?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
blame: tolerate bogus e-mail addresses a bit better
[git/jnareb-git.git] / diff.c
blob9fa841010cc21bebc464926b15232b2aace0f5ba
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 = 400;
27 static int diff_suppress_blank_empty;
28 int diff_use_color_default = -1;
29 static const char *diff_word_regex_cfg;
30 static const char *external_diff_cmd_cfg;
31 int diff_auto_refresh_index = 1;
32 static int diff_mnemonic_prefix;
33 static int diff_no_prefix;
34 static struct diff_options default_diff_options;
35
36 static char diff_colors[][COLOR_MAXLEN] = {
37 GIT_COLOR_RESET,
38 GIT_COLOR_NORMAL, /* PLAIN */
39 GIT_COLOR_BOLD, /* METAINFO */
40 GIT_COLOR_CYAN, /* FRAGINFO */
41 GIT_COLOR_RED, /* OLD */
42 GIT_COLOR_GREEN, /* NEW */
43 GIT_COLOR_YELLOW, /* COMMIT */
44 GIT_COLOR_BG_RED, /* WHITESPACE */
45 GIT_COLOR_NORMAL, /* FUNCINFO */
46 };
47
48 static int parse_diff_color_slot(const char *var, int ofs)
49 {
50 if (!strcasecmp(var+ofs, "plain"))
51 return DIFF_PLAIN;
52 if (!strcasecmp(var+ofs, "meta"))
53 return DIFF_METAINFO;
54 if (!strcasecmp(var+ofs, "frag"))
55 return DIFF_FRAGINFO;
56 if (!strcasecmp(var+ofs, "old"))
57 return DIFF_FILE_OLD;
58 if (!strcasecmp(var+ofs, "new"))
59 return DIFF_FILE_NEW;
60 if (!strcasecmp(var+ofs, "commit"))
61 return DIFF_COMMIT;
62 if (!strcasecmp(var+ofs, "whitespace"))
63 return DIFF_WHITESPACE;
64 if (!strcasecmp(var+ofs, "func"))
65 return DIFF_FUNCINFO;
66 return -1;
67 }
68
69 static int git_config_rename(const char *var, const char *value)
70 {
71 if (!value)
72 return DIFF_DETECT_RENAME;
73 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
74 return DIFF_DETECT_COPY;
75 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
76 }
77
78 /*
79 * These are to give UI layer defaults.
80 * The core-level commands such as git-diff-files should
81 * never be affected by the setting of diff.renames
82 * the user happens to have in the configuration file.
83 */
84 int git_diff_ui_config(const char *var, const char *value, void *cb)
85 {
86 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
87 diff_use_color_default = git_config_colorbool(var, value, -1);
88 return 0;
89 }
90 if (!strcmp(var, "diff.renames")) {
91 diff_detect_rename_default = git_config_rename(var, value);
92 return 0;
93 }
94 if (!strcmp(var, "diff.autorefreshindex")) {
95 diff_auto_refresh_index = git_config_bool(var, value);
96 return 0;
97 }
98 if (!strcmp(var, "diff.mnemonicprefix")) {
99 diff_mnemonic_prefix = git_config_bool(var, value);
100 return 0;
101 }
102 if (!strcmp(var, "diff.noprefix")) {
103 diff_no_prefix = git_config_bool(var, value);
104 return 0;
105 }
106 if (!strcmp(var, "diff.external"))
107 return git_config_string(&external_diff_cmd_cfg, var, value);
108 if (!strcmp(var, "diff.wordregex"))
109 return git_config_string(&diff_word_regex_cfg, var, value);
110
111 if (!strcmp(var, "diff.ignoresubmodules"))
112 handle_ignore_submodules_arg(&default_diff_options, value);
113
114 return git_diff_basic_config(var, value, cb);
115 }
116
117 int git_diff_basic_config(const char *var, const char *value, void *cb)
118 {
119 if (!strcmp(var, "diff.renamelimit")) {
120 diff_rename_limit_default = git_config_int(var, value);
121 return 0;
122 }
123
124 switch (userdiff_config(var, value)) {
125 case 0: break;
126 case -1: return -1;
127 default: return 0;
128 }
129
130 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
131 int slot = parse_diff_color_slot(var, 11);
132 if (slot < 0)
133 return 0;
134 if (!value)
135 return config_error_nonbool(var);
136 color_parse(value, var, diff_colors[slot]);
137 return 0;
138 }
139
140 /* like GNU diff's --suppress-blank-empty option */
141 if (!strcmp(var, "diff.suppressblankempty") ||
142 /* for backwards compatibility */
143 !strcmp(var, "diff.suppress-blank-empty")) {
144 diff_suppress_blank_empty = git_config_bool(var, value);
145 return 0;
146 }
147
148 if (!prefixcmp(var, "submodule."))
149 return parse_submodule_config_option(var, value);
150
151 return git_color_default_config(var, value, cb);
152 }
153
154 static char *quote_two(const char *one, const char *two)
155 {
156 int need_one = quote_c_style(one, NULL, NULL, 1);
157 int need_two = quote_c_style(two, NULL, NULL, 1);
158 struct strbuf res = STRBUF_INIT;
159
160 if (need_one + need_two) {
161 strbuf_addch(&res, '"');
162 quote_c_style(one, &res, NULL, 1);
163 quote_c_style(two, &res, NULL, 1);
164 strbuf_addch(&res, '"');
165 } else {
166 strbuf_addstr(&res, one);
167 strbuf_addstr(&res, two);
168 }
169 return strbuf_detach(&res, NULL);
170 }
171
172 static const char *external_diff(void)
173 {
174 static const char *external_diff_cmd = NULL;
175 static int done_preparing = 0;
176
177 if (done_preparing)
178 return external_diff_cmd;
179 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
180 if (!external_diff_cmd)
181 external_diff_cmd = external_diff_cmd_cfg;
182 done_preparing = 1;
183 return external_diff_cmd;
184 }
185
186 static struct diff_tempfile {
187 const char *name; /* filename external diff should read from */
188 char hex[41];
189 char mode[10];
190 char tmp_path[PATH_MAX];
191 } diff_temp[2];
192
193 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
194
195 struct emit_callback {
196 int color_diff;
197 unsigned ws_rule;
198 int blank_at_eof_in_preimage;
199 int blank_at_eof_in_postimage;
200 int lno_in_preimage;
201 int lno_in_postimage;
202 sane_truncate_fn truncate;
203 const char **label_path;
204 struct diff_words_data *diff_words;
205 struct diff_options *opt;
206 int *found_changesp;
207 struct strbuf *header;
208 };
209
210 static int count_lines(const char *data, int size)
211 {
212 int count, ch, completely_empty = 1, nl_just_seen = 0;
213 count = 0;
214 while (0 < size--) {
215 ch = *data++;
216 if (ch == '\n') {
217 count++;
218 nl_just_seen = 1;
219 completely_empty = 0;
220 }
221 else {
222 nl_just_seen = 0;
223 completely_empty = 0;
224 }
225 }
226 if (completely_empty)
227 return 0;
228 if (!nl_just_seen)
229 count++; /* no trailing newline */
230 return count;
231 }
232
233 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
234 {
235 if (!DIFF_FILE_VALID(one)) {
236 mf->ptr = (char *)""; /* does not matter */
237 mf->size = 0;
238 return 0;
239 }
240 else if (diff_populate_filespec(one, 0))
241 return -1;
242
243 mf->ptr = one->data;
244 mf->size = one->size;
245 return 0;
246 }
247
248 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
249 static unsigned long diff_filespec_size(struct diff_filespec *one)
250 {
251 if (!DIFF_FILE_VALID(one))
252 return 0;
253 diff_populate_filespec(one, 1);
254 return one->size;
255 }
256
257 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
258 {
259 char *ptr = mf->ptr;
260 long size = mf->size;
261 int cnt = 0;
262
263 if (!size)
264 return cnt;
265 ptr += size - 1; /* pointing at the very end */
266 if (*ptr != '\n')
267 ; /* incomplete line */
268 else
269 ptr--; /* skip the last LF */
270 while (mf->ptr < ptr) {
271 char *prev_eol;
272 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
273 if (*prev_eol == '\n')
274 break;
275 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
276 break;
277 cnt++;
278 ptr = prev_eol - 1;
279 }
280 return cnt;
281 }
282
283 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
284 struct emit_callback *ecbdata)
285 {
286 int l1, l2, at;
287 unsigned ws_rule = ecbdata->ws_rule;
288 l1 = count_trailing_blank(mf1, ws_rule);
289 l2 = count_trailing_blank(mf2, ws_rule);
290 if (l2 <= l1) {
291 ecbdata->blank_at_eof_in_preimage = 0;
292 ecbdata->blank_at_eof_in_postimage = 0;
293 return;
294 }
295 at = count_lines(mf1->ptr, mf1->size);
296 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
297
298 at = count_lines(mf2->ptr, mf2->size);
299 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
300 }
301
302 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
303 int first, const char *line, int len)
304 {
305 int has_trailing_newline, has_trailing_carriage_return;
306 int nofirst;
307 FILE *file = o->file;
308
309 if (o->output_prefix) {
310 struct strbuf *msg = NULL;
311 msg = o->output_prefix(o, o->output_prefix_data);
312 assert(msg);
313 fwrite(msg->buf, msg->len, 1, file);
314 }
315
316 if (len == 0) {
317 has_trailing_newline = (first == '\n');
318 has_trailing_carriage_return = (!has_trailing_newline &&
319 (first == '\r'));
320 nofirst = has_trailing_newline || has_trailing_carriage_return;
321 } else {
322 has_trailing_newline = (len > 0 && line[len-1] == '\n');
323 if (has_trailing_newline)
324 len--;
325 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
326 if (has_trailing_carriage_return)
327 len--;
328 nofirst = 0;
329 }
330
331 if (len || !nofirst) {
332 fputs(set, file);
333 if (!nofirst)
334 fputc(first, file);
335 fwrite(line, len, 1, file);
336 fputs(reset, file);
337 }
338 if (has_trailing_carriage_return)
339 fputc('\r', file);
340 if (has_trailing_newline)
341 fputc('\n', file);
342 }
343
344 static void emit_line(struct diff_options *o, const char *set, const char *reset,
345 const char *line, int len)
346 {
347 emit_line_0(o, set, reset, line[0], line+1, len-1);
348 }
349
350 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
351 {
352 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
353 ecbdata->blank_at_eof_in_preimage &&
354 ecbdata->blank_at_eof_in_postimage &&
355 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
356 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
357 return 0;
358 return ws_blank_line(line, len, ecbdata->ws_rule);
359 }
360
361 static void emit_add_line(const char *reset,
362 struct emit_callback *ecbdata,
363 const char *line, int len)
364 {
365 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
366 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
367
368 if (!*ws)
369 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
370 else if (new_blank_line_at_eof(ecbdata, line, len))
371 /* Blank line at EOF - paint '+' as well */
372 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
373 else {
374 /* Emit just the prefix, then the rest. */
375 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
376 ws_check_emit(line, len, ecbdata->ws_rule,
377 ecbdata->opt->file, set, reset, ws);
378 }
379 }
380
381 static void emit_hunk_header(struct emit_callback *ecbdata,
382 const char *line, int len)
383 {
384 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
385 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
386 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
387 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
388 static const char atat[2] = { '@', '@' };
389 const char *cp, *ep;
390 struct strbuf msgbuf = STRBUF_INIT;
391 int org_len = len;
392 int i = 1;
393
394 /*
395 * As a hunk header must begin with "@@ -<old>, +<new> @@",
396 * it always is at least 10 bytes long.
397 */
398 if (len < 10 ||
399 memcmp(line, atat, 2) ||
400 !(ep = memmem(line + 2, len - 2, atat, 2))) {
401 emit_line(ecbdata->opt, plain, reset, line, len);
402 return;
403 }
404 ep += 2; /* skip over @@ */
405
406 /* The hunk header in fraginfo color */
407 strbuf_add(&msgbuf, frag, strlen(frag));
408 strbuf_add(&msgbuf, line, ep - line);
409 strbuf_add(&msgbuf, reset, strlen(reset));
410
411 /*
412 * trailing "\r\n"
413 */
414 for ( ; i < 3; i++)
415 if (line[len - i] == '\r' || line[len - i] == '\n')
416 len--;
417
418 /* blank before the func header */
419 for (cp = ep; ep - line < len; ep++)
420 if (*ep != ' ' && *ep != '\t')
421 break;
422 if (ep != cp) {
423 strbuf_add(&msgbuf, plain, strlen(plain));
424 strbuf_add(&msgbuf, cp, ep - cp);
425 strbuf_add(&msgbuf, reset, strlen(reset));
426 }
427
428 if (ep < line + len) {
429 strbuf_add(&msgbuf, func, strlen(func));
430 strbuf_add(&msgbuf, ep, line + len - ep);
431 strbuf_add(&msgbuf, reset, strlen(reset));
432 }
433
434 strbuf_add(&msgbuf, line + len, org_len - len);
435 emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
436 strbuf_release(&msgbuf);
437 }
438
439 static struct diff_tempfile *claim_diff_tempfile(void) {
440 int i;
441 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
442 if (!diff_temp[i].name)
443 return diff_temp + i;
444 die("BUG: diff is failing to clean up its tempfiles");
445 }
446
447 static int remove_tempfile_installed;
448
449 static void remove_tempfile(void)
450 {
451 int i;
452 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
453 if (diff_temp[i].name == diff_temp[i].tmp_path)
454 unlink_or_warn(diff_temp[i].name);
455 diff_temp[i].name = NULL;
456 }
457 }
458
459 static void remove_tempfile_on_signal(int signo)
460 {
461 remove_tempfile();
462 sigchain_pop(signo);
463 raise(signo);
464 }
465
466 static void print_line_count(FILE *file, int count)
467 {
468 switch (count) {
469 case 0:
470 fprintf(file, "0,0");
471 break;
472 case 1:
473 fprintf(file, "1");
474 break;
475 default:
476 fprintf(file, "1,%d", count);
477 break;
478 }
479 }
480
481 static void emit_rewrite_lines(struct emit_callback *ecb,
482 int prefix, const char *data, int size)
483 {
484 const char *endp = NULL;
485 static const char *nneof = " No newline at end of file\n";
486 const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
487 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
488
489 while (0 < size) {
490 int len;
491
492 endp = memchr(data, '\n', size);
493 len = endp ? (endp - data + 1) : size;
494 if (prefix != '+') {
495 ecb->lno_in_preimage++;
496 emit_line_0(ecb->opt, old, reset, '-',
497 data, len);
498 } else {
499 ecb->lno_in_postimage++;
500 emit_add_line(reset, ecb, data, len);
501 }
502 size -= len;
503 data += len;
504 }
505 if (!endp) {
506 const char *plain = diff_get_color(ecb->color_diff,
507 DIFF_PLAIN);
508 emit_line_0(ecb->opt, plain, reset, '\\',
509 nneof, strlen(nneof));
510 }
511 }
512
513 static void emit_rewrite_diff(const char *name_a,
514 const char *name_b,
515 struct diff_filespec *one,
516 struct diff_filespec *two,
517 struct userdiff_driver *textconv_one,
518 struct userdiff_driver *textconv_two,
519 struct diff_options *o)
520 {
521 int lc_a, lc_b;
522 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
523 const char *name_a_tab, *name_b_tab;
524 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
525 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
526 const char *reset = diff_get_color(color_diff, DIFF_RESET);
527 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
528 const char *a_prefix, *b_prefix;
529 char *data_one, *data_two;
530 size_t size_one, size_two;
531 struct emit_callback ecbdata;
532 char *line_prefix = "";
533 struct strbuf *msgbuf;
534
535 if (o && o->output_prefix) {
536 msgbuf = o->output_prefix(o, o->output_prefix_data);
537 line_prefix = msgbuf->buf;
538 }
539
540 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
541 a_prefix = o->b_prefix;
542 b_prefix = o->a_prefix;
543 } else {
544 a_prefix = o->a_prefix;
545 b_prefix = o->b_prefix;
546 }
547
548 name_a += (*name_a == '/');
549 name_b += (*name_b == '/');
550 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
551 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
552
553 strbuf_reset(&a_name);
554 strbuf_reset(&b_name);
555 quote_two_c_style(&a_name, a_prefix, name_a, 0);
556 quote_two_c_style(&b_name, b_prefix, name_b, 0);
557
558 size_one = fill_textconv(textconv_one, one, &data_one);
559 size_two = fill_textconv(textconv_two, two, &data_two);
560
561 memset(&ecbdata, 0, sizeof(ecbdata));
562 ecbdata.color_diff = color_diff;
563 ecbdata.found_changesp = &o->found_changes;
564 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
565 ecbdata.opt = o;
566 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
567 mmfile_t mf1, mf2;
568 mf1.ptr = (char *)data_one;
569 mf2.ptr = (char *)data_two;
570 mf1.size = size_one;
571 mf2.size = size_two;
572 check_blank_at_eof(&mf1, &mf2, &ecbdata);
573 }
574 ecbdata.lno_in_preimage = 1;
575 ecbdata.lno_in_postimage = 1;
576
577 lc_a = count_lines(data_one, size_one);
578 lc_b = count_lines(data_two, size_two);
579 fprintf(o->file,
580 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
581 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
582 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
583 line_prefix, fraginfo);
584 print_line_count(o->file, lc_a);
585 fprintf(o->file, " +");
586 print_line_count(o->file, lc_b);
587 fprintf(o->file, " @@%s\n", reset);
588 if (lc_a)
589 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
590 if (lc_b)
591 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
592 if (textconv_one)
593 free((char *)data_one);
594 if (textconv_two)
595 free((char *)data_two);
596 }
597
598 struct diff_words_buffer {
599 mmfile_t text;
600 long alloc;
601 struct diff_words_orig {
602 const char *begin, *end;
603 } *orig;
604 int orig_nr, orig_alloc;
605 };
606
607 static void diff_words_append(char *line, unsigned long len,
608 struct diff_words_buffer *buffer)
609 {
610 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
611 line++;
612 len--;
613 memcpy(buffer->text.ptr + buffer->text.size, line, len);
614 buffer->text.size += len;
615 buffer->text.ptr[buffer->text.size] = '\0';
616 }
617
618 struct diff_words_style_elem {
619 const char *prefix;
620 const char *suffix;
621 const char *color; /* NULL; filled in by the setup code if
622 * color is enabled */
623 };
624
625 struct diff_words_style {
626 enum diff_words_type type;
627 struct diff_words_style_elem new, old, ctx;
628 const char *newline;
629 };
630
631 static struct diff_words_style diff_words_styles[] = {
632 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
633 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
634 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
635 };
636
637 struct diff_words_data {
638 struct diff_words_buffer minus, plus;
639 const char *current_plus;
640 int last_minus;
641 struct diff_options *opt;
642 regex_t *word_regex;
643 enum diff_words_type type;
644 struct diff_words_style *style;
645 };
646
647 static int fn_out_diff_words_write_helper(FILE *fp,
648 struct diff_words_style_elem *st_el,
649 const char *newline,
650 size_t count, const char *buf,
651 const char *line_prefix)
652 {
653 int print = 0;
654
655 while (count) {
656 char *p = memchr(buf, '\n', count);
657 if (print)
658 fputs(line_prefix, fp);
659 if (p != buf) {
660 if (st_el->color && fputs(st_el->color, fp) < 0)
661 return -1;
662 if (fputs(st_el->prefix, fp) < 0 ||
663 fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
664 fputs(st_el->suffix, fp) < 0)
665 return -1;
666 if (st_el->color && *st_el->color
667 && fputs(GIT_COLOR_RESET, fp) < 0)
668 return -1;
669 }
670 if (!p)
671 return 0;
672 if (fputs(newline, fp) < 0)
673 return -1;
674 count -= p + 1 - buf;
675 buf = p + 1;
676 print = 1;
677 }
678 return 0;
679 }
680
681 /*
682 * '--color-words' algorithm can be described as:
683 *
684 * 1. collect a the minus/plus lines of a diff hunk, divided into
685 * minus-lines and plus-lines;
686 *
687 * 2. break both minus-lines and plus-lines into words and
688 * place them into two mmfile_t with one word for each line;
689 *
690 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
691 *
692 * And for the common parts of the both file, we output the plus side text.
693 * diff_words->current_plus is used to trace the current position of the plus file
694 * which printed. diff_words->last_minus is used to trace the last minus word
695 * printed.
696 *
697 * For '--graph' to work with '--color-words', we need to output the graph prefix
698 * on each line of color words output. Generally, there are two conditions on
699 * which we should output the prefix.
700 *
701 * 1. diff_words->last_minus == 0 &&
702 * diff_words->current_plus == diff_words->plus.text.ptr
703 *
704 * that is: the plus text must start as a new line, and if there is no minus
705 * word printed, a graph prefix must be printed.
706 *
707 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
708 * *(diff_words->current_plus - 1) == '\n'
709 *
710 * that is: a graph prefix must be printed following a '\n'
711 */
712 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
713 {
714 if ((diff_words->last_minus == 0 &&
715 diff_words->current_plus == diff_words->plus.text.ptr) ||
716 (diff_words->current_plus > diff_words->plus.text.ptr &&
717 *(diff_words->current_plus - 1) == '\n')) {
718 return 1;
719 } else {
720 return 0;
721 }
722 }
723
724 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
725 {
726 struct diff_words_data *diff_words = priv;
727 struct diff_words_style *style = diff_words->style;
728 int minus_first, minus_len, plus_first, plus_len;
729 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
730 struct diff_options *opt = diff_words->opt;
731 struct strbuf *msgbuf;
732 char *line_prefix = "";
733
734 if (line[0] != '@' || parse_hunk_header(line, len,
735 &minus_first, &minus_len, &plus_first, &plus_len))
736 return;
737
738 assert(opt);
739 if (opt->output_prefix) {
740 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
741 line_prefix = msgbuf->buf;
742 }
743
744 /* POSIX requires that first be decremented by one if len == 0... */
745 if (minus_len) {
746 minus_begin = diff_words->minus.orig[minus_first].begin;
747 minus_end =
748 diff_words->minus.orig[minus_first + minus_len - 1].end;
749 } else
750 minus_begin = minus_end =
751 diff_words->minus.orig[minus_first].end;
752
753 if (plus_len) {
754 plus_begin = diff_words->plus.orig[plus_first].begin;
755 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
756 } else
757 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
758
759 if (color_words_output_graph_prefix(diff_words)) {
760 fputs(line_prefix, diff_words->opt->file);
761 }
762 if (diff_words->current_plus != plus_begin) {
763 fn_out_diff_words_write_helper(diff_words->opt->file,
764 &style->ctx, style->newline,
765 plus_begin - diff_words->current_plus,
766 diff_words->current_plus, line_prefix);
767 if (*(plus_begin - 1) == '\n')
768 fputs(line_prefix, diff_words->opt->file);
769 }
770 if (minus_begin != minus_end) {
771 fn_out_diff_words_write_helper(diff_words->opt->file,
772 &style->old, style->newline,
773 minus_end - minus_begin, minus_begin,
774 line_prefix);
775 }
776 if (plus_begin != plus_end) {
777 fn_out_diff_words_write_helper(diff_words->opt->file,
778 &style->new, style->newline,
779 plus_end - plus_begin, plus_begin,
780 line_prefix);
781 }
782
783 diff_words->current_plus = plus_end;
784 diff_words->last_minus = minus_first;
785 }
786
787 /* This function starts looking at *begin, and returns 0 iff a word was found. */
788 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
789 int *begin, int *end)
790 {
791 if (word_regex && *begin < buffer->size) {
792 regmatch_t match[1];
793 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
794 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
795 '\n', match[0].rm_eo - match[0].rm_so);
796 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
797 *begin += match[0].rm_so;
798 return *begin >= *end;
799 }
800 return -1;
801 }
802
803 /* find the next word */
804 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
805 (*begin)++;
806 if (*begin >= buffer->size)
807 return -1;
808
809 /* find the end of the word */
810 *end = *begin + 1;
811 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
812 (*end)++;
813
814 return 0;
815 }
816
817 /*
818 * This function splits the words in buffer->text, stores the list with
819 * newline separator into out, and saves the offsets of the original words
820 * in buffer->orig.
821 */
822 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
823 regex_t *word_regex)
824 {
825 int i, j;
826 long alloc = 0;
827
828 out->size = 0;
829 out->ptr = NULL;
830
831 /* fake an empty "0th" word */
832 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
833 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
834 buffer->orig_nr = 1;
835
836 for (i = 0; i < buffer->text.size; i++) {
837 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
838 return;
839
840 /* store original boundaries */
841 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
842 buffer->orig_alloc);
843 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
844 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
845 buffer->orig_nr++;
846
847 /* store one word */
848 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
849 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
850 out->ptr[out->size + j - i] = '\n';
851 out->size += j - i + 1;
852
853 i = j - 1;
854 }
855 }
856
857 /* this executes the word diff on the accumulated buffers */
858 static void diff_words_show(struct diff_words_data *diff_words)
859 {
860 xpparam_t xpp;
861 xdemitconf_t xecfg;
862 mmfile_t minus, plus;
863 struct diff_words_style *style = diff_words->style;
864
865 struct diff_options *opt = diff_words->opt;
866 struct strbuf *msgbuf;
867 char *line_prefix = "";
868
869 assert(opt);
870 if (opt->output_prefix) {
871 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
872 line_prefix = msgbuf->buf;
873 }
874
875 /* special case: only removal */
876 if (!diff_words->plus.text.size) {
877 fputs(line_prefix, diff_words->opt->file);
878 fn_out_diff_words_write_helper(diff_words->opt->file,
879 &style->old, style->newline,
880 diff_words->minus.text.size,
881 diff_words->minus.text.ptr, line_prefix);
882 diff_words->minus.text.size = 0;
883 return;
884 }
885
886 diff_words->current_plus = diff_words->plus.text.ptr;
887 diff_words->last_minus = 0;
888
889 memset(&xpp, 0, sizeof(xpp));
890 memset(&xecfg, 0, sizeof(xecfg));
891 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
892 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
893 xpp.flags = 0;
894 /* as only the hunk header will be parsed, we need a 0-context */
895 xecfg.ctxlen = 0;
896 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
897 &xpp, &xecfg);
898 free(minus.ptr);
899 free(plus.ptr);
900 if (diff_words->current_plus != diff_words->plus.text.ptr +
901 diff_words->plus.text.size) {
902 if (color_words_output_graph_prefix(diff_words))
903 fputs(line_prefix, diff_words->opt->file);
904 fn_out_diff_words_write_helper(diff_words->opt->file,
905 &style->ctx, style->newline,
906 diff_words->plus.text.ptr + diff_words->plus.text.size
907 - diff_words->current_plus, diff_words->current_plus,
908 line_prefix);
909 }
910 diff_words->minus.text.size = diff_words->plus.text.size = 0;
911 }
912
913 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
914 static void diff_words_flush(struct emit_callback *ecbdata)
915 {
916 if (ecbdata->diff_words->minus.text.size ||
917 ecbdata->diff_words->plus.text.size)
918 diff_words_show(ecbdata->diff_words);
919 }
920
921 static void free_diff_words_data(struct emit_callback *ecbdata)
922 {
923 if (ecbdata->diff_words) {
924 diff_words_flush(ecbdata);
925 free (ecbdata->diff_words->minus.text.ptr);
926 free (ecbdata->diff_words->minus.orig);
927 free (ecbdata->diff_words->plus.text.ptr);
928 free (ecbdata->diff_words->plus.orig);
929 if (ecbdata->diff_words->word_regex) {
930 regfree(ecbdata->diff_words->word_regex);
931 free(ecbdata->diff_words->word_regex);
932 }
933 free(ecbdata->diff_words);
934 ecbdata->diff_words = NULL;
935 }
936 }
937
938 const char *diff_get_color(int diff_use_color, enum color_diff ix)
939 {
940 if (diff_use_color)
941 return diff_colors[ix];
942 return "";
943 }
944
945 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
946 {
947 const char *cp;
948 unsigned long allot;
949 size_t l = len;
950
951 if (ecb->truncate)
952 return ecb->truncate(line, len);
953 cp = line;
954 allot = l;
955 while (0 < l) {
956 (void) utf8_width(&cp, &l);
957 if (!cp)
958 break; /* truncated in the middle? */
959 }
960 return allot - l;
961 }
962
963 static void find_lno(const char *line, struct emit_callback *ecbdata)
964 {
965 const char *p;
966 ecbdata->lno_in_preimage = 0;
967 ecbdata->lno_in_postimage = 0;
968 p = strchr(line, '-');
969 if (!p)
970 return; /* cannot happen */
971 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
972 p = strchr(p, '+');
973 if (!p)
974 return; /* cannot happen */
975 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
976 }
977
978 static void fn_out_consume(void *priv, char *line, unsigned long len)
979 {
980 struct emit_callback *ecbdata = priv;
981 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
982 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
983 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
984 struct diff_options *o = ecbdata->opt;
985 char *line_prefix = "";
986 struct strbuf *msgbuf;
987
988 if (o && o->output_prefix) {
989 msgbuf = o->output_prefix(o, o->output_prefix_data);
990 line_prefix = msgbuf->buf;
991 }
992
993 if (ecbdata->header) {
994 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
995 strbuf_reset(ecbdata->header);
996 ecbdata->header = NULL;
997 }
998 *(ecbdata->found_changesp) = 1;
999
1000 if (ecbdata->label_path[0]) {
1001 const char *name_a_tab, *name_b_tab;
1002
1003 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1004 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1005
1006 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1007 line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1008 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1009 line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1010 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1011 }
1012
1013 if (diff_suppress_blank_empty
1014 && len == 2 && line[0] == ' ' && line[1] == '\n') {
1015 line[0] = '\n';
1016 len = 1;
1017 }
1018
1019 if (line[0] == '@') {
1020 if (ecbdata->diff_words)
1021 diff_words_flush(ecbdata);
1022 len = sane_truncate_line(ecbdata, line, len);
1023 find_lno(line, ecbdata);
1024 emit_hunk_header(ecbdata, line, len);
1025 if (line[len-1] != '\n')
1026 putc('\n', ecbdata->opt->file);
1027 return;
1028 }
1029
1030 if (len < 1) {
1031 emit_line(ecbdata->opt, reset, reset, line, len);
1032 if (ecbdata->diff_words
1033 && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1034 fputs("~\n", ecbdata->opt->file);
1035 return;
1036 }
1037
1038 if (ecbdata->diff_words) {
1039 if (line[0] == '-') {
1040 diff_words_append(line, len,
1041 &ecbdata->diff_words->minus);
1042 return;
1043 } else if (line[0] == '+') {
1044 diff_words_append(line, len,
1045 &ecbdata->diff_words->plus);
1046 return;
1047 }
1048 diff_words_flush(ecbdata);
1049 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1050 emit_line(ecbdata->opt, plain, reset, line, len);
1051 fputs("~\n", ecbdata->opt->file);
1052 } else {
1053 /* don't print the prefix character */
1054 emit_line(ecbdata->opt, plain, reset, line+1, len-1);
1055 }
1056 return;
1057 }
1058
1059 if (line[0] != '+') {
1060 const char *color =
1061 diff_get_color(ecbdata->color_diff,
1062 line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1063 ecbdata->lno_in_preimage++;
1064 if (line[0] == ' ')
1065 ecbdata->lno_in_postimage++;
1066 emit_line(ecbdata->opt, color, reset, line, len);
1067 } else {
1068 ecbdata->lno_in_postimage++;
1069 emit_add_line(reset, ecbdata, line + 1, len - 1);
1070 }
1071 }
1072
1073 static char *pprint_rename(const char *a, const char *b)
1074 {
1075 const char *old = a;
1076 const char *new = b;
1077 struct strbuf name = STRBUF_INIT;
1078 int pfx_length, sfx_length;
1079 int len_a = strlen(a);
1080 int len_b = strlen(b);
1081 int a_midlen, b_midlen;
1082 int qlen_a = quote_c_style(a, NULL, NULL, 0);
1083 int qlen_b = quote_c_style(b, NULL, NULL, 0);
1084
1085 if (qlen_a || qlen_b) {
1086 quote_c_style(a, &name, NULL, 0);
1087 strbuf_addstr(&name, " => ");
1088 quote_c_style(b, &name, NULL, 0);
1089 return strbuf_detach(&name, NULL);
1090 }
1091
1092 /* Find common prefix */
1093 pfx_length = 0;
1094 while (*old && *new && *old == *new) {
1095 if (*old == '/')
1096 pfx_length = old - a + 1;
1097 old++;
1098 new++;
1099 }
1100
1101 /* Find common suffix */
1102 old = a + len_a;
1103 new = b + len_b;
1104 sfx_length = 0;
1105 while (a <= old && b <= new && *old == *new) {
1106 if (*old == '/')
1107 sfx_length = len_a - (old - a);
1108 old--;
1109 new--;
1110 }
1111
1112 /*
1113 * pfx{mid-a => mid-b}sfx
1114 * {pfx-a => pfx-b}sfx
1115 * pfx{sfx-a => sfx-b}
1116 * name-a => name-b
1117 */
1118 a_midlen = len_a - pfx_length - sfx_length;
1119 b_midlen = len_b - pfx_length - sfx_length;
1120 if (a_midlen < 0)
1121 a_midlen = 0;
1122 if (b_midlen < 0)
1123 b_midlen = 0;
1124
1125 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1126 if (pfx_length + sfx_length) {
1127 strbuf_add(&name, a, pfx_length);
1128 strbuf_addch(&name, '{');
1129 }
1130 strbuf_add(&name, a + pfx_length, a_midlen);
1131 strbuf_addstr(&name, " => ");
1132 strbuf_add(&name, b + pfx_length, b_midlen);
1133 if (pfx_length + sfx_length) {
1134 strbuf_addch(&name, '}');
1135 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1136 }
1137 return strbuf_detach(&name, NULL);
1138 }
1139
1140 struct diffstat_t {
1141 int nr;
1142 int alloc;
1143 struct diffstat_file {
1144 char *from_name;
1145 char *name;
1146 char *print_name;
1147 unsigned is_unmerged:1;
1148 unsigned is_binary:1;
1149 unsigned is_renamed:1;
1150 uintmax_t added, deleted;
1151 } **files;
1152 };
1153
1154 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1155 const char *name_a,
1156 const char *name_b)
1157 {
1158 struct diffstat_file *x;
1159 x = xcalloc(sizeof (*x), 1);
1160 if (diffstat->nr == diffstat->alloc) {
1161 diffstat->alloc = alloc_nr(diffstat->alloc);
1162 diffstat->files = xrealloc(diffstat->files,
1163 diffstat->alloc * sizeof(x));
1164 }
1165 diffstat->files[diffstat->nr++] = x;
1166 if (name_b) {
1167 x->from_name = xstrdup(name_a);
1168 x->name = xstrdup(name_b);
1169 x->is_renamed = 1;
1170 }
1171 else {
1172 x->from_name = NULL;
1173 x->name = xstrdup(name_a);
1174 }
1175 return x;
1176 }
1177
1178 static void diffstat_consume(void *priv, char *line, unsigned long len)
1179 {
1180 struct diffstat_t *diffstat = priv;
1181 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1182
1183 if (line[0] == '+')
1184 x->added++;
1185 else if (line[0] == '-')
1186 x->deleted++;
1187 }
1188
1189 const char mime_boundary_leader[] = "------------";
1190
1191 static int scale_linear(int it, int width, int max_change)
1192 {
1193 /*
1194 * make sure that at least one '-' is printed if there were deletions,
1195 * and likewise for '+'.
1196 */
1197 if (max_change < 2)
1198 return it;
1199 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1200 }
1201
1202 static void show_name(FILE *file,
1203 const char *prefix, const char *name, int len)
1204 {
1205 fprintf(file, " %s%-*s |", prefix, len, name);
1206 }
1207
1208 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1209 {
1210 if (cnt <= 0)
1211 return;
1212 fprintf(file, "%s", set);
1213 while (cnt--)
1214 putc(ch, file);
1215 fprintf(file, "%s", reset);
1216 }
1217
1218 static void fill_print_name(struct diffstat_file *file)
1219 {
1220 char *pname;
1221
1222 if (file->print_name)
1223 return;
1224
1225 if (!file->is_renamed) {
1226 struct strbuf buf = STRBUF_INIT;
1227 if (quote_c_style(file->name, &buf, NULL, 0)) {
1228 pname = strbuf_detach(&buf, NULL);
1229 } else {
1230 pname = file->name;
1231 strbuf_release(&buf);
1232 }
1233 } else {
1234 pname = pprint_rename(file->from_name, file->name);
1235 }
1236 file->print_name = pname;
1237 }
1238
1239 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1240 {
1241 int i, len, add, del, adds = 0, dels = 0;
1242 uintmax_t max_change = 0, max_len = 0;
1243 int total_files = data->nr;
1244 int width, name_width;
1245 const char *reset, *add_c, *del_c;
1246 const char *line_prefix = "";
1247 struct strbuf *msg = NULL;
1248
1249 if (data->nr == 0)
1250 return;
1251
1252 if (options->output_prefix) {
1253 msg = options->output_prefix(options, options->output_prefix_data);
1254 line_prefix = msg->buf;
1255 }
1256
1257 width = options->stat_width ? options->stat_width : 80;
1258 name_width = options->stat_name_width ? options->stat_name_width : 50;
1259
1260 /* Sanity: give at least 5 columns to the graph,
1261 * but leave at least 10 columns for the name.
1262 */
1263 if (width < 25)
1264 width = 25;
1265 if (name_width < 10)
1266 name_width = 10;
1267 else if (width < name_width + 15)
1268 name_width = width - 15;
1269
1270 /* Find the longest filename and max number of changes */
1271 reset = diff_get_color_opt(options, DIFF_RESET);
1272 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1273 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1274
1275 for (i = 0; i < data->nr; i++) {
1276 struct diffstat_file *file = data->files[i];
1277 uintmax_t change = file->added + file->deleted;
1278 fill_print_name(file);
1279 len = strlen(file->print_name);
1280 if (max_len < len)
1281 max_len = len;
1282
1283 if (file->is_binary || file->is_unmerged)
1284 continue;
1285 if (max_change < change)
1286 max_change = change;
1287 }
1288
1289 /* Compute the width of the graph part;
1290 * 10 is for one blank at the beginning of the line plus
1291 * " | count " between the name and the graph.
1292 *
1293 * From here on, name_width is the width of the name area,
1294 * and width is the width of the graph area.
1295 */
1296 name_width = (name_width < max_len) ? name_width : max_len;
1297 if (width < (name_width + 10) + max_change)
1298 width = width - (name_width + 10);
1299 else
1300 width = max_change;
1301
1302 for (i = 0; i < data->nr; i++) {
1303 const char *prefix = "";
1304 char *name = data->files[i]->print_name;
1305 uintmax_t added = data->files[i]->added;
1306 uintmax_t deleted = data->files[i]->deleted;
1307 int name_len;
1308
1309 /*
1310 * "scale" the filename
1311 */
1312 len = name_width;
1313 name_len = strlen(name);
1314 if (name_width < name_len) {
1315 char *slash;
1316 prefix = "...";
1317 len -= 3;
1318 name += name_len - len;
1319 slash = strchr(name, '/');
1320 if (slash)
1321 name = slash;
1322 }
1323
1324 if (data->files[i]->is_binary) {
1325 fprintf(options->file, "%s", line_prefix);
1326 show_name(options->file, prefix, name, len);
1327 fprintf(options->file, " Bin ");
1328 fprintf(options->file, "%s%"PRIuMAX"%s",
1329 del_c, deleted, reset);
1330 fprintf(options->file, " -> ");
1331 fprintf(options->file, "%s%"PRIuMAX"%s",
1332 add_c, added, reset);
1333 fprintf(options->file, " bytes");
1334 fprintf(options->file, "\n");
1335 continue;
1336 }
1337 else if (data->files[i]->is_unmerged) {
1338 fprintf(options->file, "%s", line_prefix);
1339 show_name(options->file, prefix, name, len);
1340 fprintf(options->file, " Unmerged\n");
1341 continue;
1342 }
1343 else if (!data->files[i]->is_renamed &&
1344 (added + deleted == 0)) {
1345 total_files--;
1346 continue;
1347 }
1348
1349 /*
1350 * scale the add/delete
1351 */
1352 add = added;
1353 del = deleted;
1354 adds += add;
1355 dels += del;
1356
1357 if (width <= max_change) {
1358 add = scale_linear(add, width, max_change);
1359 del = scale_linear(del, width, max_change);
1360 }
1361 fprintf(options->file, "%s", line_prefix);
1362 show_name(options->file, prefix, name, len);
1363 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1364 added + deleted ? " " : "");
1365 show_graph(options->file, '+', add, add_c, reset);
1366 show_graph(options->file, '-', del, del_c, reset);
1367 fprintf(options->file, "\n");
1368 }
1369 fprintf(options->file, "%s", line_prefix);
1370 fprintf(options->file,
1371 " %d files changed, %d insertions(+), %d deletions(-)\n",
1372 total_files, adds, dels);
1373 }
1374
1375 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1376 {
1377 int i, adds = 0, dels = 0, total_files = data->nr;
1378
1379 if (data->nr == 0)
1380 return;
1381
1382 for (i = 0; i < data->nr; i++) {
1383 if (!data->files[i]->is_binary &&
1384 !data->files[i]->is_unmerged) {
1385 int added = data->files[i]->added;
1386 int deleted= data->files[i]->deleted;
1387 if (!data->files[i]->is_renamed &&
1388 (added + deleted == 0)) {
1389 total_files--;
1390 } else {
1391 adds += added;
1392 dels += deleted;
1393 }
1394 }
1395 }
1396 if (options->output_prefix) {
1397 struct strbuf *msg = NULL;
1398 msg = options->output_prefix(options,
1399 options->output_prefix_data);
1400 fprintf(options->file, "%s", msg->buf);
1401 }
1402 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1403 total_files, adds, dels);
1404 }
1405
1406 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1407 {
1408 int i;
1409
1410 if (data->nr == 0)
1411 return;
1412
1413 for (i = 0; i < data->nr; i++) {
1414 struct diffstat_file *file = data->files[i];
1415
1416 if (options->output_prefix) {
1417 struct strbuf *msg = NULL;
1418 msg = options->output_prefix(options,
1419 options->output_prefix_data);
1420 fprintf(options->file, "%s", msg->buf);
1421 }
1422
1423 if (file->is_binary)
1424 fprintf(options->file, "-\t-\t");
1425 else
1426 fprintf(options->file,
1427 "%"PRIuMAX"\t%"PRIuMAX"\t",
1428 file->added, file->deleted);
1429 if (options->line_termination) {
1430 fill_print_name(file);
1431 if (!file->is_renamed)
1432 write_name_quoted(file->name, options->file,
1433 options->line_termination);
1434 else {
1435 fputs(file->print_name, options->file);
1436 putc(options->line_termination, options->file);
1437 }
1438 } else {
1439 if (file->is_renamed) {
1440 putc('\0', options->file);
1441 write_name_quoted(file->from_name, options->file, '\0');
1442 }
1443 write_name_quoted(file->name, options->file, '\0');
1444 }
1445 }
1446 }
1447
1448 struct dirstat_file {
1449 const char *name;
1450 unsigned long changed;
1451 };
1452
1453 struct dirstat_dir {
1454 struct dirstat_file *files;
1455 int alloc, nr, percent, cumulative;
1456 };
1457
1458 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1459 unsigned long changed, const char *base, int baselen)
1460 {
1461 unsigned long this_dir = 0;
1462 unsigned int sources = 0;
1463 const char *line_prefix = "";
1464 struct strbuf *msg = NULL;
1465
1466 if (opt->output_prefix) {
1467 msg = opt->output_prefix(opt, opt->output_prefix_data);
1468 line_prefix = msg->buf;
1469 }
1470
1471 while (dir->nr) {
1472 struct dirstat_file *f = dir->files;
1473 int namelen = strlen(f->name);
1474 unsigned long this;
1475 char *slash;
1476
1477 if (namelen < baselen)
1478 break;
1479 if (memcmp(f->name, base, baselen))
1480 break;
1481 slash = strchr(f->name + baselen, '/');
1482 if (slash) {
1483 int newbaselen = slash + 1 - f->name;
1484 this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1485 sources++;
1486 } else {
1487 this = f->changed;
1488 dir->files++;
1489 dir->nr--;
1490 sources += 2;
1491 }
1492 this_dir += this;
1493 }
1494
1495 /*
1496 * We don't report dirstat's for
1497 * - the top level
1498 * - or cases where everything came from a single directory
1499 * under this directory (sources == 1).
1500 */
1501 if (baselen && sources != 1) {
1502 int permille = this_dir * 1000 / changed;
1503 if (permille) {
1504 int percent = permille / 10;
1505 if (percent >= dir->percent) {
1506 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1507 percent, permille % 10, baselen, base);
1508 if (!dir->cumulative)
1509 return 0;
1510 }
1511 }
1512 }
1513 return this_dir;
1514 }
1515
1516 static int dirstat_compare(const void *_a, const void *_b)
1517 {
1518 const struct dirstat_file *a = _a;
1519 const struct dirstat_file *b = _b;
1520 return strcmp(a->name, b->name);
1521 }
1522
1523 static void show_dirstat(struct diff_options *options)
1524 {
1525 int i;
1526 unsigned long changed;
1527 struct dirstat_dir dir;
1528 struct diff_queue_struct *q = &diff_queued_diff;
1529
1530 dir.files = NULL;
1531 dir.alloc = 0;
1532 dir.nr = 0;
1533 dir.percent = options->dirstat_percent;
1534 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1535
1536 changed = 0;
1537 for (i = 0; i < q->nr; i++) {
1538 struct diff_filepair *p = q->queue[i];
1539 const char *name;
1540 unsigned long copied, added, damage;
1541
1542 name = p->one->path ? p->one->path : p->two->path;
1543
1544 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1545 diff_populate_filespec(p->one, 0);
1546 diff_populate_filespec(p->two, 0);
1547 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1548 &copied, &added);
1549 diff_free_filespec_data(p->one);
1550 diff_free_filespec_data(p->two);
1551 } else if (DIFF_FILE_VALID(p->one)) {
1552 diff_populate_filespec(p->one, 1);
1553 copied = added = 0;
1554 diff_free_filespec_data(p->one);
1555 } else if (DIFF_FILE_VALID(p->two)) {
1556 diff_populate_filespec(p->two, 1);
1557 copied = 0;
1558 added = p->two->size;
1559 diff_free_filespec_data(p->two);
1560 } else
1561 continue;
1562
1563 /*
1564 * Original minus copied is the removed material,
1565 * added is the new material. They are both damages
1566 * made to the preimage. In --dirstat-by-file mode, count
1567 * damaged files, not damaged lines. This is done by
1568 * counting only a single damaged line per file.
1569 */
1570 damage = (p->one->size - copied) + added;
1571 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1572 damage = 1;
1573
1574 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1575 dir.files[dir.nr].name = name;
1576 dir.files[dir.nr].changed = damage;
1577 changed += damage;
1578 dir.nr++;
1579 }
1580
1581 /* This can happen even with many files, if everything was renames */
1582 if (!changed)
1583 return;
1584
1585 /* Show all directories with more than x% of the changes */
1586 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1587 gather_dirstat(options, &dir, changed, "", 0);
1588 }
1589
1590 static void free_diffstat_info(struct diffstat_t *diffstat)
1591 {
1592 int i;
1593 for (i = 0; i < diffstat->nr; i++) {
1594 struct diffstat_file *f = diffstat->files[i];
1595 if (f->name != f->print_name)
1596 free(f->print_name);
1597 free(f->name);
1598 free(f->from_name);
1599 free(f);
1600 }
1601 free(diffstat->files);
1602 }
1603
1604 struct checkdiff_t {
1605 const char *filename;
1606 int lineno;
1607 int conflict_marker_size;
1608 struct diff_options *o;
1609 unsigned ws_rule;
1610 unsigned status;
1611 };
1612
1613 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1614 {
1615 char firstchar;
1616 int cnt;
1617
1618 if (len < marker_size + 1)
1619 return 0;
1620 firstchar = line[0];
1621 switch (firstchar) {
1622 case '=': case '>': case '<': case '|':
1623 break;
1624 default:
1625 return 0;
1626 }
1627 for (cnt = 1; cnt < marker_size; cnt++)
1628 if (line[cnt] != firstchar)
1629 return 0;
1630 /* line[1] thru line[marker_size-1] are same as firstchar */
1631 if (len < marker_size + 1 || !isspace(line[marker_size]))
1632 return 0;
1633 return 1;
1634 }
1635
1636 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1637 {
1638 struct checkdiff_t *data = priv;
1639 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1640 int marker_size = data->conflict_marker_size;
1641 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1642 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1643 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1644 char *err;
1645 char *line_prefix = "";
1646 struct strbuf *msgbuf;
1647
1648 assert(data->o);
1649 if (data->o->output_prefix) {
1650 msgbuf = data->o->output_prefix(data->o,
1651 data->o->output_prefix_data);
1652 line_prefix = msgbuf->buf;
1653 }
1654
1655 if (line[0] == '+') {
1656 unsigned bad;
1657 data->lineno++;
1658 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1659 data->status |= 1;
1660 fprintf(data->o->file,
1661 "%s%s:%d: leftover conflict marker\n",
1662 line_prefix, data->filename, data->lineno);
1663 }
1664 bad = ws_check(line + 1, len - 1, data->ws_rule);
1665 if (!bad)
1666 return;
1667 data->status |= bad;
1668 err = whitespace_error_string(bad);
1669 fprintf(data->o->file, "%s%s:%d: %s.\n",
1670 line_prefix, data->filename, data->lineno, err);
1671 free(err);
1672 emit_line(data->o, set, reset, line, 1);
1673 ws_check_emit(line + 1, len - 1, data->ws_rule,
1674 data->o->file, set, reset, ws);
1675 } else if (line[0] == ' ') {
1676 data->lineno++;
1677 } else if (line[0] == '@') {
1678 char *plus = strchr(line, '+');
1679 if (plus)
1680 data->lineno = strtol(plus, NULL, 10) - 1;
1681 else
1682 die("invalid diff");
1683 }
1684 }
1685
1686 static unsigned char *deflate_it(char *data,
1687 unsigned long size,
1688 unsigned long *result_size)
1689 {
1690 int bound;
1691 unsigned char *deflated;
1692 z_stream stream;
1693
1694 memset(&stream, 0, sizeof(stream));
1695 deflateInit(&stream, zlib_compression_level);
1696 bound = deflateBound(&stream, size);
1697 deflated = xmalloc(bound);
1698 stream.next_out = deflated;
1699 stream.avail_out = bound;
1700
1701 stream.next_in = (unsigned char *)data;
1702 stream.avail_in = size;
1703 while (deflate(&stream, Z_FINISH) == Z_OK)
1704 ; /* nothing */
1705 deflateEnd(&stream);
1706 *result_size = stream.total_out;
1707 return deflated;
1708 }
1709
1710 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1711 {
1712 void *cp;
1713 void *delta;
1714 void *deflated;
1715 void *data;
1716 unsigned long orig_size;
1717 unsigned long delta_size;
1718 unsigned long deflate_size;
1719 unsigned long data_size;
1720
1721 /* We could do deflated delta, or we could do just deflated two,
1722 * whichever is smaller.
1723 */
1724 delta = NULL;
1725 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1726 if (one->size && two->size) {
1727 delta = diff_delta(one->ptr, one->size,
1728 two->ptr, two->size,
1729 &delta_size, deflate_size);
1730 if (delta) {
1731 void *to_free = delta;
1732 orig_size = delta_size;
1733 delta = deflate_it(delta, delta_size, &delta_size);
1734 free(to_free);
1735 }
1736 }
1737
1738 if (delta && delta_size < deflate_size) {
1739 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
1740 free(deflated);
1741 data = delta;
1742 data_size = delta_size;
1743 }
1744 else {
1745 fprintf(file, "%sliteral %lu\n", prefix, two->size);
1746 free(delta);
1747 data = deflated;
1748 data_size = deflate_size;
1749 }
1750
1751 /* emit data encoded in base85 */
1752 cp = data;
1753 while (data_size) {
1754 int bytes = (52 < data_size) ? 52 : data_size;
1755 char line[70];
1756 data_size -= bytes;
1757 if (bytes <= 26)
1758 line[0] = bytes + 'A' - 1;
1759 else
1760 line[0] = bytes - 26 + 'a' - 1;
1761 encode_85(line + 1, cp, bytes);
1762 cp = (char *) cp + bytes;
1763 fprintf(file, "%s", prefix);
1764 fputs(line, file);
1765 fputc('\n', file);
1766 }
1767 fprintf(file, "%s\n", prefix);
1768 free(data);
1769 }
1770
1771 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1772 {
1773 fprintf(file, "%sGIT binary patch\n", prefix);
1774 emit_binary_diff_body(file, one, two, prefix);
1775 emit_binary_diff_body(file, two, one, prefix);
1776 }
1777
1778 static void diff_filespec_load_driver(struct diff_filespec *one)
1779 {
1780 /* Use already-loaded driver */
1781 if (one->driver)
1782 return;
1783
1784 if (S_ISREG(one->mode))
1785 one->driver = userdiff_find_by_path(one->path);
1786
1787 /* Fallback to default settings */
1788 if (!one->driver)
1789 one->driver = userdiff_find_by_name("default");
1790 }
1791
1792 int diff_filespec_is_binary(struct diff_filespec *one)
1793 {
1794 if (one->is_binary == -1) {
1795 diff_filespec_load_driver(one);
1796 if (one->driver->binary != -1)
1797 one->is_binary = one->driver->binary;
1798 else {
1799 if (!one->data && DIFF_FILE_VALID(one))
1800 diff_populate_filespec(one, 0);
1801 if (one->data)
1802 one->is_binary = buffer_is_binary(one->data,
1803 one->size);
1804 if (one->is_binary == -1)
1805 one->is_binary = 0;
1806 }
1807 }
1808 return one->is_binary;
1809 }
1810
1811 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1812 {
1813 diff_filespec_load_driver(one);
1814 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1815 }
1816
1817 static const char *userdiff_word_regex(struct diff_filespec *one)
1818 {
1819 diff_filespec_load_driver(one);
1820 return one->driver->word_regex;
1821 }
1822
1823 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1824 {
1825 if (!options->a_prefix)
1826 options->a_prefix = a;
1827 if (!options->b_prefix)
1828 options->b_prefix = b;
1829 }
1830
1831 struct userdiff_driver *get_textconv(struct diff_filespec *one)
1832 {
1833 if (!DIFF_FILE_VALID(one))
1834 return NULL;
1835
1836 diff_filespec_load_driver(one);
1837 if (!one->driver->textconv)
1838 return NULL;
1839
1840 if (one->driver->textconv_want_cache && !one->driver->textconv_cache) {
1841 struct notes_cache *c = xmalloc(sizeof(*c));
1842 struct strbuf name = STRBUF_INIT;
1843
1844 strbuf_addf(&name, "textconv/%s", one->driver->name);
1845 notes_cache_init(c, name.buf, one->driver->textconv);
1846 one->driver->textconv_cache = c;
1847 }
1848
1849 return one->driver;
1850 }
1851
1852 static void builtin_diff(const char *name_a,
1853 const char *name_b,
1854 struct diff_filespec *one,
1855 struct diff_filespec *two,
1856 const char *xfrm_msg,
1857 int must_show_header,
1858 struct diff_options *o,
1859 int complete_rewrite)
1860 {
1861 mmfile_t mf1, mf2;
1862 const char *lbl[2];
1863 char *a_one, *b_two;
1864 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1865 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1866 const char *a_prefix, *b_prefix;
1867 struct userdiff_driver *textconv_one = NULL;
1868 struct userdiff_driver *textconv_two = NULL;
1869 struct strbuf header = STRBUF_INIT;
1870 struct strbuf *msgbuf;
1871 char *line_prefix = "";
1872
1873 if (o->output_prefix) {
1874 msgbuf = o->output_prefix(o, o->output_prefix_data);
1875 line_prefix = msgbuf->buf;
1876 }
1877
1878 if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1879 (!one->mode || S_ISGITLINK(one->mode)) &&
1880 (!two->mode || S_ISGITLINK(two->mode))) {
1881 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1882 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1883 show_submodule_summary(o->file, one ? one->path : two->path,
1884 one->sha1, two->sha1, two->dirty_submodule,
1885 del, add, reset);
1886 return;
1887 }
1888
1889 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1890 textconv_one = get_textconv(one);
1891 textconv_two = get_textconv(two);
1892 }
1893
1894 diff_set_mnemonic_prefix(o, "a/", "b/");
1895 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1896 a_prefix = o->b_prefix;
1897 b_prefix = o->a_prefix;
1898 } else {
1899 a_prefix = o->a_prefix;
1900 b_prefix = o->b_prefix;
1901 }
1902
1903 /* Never use a non-valid filename anywhere if at all possible */
1904 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1905 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1906
1907 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1908 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1909 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1910 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1911 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
1912 if (lbl[0][0] == '/') {
1913 /* /dev/null */
1914 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
1915 if (xfrm_msg)
1916 strbuf_addstr(&header, xfrm_msg);
1917 must_show_header = 1;
1918 }
1919 else if (lbl[1][0] == '/') {
1920 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
1921 if (xfrm_msg)
1922 strbuf_addstr(&header, xfrm_msg);
1923 must_show_header = 1;
1924 }
1925 else {
1926 if (one->mode != two->mode) {
1927 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
1928 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
1929 must_show_header = 1;
1930 }
1931 if (xfrm_msg)
1932 strbuf_addstr(&header, xfrm_msg);
1933
1934 /*
1935 * we do not run diff between different kind
1936 * of objects.
1937 */
1938 if ((one->mode ^ two->mode) & S_IFMT)
1939 goto free_ab_and_return;
1940 if (complete_rewrite &&
1941 (textconv_one || !diff_filespec_is_binary(one)) &&
1942 (textconv_two || !diff_filespec_is_binary(two))) {
1943 fprintf(o->file, "%s", header.buf);
1944 strbuf_reset(&header);
1945 emit_rewrite_diff(name_a, name_b, one, two,
1946 textconv_one, textconv_two, o);
1947 o->found_changes = 1;
1948 goto free_ab_and_return;
1949 }
1950 }
1951
1952 if (!DIFF_OPT_TST(o, TEXT) &&
1953 ( (!textconv_one && diff_filespec_is_binary(one)) ||
1954 (!textconv_two && diff_filespec_is_binary(two)) )) {
1955 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1956 die("unable to read files to diff");
1957 /* Quite common confusing case */
1958 if (mf1.size == mf2.size &&
1959 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
1960 if (must_show_header)
1961 fprintf(o->file, "%s", header.buf);
1962 goto free_ab_and_return;
1963 }
1964 fprintf(o->file, "%s", header.buf);
1965 strbuf_reset(&header);
1966 if (DIFF_OPT_TST(o, BINARY))
1967 emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
1968 else
1969 fprintf(o->file, "%sBinary files %s and %s differ\n",
1970 line_prefix, lbl[0], lbl[1]);
1971 o->found_changes = 1;
1972 }
1973 else {
1974 /* Crazy xdl interfaces.. */
1975 const char *diffopts = getenv("GIT_DIFF_OPTS");
1976 xpparam_t xpp;
1977 xdemitconf_t xecfg;
1978 struct emit_callback ecbdata;
1979 const struct userdiff_funcname *pe;
1980
1981 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || must_show_header) {
1982 fprintf(o->file, "%s", header.buf);
1983 strbuf_reset(&header);
1984 }
1985
1986 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
1987 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
1988
1989 pe = diff_funcname_pattern(one);
1990 if (!pe)
1991 pe = diff_funcname_pattern(two);
1992
1993 memset(&xpp, 0, sizeof(xpp));
1994 memset(&xecfg, 0, sizeof(xecfg));
1995 memset(&ecbdata, 0, sizeof(ecbdata));
1996 ecbdata.label_path = lbl;
1997 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1998 ecbdata.found_changesp = &o->found_changes;
1999 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
2000 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2001 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2002 ecbdata.opt = o;
2003 ecbdata.header = header.len ? &header : NULL;
2004 xpp.flags = o->xdl_opts;
2005 xecfg.ctxlen = o->context;
2006 xecfg.interhunkctxlen = o->interhunkcontext;
2007 xecfg.flags = XDL_EMIT_FUNCNAMES;
2008 if (pe)
2009 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2010 if (!diffopts)
2011 ;
2012 else if (!prefixcmp(diffopts, "--unified="))
2013 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2014 else if (!prefixcmp(diffopts, "-u"))
2015 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2016 if (o->word_diff) {
2017 int i;
2018
2019 ecbdata.diff_words =
2020 xcalloc(1, sizeof(struct diff_words_data));
2021 ecbdata.diff_words->type = o->word_diff;
2022 ecbdata.diff_words->opt = o;
2023 if (!o->word_regex)
2024 o->word_regex = userdiff_word_regex(one);
2025 if (!o->word_regex)
2026 o->word_regex = userdiff_word_regex(two);
2027 if (!o->word_regex)
2028 o->word_regex = diff_word_regex_cfg;
2029 if (o->word_regex) {
2030 ecbdata.diff_words->word_regex = (regex_t *)
2031 xmalloc(sizeof(regex_t));
2032 if (regcomp(ecbdata.diff_words->word_regex,
2033 o->word_regex,
2034 REG_EXTENDED | REG_NEWLINE))
2035 die ("Invalid regular expression: %s",
2036 o->word_regex);
2037 }
2038 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2039 if (o->word_diff == diff_words_styles[i].type) {
2040 ecbdata.diff_words->style =
2041 &diff_words_styles[i];
2042 break;
2043 }
2044 }
2045 if (DIFF_OPT_TST(o, COLOR_DIFF)) {
2046 struct diff_words_style *st = ecbdata.diff_words->style;
2047 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2048 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2049 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
2050 }
2051 }
2052 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2053 &xpp, &xecfg);
2054 if (o->word_diff)
2055 free_diff_words_data(&ecbdata);
2056 if (textconv_one)
2057 free(mf1.ptr);
2058 if (textconv_two)
2059 free(mf2.ptr);
2060 xdiff_clear_find_func(&xecfg);
2061 }
2062
2063 free_ab_and_return:
2064 strbuf_release(&header);
2065 diff_free_filespec_data(one);
2066 diff_free_filespec_data(two);
2067 free(a_one);
2068 free(b_two);
2069 return;
2070 }
2071
2072 static void builtin_diffstat(const char *name_a, const char *name_b,
2073 struct diff_filespec *one,
2074 struct diff_filespec *two,
2075 struct diffstat_t *diffstat,
2076 struct diff_options *o,
2077 int complete_rewrite)
2078 {
2079 mmfile_t mf1, mf2;
2080 struct diffstat_file *data;
2081
2082 data = diffstat_add(diffstat, name_a, name_b);
2083
2084 if (!one || !two) {
2085 data->is_unmerged = 1;
2086 return;
2087 }
2088
2089 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2090 data->is_binary = 1;
2091 data->added = diff_filespec_size(two);
2092 data->deleted = diff_filespec_size(one);
2093 }
2094
2095 else if (complete_rewrite) {
2096 diff_populate_filespec(one, 0);
2097 diff_populate_filespec(two, 0);
2098 data->deleted = count_lines(one->data, one->size);
2099 data->added = count_lines(two->data, two->size);
2100 }
2101
2102 else {
2103 /* Crazy xdl interfaces.. */
2104 xpparam_t xpp;
2105 xdemitconf_t xecfg;
2106
2107 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2108 die("unable to read files to diff");
2109
2110 memset(&xpp, 0, sizeof(xpp));
2111 memset(&xecfg, 0, sizeof(xecfg));
2112 xpp.flags = o->xdl_opts;
2113 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2114 &xpp, &xecfg);
2115 }
2116
2117 diff_free_filespec_data(one);
2118 diff_free_filespec_data(two);
2119 }
2120
2121 static void builtin_checkdiff(const char *name_a, const char *name_b,
2122 const char *attr_path,
2123 struct diff_filespec *one,
2124 struct diff_filespec *two,
2125 struct diff_options *o)
2126 {
2127 mmfile_t mf1, mf2;
2128 struct checkdiff_t data;
2129
2130 if (!two)
2131 return;
2132
2133 memset(&data, 0, sizeof(data));
2134 data.filename = name_b ? name_b : name_a;
2135 data.lineno = 0;
2136 data.o = o;
2137 data.ws_rule = whitespace_rule(attr_path);
2138 data.conflict_marker_size = ll_merge_marker_size(attr_path);
2139
2140 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2141 die("unable to read files to diff");
2142
2143 /*
2144 * All the other codepaths check both sides, but not checking
2145 * the "old" side here is deliberate. We are checking the newly
2146 * introduced changes, and as long as the "new" side is text, we
2147 * can and should check what it introduces.
2148 */
2149 if (diff_filespec_is_binary(two))
2150 goto free_and_return;
2151 else {
2152 /* Crazy xdl interfaces.. */
2153 xpparam_t xpp;
2154 xdemitconf_t xecfg;
2155
2156 memset(&xpp, 0, sizeof(xpp));
2157 memset(&xecfg, 0, sizeof(xecfg));
2158 xecfg.ctxlen = 1; /* at least one context line */
2159 xpp.flags = 0;
2160 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2161 &xpp, &xecfg);
2162
2163 if (data.ws_rule & WS_BLANK_AT_EOF) {
2164 struct emit_callback ecbdata;
2165 int blank_at_eof;
2166
2167 ecbdata.ws_rule = data.ws_rule;
2168 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2169 blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2170
2171 if (blank_at_eof) {
2172 static char *err;
2173 if (!err)
2174 err = whitespace_error_string(WS_BLANK_AT_EOF);
2175 fprintf(o->file, "%s:%d: %s.\n",
2176 data.filename, blank_at_eof, err);
2177 data.status = 1; /* report errors */
2178 }
2179 }
2180 }
2181 free_and_return:
2182 diff_free_filespec_data(one);
2183 diff_free_filespec_data(two);
2184 if (data.status)
2185 DIFF_OPT_SET(o, CHECK_FAILED);
2186 }
2187
2188 struct diff_filespec *alloc_filespec(const char *path)
2189 {
2190 int namelen = strlen(path);
2191 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2192
2193 memset(spec, 0, sizeof(*spec));
2194 spec->path = (char *)(spec + 1);
2195 memcpy(spec->path, path, namelen+1);
2196 spec->count = 1;
2197 spec->is_binary = -1;
2198 return spec;
2199 }
2200
2201 void free_filespec(struct diff_filespec *spec)
2202 {
2203 if (!--spec->count) {
2204 diff_free_filespec_data(spec);
2205 free(spec);
2206 }
2207 }
2208
2209 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2210 unsigned short mode)
2211 {
2212 if (mode) {
2213 spec->mode = canon_mode(mode);
2214 hashcpy(spec->sha1, sha1);
2215 spec->sha1_valid = !is_null_sha1(sha1);
2216 }
2217 }
2218
2219 /*
2220 * Given a name and sha1 pair, if the index tells us the file in
2221 * the work tree has that object contents, return true, so that
2222 * prepare_temp_file() does not have to inflate and extract.
2223 */
2224 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2225 {
2226 struct cache_entry *ce;
2227 struct stat st;
2228 int pos, len;
2229
2230 /*
2231 * We do not read the cache ourselves here, because the
2232 * benchmark with my previous version that always reads cache
2233 * shows that it makes things worse for diff-tree comparing
2234 * two linux-2.6 kernel trees in an already checked out work
2235 * tree. This is because most diff-tree comparisons deal with
2236 * only a small number of files, while reading the cache is
2237 * expensive for a large project, and its cost outweighs the
2238 * savings we get by not inflating the object to a temporary
2239 * file. Practically, this code only helps when we are used
2240 * by diff-cache --cached, which does read the cache before
2241 * calling us.
2242 */
2243 if (!active_cache)
2244 return 0;
2245
2246 /* We want to avoid the working directory if our caller
2247 * doesn't need the data in a normal file, this system
2248 * is rather slow with its stat/open/mmap/close syscalls,
2249 * and the object is contained in a pack file. The pack
2250 * is probably already open and will be faster to obtain
2251 * the data through than the working directory. Loose
2252 * objects however would tend to be slower as they need
2253 * to be individually opened and inflated.
2254 */
2255 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2256 return 0;
2257
2258 len = strlen(name);
2259 pos = cache_name_pos(name, len);
2260 if (pos < 0)
2261 return 0;
2262 ce = active_cache[pos];
2263
2264 /*
2265 * This is not the sha1 we are looking for, or
2266 * unreusable because it is not a regular file.
2267 */
2268 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2269 return 0;
2270
2271 /*
2272 * If ce is marked as "assume unchanged", there is no
2273 * guarantee that work tree matches what we are looking for.
2274 */
2275 if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2276 return 0;
2277
2278 /*
2279 * If ce matches the file in the work tree, we can reuse it.
2280 */
2281 if (ce_uptodate(ce) ||
2282 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2283 return 1;
2284
2285 return 0;
2286 }
2287
2288 static int populate_from_stdin(struct diff_filespec *s)
2289 {
2290 struct strbuf buf = STRBUF_INIT;
2291 size_t size = 0;
2292
2293 if (strbuf_read(&buf, 0, 0) < 0)
2294 return error("error while reading from stdin %s",
2295 strerror(errno));
2296
2297 s->should_munmap = 0;
2298 s->data = strbuf_detach(&buf, &size);
2299 s->size = size;
2300 s->should_free = 1;
2301 return 0;
2302 }
2303
2304 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2305 {
2306 int len;
2307 char *data = xmalloc(100), *dirty = "";
2308
2309 /* Are we looking at the work tree? */
2310 if (s->dirty_submodule)
2311 dirty = "-dirty";
2312
2313 len = snprintf(data, 100,
2314 "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2315 s->data = data;
2316 s->size = len;
2317 s->should_free = 1;
2318 if (size_only) {
2319 s->data = NULL;
2320 free(data);
2321 }
2322 return 0;
2323 }
2324
2325 /*
2326 * While doing rename detection and pickaxe operation, we may need to
2327 * grab the data for the blob (or file) for our own in-core comparison.
2328 * diff_filespec has data and size fields for this purpose.
2329 */
2330 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2331 {
2332 int err = 0;
2333 if (!DIFF_FILE_VALID(s))
2334 die("internal error: asking to populate invalid file.");
2335 if (S_ISDIR(s->mode))
2336 return -1;
2337
2338 if (s->data)
2339 return 0;
2340
2341 if (size_only && 0 < s->size)
2342 return 0;
2343
2344 if (S_ISGITLINK(s->mode))
2345 return diff_populate_gitlink(s, size_only);
2346
2347 if (!s->sha1_valid ||
2348 reuse_worktree_file(s->path, s->sha1, 0)) {
2349 struct strbuf buf = STRBUF_INIT;
2350 struct stat st;
2351 int fd;
2352
2353 if (!strcmp(s->path, "-"))
2354 return populate_from_stdin(s);
2355
2356 if (lstat(s->path, &st) < 0) {
2357 if (errno == ENOENT) {
2358 err_empty:
2359 err = -1;
2360 empty:
2361 s->data = (char *)"";
2362 s->size = 0;
2363 return err;
2364 }
2365 }
2366 s->size = xsize_t(st.st_size);
2367 if (!s->size)
2368 goto empty;
2369 if (S_ISLNK(st.st_mode)) {
2370 struct strbuf sb = STRBUF_INIT;
2371
2372 if (strbuf_readlink(&sb, s->path, s->size))
2373 goto err_empty;
2374 s->size = sb.len;
2375 s->data = strbuf_detach(&sb, NULL);
2376 s->should_free = 1;
2377 return 0;
2378 }
2379 if (size_only)
2380 return 0;
2381 fd = open(s->path, O_RDONLY);
2382 if (fd < 0)
2383 goto err_empty;
2384 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2385 close(fd);
2386 s->should_munmap = 1;
2387
2388 /*
2389 * Convert from working tree format to canonical git format
2390 */
2391 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2392 size_t size = 0;
2393 munmap(s->data, s->size);
2394 s->should_munmap = 0;
2395 s->data = strbuf_detach(&buf, &size);
2396 s->size = size;
2397 s->should_free = 1;
2398 }
2399 }
2400 else {
2401 enum object_type type;
2402 if (size_only) {
2403 type = sha1_object_info(s->sha1, &s->size);
2404 if (type < 0)
2405 die("unable to read %s", sha1_to_hex(s->sha1));
2406 } else {
2407 s->data = read_sha1_file(s->sha1, &type, &s->size);
2408 if (!s->data)
2409 die("unable to read %s", sha1_to_hex(s->sha1));
2410 s->should_free = 1;
2411 }
2412 }
2413 return 0;
2414 }
2415
2416 void diff_free_filespec_blob(struct diff_filespec *s)
2417 {
2418 if (s->should_free)
2419 free(s->data);
2420 else if (s->should_munmap)
2421 munmap(s->data, s->size);
2422
2423 if (s->should_free || s->should_munmap) {
2424 s->should_free = s->should_munmap = 0;
2425 s->data = NULL;
2426 }
2427 }
2428
2429 void diff_free_filespec_data(struct diff_filespec *s)
2430 {
2431 diff_free_filespec_blob(s);
2432 free(s->cnt_data);
2433 s->cnt_data = NULL;
2434 }
2435
2436 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2437 void *blob,
2438 unsigned long size,
2439 const unsigned char *sha1,
2440 int mode)
2441 {
2442 int fd;
2443 struct strbuf buf = STRBUF_INIT;
2444 struct strbuf template = STRBUF_INIT;
2445 char *path_dup = xstrdup(path);
2446 const char *base = basename(path_dup);
2447
2448 /* Generate "XXXXXX_basename.ext" */
2449 strbuf_addstr(&template, "XXXXXX_");
2450 strbuf_addstr(&template, base);
2451
2452 fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2453 strlen(base) + 1);
2454 if (fd < 0)
2455 die_errno("unable to create temp-file");
2456 if (convert_to_working_tree(path,
2457 (const char *)blob, (size_t)size, &buf)) {
2458 blob = buf.buf;
2459 size = buf.len;
2460 }
2461 if (write_in_full(fd, blob, size) != size)
2462 die_errno("unable to write temp-file");
2463 close(fd);
2464 temp->name = temp->tmp_path;
2465 strcpy(temp->hex, sha1_to_hex(sha1));
2466 temp->hex[40] = 0;
2467 sprintf(temp->mode, "%06o", mode);
2468 strbuf_release(&buf);
2469 strbuf_release(&template);
2470 free(path_dup);
2471 }
2472
2473 static struct diff_tempfile *prepare_temp_file(const char *name,
2474 struct diff_filespec *one)
2475 {
2476 struct diff_tempfile *temp = claim_diff_tempfile();
2477
2478 if (!DIFF_FILE_VALID(one)) {
2479 not_a_valid_file:
2480 /* A '-' entry produces this for file-2, and
2481 * a '+' entry produces this for file-1.
2482 */
2483 temp->name = "/dev/null";
2484 strcpy(temp->hex, ".");
2485 strcpy(temp->mode, ".");
2486 return temp;
2487 }
2488
2489 if (!remove_tempfile_installed) {
2490 atexit(remove_tempfile);
2491 sigchain_push_common(remove_tempfile_on_signal);
2492 remove_tempfile_installed = 1;
2493 }
2494
2495 if (!one->sha1_valid ||
2496 reuse_worktree_file(name, one->sha1, 1)) {
2497 struct stat st;
2498 if (lstat(name, &st) < 0) {
2499 if (errno == ENOENT)
2500 goto not_a_valid_file;
2501 die_errno("stat(%s)", name);
2502 }
2503 if (S_ISLNK(st.st_mode)) {
2504 struct strbuf sb = STRBUF_INIT;
2505 if (strbuf_readlink(&sb, name, st.st_size) < 0)
2506 die_errno("readlink(%s)", name);
2507 prep_temp_blob(name, temp, sb.buf, sb.len,
2508 (one->sha1_valid ?
2509 one->sha1 : null_sha1),
2510 (one->sha1_valid ?
2511 one->mode : S_IFLNK));
2512 strbuf_release(&sb);
2513 }
2514 else {
2515 /* we can borrow from the file in the work tree */
2516 temp->name = name;
2517 if (!one->sha1_valid)
2518 strcpy(temp->hex, sha1_to_hex(null_sha1));
2519 else
2520 strcpy(temp->hex, sha1_to_hex(one->sha1));
2521 /* Even though we may sometimes borrow the
2522 * contents from the work tree, we always want
2523 * one->mode. mode is trustworthy even when
2524 * !(one->sha1_valid), as long as
2525 * DIFF_FILE_VALID(one).
2526 */
2527 sprintf(temp->mode, "%06o", one->mode);
2528 }
2529 return temp;
2530 }
2531 else {
2532 if (diff_populate_filespec(one, 0))
2533 die("cannot read data blob for %s", one->path);
2534 prep_temp_blob(name, temp, one->data, one->size,
2535 one->sha1, one->mode);
2536 }
2537 return temp;
2538 }
2539
2540 /* An external diff command takes:
2541 *
2542 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2543 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2544 *
2545 */
2546 static void run_external_diff(const char *pgm,
2547 const char *name,
2548 const char *other,
2549 struct diff_filespec *one,
2550 struct diff_filespec *two,
2551 const char *xfrm_msg,
2552 int complete_rewrite)
2553 {
2554 const char *spawn_arg[10];
2555 int retval;
2556 const char **arg = &spawn_arg[0];
2557
2558 if (one && two) {
2559 struct diff_tempfile *temp_one, *temp_two;
2560 const char *othername = (other ? other : name);
2561 temp_one = prepare_temp_file(name, one);
2562 temp_two = prepare_temp_file(othername, two);
2563 *arg++ = pgm;
2564 *arg++ = name;
2565 *arg++ = temp_one->name;
2566 *arg++ = temp_one->hex;
2567 *arg++ = temp_one->mode;
2568 *arg++ = temp_two->name;
2569 *arg++ = temp_two->hex;
2570 *arg++ = temp_two->mode;
2571 if (other) {
2572 *arg++ = other;
2573 *arg++ = xfrm_msg;
2574 }
2575 } else {
2576 *arg++ = pgm;
2577 *arg++ = name;
2578 }
2579 *arg = NULL;
2580 fflush(NULL);
2581 retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2582 remove_tempfile();
2583 if (retval) {
2584 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2585 exit(1);
2586 }
2587 }
2588
2589 static int similarity_index(struct diff_filepair *p)
2590 {
2591 return p->score * 100 / MAX_SCORE;
2592 }
2593
2594 static void fill_metainfo(struct strbuf *msg,
2595 const char *name,
2596 const char *other,
2597 struct diff_filespec *one,
2598 struct diff_filespec *two,
2599 struct diff_options *o,
2600 struct diff_filepair *p,
2601 int *must_show_header,
2602 int use_color)
2603 {
2604 const char *set = diff_get_color(use_color, DIFF_METAINFO);
2605 const char *reset = diff_get_color(use_color, DIFF_RESET);
2606 struct strbuf *msgbuf;
2607 char *line_prefix = "";
2608
2609 *must_show_header = 1;
2610 if (o->output_prefix) {
2611 msgbuf = o->output_prefix(o, o->output_prefix_data);
2612 line_prefix = msgbuf->buf;
2613 }
2614 strbuf_init(msg, PATH_MAX * 2 + 300);
2615 switch (p->status) {
2616 case DIFF_STATUS_COPIED:
2617 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2618 line_prefix, set, similarity_index(p));
2619 strbuf_addf(msg, "%s\n%s%scopy from ",
2620 reset, line_prefix, set);
2621 quote_c_style(name, msg, NULL, 0);
2622 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2623 quote_c_style(other, msg, NULL, 0);
2624 strbuf_addf(msg, "%s\n", reset);
2625 break;
2626 case DIFF_STATUS_RENAMED:
2627 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2628 line_prefix, set, similarity_index(p));
2629 strbuf_addf(msg, "%s\n%s%srename from ",
2630 reset, line_prefix, set);
2631 quote_c_style(name, msg, NULL, 0);
2632 strbuf_addf(msg, "%s\n%s%srename to ",
2633 reset, line_prefix, set);
2634 quote_c_style(other, msg, NULL, 0);
2635 strbuf_addf(msg, "%s\n", reset);
2636 break;
2637 case DIFF_STATUS_MODIFIED:
2638 if (p->score) {
2639 strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2640 line_prefix,
2641 set, similarity_index(p), reset);
2642 break;
2643 }
2644 /* fallthru */
2645 default:
2646 *must_show_header = 0;
2647 }
2648 if (one && two && hashcmp(one->sha1, two->sha1)) {
2649 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2650
2651 if (DIFF_OPT_TST(o, BINARY)) {
2652 mmfile_t mf;
2653 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2654 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2655 abbrev = 40;
2656 }
2657 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2658 find_unique_abbrev(one->sha1, abbrev));
2659 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2660 if (one->mode == two->mode)
2661 strbuf_addf(msg, " %06o", one->mode);
2662 strbuf_addf(msg, "%s\n", reset);
2663 }
2664 }
2665
2666 static void run_diff_cmd(const char *pgm,
2667 const char *name,
2668 const char *other,
2669 const char *attr_path,
2670 struct diff_filespec *one,
2671 struct diff_filespec *two,
2672 struct strbuf *msg,
2673 struct diff_options *o,
2674 struct diff_filepair *p)
2675 {
2676 const char *xfrm_msg = NULL;
2677 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2678 int must_show_header = 0;
2679
2680 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2681 pgm = NULL;
2682 else {
2683 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2684 if (drv && drv->external)
2685 pgm = drv->external;
2686 }
2687
2688 if (msg) {
2689 /*
2690 * don't use colors when the header is intended for an
2691 * external diff driver
2692 */
2693 fill_metainfo(msg, name, other, one, two, o, p,
2694 &must_show_header,
2695 DIFF_OPT_TST(o, COLOR_DIFF) && !pgm);
2696 xfrm_msg = msg->len ? msg->buf : NULL;
2697 }
2698
2699 if (pgm) {
2700 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2701 complete_rewrite);
2702 return;
2703 }
2704 if (one && two)
2705 builtin_diff(name, other ? other : name,
2706 one, two, xfrm_msg, must_show_header,
2707 o, complete_rewrite);
2708 else
2709 fprintf(o->file, "* Unmerged path %s\n", name);
2710 }
2711
2712 static void diff_fill_sha1_info(struct diff_filespec *one)
2713 {
2714 if (DIFF_FILE_VALID(one)) {
2715 if (!one->sha1_valid) {
2716 struct stat st;
2717 if (!strcmp(one->path, "-")) {
2718 hashcpy(one->sha1, null_sha1);
2719 return;
2720 }
2721 if (lstat(one->path, &st) < 0)
2722 die_errno("stat '%s'", one->path);
2723 if (index_path(one->sha1, one->path, &st, 0))
2724 die("cannot hash %s", one->path);
2725 }
2726 }
2727 else
2728 hashclr(one->sha1);
2729 }
2730
2731 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2732 {
2733 /* Strip the prefix but do not molest /dev/null and absolute paths */
2734 if (*namep && **namep != '/') {
2735 *namep += prefix_length;
2736 if (**namep == '/')
2737 ++*namep;
2738 }
2739 if (*otherp && **otherp != '/') {
2740 *otherp += prefix_length;
2741 if (**otherp == '/')
2742 ++*otherp;
2743 }
2744 }
2745
2746 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2747 {
2748 const char *pgm = external_diff();
2749 struct strbuf msg;
2750 struct diff_filespec *one = p->one;
2751 struct diff_filespec *two = p->two;
2752 const char *name;
2753 const char *other;
2754 const char *attr_path;
2755
2756 name = p->one->path;
2757 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2758 attr_path = name;
2759 if (o->prefix_length)
2760 strip_prefix(o->prefix_length, &name, &other);
2761
2762 if (DIFF_PAIR_UNMERGED(p)) {
2763 run_diff_cmd(pgm, name, NULL, attr_path,
2764 NULL, NULL, NULL, o, p);
2765 return;
2766 }
2767
2768 diff_fill_sha1_info(one);
2769 diff_fill_sha1_info(two);
2770
2771 if (!pgm &&
2772 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2773 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2774 /*
2775 * a filepair that changes between file and symlink
2776 * needs to be split into deletion and creation.
2777 */
2778 struct diff_filespec *null = alloc_filespec(two->path);
2779 run_diff_cmd(NULL, name, other, attr_path,
2780 one, null, &msg, o, p);
2781 free(null);
2782 strbuf_release(&msg);
2783
2784 null = alloc_filespec(one->path);
2785 run_diff_cmd(NULL, name, other, attr_path,
2786 null, two, &msg, o, p);
2787 free(null);
2788 }
2789 else
2790 run_diff_cmd(pgm, name, other, attr_path,
2791 one, two, &msg, o, p);
2792
2793 strbuf_release(&msg);
2794 }
2795
2796 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2797 struct diffstat_t *diffstat)
2798 {
2799 const char *name;
2800 const char *other;
2801 int complete_rewrite = 0;
2802
2803 if (DIFF_PAIR_UNMERGED(p)) {
2804 /* unmerged */
2805 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2806 return;
2807 }
2808
2809 name = p->one->path;
2810 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2811
2812 if (o->prefix_length)
2813 strip_prefix(o->prefix_length, &name, &other);
2814
2815 diff_fill_sha1_info(p->one);
2816 diff_fill_sha1_info(p->two);
2817
2818 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2819 complete_rewrite = 1;
2820 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2821 }
2822
2823 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2824 {
2825 const char *name;
2826 const char *other;
2827 const char *attr_path;
2828
2829 if (DIFF_PAIR_UNMERGED(p)) {
2830 /* unmerged */
2831 return;
2832 }
2833
2834 name = p->one->path;
2835 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2836 attr_path = other ? other : name;
2837
2838 if (o->prefix_length)
2839 strip_prefix(o->prefix_length, &name, &other);
2840
2841 diff_fill_sha1_info(p->one);
2842 diff_fill_sha1_info(p->two);
2843
2844 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2845 }
2846
2847 void diff_setup(struct diff_options *options)
2848 {
2849 memcpy(options, &default_diff_options, sizeof(*options));
2850
2851 options->file = stdout;
2852
2853 options->line_termination = '\n';
2854 options->break_opt = -1;
2855 options->rename_limit = -1;
2856 options->dirstat_percent = 3;
2857 options->context = 3;
2858
2859 options->change = diff_change;
2860 options->add_remove = diff_addremove;
2861 if (diff_use_color_default > 0)
2862 DIFF_OPT_SET(options, COLOR_DIFF);
2863 options->detect_rename = diff_detect_rename_default;
2864
2865 if (diff_no_prefix) {
2866 options->a_prefix = options->b_prefix = "";
2867 } else if (!diff_mnemonic_prefix) {
2868 options->a_prefix = "a/";
2869 options->b_prefix = "b/";
2870 }
2871 }
2872
2873 int diff_setup_done(struct diff_options *options)
2874 {
2875 int count = 0;
2876
2877 if (options->output_format & DIFF_FORMAT_NAME)
2878 count++;
2879 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2880 count++;
2881 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2882 count++;
2883 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2884 count++;
2885 if (count > 1)
2886 die("--name-only, --name-status, --check and -s are mutually exclusive");
2887
2888 /*
2889 * Most of the time we can say "there are changes"
2890 * only by checking if there are changed paths, but
2891 * --ignore-whitespace* options force us to look
2892 * inside contents.
2893 */
2894
2895 if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2896 DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2897 DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2898 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2899 else
2900 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2901
2902 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2903 options->detect_rename = DIFF_DETECT_COPY;
2904
2905 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2906 options->prefix = NULL;
2907 if (options->prefix)
2908 options->prefix_length = strlen(options->prefix);
2909 else
2910 options->prefix_length = 0;
2911
2912 if (options->output_format & (DIFF_FORMAT_NAME |
2913 DIFF_FORMAT_NAME_STATUS |
2914 DIFF_FORMAT_CHECKDIFF |
2915 DIFF_FORMAT_NO_OUTPUT))
2916 options->output_format &= ~(DIFF_FORMAT_RAW |
2917 DIFF_FORMAT_NUMSTAT |
2918 DIFF_FORMAT_DIFFSTAT |
2919 DIFF_FORMAT_SHORTSTAT |
2920 DIFF_FORMAT_DIRSTAT |
2921 DIFF_FORMAT_SUMMARY |
2922 DIFF_FORMAT_PATCH);
2923
2924 /*
2925 * These cases always need recursive; we do not drop caller-supplied
2926 * recursive bits for other formats here.
2927 */
2928 if (options->output_format & (DIFF_FORMAT_PATCH |
2929 DIFF_FORMAT_NUMSTAT |
2930 DIFF_FORMAT_DIFFSTAT |
2931 DIFF_FORMAT_SHORTSTAT |
2932 DIFF_FORMAT_DIRSTAT |
2933 DIFF_FORMAT_SUMMARY |
2934 DIFF_FORMAT_CHECKDIFF))
2935 DIFF_OPT_SET(options, RECURSIVE);
2936 /*
2937 * Also pickaxe would not work very well if you do not say recursive
2938 */
2939 if (options->pickaxe)
2940 DIFF_OPT_SET(options, RECURSIVE);
2941 /*
2942 * When patches are generated, submodules diffed against the work tree
2943 * must be checked for dirtiness too so it can be shown in the output
2944 */
2945 if (options->output_format & DIFF_FORMAT_PATCH)
2946 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2947
2948 if (options->detect_rename && options->rename_limit < 0)
2949 options->rename_limit = diff_rename_limit_default;
2950 if (options->setup & DIFF_SETUP_USE_CACHE) {
2951 if (!active_cache)
2952 /* read-cache does not die even when it fails
2953 * so it is safe for us to do this here. Also
2954 * it does not smudge active_cache or active_nr
2955 * when it fails, so we do not have to worry about
2956 * cleaning it up ourselves either.
2957 */
2958 read_cache();
2959 }
2960 if (options->abbrev <= 0 || 40 < options->abbrev)
2961 options->abbrev = 40; /* full */
2962
2963 /*
2964 * It does not make sense to show the first hit we happened
2965 * to have found. It does not make sense not to return with
2966 * exit code in such a case either.
2967 */
2968 if (DIFF_OPT_TST(options, QUICK)) {
2969 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2970 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2971 }
2972
2973 return 0;
2974 }
2975
2976 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2977 {
2978 char c, *eq;
2979 int len;
2980
2981 if (*arg != '-')
2982 return 0;
2983 c = *++arg;
2984 if (!c)
2985 return 0;
2986 if (c == arg_short) {
2987 c = *++arg;
2988 if (!c)
2989 return 1;
2990 if (val && isdigit(c)) {
2991 char *end;
2992 int n = strtoul(arg, &end, 10);
2993 if (*end)
2994 return 0;
2995 *val = n;
2996 return 1;
2997 }
2998 return 0;
2999 }
3000 if (c != '-')
3001 return 0;
3002 arg++;
3003 eq = strchr(arg, '=');
3004 if (eq)
3005 len = eq - arg;
3006 else
3007 len = strlen(arg);
3008 if (!len || strncmp(arg, arg_long, len))
3009 return 0;
3010 if (eq) {
3011 int n;
3012 char *end;
3013 if (!isdigit(*++eq))
3014 return 0;
3015 n = strtoul(eq, &end, 10);
3016 if (*end)
3017 return 0;
3018 *val = n;
3019 }
3020 return 1;
3021 }
3022
3023 static int diff_scoreopt_parse(const char *opt);
3024
3025 static inline int short_opt(char opt, const char **argv,
3026 const char **optarg)
3027 {
3028 const char *arg = argv[0];
3029 if (arg[0] != '-' || arg[1] != opt)
3030 return 0;
3031 if (arg[2] != '\0') {
3032 *optarg = arg + 2;
3033 return 1;
3034 }
3035 if (!argv[1])
3036 die("Option '%c' requires a value", opt);
3037 *optarg = argv[1];
3038 return 2;
3039 }
3040
3041 int parse_long_opt(const char *opt, const char **argv,
3042 const char **optarg)
3043 {
3044 const char *arg = argv[0];
3045 if (arg[0] != '-' || arg[1] != '-')
3046 return 0;
3047 arg += strlen("--");
3048 if (prefixcmp(arg, opt))
3049 return 0;
3050 arg += strlen(opt);
3051 if (*arg == '=') { /* sticked form: --option=value */
3052 *optarg = arg + 1;
3053 return 1;
3054 }
3055 if (*arg != '\0')
3056 return 0;
3057 /* separate form: --option value */
3058 if (!argv[1])
3059 die("Option '--%s' requires a value", opt);
3060 *optarg = argv[1];
3061 return 2;
3062 }
3063
3064 static int stat_opt(struct diff_options *options, const char **av)
3065 {
3066 const char *arg = av[0];
3067 char *end;
3068 int width = options->stat_width;
3069 int name_width = options->stat_name_width;
3070 int argcount = 1;
3071
3072 arg += strlen("--stat");
3073 end = (char *)arg;
3074
3075 switch (*arg) {
3076 case '-':
3077 if (!prefixcmp(arg, "-width")) {
3078 arg += strlen("-width");
3079 if (*arg == '=')
3080 width = strtoul(arg + 1, &end, 10);
3081 else if (!*arg && !av[1])
3082 die("Option '--stat-width' requires a value");
3083 else if (!*arg) {
3084 width = strtoul(av[1], &end, 10);
3085 argcount = 2;
3086 }
3087 } else if (!prefixcmp(arg, "-name-width")) {
3088 arg += strlen("-name-width");
3089 if (*arg == '=')
3090 name_width = strtoul(arg + 1, &end, 10);
3091 else if (!*arg && !av[1])
3092 die("Option '--stat-name-width' requires a value");
3093 else if (!*arg) {
3094 name_width = strtoul(av[1], &end, 10);
3095 argcount = 2;
3096 }
3097 }
3098 break;
3099 case '=':
3100 width = strtoul(arg+1, &end, 10);
3101 if (*end == ',')
3102 name_width = strtoul(end+1, &end, 10);
3103 }
3104
3105 /* Important! This checks all the error cases! */
3106 if (*end)
3107 return 0;
3108 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3109 options->stat_name_width = name_width;
3110 options->stat_width = width;
3111 return argcount;
3112 }
3113
3114 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3115 {
3116 const char *arg = av[0];
3117 const char *optarg;
3118 int argcount;
3119
3120 /* Output format options */
3121 if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3122 options->output_format |= DIFF_FORMAT_PATCH;
3123 else if (opt_arg(arg, 'U', "unified", &options->context))
3124 options->output_format |= DIFF_FORMAT_PATCH;
3125 else if (!strcmp(arg, "--raw"))
3126 options->output_format |= DIFF_FORMAT_RAW;
3127 else if (!strcmp(arg, "--patch-with-raw"))
3128 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3129 else if (!strcmp(arg, "--numstat"))
3130 options->output_format |= DIFF_FORMAT_NUMSTAT;
3131 else if (!strcmp(arg, "--shortstat"))
3132 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3133 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
3134 options->output_format |= DIFF_FORMAT_DIRSTAT;
3135 else if (!strcmp(arg, "--cumulative")) {
3136 options->output_format |= DIFF_FORMAT_DIRSTAT;
3137 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
3138 } else if (opt_arg(arg, 0, "dirstat-by-file",
3139 &options->dirstat_percent)) {
3140 options->output_format |= DIFF_FORMAT_DIRSTAT;
3141 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
3142 }
3143 else if (!strcmp(arg, "--check"))
3144 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3145 else if (!strcmp(arg, "--summary"))
3146 options->output_format |= DIFF_FORMAT_SUMMARY;
3147 else if (!strcmp(arg, "--patch-with-stat"))
3148 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3149 else if (!strcmp(arg, "--name-only"))
3150 options->output_format |= DIFF_FORMAT_NAME;
3151 else if (!strcmp(arg, "--name-status"))
3152 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3153 else if (!strcmp(arg, "-s"))
3154 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3155 else if (!prefixcmp(arg, "--stat"))
3156 /* --stat, --stat-width, or --stat-name-width */
3157 return stat_opt(options, av);
3158
3159 /* renames options */
3160 else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3161 !strcmp(arg, "--break-rewrites")) {
3162 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3163 return error("invalid argument to -B: %s", arg+2);
3164 }
3165 else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3166 !strcmp(arg, "--find-renames")) {
3167 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3168 return error("invalid argument to -M: %s", arg+2);
3169 options->detect_rename = DIFF_DETECT_RENAME;
3170 }
3171 else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3172 !strcmp(arg, "--find-copies")) {
3173 if (options->detect_rename == DIFF_DETECT_COPY)
3174 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3175 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3176 return error("invalid argument to -C: %s", arg+2);
3177 options->detect_rename = DIFF_DETECT_COPY;
3178 }
3179 else if (!strcmp(arg, "--no-renames"))
3180 options->detect_rename = 0;
3181 else if (!strcmp(arg, "--relative"))
3182 DIFF_OPT_SET(options, RELATIVE_NAME);
3183 else if (!prefixcmp(arg, "--relative=")) {
3184 DIFF_OPT_SET(options, RELATIVE_NAME);
3185 options->prefix = arg + 11;
3186 }
3187
3188 /* xdiff options */
3189 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3190 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3191 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3192 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3193 else if (!strcmp(arg, "--ignore-space-at-eol"))
3194 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3195 else if (!strcmp(arg, "--patience"))
3196 DIFF_XDL_SET(options, PATIENCE_DIFF);
3197
3198 /* flags options */
3199 else if (!strcmp(arg, "--binary")) {
3200 options->output_format |= DIFF_FORMAT_PATCH;
3201 DIFF_OPT_SET(options, BINARY);
3202 }
3203 else if (!strcmp(arg, "--full-index"))
3204 DIFF_OPT_SET(options, FULL_INDEX);
3205 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3206 DIFF_OPT_SET(options, TEXT);
3207 else if (!strcmp(arg, "-R"))
3208 DIFF_OPT_SET(options, REVERSE_DIFF);
3209 else if (!strcmp(arg, "--find-copies-harder"))
3210 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3211 else if (!strcmp(arg, "--follow"))
3212 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3213 else if (!strcmp(arg, "--color"))
3214 DIFF_OPT_SET(options, COLOR_DIFF);
3215 else if (!prefixcmp(arg, "--color=")) {
3216 int value = git_config_colorbool(NULL, arg+8, -1);
3217 if (value == 0)
3218 DIFF_OPT_CLR(options, COLOR_DIFF);
3219 else if (value > 0)
3220 DIFF_OPT_SET(options, COLOR_DIFF);
3221 else
3222 return error("option `color' expects \"always\", \"auto\", or \"never\"");
3223 }
3224 else if (!strcmp(arg, "--no-color"))
3225 DIFF_OPT_CLR(options, COLOR_DIFF);
3226 else if (!strcmp(arg, "--color-words")) {
3227 DIFF_OPT_SET(options, COLOR_DIFF);
3228 options->word_diff = DIFF_WORDS_COLOR;
3229 }
3230 else if (!prefixcmp(arg, "--color-words=")) {
3231 DIFF_OPT_SET(options, COLOR_DIFF);
3232 options->word_diff = DIFF_WORDS_COLOR;
3233 options->word_regex = arg + 14;
3234 }
3235 else if (!strcmp(arg, "--word-diff")) {
3236 if (options->word_diff == DIFF_WORDS_NONE)
3237 options->word_diff = DIFF_WORDS_PLAIN;
3238 }
3239 else if (!prefixcmp(arg, "--word-diff=")) {
3240 const char *type = arg + 12;
3241 if (!strcmp(type, "plain"))
3242 options->word_diff = DIFF_WORDS_PLAIN;
3243 else if (!strcmp(type, "color")) {
3244 DIFF_OPT_SET(options, COLOR_DIFF);
3245 options->word_diff = DIFF_WORDS_COLOR;
3246 }
3247 else if (!strcmp(type, "porcelain"))
3248 options->word_diff = DIFF_WORDS_PORCELAIN;
3249 else if (!strcmp(type, "none"))
3250 options->word_diff = DIFF_WORDS_NONE;
3251 else
3252 die("bad --word-diff argument: %s", type);
3253 }
3254 else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3255 if (options->word_diff == DIFF_WORDS_NONE)
3256 options->word_diff = DIFF_WORDS_PLAIN;
3257 options->word_regex = optarg;
3258 return argcount;
3259 }
3260 else if (!strcmp(arg, "--exit-code"))
3261 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3262 else if (!strcmp(arg, "--quiet"))
3263 DIFF_OPT_SET(options, QUICK);
3264 else if (!strcmp(arg, "--ext-diff"))
3265 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3266 else if (!strcmp(arg, "--no-ext-diff"))
3267 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3268 else if (!strcmp(arg, "--textconv"))
3269 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3270 else if (!strcmp(arg, "--no-textconv"))
3271 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3272 else if (!strcmp(arg, "--ignore-submodules")) {
3273 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3274 handle_ignore_submodules_arg(options, "all");
3275 } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3276 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3277 handle_ignore_submodules_arg(options, arg + 20);
3278 } else if (!strcmp(arg, "--submodule"))
3279 DIFF_OPT_SET(options, SUBMODULE_LOG);
3280 else if (!prefixcmp(arg, "--submodule=")) {
3281 if (!strcmp(arg + 12, "log"))
3282 DIFF_OPT_SET(options, SUBMODULE_LOG);
3283 }
3284
3285 /* misc options */
3286 else if (!strcmp(arg, "-z"))
3287 options->line_termination = 0;
3288 else if ((argcount = short_opt('l', av, &optarg))) {
3289 options->rename_limit = strtoul(optarg, NULL, 10);
3290 return argcount;
3291 }
3292 else if ((argcount = short_opt('S', av, &optarg))) {
3293 options->pickaxe = optarg;
3294 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3295 return argcount;
3296 } else if ((argcount = short_opt('G', av, &optarg))) {
3297 options->pickaxe = optarg;
3298 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3299 return argcount;
3300 }
3301 else if (!strcmp(arg, "--pickaxe-all"))
3302 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3303 else if (!strcmp(arg, "--pickaxe-regex"))
3304 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3305 else if ((argcount = short_opt('O', av, &optarg))) {
3306 options->orderfile = optarg;
3307 return argcount;
3308 }
3309 else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3310 options->filter = optarg;
3311 return argcount;
3312 }
3313 else if (!strcmp(arg, "--abbrev"))
3314 options->abbrev = DEFAULT_ABBREV;
3315 else if (!prefixcmp(arg, "--abbrev=")) {
3316 options->abbrev = strtoul(arg + 9, NULL, 10);
3317 if (options->abbrev < MINIMUM_ABBREV)
3318 options->abbrev = MINIMUM_ABBREV;
3319 else if (40 < options->abbrev)
3320 options->abbrev = 40;
3321 }
3322 else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3323 options->a_prefix = optarg;
3324 return argcount;
3325 }
3326 else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3327 options->b_prefix = optarg;
3328 return argcount;
3329 }
3330 else if (!strcmp(arg, "--no-prefix"))
3331 options->a_prefix = options->b_prefix = "";
3332 else if (opt_arg(arg, '\0', "inter-hunk-context",
3333 &options->interhunkcontext))
3334 ;
3335 else if ((argcount = parse_long_opt("output", av, &optarg))) {
3336 options->file = fopen(optarg, "w");
3337 if (!options->file)
3338 die_errno("Could not open '%s'", optarg);
3339 options->close_file = 1;
3340 return argcount;
3341 } else
3342 return 0;
3343 return 1;
3344 }
3345
3346 int parse_rename_score(const char **cp_p)
3347 {
3348 unsigned long num, scale;
3349 int ch, dot;
3350 const char *cp = *cp_p;
3351
3352 num = 0;
3353 scale = 1;
3354 dot = 0;
3355 for (;;) {
3356 ch = *cp;
3357 if ( !dot && ch == '.' ) {
3358 scale = 1;
3359 dot = 1;
3360 } else if ( ch == '%' ) {
3361 scale = dot ? scale*100 : 100;
3362 cp++; /* % is always at the end */
3363 break;
3364 } else if ( ch >= '0' && ch <= '9' ) {
3365 if ( scale < 100000 ) {
3366 scale *= 10;
3367 num = (num*10) + (ch-'0');
3368 }
3369 } else {
3370 break;
3371 }
3372 cp++;
3373 }
3374 *cp_p = cp;
3375
3376 /* user says num divided by scale and we say internally that
3377 * is MAX_SCORE * num / scale.
3378 */
3379 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3380 }
3381
3382 static int diff_scoreopt_parse(const char *opt)
3383 {
3384 int opt1, opt2, cmd;
3385
3386 if (*opt++ != '-')
3387 return -1;
3388 cmd = *opt++;
3389 if (cmd == '-') {
3390 /* convert the long-form arguments into short-form versions */
3391 if (!prefixcmp(opt, "break-rewrites")) {
3392 opt += strlen("break-rewrites");
3393 if (*opt == 0 || *opt++ == '=')
3394 cmd = 'B';
3395 } else if (!prefixcmp(opt, "find-copies")) {
3396 opt += strlen("find-copies");
3397 if (*opt == 0 || *opt++ == '=')
3398 cmd = 'C';
3399 } else if (!prefixcmp(opt, "find-renames")) {
3400 opt += strlen("find-renames");
3401 if (*opt == 0 || *opt++ == '=')
3402 cmd = 'M';
3403 }
3404 }
3405 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3406 return -1; /* that is not a -M, -C nor -B option */
3407
3408 opt1 = parse_rename_score(&opt);
3409 if (cmd != 'B')
3410 opt2 = 0;
3411 else {
3412 if (*opt == 0)
3413 opt2 = 0;
3414 else if (*opt != '/')
3415 return -1; /* we expect -B80/99 or -B80 */
3416 else {
3417 opt++;
3418 opt2 = parse_rename_score(&opt);
3419 }
3420 }
3421 if (*opt != 0)
3422 return -1;
3423 return opt1 | (opt2 << 16);
3424 }
3425
3426 struct diff_queue_struct diff_queued_diff;
3427
3428 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3429 {
3430 if (queue->alloc <= queue->nr) {
3431 queue->alloc = alloc_nr(queue->alloc);
3432 queue->queue = xrealloc(queue->queue,
3433 sizeof(dp) * queue->alloc);
3434 }
3435 queue->queue[queue->nr++] = dp;
3436 }
3437
3438 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3439 struct diff_filespec *one,
3440 struct diff_filespec *two)
3441 {
3442 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3443 dp->one = one;
3444 dp->two = two;
3445 if (queue)
3446 diff_q(queue, dp);
3447 return dp;
3448 }
3449
3450 void diff_free_filepair(struct diff_filepair *p)
3451 {
3452 free_filespec(p->one);
3453 free_filespec(p->two);
3454 free(p);
3455 }
3456
3457 /* This is different from find_unique_abbrev() in that
3458 * it stuffs the result with dots for alignment.
3459 */
3460 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3461 {
3462 int abblen;
3463 const char *abbrev;
3464 if (len == 40)
3465 return sha1_to_hex(sha1);
3466
3467 abbrev = find_unique_abbrev(sha1, len);
3468 abblen = strlen(abbrev);
3469 if (abblen < 37) {
3470 static char hex[41];
3471 if (len < abblen && abblen <= len + 2)
3472 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3473 else
3474 sprintf(hex, "%s...", abbrev);
3475 return hex;
3476 }
3477 return sha1_to_hex(sha1);
3478 }
3479
3480 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3481 {
3482 int line_termination = opt->line_termination;
3483 int inter_name_termination = line_termination ? '\t' : '\0';
3484 if (opt->output_prefix) {
3485 struct strbuf *msg = NULL;
3486 msg = opt->output_prefix(opt, opt->output_prefix_data);
3487 fprintf(opt->file, "%s", msg->buf);
3488 }
3489
3490 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3491 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3492 diff_unique_abbrev(p->one->sha1, opt->abbrev));
3493 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3494 }
3495 if (p->score) {
3496 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3497 inter_name_termination);
3498 } else {
3499 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3500 }
3501
3502 if (p->status == DIFF_STATUS_COPIED ||
3503 p->status == DIFF_STATUS_RENAMED) {
3504 const char *name_a, *name_b;
3505 name_a = p->one->path;
3506 name_b = p->two->path;
3507 strip_prefix(opt->prefix_length, &name_a, &name_b);
3508 write_name_quoted(name_a, opt->file, inter_name_termination);
3509 write_name_quoted(name_b, opt->file, line_termination);
3510 } else {
3511 const char *name_a, *name_b;
3512 name_a = p->one->mode ? p->one->path : p->two->path;
3513 name_b = NULL;
3514 strip_prefix(opt->prefix_length, &name_a, &name_b);
3515 write_name_quoted(name_a, opt->file, line_termination);
3516 }
3517 }
3518
3519 int diff_unmodified_pair(struct diff_filepair *p)
3520 {
3521 /* This function is written stricter than necessary to support
3522 * the currently implemented transformers, but the idea is to
3523 * let transformers to produce diff_filepairs any way they want,
3524 * and filter and clean them up here before producing the output.
3525 */
3526 struct diff_filespec *one = p->one, *two = p->two;
3527
3528 if (DIFF_PAIR_UNMERGED(p))
3529 return 0; /* unmerged is interesting */
3530
3531 /* deletion, addition, mode or type change
3532 * and rename are all interesting.
3533 */
3534 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3535 DIFF_PAIR_MODE_CHANGED(p) ||
3536 strcmp(one->path, two->path))
3537 return 0;
3538
3539 /* both are valid and point at the same path. that is, we are
3540 * dealing with a change.
3541 */
3542 if (one->sha1_valid && two->sha1_valid &&
3543 !hashcmp(one->sha1, two->sha1) &&
3544 !one->dirty_submodule && !two->dirty_submodule)
3545 return 1; /* no change */
3546 if (!one->sha1_valid && !two->sha1_valid)
3547 return 1; /* both look at the same file on the filesystem. */
3548 return 0;
3549 }
3550
3551 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3552 {
3553 if (diff_unmodified_pair(p))
3554 return;
3555
3556 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3557 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3558 return; /* no tree diffs in patch format */
3559
3560 run_diff(p, o);
3561 }
3562
3563 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3564 struct diffstat_t *diffstat)
3565 {
3566 if (diff_unmodified_pair(p))
3567 return;
3568
3569 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3570 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3571 return; /* no useful stat for tree diffs */
3572
3573 run_diffstat(p, o, diffstat);
3574 }
3575
3576 static void diff_flush_checkdiff(struct diff_filepair *p,
3577 struct diff_options *o)
3578 {
3579 if (diff_unmodified_pair(p))
3580 return;
3581
3582 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3583 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3584 return; /* nothing to check in tree diffs */
3585
3586 run_checkdiff(p, o);
3587 }
3588
3589 int diff_queue_is_empty(void)
3590 {
3591 struct diff_queue_struct *q = &diff_queued_diff;
3592 int i;
3593 for (i = 0; i < q->nr; i++)
3594 if (!diff_unmodified_pair(q->queue[i]))
3595 return 0;
3596 return 1;
3597 }
3598
3599 #if DIFF_DEBUG
3600 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3601 {
3602 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3603 x, one ? one : "",
3604 s->path,
3605 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3606 s->mode,
3607 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3608 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3609 x, one ? one : "",
3610 s->size, s->xfrm_flags);
3611 }
3612
3613 void diff_debug_filepair(const struct diff_filepair *p, int i)
3614 {
3615 diff_debug_filespec(p->one, i, "one");
3616 diff_debug_filespec(p->two, i, "two");
3617 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3618 p->score, p->status ? p->status : '?',
3619 p->one->rename_used, p->broken_pair);
3620 }
3621
3622 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3623 {
3624 int i;
3625 if (msg)
3626 fprintf(stderr, "%s\n", msg);
3627 fprintf(stderr, "q->nr = %d\n", q->nr);
3628 for (i = 0; i < q->nr; i++) {
3629 struct diff_filepair *p = q->queue[i];
3630 diff_debug_filepair(p, i);
3631 }
3632 }
3633 #endif
3634
3635 static void diff_resolve_rename_copy(void)
3636 {
3637 int i;
3638 struct diff_filepair *p;
3639 struct diff_queue_struct *q = &diff_queued_diff;
3640
3641 diff_debug_queue("resolve-rename-copy", q);
3642
3643 for (i = 0; i < q->nr; i++) {
3644 p = q->queue[i];
3645 p->status = 0; /* undecided */
3646 if (DIFF_PAIR_UNMERGED(p))
3647 p->status = DIFF_STATUS_UNMERGED;
3648 else if (!DIFF_FILE_VALID(p->one))
3649 p->status = DIFF_STATUS_ADDED;
3650 else if (!DIFF_FILE_VALID(p->two))
3651 p->status = DIFF_STATUS_DELETED;
3652 else if (DIFF_PAIR_TYPE_CHANGED(p))
3653 p->status = DIFF_STATUS_TYPE_CHANGED;
3654
3655 /* from this point on, we are dealing with a pair
3656 * whose both sides are valid and of the same type, i.e.
3657 * either in-place edit or rename/copy edit.
3658 */
3659 else if (DIFF_PAIR_RENAME(p)) {
3660 /*
3661 * A rename might have re-connected a broken
3662 * pair up, causing the pathnames to be the
3663 * same again. If so, that's not a rename at
3664 * all, just a modification..
3665 *
3666 * Otherwise, see if this source was used for
3667 * multiple renames, in which case we decrement
3668 * the count, and call it a copy.
3669 */
3670 if (!strcmp(p->one->path, p->two->path))
3671 p->status = DIFF_STATUS_MODIFIED;
3672 else if (--p->one->rename_used > 0)
3673 p->status = DIFF_STATUS_COPIED;
3674 else
3675 p->status = DIFF_STATUS_RENAMED;
3676 }
3677 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3678 p->one->mode != p->two->mode ||
3679 p->one->dirty_submodule ||
3680 p->two->dirty_submodule ||
3681 is_null_sha1(p->one->sha1))
3682 p->status = DIFF_STATUS_MODIFIED;
3683 else {
3684 /* This is a "no-change" entry and should not
3685 * happen anymore, but prepare for broken callers.
3686 */
3687 error("feeding unmodified %s to diffcore",
3688 p->one->path);
3689 p->status = DIFF_STATUS_UNKNOWN;
3690 }
3691 }
3692 diff_debug_queue("resolve-rename-copy done", q);
3693 }
3694
3695 static int check_pair_status(struct diff_filepair *p)
3696 {
3697 switch (p->status) {
3698 case DIFF_STATUS_UNKNOWN:
3699 return 0;
3700 case 0:
3701 die("internal error in diff-resolve-rename-copy");
3702 default:
3703 return 1;
3704 }
3705 }
3706
3707 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3708 {
3709 int fmt = opt->output_format;
3710
3711 if (fmt & DIFF_FORMAT_CHECKDIFF)
3712 diff_flush_checkdiff(p, opt);
3713 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3714 diff_flush_raw(p, opt);
3715 else if (fmt & DIFF_FORMAT_NAME) {
3716 const char *name_a, *name_b;
3717 name_a = p->two->path;
3718 name_b = NULL;
3719 strip_prefix(opt->prefix_length, &name_a, &name_b);
3720 write_name_quoted(name_a, opt->file, opt->line_termination);
3721 }
3722 }
3723
3724 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3725 {
3726 if (fs->mode)
3727 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3728 else
3729 fprintf(file, " %s ", newdelete);
3730 write_name_quoted(fs->path, file, '\n');
3731 }
3732
3733
3734 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
3735 const char *line_prefix)
3736 {
3737 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3738 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
3739 p->two->mode, show_name ? ' ' : '\n');
3740 if (show_name) {
3741 write_name_quoted(p->two->path, file, '\n');
3742 }
3743 }
3744 }
3745
3746 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
3747 const char *line_prefix)
3748 {
3749 char *names = pprint_rename(p->one->path, p->two->path);
3750
3751 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3752 free(names);
3753 show_mode_change(file, p, 0, line_prefix);
3754 }
3755
3756 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
3757 {
3758 FILE *file = opt->file;
3759 char *line_prefix = "";
3760
3761 if (opt->output_prefix) {
3762 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
3763 line_prefix = buf->buf;
3764 }
3765
3766 switch(p->status) {
3767 case DIFF_STATUS_DELETED:
3768 fputs(line_prefix, file);
3769 show_file_mode_name(file, "delete", p->one);
3770 break;
3771 case DIFF_STATUS_ADDED:
3772 fputs(line_prefix, file);
3773 show_file_mode_name(file, "create", p->two);
3774 break;
3775 case DIFF_STATUS_COPIED:
3776 fputs(line_prefix, file);
3777 show_rename_copy(file, "copy", p, line_prefix);
3778 break;
3779 case DIFF_STATUS_RENAMED:
3780 fputs(line_prefix, file);
3781 show_rename_copy(file, "rename", p, line_prefix);
3782 break;
3783 default:
3784 if (p->score) {
3785 fprintf(file, "%s rewrite ", line_prefix);
3786 write_name_quoted(p->two->path, file, ' ');
3787 fprintf(file, "(%d%%)\n", similarity_index(p));
3788 }
3789 show_mode_change(file, p, !p->score, line_prefix);
3790 break;
3791 }
3792 }
3793
3794 struct patch_id_t {
3795 git_SHA_CTX *ctx;
3796 int patchlen;
3797 };
3798
3799 static int remove_space(char *line, int len)
3800 {
3801 int i;
3802 char *dst = line;
3803 unsigned char c;
3804
3805 for (i = 0; i < len; i++)
3806 if (!isspace((c = line[i])))
3807 *dst++ = c;
3808
3809 return dst - line;
3810 }
3811
3812 static void patch_id_consume(void *priv, char *line, unsigned long len)
3813 {
3814 struct patch_id_t *data = priv;
3815 int new_len;
3816
3817 /* Ignore line numbers when computing the SHA1 of the patch */
3818 if (!prefixcmp(line, "@@ -"))
3819 return;
3820
3821 new_len = remove_space(line, len);
3822
3823 git_SHA1_Update(data->ctx, line, new_len);
3824 data->patchlen += new_len;
3825 }
3826
3827 /* returns 0 upon success, and writes result into sha1 */
3828 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3829 {
3830 struct diff_queue_struct *q = &diff_queued_diff;
3831 int i;
3832 git_SHA_CTX ctx;
3833 struct patch_id_t data;
3834 char buffer[PATH_MAX * 4 + 20];
3835
3836 git_SHA1_Init(&ctx);
3837 memset(&data, 0, sizeof(struct patch_id_t));
3838 data.ctx = &ctx;
3839
3840 for (i = 0; i < q->nr; i++) {
3841 xpparam_t xpp;
3842 xdemitconf_t xecfg;
3843 mmfile_t mf1, mf2;
3844 struct diff_filepair *p = q->queue[i];
3845 int len1, len2;
3846
3847 memset(&xpp, 0, sizeof(xpp));
3848 memset(&xecfg, 0, sizeof(xecfg));
3849 if (p->status == 0)
3850 return error("internal diff status error");
3851 if (p->status == DIFF_STATUS_UNKNOWN)
3852 continue;
3853 if (diff_unmodified_pair(p))
3854 continue;
3855 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3856 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3857 continue;
3858 if (DIFF_PAIR_UNMERGED(p))
3859 continue;
3860
3861 diff_fill_sha1_info(p->one);
3862 diff_fill_sha1_info(p->two);
3863 if (fill_mmfile(&mf1, p->one) < 0 ||
3864 fill_mmfile(&mf2, p->two) < 0)
3865 return error("unable to read files to diff");
3866
3867 len1 = remove_space(p->one->path, strlen(p->one->path));
3868 len2 = remove_space(p->two->path, strlen(p->two->path));
3869 if (p->one->mode == 0)
3870 len1 = snprintf(buffer, sizeof(buffer),
3871 "diff--gita/%.*sb/%.*s"
3872 "newfilemode%06o"
3873 "---/dev/null"
3874 "+++b/%.*s",
3875 len1, p->one->path,
3876 len2, p->two->path,
3877 p->two->mode,
3878 len2, p->two->path);
3879 else if (p->two->mode == 0)
3880 len1 = snprintf(buffer, sizeof(buffer),
3881 "diff--gita/%.*sb/%.*s"
3882 "deletedfilemode%06o"
3883 "---a/%.*s"
3884 "+++/dev/null",
3885 len1, p->one->path,
3886 len2, p->two->path,
3887 p->one->mode,
3888 len1, p->one->path);
3889 else
3890 len1 = snprintf(buffer, sizeof(buffer),
3891 "diff--gita/%.*sb/%.*s"
3892 "---a/%.*s"
3893 "+++b/%.*s",
3894 len1, p->one->path,
3895 len2, p->two->path,
3896 len1, p->one->path,
3897 len2, p->two->path);
3898 git_SHA1_Update(&ctx, buffer, len1);
3899
3900 if (diff_filespec_is_binary(p->one) ||
3901 diff_filespec_is_binary(p->two)) {
3902 git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
3903 git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
3904 continue;
3905 }
3906
3907 xpp.flags = 0;
3908 xecfg.ctxlen = 3;
3909 xecfg.flags = 0;
3910 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3911 &xpp, &xecfg);
3912 }
3913
3914 git_SHA1_Final(sha1, &ctx);
3915 return 0;
3916 }
3917
3918 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3919 {
3920 struct diff_queue_struct *q = &diff_queued_diff;
3921 int i;
3922 int result = diff_get_patch_id(options, sha1);
3923
3924 for (i = 0; i < q->nr; i++)
3925 diff_free_filepair(q->queue[i]);
3926
3927 free(q->queue);
3928 DIFF_QUEUE_CLEAR(q);
3929
3930 return result;
3931 }
3932
3933 static int is_summary_empty(const struct diff_queue_struct *q)
3934 {
3935 int i;
3936
3937 for (i = 0; i < q->nr; i++) {
3938 const struct diff_filepair *p = q->queue[i];
3939
3940 switch (p->status) {
3941 case DIFF_STATUS_DELETED:
3942 case DIFF_STATUS_ADDED:
3943 case DIFF_STATUS_COPIED:
3944 case DIFF_STATUS_RENAMED:
3945 return 0;
3946 default:
3947 if (p->score)
3948 return 0;
3949 if (p->one->mode && p->two->mode &&
3950 p->one->mode != p->two->mode)
3951 return 0;
3952 break;
3953 }
3954 }
3955 return 1;
3956 }
3957
3958 void diff_flush(struct diff_options *options)
3959 {
3960 struct diff_queue_struct *q = &diff_queued_diff;
3961 int i, output_format = options->output_format;
3962 int separator = 0;
3963
3964 /*
3965 * Order: raw, stat, summary, patch
3966 * or: name/name-status/checkdiff (other bits clear)
3967 */
3968 if (!q->nr)
3969 goto free_queue;
3970
3971 if (output_format & (DIFF_FORMAT_RAW |
3972 DIFF_FORMAT_NAME |
3973 DIFF_FORMAT_NAME_STATUS |
3974 DIFF_FORMAT_CHECKDIFF)) {
3975 for (i = 0; i < q->nr; i++) {
3976 struct diff_filepair *p = q->queue[i];
3977 if (check_pair_status(p))
3978 flush_one_pair(p, options);
3979 }
3980 separator++;
3981 }
3982
3983 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3984 struct diffstat_t diffstat;
3985
3986 memset(&diffstat, 0, sizeof(struct diffstat_t));
3987 for (i = 0; i < q->nr; i++) {
3988 struct diff_filepair *p = q->queue[i];
3989 if (check_pair_status(p))
3990 diff_flush_stat(p, options, &diffstat);
3991 }
3992 if (output_format & DIFF_FORMAT_NUMSTAT)
3993 show_numstat(&diffstat, options);
3994 if (output_format & DIFF_FORMAT_DIFFSTAT)
3995 show_stats(&diffstat, options);
3996 if (output_format & DIFF_FORMAT_SHORTSTAT)
3997 show_shortstats(&diffstat, options);
3998 free_diffstat_info(&diffstat);
3999 separator++;
4000 }
4001 if (output_format & DIFF_FORMAT_DIRSTAT)
4002 show_dirstat(options);
4003
4004 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4005 for (i = 0; i < q->nr; i++) {
4006 diff_summary(options, q->queue[i]);
4007 }
4008 separator++;
4009 }
4010
4011 if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4012 DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4013 DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4014 /*
4015 * run diff_flush_patch for the exit status. setting
4016 * options->file to /dev/null should be safe, becaue we
4017 * aren't supposed to produce any output anyway.
4018 */
4019 if (options->close_file)
4020 fclose(options->file);
4021 options->file = fopen("/dev/null", "w");
4022 if (!options->file)
4023 die_errno("Could not open /dev/null");
4024 options->close_file = 1;
4025 for (i = 0; i < q->nr; i++) {
4026 struct diff_filepair *p = q->queue[i];
4027 if (check_pair_status(p))
4028 diff_flush_patch(p, options);
4029 if (options->found_changes)
4030 break;
4031 }
4032 }
4033
4034 if (output_format & DIFF_FORMAT_PATCH) {
4035 if (separator) {
4036 putc(options->line_termination, options->file);
4037 if (options->stat_sep) {
4038 /* attach patch instead of inline */
4039 fputs(options->stat_sep, options->file);
4040 }
4041 }
4042
4043 for (i = 0; i < q->nr; i++) {
4044 struct diff_filepair *p = q->queue[i];
4045 if (check_pair_status(p))
4046 diff_flush_patch(p, options);
4047 }
4048 }
4049
4050 if (output_format & DIFF_FORMAT_CALLBACK)
4051 options->format_callback(q, options, options->format_callback_data);
4052
4053 for (i = 0; i < q->nr; i++)
4054 diff_free_filepair(q->queue[i]);
4055 free_queue:
4056 free(q->queue);
4057 DIFF_QUEUE_CLEAR(q);
4058 if (options->close_file)
4059 fclose(options->file);
4060
4061 /*
4062 * Report the content-level differences with HAS_CHANGES;
4063 * diff_addremove/diff_change does not set the bit when
4064 * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4065 */
4066 if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4067 if (options->found_changes)
4068 DIFF_OPT_SET(options, HAS_CHANGES);
4069 else
4070 DIFF_OPT_CLR(options, HAS_CHANGES);
4071 }
4072 }
4073
4074 static void diffcore_apply_filter(const char *filter)
4075 {
4076 int i;
4077 struct diff_queue_struct *q = &diff_queued_diff;
4078 struct diff_queue_struct outq;
4079 DIFF_QUEUE_CLEAR(&outq);
4080
4081 if (!filter)
4082 return;
4083
4084 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4085 int found;
4086 for (i = found = 0; !found && i < q->nr; i++) {
4087 struct diff_filepair *p = q->queue[i];
4088 if (((p->status == DIFF_STATUS_MODIFIED) &&
4089 ((p->score &&
4090 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4091 (!p->score &&
4092 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4093 ((p->status != DIFF_STATUS_MODIFIED) &&
4094 strchr(filter, p->status)))
4095 found++;
4096 }
4097 if (found)
4098 return;
4099
4100 /* otherwise we will clear the whole queue
4101 * by copying the empty outq at the end of this
4102 * function, but first clear the current entries
4103 * in the queue.
4104 */
4105 for (i = 0; i < q->nr; i++)
4106 diff_free_filepair(q->queue[i]);
4107 }
4108 else {
4109 /* Only the matching ones */
4110 for (i = 0; i < q->nr; i++) {
4111 struct diff_filepair *p = q->queue[i];
4112
4113 if (((p->status == DIFF_STATUS_MODIFIED) &&
4114 ((p->score &&
4115 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4116 (!p->score &&
4117 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4118 ((p->status != DIFF_STATUS_MODIFIED) &&
4119 strchr(filter, p->status)))
4120 diff_q(&outq, p);
4121 else
4122 diff_free_filepair(p);
4123 }
4124 }
4125 free(q->queue);
4126 *q = outq;
4127 }
4128
4129 /* Check whether two filespecs with the same mode and size are identical */
4130 static int diff_filespec_is_identical(struct diff_filespec *one,
4131 struct diff_filespec *two)
4132 {
4133 if (S_ISGITLINK(one->mode))
4134 return 0;
4135 if (diff_populate_filespec(one, 0))
4136 return 0;
4137 if (diff_populate_filespec(two, 0))
4138 return 0;
4139 return !memcmp(one->data, two->data, one->size);
4140 }
4141
4142 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4143 {
4144 int i;
4145 struct diff_queue_struct *q = &diff_queued_diff;
4146 struct diff_queue_struct outq;
4147 DIFF_QUEUE_CLEAR(&outq);
4148
4149 for (i = 0; i < q->nr; i++) {
4150 struct diff_filepair *p = q->queue[i];
4151
4152 /*
4153 * 1. Entries that come from stat info dirtiness
4154 * always have both sides (iow, not create/delete),
4155 * one side of the object name is unknown, with
4156 * the same mode and size. Keep the ones that
4157 * do not match these criteria. They have real
4158 * differences.
4159 *
4160 * 2. At this point, the file is known to be modified,
4161 * with the same mode and size, and the object
4162 * name of one side is unknown. Need to inspect
4163 * the identical contents.
4164 */
4165 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4166 !DIFF_FILE_VALID(p->two) ||
4167 (p->one->sha1_valid && p->two->sha1_valid) ||
4168 (p->one->mode != p->two->mode) ||
4169 diff_populate_filespec(p->one, 1) ||
4170 diff_populate_filespec(p->two, 1) ||
4171 (p->one->size != p->two->size) ||
4172 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4173 diff_q(&outq, p);
4174 else {
4175 /*
4176 * The caller can subtract 1 from skip_stat_unmatch
4177 * to determine how many paths were dirty only
4178 * due to stat info mismatch.
4179 */
4180 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4181 diffopt->skip_stat_unmatch++;
4182 diff_free_filepair(p);
4183 }
4184 }
4185 free(q->queue);
4186 *q = outq;
4187 }
4188
4189 static int diffnamecmp(const void *a_, const void *b_)
4190 {
4191 const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4192 const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4193 const char *name_a, *name_b;
4194
4195 name_a = a->one ? a->one->path : a->two->path;
4196 name_b = b->one ? b->one->path : b->two->path;
4197 return strcmp(name_a, name_b);
4198 }
4199
4200 void diffcore_fix_diff_index(struct diff_options *options)
4201 {
4202 struct diff_queue_struct *q = &diff_queued_diff;
4203 qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4204 }
4205
4206 void diffcore_std(struct diff_options *options)
4207 {
4208 if (options->skip_stat_unmatch)
4209 diffcore_skip_stat_unmatch(options);
4210 if (!options->found_follow) {
4211 /* See try_to_follow_renames() in tree-diff.c */
4212 if (options->break_opt != -1)
4213 diffcore_break(options->break_opt);
4214 if (options->detect_rename)
4215 diffcore_rename(options);
4216 if (options->break_opt != -1)
4217 diffcore_merge_broken();
4218 }
4219 if (options->pickaxe)
4220 diffcore_pickaxe(options);
4221 if (options->orderfile)
4222 diffcore_order(options->orderfile);
4223 if (!options->found_follow)
4224 /* See try_to_follow_renames() in tree-diff.c */
4225 diff_resolve_rename_copy();
4226 diffcore_apply_filter(options->filter);
4227
4228 if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4229 DIFF_OPT_SET(options, HAS_CHANGES);
4230 else
4231 DIFF_OPT_CLR(options, HAS_CHANGES);
4232
4233 options->found_follow = 0;
4234 }
4235
4236 int diff_result_code(struct diff_options *opt, int status)
4237 {
4238 int result = 0;
4239 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4240 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4241 return status;
4242 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4243 DIFF_OPT_TST(opt, HAS_CHANGES))
4244 result |= 01;
4245 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4246 DIFF_OPT_TST(opt, CHECK_FAILED))
4247 result |= 02;
4248 return result;
4249 }
4250
4251 /*
4252 * Shall changes to this submodule be ignored?
4253 *
4254 * Submodule changes can be configured to be ignored separately for each path,
4255 * but that configuration can be overridden from the command line.
4256 */
4257 static int is_submodule_ignored(const char *path, struct diff_options *options)
4258 {
4259 int ignored = 0;
4260 unsigned orig_flags = options->flags;
4261 if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4262 set_diffopt_flags_from_submodule_config(options, path);
4263 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4264 ignored = 1;
4265 options->flags = orig_flags;
4266 return ignored;
4267 }
4268
4269 void diff_addremove(struct diff_options *options,
4270 int addremove, unsigned mode,
4271 const unsigned char *sha1,
4272 const char *concatpath, unsigned dirty_submodule)
4273 {
4274 struct diff_filespec *one, *two;
4275
4276 if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4277 return;
4278
4279 /* This may look odd, but it is a preparation for
4280 * feeding "there are unchanged files which should
4281 * not produce diffs, but when you are doing copy
4282 * detection you would need them, so here they are"
4283 * entries to the diff-core. They will be prefixed
4284 * with something like '=' or '*' (I haven't decided
4285 * which but should not make any difference).
4286 * Feeding the same new and old to diff_change()
4287 * also has the same effect.
4288 * Before the final output happens, they are pruned after
4289 * merged into rename/copy pairs as appropriate.
4290 */
4291 if (DIFF_OPT_TST(options, REVERSE_DIFF))
4292 addremove = (addremove == '+' ? '-' :
4293 addremove == '-' ? '+' : addremove);
4294
4295 if (options->prefix &&
4296 strncmp(concatpath, options->prefix, options->prefix_length))
4297 return;
4298
4299 one = alloc_filespec(concatpath);
4300 two = alloc_filespec(concatpath);
4301
4302 if (addremove != '+')
4303 fill_filespec(one, sha1, mode);
4304 if (addremove != '-') {
4305 fill_filespec(two, sha1, mode);
4306 two->dirty_submodule = dirty_submodule;
4307 }
4308
4309 diff_queue(&diff_queued_diff, one, two);
4310 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4311 DIFF_OPT_SET(options, HAS_CHANGES);
4312 }
4313
4314 void diff_change(struct diff_options *options,
4315 unsigned old_mode, unsigned new_mode,
4316 const unsigned char *old_sha1,
4317 const unsigned char *new_sha1,
4318 const char *concatpath,
4319 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4320 {
4321 struct diff_filespec *one, *two;
4322
4323 if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4324 is_submodule_ignored(concatpath, options))
4325 return;
4326
4327 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4328 unsigned tmp;
4329 const unsigned char *tmp_c;
4330 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4331 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4332 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4333 new_dirty_submodule = tmp;
4334 }
4335
4336 if (options->prefix &&
4337 strncmp(concatpath, options->prefix, options->prefix_length))
4338 return;
4339
4340 one = alloc_filespec(concatpath);
4341 two = alloc_filespec(concatpath);
4342 fill_filespec(one, old_sha1, old_mode);
4343 fill_filespec(two, new_sha1, new_mode);
4344 one->dirty_submodule = old_dirty_submodule;
4345 two->dirty_submodule = new_dirty_submodule;
4346
4347 diff_queue(&diff_queued_diff, one, two);
4348 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4349 DIFF_OPT_SET(options, HAS_CHANGES);
4350 }
4351
4352 void diff_unmerge(struct diff_options *options,
4353 const char *path,
4354 unsigned mode, const unsigned char *sha1)
4355 {
4356 struct diff_filespec *one, *two;
4357
4358 if (options->prefix &&
4359 strncmp(path, options->prefix, options->prefix_length))
4360 return;
4361
4362 one = alloc_filespec(path);
4363 two = alloc_filespec(path);
4364 fill_filespec(one, sha1, mode);
4365 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
4366 }
4367
4368 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4369 size_t *outsize)
4370 {
4371 struct diff_tempfile *temp;
4372 const char *argv[3];
4373 const char **arg = argv;
4374 struct child_process child;
4375 struct strbuf buf = STRBUF_INIT;
4376 int err = 0;
4377
4378 temp = prepare_temp_file(spec->path, spec);
4379 *arg++ = pgm;
4380 *arg++ = temp->name;
4381 *arg = NULL;
4382
4383 memset(&child, 0, sizeof(child));
4384 child.use_shell = 1;
4385 child.argv = argv;
4386 child.out = -1;
4387 if (start_command(&child)) {
4388 remove_tempfile();
4389 return NULL;
4390 }
4391
4392 if (strbuf_read(&buf, child.out, 0) < 0)
4393 err = error("error reading from textconv command '%s'", pgm);
4394 close(child.out);
4395
4396 if (finish_command(&child) || err) {
4397 strbuf_release(&buf);
4398 remove_tempfile();
4399 return NULL;
4400 }
4401 remove_tempfile();
4402
4403 return strbuf_detach(&buf, outsize);
4404 }
4405
4406 size_t fill_textconv(struct userdiff_driver *driver,
4407 struct diff_filespec *df,
4408 char **outbuf)
4409 {
4410 size_t size;
4411
4412 if (!driver || !driver->textconv) {
4413 if (!DIFF_FILE_VALID(df)) {
4414 *outbuf = "";
4415 return 0;
4416 }
4417 if (diff_populate_filespec(df, 0))
4418 die("unable to read files to diff");
4419 *outbuf = df->data;
4420 return df->size;
4421 }
4422
4423 if (driver->textconv_cache && df->sha1_valid) {
4424 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4425 &size);
4426 if (*outbuf)
4427 return size;
4428 }
4429
4430 *outbuf = run_textconv(driver->textconv, df, &size);
4431 if (!*outbuf)
4432 die("unable to read files to diff");
4433
4434 if (driver->textconv_cache && df->sha1_valid) {
4435 /* ignore errors, as we might be in a readonly repository */
4436 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4437 size);
4438 /*
4439 * we could save up changes and flush them all at the end,
4440 * but we would need an extra call after all diffing is done.
4441 * Since generating a cache entry is the slow path anyway,
4442 * this extra overhead probably isn't a big deal.
4443 */
4444 notes_cache_write(driver->textconv_cache);
4445 }
4446
4447 return size;
4448 }
git with Jakub Narębski modifications