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