusive");
2863
2864 /*
2865 * Most of the time we can say "there are changes"
2866 * only by checking if there are changed paths, but
2867 * --ignore-whitespace* options force us to look
2868 * inside contents.
2869 */
2870
2871 if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2872 DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2873 DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2874 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2875 else
2876 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2877
2878 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2879 options->detect_rename = DIFF_DETECT_COPY;
2880
2881 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2882 options->prefix = NULL;
2883 if (options->prefix)
2884 options->prefix_length = strlen(options->prefix);
2885 else
2886 options->prefix_length = 0;
2887
2888 if (options->output_format & (DIFF_FORMAT_NAME |
2889 DIFF_FORMAT_NAME_STATUS |
2890 DIFF_FORMAT_CHECKDIFF |
2891 DIFF_FORMAT_NO_OUTPUT))
2892 options->output_format &= ~(DIFF_FORMAT_RAW |
2893 DIFF_FORMAT_NUMSTAT |
2894 DIFF_FORMAT_DIFFSTAT |
2895 DIFF_FORMAT_SHORTSTAT |
2896 DIFF_FORMAT_DIRSTAT |
2897 DIFF_FORMAT_SUMMARY |
2898 DIFF_FORMAT_PATCH);
2899
2900 /*
2901 * These cases always need recursive; we do not drop caller-supplied
2902 * recursive bits for other formats here.
2903 */
2904 if (options->output_format & (DIFF_FORMAT_PATCH |
2905 DIFF_FORMAT_NUMSTAT |
2906 DIFF_FORMAT_DIFFSTAT |
2907 DIFF_FORMAT_SHORTSTAT |
2908 DIFF_FORMAT_DIRSTAT |
2909 DIFF_FORMAT_SUMMARY |
2910 DIFF_FORMAT_CHECKDIFF))
2911 DIFF_OPT_SET(options, RECURSIVE);
2912 /*
2913 * Also pickaxe would not work very well if you do not say recursive
2914 */
2915 if (options->pickaxe)
2916 DIFF_OPT_SET(options, RECURSIVE);
2917 /*
2918 * When patches are generated, submodules diffed against the work tree
2919 * must be checked for dirtiness too so it can be shown in the output
2920 */
2921 if (options->output_format & DIFF_FORMAT_PATCH)
2922 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2923
2924 if (options->detect_rename && options->rename_limit < 0)
2925 options->rename_limit = diff_rename_limit_default;
2926 if (options->setup & DIFF_SETUP_USE_CACHE) {
2927 if (!active_cache)
2928 /* read-cache does not die even when it fails
2929 * so it is safe for us to do this here. Also
2930 * it does not smudge active_cache or active_nr
2931 * when it fails, so we do not have to worry about
2932 * cleaning it up ourselves either.
2933 */
2934 read_cache();
2935 }
2936 if (options->abbrev <= 0 || 40 < options->abbrev)
2937 options->abbrev = 40; /* full */
2938
2939 /*
2940 * It does not make sense to show the first hit we happened
2941 * to have found. It does not make sense not to return with
2942 * exit code in such a case either.
2943 */
2944 if (DIFF_OPT_TST(options, QUICK)) {
2945 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2946 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2947 }
2948
2949 return 0;
2950 }
2951
2952 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2953 {
2954 char c, *eq;
2955 int len;
2956
2957 if (*arg != '-')
2958 return 0;
2959 c = *++arg;
2960 if (!c)
2961 return 0;
2962 if (c == arg_short) {
2963 c = *++arg;
2964 if (!c)
2965 return 1;
2966 if (val && isdigit(c)) {
2967 char *end;
2968 int n = strtoul(arg, &end, 10);
2969 if (*end)
2970 return 0;
2971 *val = n;
2972 return 1;
2973 }
2974 return 0;
2975 }
2976 if (c != '-')
2977 return 0;
2978 arg++;
2979 eq = strchr(arg, '=');
2980 if (eq)
2981 len = eq - arg;
2982 else
2983 len = strlen(arg);
2984 if (!len || strncmp(arg, arg_long, len))
2985 return 0;
2986 if (eq) {
2987 int n;
2988 char *end;
2989 if (!isdigit(*++eq))
2990 return 0;
2991 n = strtoul(eq, &end, 10);
2992 if (*end)
2993 return 0;
2994 *val = n;
2995 }
2996 return 1;
2997 }
2998
2999 static int diff_scoreopt_parse(const char *opt);
3000
3001 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3002 {
3003 const char *arg = av[0];
3004
3005 /* Output format options */
3006 if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3007 options->output_format |= DIFF_FORMAT_PATCH;
3008 else if (opt_arg(arg, 'U', "unified", &options->context))
3009 options->output_format |= DIFF_FORMAT_PATCH;
3010 else if (!strcmp(arg, "--raw"))
3011 options->output_format |= DIFF_FORMAT_RAW;
3012 else if (!strcmp(arg, "--patch-with-raw"))
3013 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3014 else if (!strcmp(arg, "--numstat"))
3015 options->output_format |= DIFF_FORMAT_NUMSTAT;
3016 else if (!strcmp(arg, "--shortstat"))
3017 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3018 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
3019 options->output_format |= DIFF_FORMAT_DIRSTAT;
3020 else if (!strcmp(arg, "--cumulative")) {
3021 options->output_format |= DIFF_FORMAT_DIRSTAT;
3022 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
3023 } else if (opt_arg(arg, 0, "dirstat-by-file",
3024 &options->dirstat_percent)) {
3025 options->output_format |= DIFF_FORMAT_DIRSTAT;
3026 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
3027 }
3028 else if (!strcmp(arg, "--check"))
3029 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3030 else if (!strcmp(arg, "--summary"))
3031 options->output_format |= DIFF_FORMAT_SUMMARY;
3032 else if (!strcmp(arg, "--patch-with-stat"))
3033 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3034 else if (!strcmp(arg, "--name-only"))
3035 options->output_format |= DIFF_FORMAT_NAME;
3036 else if (!strcmp(arg, "--name-status"))
3037 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3038 else if (!strcmp(arg, "-s"))
3039 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3040 else if (!prefixcmp(arg, "--stat")) {
3041 char *end;
3042 int width = options->stat_width;
3043 int name_width = options->stat_name_width;
3044 arg += 6;
3045 end = (char *)arg;
3046
3047 switch (*arg) {
3048 case '-':
3049 if (!prefixcmp(arg, "-width="))
3050 width = strtoul(arg + 7, &end, 10);
3051 else if (!prefixcmp(arg, "-name-width="))
3052 name_width = strtoul(arg + 12, &end, 10);
3053 break;
3054 case '=':
3055 width = strtoul(arg+1, &end, 10);
3056 if (*end == ',')
3057 name_width = strtoul(end+1, &end, 10);
3058 }
3059
3060 /* Important! This checks all the error cases! */
3061 if (*end)
3062 return 0;
3063 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3064 options->stat_name_width = name_width;
3065 options->stat_width = width;
3066 }
3067
3068 /* renames options */
3069 else if (!prefixcmp(arg, "-B")) {
3070 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3071 return -1;
3072 }
3073 else if (!prefixcmp(arg, "-M")) {
3074 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3075 return -1;
3076 options->detect_rename = DIFF_DETECT_RENAME;
3077 }
3078 else if (!prefixcmp(arg, "-C")) {
3079 if (options->detect_rename == DIFF_DETECT_COPY)
3080 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3081 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3082 return -1;
3083 options->detect_rename = DIFF_DETECT_COPY;
3084 }
3085 else if (!strcmp(arg, "--no-renames"))
3086 options->detect_rename = 0;
3087 else if (!strcmp(arg, "--relative"))
3088 DIFF_OPT_SET(options, RELATIVE_NAME);
3089 else if (!prefixcmp(arg, "--relative=")) {
3090 DIFF_OPT_SET(options, RELATIVE_NAME);
3091 options->prefix = arg + 11;
3092 }
3093
3094 /* xdiff options */
3095 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3096 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3097 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3098 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3099 else if (!strcmp(arg, "--ignore-space-at-eol"))
3100 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3101 else if (!strcmp(arg, "--patience"))
3102 DIFF_XDL_SET(options, PATIENCE_DIFF);
3103
3104 /* flags options */
3105 else if (!strcmp(arg, "--binary")) {
3106 options->output_format |= DIFF_FORMAT_PATCH;
3107 DIFF_OPT_SET(options, BINARY);
3108 }
3109 else if (!strcmp(arg, "--full-index"))
3110 DIFF_OPT_SET(options, FULL_INDEX);
3111 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3112 DIFF_OPT_SET(options, TEXT);
3113 else if (!strcmp(arg, "-R"))
3114 DIFF_OPT_SET(options, REVERSE_DIFF);
3115 else if (!strcmp(arg, "--find-copies-harder"))
3116 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3117 else if (!strcmp(arg, "--follow"))
3118 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3119 else if (!strcmp(arg, "--color"))
3120 DIFF_OPT_SET(options, COLOR_DIFF);
3121 else if (!prefixcmp(arg, "--color=")) {
3122 int value = git_config_colorbool(NULL, arg+8, -1);
3123 if (value == 0)
3124 DIFF_OPT_CLR(options, COLOR_DIFF);
3125 else if (value > 0)
3126 DIFF_OPT_SET(options, COLOR_DIFF);
3127 else
3128 return error("option `color' expects \"always\", \"auto\", or \"never\"");
3129 }
3130 else if (!strcmp(arg, "--no-color"))
3131 DIFF_OPT_CLR(options, COLOR_DIFF);
3132 else if (!strcmp(arg, "--color-words")) {
3133 DIFF_OPT_SET(options, COLOR_DIFF);
3134 options->word_diff = DIFF_WORDS_COLOR;
3135 }
3136 else if (!prefixcmp(arg, "--color-words=")) {
3137 DIFF_OPT_SET(options, COLOR_DIFF);
3138 options->word_diff = DIFF_WORDS_COLOR;
3139 options->word_regex = arg + 14;
3140 }
3141 else if (!strcmp(arg, "--word-diff")) {
3142 if (options->word_diff == DIFF_WORDS_NONE)
3143 options->word_diff = DIFF_WORDS_PLAIN;
3144 }
3145 else if (!prefixcmp(arg, "--word-diff=")) {
3146 const char *type = arg + 12;
3147 if (!strcmp(type, "plain"))
3148 options->word_diff = DIFF_WORDS_PLAIN;
3149 else if (!strcmp(type, "color")) {
3150 DIFF_OPT_SET(options, COLOR_DIFF);
3151 options->word_diff = DIFF_WORDS_COLOR;
3152 }
3153 else if (!strcmp(type, "porcelain"))
3154 options->word_diff = DIFF_WORDS_PORCELAIN;
3155 else if (!strcmp(type, "none"))
3156 options->word_diff = DIFF_WORDS_NONE;
3157 else
3158 die("bad --word-diff argument: %s", type);
3159 }
3160 else if (!prefixcmp(arg, "--word-diff-regex=")) {
3161 if (options->word_diff == DIFF_WORDS_NONE)
3162 options->word_diff = DIFF_WORDS_PLAIN;
3163 options->word_regex = arg + 18;
3164 }
3165 else if (!strcmp(arg, "--exit-code"))
3166 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3167 else if (!strcmp(arg, "--quiet"))
3168 DIFF_OPT_SET(options, QUICK);
3169 else if (!strcmp(arg, "--ext-diff"))
3170 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3171 else if (!strcmp(arg, "--no-ext-diff"))
3172 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3173 else if (!strcmp(arg, "--textconv"))
3174 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3175 else if (!strcmp(arg, "--no-textconv"))
3176 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3177 else if (!strcmp(arg, "--ignore-submodules"))
3178 handle_ignore_submodules_arg(options, "all");
3179 else if (!prefixcmp(arg, "--ignore-submodules="))
3180 handle_ignore_submodules_arg(options, arg + 20);
3181 else if (!strcmp(arg, "--submodule"))
3182 DIFF_OPT_SET(options, SUBMODULE_LOG);
3183 else if (!prefixcmp(arg, "--submodule=")) {
3184 if (!strcmp(arg + 12, "log"))
3185 DIFF_OPT_SET(options, SUBMODULE_LOG);
3186 }
3187
3188 /* misc options */
3189 else if (!strcmp(arg, "-z"))
3190 options->line_termination = 0;
3191 else if (!prefixcmp(arg, "-l"))
3192 options->rename_limit = strtoul(arg+2, NULL, 10);
3193 else if (!prefixcmp(arg, "-S"))
3194 options->pickaxe = arg + 2;
3195 else if (!strcmp(arg, "--pickaxe-all"))
3196 options->pickaxe_opts = DIFF_PICKAXE_ALL;
3197 else if (!strcmp(arg, "--pickaxe-regex"))
3198 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
3199 else if (!prefixcmp(arg, "-O"))
3200 options->orderfile = arg + 2;
3201 else if (!prefixcmp(arg, "--diff-filter="))
3202 options->filter = arg + 14;
3203 else if (!strcmp(arg, "--abbrev"))
3204 options->abbrev = DEFAULT_ABBREV;
3205 else if (!prefixcmp(arg, "--abbrev=")) {
3206 options->abbrev = strtoul(arg + 9, NULL, 10);
3207 if (options->abbrev < MINIMUM_ABBREV)
3208 options->abbrev = MINIMUM_ABBREV;
3209 else if (40 < options->abbrev)
3210 options->abbrev = 40;
3211 }
3212 else if (!prefixcmp(arg, "--src-prefix="))
3213 options->a_prefix = arg + 13;
3214 else if (!prefixcmp(arg, "--dst-prefix="))
3215 options->b_prefix = arg + 13;
3216 else if (!strcmp(arg, "--no-prefix"))
3217 options->a_prefix = options->b_prefix = "";
3218 else if (opt_arg(arg, '\0', "inter-hunk-context",
3219 &options->interhunkcontext))
3220 ;
3221 else if (!prefixcmp(arg, "--output=")) {
3222 options->file = fopen(arg + strlen("--output="), "w");
3223 if (!options->file)
3224 die_errno("Could not open '%s'", arg + strlen("--output="));
3225 options->close_file = 1;
3226 } else
3227 return 0;
3228 return 1;
3229 }
3230
3231 static int parse_num(const char **cp_p)
3232 {
3233 unsigned long num, scale;
3234 int ch, dot;
3235 const char *cp = *cp_p;
3236
3237 num = 0;
3238 scale = 1;
3239 dot = 0;
3240 for (;;) {
3241 ch = *cp;
3242 if ( !dot && ch == '.' ) {
3243 scale = 1;
3244 dot = 1;
3245 } else if ( ch == '%' ) {
3246 scale = dot ? scale*100 : 100;
3247 cp++; /* % is always at the end */
3248 break;
3249 } else if ( ch >= '0' && ch <= '9' ) {
3250 if ( scale < 100000 ) {
3251 scale *= 10;
3252 num = (num*10) + (ch-'0');
3253 }
3254 } else {
3255 break;
3256 }
3257 cp++;
3258 }
3259 *cp_p = cp;
3260
3261 /* user says num divided by scale and we say internally that
3262 * is MAX_SCORE * num / scale.
3263 */
3264 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3265 }
3266
3267 static int diff_scoreopt_parse(const char *opt)
3268 {
3269 int opt1, opt2, cmd;
3270
3271 if (*opt++ != '-')
3272 return -1;
3273 cmd = *opt++;
3274 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3275 return -1; /* that is not a -M, -C nor -B option */
3276
3277 opt1 = parse_num(&opt);
3278 if (cmd != 'B')
3279 opt2 = 0;
3280 else {
3281 if (*opt == 0)
3282 opt2 = 0;
3283 else if (*opt != '/')
3284 return -1; /* we expect -B80/99 or -B80 */
3285 else {
3286 opt++;
3287 opt2 = parse_num(&opt);
3288 }
3289 }
3290 if (*opt != 0)
3291 return -1;
3292 return opt1 | (opt2 << 16);
3293 }
3294
3295 struct diff_queue_struct diff_queued_diff;
3296
3297 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3298 {
3299 if (queue->alloc <= queue->nr) {
3300 queue->alloc = alloc_nr(queue->alloc);
3301 queue->queue = xrealloc(queue->queue,
3302 sizeof(dp) * queue->alloc);
3303 }
3304 queue->queue[queue->nr++] = dp;
3305 }
3306
3307 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3308 struct diff_filespec *one,
3309 struct diff_filespec *two)
3310 {
3311 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3312 dp->one = one;
3313 dp->two = two;
3314 if (queue)
3315 diff_q(queue, dp);
3316 return dp;
3317 }
3318
3319 void diff_free_filepair(struct diff_filepair *p)
3320 {
3321 free_filespec(p->one);
3322 free_filespec(p->two);
3323 free(p);
3324 }
3325
3326 /* This is different from find_unique_abbrev() in that
3327 * it stuffs the result with dots for alignment.
3328 */
3329 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3330 {
3331 int abblen;
3332 const char *abbrev;
3333 if (len == 40)
3334 return sha1_to_hex(sha1);
3335
3336 abbrev = find_unique_abbrev(sha1, len);
3337 abblen = strlen(abbrev);
3338 if (abblen < 37) {
3339 static char hex[41];
3340 if (len < abblen && abblen <= len + 2)
3341 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3342 else
3343 sprintf(hex, "%s...", abbrev);
3344 return hex;
3345 }
3346 return sha1_to_hex(sha1);
3347 }
3348
3349 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3350 {
3351 int line_termination = opt->line_termination;
3352 int inter_name_termination = line_termination ? '\t' : '\0';
3353 if (opt->output_prefix) {
3354 struct strbuf *msg = NULL;
3355 msg = opt->output_prefix(opt, opt->output_prefix_data);
3356 fprintf(opt->file, "%s", msg->buf);
3357 }
3358
3359 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3360 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3361 diff_unique_abbrev(p->one->sha1, opt->abbrev));
3362 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3363 }
3364 if (p->score) {
3365 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3366 inter_name_termination);
3367 } else {
3368 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3369 }
3370
3371 if (p->status == DIFF_STATUS_COPIED ||
3372 p->status == DIFF_STATUS_RENAMED) {
3373 const char *name_a, *name_b;
3374 name_a = p->one->path;
3375 name_b = p->two->path;
3376 strip_prefix(opt->prefix_length, &name_a, &name_b);
3377 write_name_quoted(name_a, opt->file, inter_name_termination);
3378 write_name_quoted(name_b, opt->file, line_termination);
3379 } else {
3380 const char *name_a, *name_b;
3381 name_a = p->one->mode ? p->one->path : p->two->path;
3382 name_b = NULL;
3383 strip_prefix(opt->prefix_length, &name_a, &name_b);
3384 write_name_quoted(name_a, opt->file, line_termination);
3385 }
3386 }
3387
3388 int diff_unmodified_pair(struct diff_filepair *p)
3389 {
3390 /* This function is written stricter than necessary to support
3391 * the currently implemented transformers, but the idea is to
3392 * let transformers to produce diff_filepairs any way they want,
3393 * and filter and clean them up here before producing the output.
3394 */
3395 struct diff_filespec *one = p->one, *two = p->two;
3396
3397 if (DIFF_PAIR_UNMERGED(p))
3398 return 0; /* unmerged is interesting */
3399
3400 /* deletion, addition, mode or type change
3401 * and rename are all interesting.
3402 */
3403 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3404 DIFF_PAIR_MODE_CHANGED(p) ||
3405 strcmp(one->path, two->path))
3406 return 0;
3407
3408 /* both are valid and point at the same path. that is, we are
3409 * dealing with a change.
3410 */
3411 if (one->sha1_valid && two->sha1_valid &&
3412 !hashcmp(one->sha1, two->sha1) &&
3413 !one->dirty_submodule && !two->dirty_submodule)
3414 return 1; /* no change */
3415 if (!one->sha1_valid && !two->sha1_valid)
3416 return 1; /* both look at the same file on the filesystem. */
3417 return 0;
3418 }
3419
3420 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3421 {
3422 if (diff_unmodified_pair(p))
3423 return;
3424
3425 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3426 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3427 return; /* no tree diffs in patch format */
3428
3429 run_diff(p, o);
3430 }
3431
3432 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3433 struct diffstat_t *diffstat)
3434 {
3435 if (diff_unmodified_pair(p))
3436 return;
3437
3438 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3439 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3440 return; /* no tree diffs in patch format */
3441
3442 run_diffstat(p, o, diffstat);
3443 }
3444
3445 static void diff_flush_checkdiff(struct diff_filepair *p,
3446 struct diff_options *o)
3447 {
3448 if (diff_unmodified_pair(p))
3449 return;
3450
3451 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3452 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3453 return; /* no tree diffs in patch format */
3454
3455 run_checkdiff(p, o);
3456 }
3457
3458 int diff_queue_is_empty(void)
3459 {
3460 struct diff_queue_struct *q = &diff_queued_diff;
3461 int i;
3462 for (i = 0; i < q->nr; i++)
3463 if (!diff_unmodified_pair(q->queue[i]))
3464 return 0;
3465 return 1;
3466 }
3467
3468 #if DIFF_DEBUG
3469 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3470 {
3471 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3472 x, one ? one : "",
3473 s->path,
3474 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3475 s->mode,
3476 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3477 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3478 x, one ? one : "",
3479 s->size, s->xfrm_flags);
3480 }
3481
3482 void diff_debug_filepair(const struct diff_filepair *p, int i)
3483 {
3484 diff_debug_filespec(p->one, i, "one");
3485 diff_debug_filespec(p->two, i, "two");
3486 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3487 p->score, p->status ? p->status : '?',
3488 p->one->rename_used, p->broken_pair);
3489 }
3490
3491 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3492 {
3493 int i;
3494 if (msg)
3495 fprintf(stderr, "%s\n", msg);
3496 fprintf(stderr, "q->nr = %d\n", q->nr);
3497 for (i = 0; i < q->nr; i++) {
3498 struct diff_filepair *p = q->queue[i];
3499 diff_debug_filepair(p, i);
3500 }
3501 }
3502 #endif
3503
3504 static void diff_resolve_rename_copy(void)
3505 {
3506 int i;
3507 struct diff_filepair *p;
3508 struct diff_queue_struct *q = &diff_queued_diff;
3509
3510 diff_debug_queue("resolve-rename-copy", q);
3511
3512 for (i = 0; i < q->nr; i++) {
3513 p = q->queue[i];
3514 p->status = 0; /* undecided */
3515 if (DIFF_PAIR_UNMERGED(p))
3516 p->status = DIFF_STATUS_UNMERGED;
3517 else if (!DIFF_FILE_VALID(p->one))
3518 p->status = DIFF_STATUS_ADDED;
3519 else if (!DIFF_FILE_VALID(p->two))
3520 p->status = DIFF_STATUS_DELETED;
3521 else if (DIFF_PAIR_TYPE_CHANGED(p))
3522 p->status = DIFF_STATUS_TYPE_CHANGED;
3523
3524 /* from this point on, we are dealing with a pair
3525 * whose both sides are valid and of the same type, i.e.
3526 * either in-place edit or rename/copy edit.
3527 */
3528 else if (DIFF_PAIR_RENAME(p)) {
3529 /*
3530 * A rename might have re-connected a broken
3531 * pair up, causing the pathnames to be the
3532 * same again. If so, that's not a rename at
3533 * all, just a modification..
3534 *
3535 * Otherwise, see if this source was used for
3536 * multiple renames, in which case we decrement
3537 * the count, and call it a copy.
3538 */
3539 if (!strcmp(p->one->path, p->two->path))
3540 p->status = DIFF_STATUS_MODIFIED;
3541 else if (--p->one->rename_used > 0)
3542 p->status = DIFF_STATUS_COPIED;
3543 else
3544 p->status = DIFF_STATUS_RENAMED;
3545 }
3546 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3547 p->one->mode != p->two->mode ||
3548 p->one->dirty_submodule ||
3549 p->two->dirty_submodule ||
3550 is_null_sha1(p->one->sha1))
3551 p->status = DIFF_STATUS_MODIFIED;
3552 else {
3553 /* This is a "no-change" entry and should not
3554 * happen anymore, but prepare for broken callers.
3555 */
3556 error("feeding unmodified %s to diffcore",
3557 p->one->path);
3558 p->status = DIFF_STATUS_UNKNOWN;
3559 }
3560 }
3561 diff_debug_queue("resolve-rename-copy done", q);
3562 }
3563
3564 static int check_pair_status(struct diff_filepair *p)
3565 {
3566 switch (p->status) {
3567 case DIFF_STATUS_UNKNOWN:
3568 return 0;
3569 case 0:
3570 die("internal error in diff-resolve-rename-copy");
3571 default:
3572 return 1;
3573 }
3574 }
3575
3576 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3577 {
3578 int fmt = opt->output_format;
3579
3580 if (fmt & DIFF_FORMAT_CHECKDIFF)
3581 diff_flush_checkdiff(p, opt);
3582 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3583 diff_flush_raw(p, opt);
3584 else if (fmt & DIFF_FORMAT_NAME) {
3585 const char *name_a, *name_b;
3586 name_a = p->two->path;
3587 name_b = NULL;
3588 strip_prefix(opt->prefix_length, &name_a, &name_b);
3589 write_name_quoted(name_a, opt->file, opt->line_termination);
3590 }
3591 }
3592
3593 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3594 {
3595 if (fs->mode)
3596 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3597 else
3598 fprintf(file, " %s ", newdelete);
3599 write_name_quoted(fs->path, file, '\n');
3600 }
3601
3602
3603 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
3604 const char *line_prefix)
3605 {
3606 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3607 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
3608 p->two->mode, show_name ? ' ' : '\n');
3609 if (show_name) {
3610 write_name_quoted(p->two->path, file, '\n');
3611 }
3612 }
3613 }
3614
3615 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
3616 const char *line_prefix)
3617 {
3618 char *names = pprint_rename(p->one->path, p->two->path);
3619
3620 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3621 free(names);
3622 show_mode_change(file, p, 0, line_prefix);
3623 }
3624
3625 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
3626 {
3627 FILE *file = opt->file;
3628 char *line_prefix = "";
3629
3630 if (opt->output_prefix) {
3631 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
3632 line_prefix = buf->buf;
3633 }
3634
3635 switch(p->status) {
3636 case DIFF_STATUS_DELETED:
3637 fputs(line_prefix, file);
3638 show_file_mode_name(file, "delete", p->one);
3639 break;
3640 case DIFF_STATUS_ADDED:
3641 fputs(line_prefix, file);
3642 show_file_mode_name(file, "create", p->two);
3643 break;
3644 case DIFF_STATUS_COPIED:
3645 fputs(line_prefix, file);
3646 show_rename_copy(file, "copy", p, line_prefix);
3647 break;
3648 case DIFF_STATUS_RENAMED:
3649 fputs(line_prefix, file);
3650 show_rename_copy(file, "rename", p, line_prefix);
3651 break;
3652 default:
3653 if (p->score) {
3654 fprintf(file, "%s rewrite ", line_prefix);
3655 write_name_quoted(p->two->path, file, ' ');
3656 fprintf(file, "(%d%%)\n", similarity_index(p));
3657 }
3658 show_mode_change(file, p, !p->score, line_prefix);
3659 break;
3660 }
3661 }
3662
3663 struct patch_id_t {
3664 git_SHA_CTX *ctx;
3665 int patchlen;
3666 };
3667
3668 static int remove_space(char *line, int len)
3669 {
3670 int i;
3671 char *dst = line;
3672 unsigned char c;
3673
3674 for (i = 0; i < len; i++)
3675 if (!isspace((c = line[i])))
3676 *dst++ = c;
3677
3678 return dst - line;
3679 }
3680
3681 static void patch_id_consume(void *priv, char *line, unsigned long len)
3682 {
3683 struct patch_id_t *data = priv;
3684 int new_len;
3685
3686 /* Ignore line numbers when computing the SHA1 of the patch */
3687 if (!prefixcmp(line, "@@ -"))
3688 return;
3689
3690 new_len = remove_space(line, len);
3691
3692 git_SHA1_Update(data->ctx, line, new_len);
3693 data->patchlen += new_len;
3694 }
3695
3696 /* returns 0 upon success, and writes result into sha1 */
3697 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3698 {
3699 struct diff_queue_struct *q = &diff_queued_diff;
3700 int i;
3701 git_SHA_CTX ctx;
3702 struct patch_id_t data;
3703 char buffer[PATH_MAX * 4 + 20];
3704
3705 git_SHA1_Init(&ctx);
3706 memset(&data, 0, sizeof(struct patch_id_t));
3707 data.ctx = &ctx;
3708
3709 for (i = 0; i < q->nr; i++) {
3710 xpparam_t xpp;
3711 xdemitconf_t xecfg;
3712 mmfile_t mf1, mf2;
3713 struct diff_filepair *p = q->queue[i];
3714 int len1, len2;
3715
3716 memset(&xpp, 0, sizeof(xpp));
3717 memset(&xecfg, 0, sizeof(xecfg));
3718 if (p->status == 0)
3719 return error("internal diff status error");
3720 if (p->status == DIFF_STATUS_UNKNOWN)
3721 continue;
3722 if (diff_unmodified_pair(p))
3723 continue;
3724 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3725 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3726 continue;
3727 if (DIFF_PAIR_UNMERGED(p))
3728 continue;
3729
3730 diff_fill_sha1_info(p->one);
3731 diff_fill_sha1_info(p->two);
3732 if (fill_mmfile(&mf1, p->one) < 0 ||
3733 fill_mmfile(&mf2, p->two) < 0)
3734 return error("unable to read files to diff");
3735
3736 len1 = remove_space(p->one->path, strlen(p->one->path));
3737 len2 = remove_space(p->two->path, strlen(p->two->path));
3738 if (p->one->mode == 0)
3739 len1 = snprintf(buffer, sizeof(buffer),
3740 "diff--gita/%.*sb/%.*s"
3741 "newfilemode%06o"
3742 "---/dev/null"
3743 "+++b/%.*s",
3744 len1, p->one->path,
3745 len2, p->two->path,
3746 p->two->mode,
3747 len2, p->two->path);
3748 else if (p->two->mode == 0)
3749 len1 = snprintf(buffer, sizeof(buffer),
3750 "diff--gita/%.*sb/%.*s"
3751 "deletedfilemode%06o"
3752 "---a/%.*s"
3753 "+++/dev/null",
3754 len1, p->one->path,
3755 len2, p->two->path,
3756 p->one->mode,
3757 len1, p->one->path);
3758 else
3759 len1 = snprintf(buffer, sizeof(buffer),
3760 "diff--gita/%.*sb/%.*s"
3761 "---a/%.*s"
3762 "+++b/%.*s",
3763 len1, p->one->path,
3764 len2, p->two->path,
3765 len1, p->one->path,
3766 len2, p->two->path);
3767 git_SHA1_Update(&ctx, buffer, len1);
3768
3769 xpp.flags = 0;
3770 xecfg.ctxlen = 3;
3771 xecfg.flags = XDL_EMIT_FUNCNAMES;
3772 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3773 &xpp, &xecfg);
3774 }
3775
3776 git_SHA1_Final(sha1, &ctx);
3777 return 0;
3778 }
3779
3780 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3781 {
3782 struct diff_queue_struct *q = &diff_queued_diff;
3783 int i;
3784 int result = diff_get_patch_id(options, sha1);
3785
3786 for (i = 0; i < q->nr; i++)
3787 diff_free_filepair(q->queue[i]);
3788
3789 free(q->queue);
3790 DIFF_QUEUE_CLEAR(q);
3791
3792 return result;
3793 }
3794
3795 static int is_summary_empty(const struct diff_queue_struct *q)
3796 {
3797 int i;
3798
3799 for (i = 0; i < q->nr; i++) {
3800 const struct diff_filepair *p = q->queue[i];
3801
3802 switch (p->status) {
3803 case DIFF_STATUS_DELETED:
3804 case DIFF_STATUS_ADDED:
3805 case DIFF_STATUS_COPIED:
3806 case DIFF_STATUS_RENAMED:
3807 return 0;
3808 default:
3809 if (p->score)
3810 return 0;
3811 if (p->one->mode && p->two->mode &&
3812 p->one->mode != p->two->mode)
3813 return 0;
3814 break;
3815 }
3816 }
3817 return 1;
3818 }
3819
3820 void diff_flush(struct diff_options *options)
3821 {
3822 struct diff_queue_struct *q = &diff_queued_diff;
3823 int i, output_format = options->output_format;
3824 int separator = 0;
3825
3826 /*
3827 * Order: raw, stat, summary, patch
3828 * or: name/name-status/checkdiff (other bits clear)
3829 */
3830 if (!q->nr)
3831 goto free_queue;
3832
3833 if (output_format & (DIFF_FORMAT_RAW |
3834 DIFF_FORMAT_NAME |
3835 DIFF_FORMAT_NAME_STATUS |
3836 DIFF_FORMAT_CHECKDIFF)) {
3837 for (i = 0; i < q->nr; i++) {
3838 struct diff_filepair *p = q->queue[i];
3839 if (check_pair_status(p))
3840 flush_one_pair(p, options);
3841 }
3842 separator++;
3843 }
3844
3845 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3846 struct diffstat_t diffstat;
3847
3848 memset(&diffstat, 0, sizeof(struct diffstat_t));
3849 for (i = 0; i < q->nr; i++) {
3850 struct diff_filepair *p = q->queue[i];
3851 if (check_pair_status(p))
3852 diff_flush_stat(p, options, &diffstat);
3853 }
3854 if (output_format & DIFF_FORMAT_NUMSTAT)
3855 show_numstat(&diffstat, options);
3856 if (output_format & DIFF_FORMAT_DIFFSTAT)
3857 show_stats(&diffstat, options);
3858 if (output_format & DIFF_FORMAT_SHORTSTAT)
3859 show_shortstats(&diffstat, options);
3860 free_diffstat_info(&diffstat);
3861 separator++;
3862 }
3863 if (output_format & DIFF_FORMAT_DIRSTAT)
3864 show_dirstat(options);
3865
3866 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3867 for (i = 0; i < q->nr; i++) {
3868 diff_summary(options, q->queue[i]);
3869 }
3870 separator++;
3871 }
3872
3873 if (output_format & DIFF_FORMAT_NO_OUTPUT &&
3874 DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
3875 DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3876 /*
3877 * run diff_flush_patch for the exit status. setting
3878 * options->file to /dev/null should be safe, becaue we
3879 * aren't supposed to produce any output anyway.
3880 */
3881 if (options->close_file)
3882 fclose(options->file);
3883 options->file = fopen("/dev/null", "w");
3884 if (!options->file)
3885 die_errno("Could not open /dev/null");
3886 options->close_file = 1;
3887 for (i = 0; i < q->nr; i++) {
3888 struct diff_filepair *p = q->queue[i];
3889 if (check_pair_status(p))
3890 diff_flush_patch(p, options);
3891 if (options->found_changes)
3892 break;
3893 }
3894 }
3895
3896 if (output_format & DIFF_FORMAT_PATCH) {
3897 if (separator) {
3898 putc(options->line_termination, options->file);
3899 if (options->stat_sep) {
3900 /* attach patch instead of inline */
3901 fputs(options->stat_sep, options->file);
3902 }
3903 }
3904
3905 for (i = 0; i < q->nr; i++) {
3906 struct diff_filepair *p = q->queue[i];
3907 if (check_pair_status(p))
3908 diff_flush_patch(p, options);
3909 }
3910 }
3911
3912 if (output_format & DIFF_FORMAT_CALLBACK)
3913 options->format_callback(q, options, options->format_callback_data);
3914
3915 for (i = 0; i < q->nr; i++)
3916 diff_free_filepair(q->queue[i]);
3917 free_queue:
3918 free(q->queue);
3919 DIFF_QUEUE_CLEAR(q);
3920 if (options->close_file)
3921 fclose(options->file);
3922
3923 /*
3924 * Report the content-level differences with HAS_CHANGES;
3925 * diff_addremove/diff_change does not set the bit when
3926 * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
3927 */
3928 if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
3929 if (options->found_changes)
3930 DIFF_OPT_SET(options, HAS_CHANGES);
3931 else
3932 DIFF_OPT_CLR(options, HAS_CHANGES);
3933 }
3934 }
3935
3936 static void diffcore_apply_filter(const char *filter)
3937 {
3938 int i;
3939 struct diff_queue_struct *q = &diff_queued_diff;
3940 struct diff_queue_struct outq;
3941 DIFF_QUEUE_CLEAR(&outq);
3942
3943 if (!filter)
3944 return;
3945
3946 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3947 int found;
3948 for (i = found = 0; !found && i < q->nr; i++) {
3949 struct diff_filepair *p = q->queue[i];
3950 if (((p->status == DIFF_STATUS_MODIFIED) &&
3951 ((p->score &&
3952 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3953 (!p->score &&
3954 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3955 ((p->status != DIFF_STATUS_MODIFIED) &&
3956 strchr(filter, p->status)))
3957 found++;
3958 }
3959 if (found)
3960 return;
3961
3962 /* otherwise we will clear the whole queue
3963 * by copying the empty outq at the end of this
3964 * function, but first clear the current entries
3965 * in the queue.
3966 */
3967 for (i = 0; i < q->nr; i++)
3968 diff_free_filepair(q->queue[i]);
3969 }
3970 else {
3971 /* Only the matching ones */
3972 for (i = 0; i < q->nr; i++) {
3973 struct diff_filepair *p = q->queue[i];
3974
3975 if (((p->status == DIFF_STATUS_MODIFIED) &&
3976 ((p->score &&
3977 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3978 (!p->score &&
3979 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3980 ((p->status != DIFF_STATUS_MODIFIED) &&
3981 strchr(filter, p->status)))
3982 diff_q(&outq, p);
3983 else
3984 diff_free_filepair(p);
3985 }
3986 }
3987 free(q->queue);
3988 *q = outq;
3989 }
3990
3991 /* Check whether two filespecs with the same mode and size are identical */
3992 static int diff_filespec_is_identical(struct diff_filespec *one,
3993 struct diff_filespec *two)
3994 {
3995 if (S_ISGITLINK(one->mode))
3996 return 0;
3997 if (diff_populate_filespec(one, 0))
3998 return 0;
3999 if (diff_populate_filespec(two, 0))
4000 return 0;
4001 return !memcmp(one->data, two->data, one->size);
4002 }
4003
4004 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4005 {
4006 int i;
4007 struct diff_queue_struct *q = &diff_queued_diff;
4008 struct diff_queue_struct outq;
4009 DIFF_QUEUE_CLEAR(&outq);
4010
4011 for (i = 0; i < q->nr; i++) {
4012 struct diff_filepair *p = q->queue[i];
4013
4014 /*
4015 * 1. Entries that come from stat info dirtiness
4016 * always have both sides (iow, not create/delete),
4017 * one side of the object name is unknown, with
4018 * the same mode and size. Keep the ones that
4019 * do not match these criteria. They have real
4020 * differences.
4021 *
4022 * 2. At this point, the file is known to be modified,
4023 * with the same mode and size, and the object
4024 * name of one side is unknown. Need to inspect
4025 * the identical contents.
4026 */
4027 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4028 !DIFF_FILE_VALID(p->two) ||
4029 (p->one->sha1_valid && p->two->sha1_valid) ||
4030 (p->one->mode != p->two->mode) ||
4031 diff_populate_filespec(p->one, 1) ||
4032 diff_populate_filespec(p->two, 1) ||
4033 (p->one->size != p->two->size) ||
4034 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4035 diff_q(&outq, p);
4036 else {
4037 /*
4038 * The caller can subtract 1 from skip_stat_unmatch
4039 * to determine how many paths were dirty only
4040 * due to stat info mismatch.
4041 */
4042 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4043 diffopt->skip_stat_unmatch++;
4044 diff_free_filepair(p);
4045 }
4046 }
4047 free(q->queue);
4048 *q = outq;
4049 }
4050
4051 static int diffnamecmp(const void *a_, const void *b_)
4052 {
4053 const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4054 const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4055 const char *name_a, *name_b;
4056
4057 name_a = a->one ? a->one->path : a->two->path;
4058 name_b = b->one ? b->one->path : b->two->path;
4059 return strcmp(name_a, name_b);
4060 }
4061
4062 void diffcore_fix_diff_index(struct diff_options *options)
4063 {
4064 struct diff_queue_struct *q = &diff_queued_diff;
4065 qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4066 }
4067
4068 void diffcore_std(struct diff_options *options)
4069 {
4070 if (options->skip_stat_unmatch)
4071 diffcore_skip_stat_unmatch(options);
4072 if (!options->found_follow) {
4073 /* See try_to_follow_renames() in tree-diff.c */
4074 if (options->break_opt != -1)
4075 diffcore_break(options->break_opt);
4076 if (options->detect_rename)
4077 diffcore_rename(options);
4078 if (options->break_opt != -1)
4079 diffcore_merge_broken();
4080 }
4081 if (options->pickaxe)
4082 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
4083 if (options->orderfile)
4084 diffcore_order(options->orderfile);
4085 if (!options->found_follow)
4086 /* See try_to_follow_renames() in tree-diff.c */
4087 diff_resolve_rename_copy();
4088 diffcore_apply_filter(options->filter);
4089
4090 if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4091 DIFF_OPT_SET(options, HAS_CHANGES);
4092 else
4093 DIFF_OPT_CLR(options, HAS_CHANGES);
4094
4095 options->found_follow = 0;
4096 }
4097
4098 int diff_result_code(struct diff_options *opt, int status)
4099 {
4100 int result = 0;
4101 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4102 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4103 return status;
4104 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4105 DIFF_OPT_TST(opt, HAS_CHANGES))
4106 result |= 01;
4107 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4108 DIFF_OPT_TST(opt, CHECK_FAILED))
4109 result |= 02;
4110 return result;
4111 }
4112
4113 void diff_addremove(struct diff_options *options,
4114 int addremove, unsigned mode,
4115 const unsigned char *sha1,
4116 const char *concatpath, unsigned dirty_submodule)
4117 {
4118 struct diff_filespec *one, *two;
4119
4120 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
4121 return;
4122
4123 /* This may look odd, but it is a preparation for
4124 * feeding "there are unchanged files which should
4125 * not produce diffs, but when you are doing copy
4126 * detection you would need them, so here they are"
4127 * entries to the diff-core. They will be prefixed
4128 * with something like '=' or '*' (I haven't decided
4129 * which but should not make any difference).
4130 * Feeding the same new and old to diff_change()
4131 * also has the same effect.
4132 * Before the final output happens, they are pruned after
4133 * merged into rename/copy pairs as appropriate.
4134 */
4135 if (DIFF_OPT_TST(options, REVERSE_DIFF))
4136 addremove = (addremove == '+' ? '-' :
4137 addremove == '-' ? '+' : addremove);
4138
4139 if (options->prefix &&
4140 strncmp(concatpath, options->prefix, options->prefix_length))
4141 return;
4142
4143 one = alloc_filespec(concatpath);
4144 two = alloc_filespec(concatpath);
4145
4146 if (addremove != '+')
4147 fill_filespec(one, sha1, mode);
4148 if (addremove != '-') {
4149 fill_filespec(two, sha1, mode);
4150 two->dirty_submodule = dirty_submodule;
4151 }
4152
4153 diff_queue(&diff_queued_diff, one, two);
4154 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4155 DIFF_OPT_SET(options, HAS_CHANGES);
4156 }
4157
4158 void diff_change(struct diff_options *options,
4159 unsigned old_mode, unsigned new_mode,
4160 const unsigned char *old_sha1,
4161 const unsigned char *new_sha1,
4162 const char *concatpath,
4163 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4164 {
4165 struct diff_filespec *one, *two;
4166
4167 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
4168 && S_ISGITLINK(new_mode))
4169 return;
4170
4171 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4172 unsigned tmp;
4173 const unsigned char *tmp_c;
4174 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4175 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4176 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4177 new_dirty_submodule = tmp;
4178 }
4179
4180 if (options->prefix &&
4181 strncmp(concatpath, options->prefix, options->prefix_length))
4182 return;
4183
4184 one = alloc_filespec(concatpath);
4185 two = alloc_filespec(concatpath);
4186 fill_filespec(one, old_sha1, old_mode);
4187 fill_filespec(two, new_sha1, new_mode);
4188 one->dirty_submodule = old_dirty_submodule;
4189 two->dirty_submodule = new_dirty_submodule;
4190
4191 diff_queue(&diff_queued_diff, one, two);
4192 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4193 DIFF_OPT_SET(options, HAS_CHANGES);
4194 }
4195
4196 void diff_unmerge(struct diff_options *options,
4197 const char *path,
4198 unsigned mode, const unsigned char *sha1)
4199 {
4200 struct diff_filespec *one, *two;
4201
4202 if (options->prefix &&
4203 strncmp(path, options->prefix, options->prefix_length))
4204 return;
4205
4206 one = alloc_filespec(path);
4207 two = alloc_filespec(path);
4208 fill_filespec(one, sha1, mode);
4209 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
4210 }
4211
4212 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4213 size_t *outsize)
4214 {
4215 struct diff_tempfile *temp;
4216 const char *argv[3];
4217 const char **arg = argv;
4218 struct child_process child;
4219 struct strbuf buf = STRBUF_INIT;
4220 int err = 0;
4221
4222 temp = prepare_temp_file(spec->path, spec);
4223 *arg++ = pgm;
4224 *arg++ = temp->name;
4225 *arg = NULL;
4226
4227 memset(&child, 0, sizeof(child));
4228 child.use_shell = 1;
4229 child.argv = argv;
4230 child.out = -1;
4231 if (start_command(&child)) {
4232 remove_tempfile();
4233 return NULL;
4234 }
4235
4236 if (strbuf_read(&buf, child.out, 0) < 0)
4237 err = error("error reading from textconv command '%s'", pgm);
4238 close(child.out);
4239
4240 if (finish_command(&child) || err) {
4241 strbuf_release(&buf);
4242 remove_tempfile();
4243 return NULL;
4244 }
4245 remove_tempfile();
4246
4247 return strbuf_detach(&buf, outsize);
4248 }
4249
4250 size_t fill_textconv(struct userdiff_driver *driver,
4251 struct diff_filespec *df,
4252 char **outbuf)
4253 {
4254 size_t size;
4255
4256 if (!driver || !driver->textconv) {
4257 if (!DIFF_FILE_VALID(df)) {
4258 *outbuf = "";
4259 return 0;
4260 }
4261 if (diff_populate_filespec(df, 0))
4262 die("unable to read files to diff");
4263 *outbuf = df->data;
4264 return df->size;
4265 }
4266
4267 if (driver->textconv_cache) {
4268 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4269 &size);
4270 if (*outbuf)
4271 return size;
4272 }
4273
4274 *outbuf = run_textconv(driver->textconv, df, &size);
4275 if (!*outbuf)
4276 die("unable to read files to diff");
4277
4278 if (driver->textconv_cache) {
4279 /* ignore errors, as we might be in a readonly repository */
4280 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4281 size);
4282 /*
4283 * we could save up changes and flush them all at the end,
4284 * but we would need an extra call after all diffing is done.
4285 * Since generating a cache entry is the slow path anyway,
4286 * this extra overhead probably isn't a big deal.
4287 */
4288 notes_cache_write(driver->textconv_cache);
4289 }
4290
4291 return size;
4292 }
Jonathan Nieder’s miscellaneous git branches