diff: do not use null sha1 as a sentinel value
[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 } else if (!prefixcmp(line, "\\ ")) {
1117 /*
1118 * Eat the "no newline at eof" marker as if we
1119 * saw a "+" or "-" line with nothing on it,
1120 * and return without diff_words_flush() to
1121 * defer processing. If this is the end of
1122 * preimage, more "+" lines may come after it.
1123 */
1124 return;
1125 }
1126 diff_words_flush(ecbdata);
1127 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1128 emit_line(ecbdata->opt, plain, reset, line, len);
1129 fputs("~\n", ecbdata->opt->file);
1130 } else {
1131 /*
1132 * Skip the prefix character, if any. With
1133 * diff_suppress_blank_empty, there may be
1134 * none.
1135 */
1136 if (line[0] != '\n') {
1137 line++;
1138 len--;
1139 }
1140 emit_line(ecbdata->opt, plain, reset, line, len);
1141 }
1142 return;
1143 }
1144
1145 if (line[0] != '+') {
1146 const char *color =
1147 diff_get_color(ecbdata->color_diff,
1148 line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1149 ecbdata->lno_in_preimage++;
1150 if (line[0] == ' ')
1151 ecbdata->lno_in_postimage++;
1152 emit_line(ecbdata->opt, color, reset, line, len);
1153 } else {
1154 ecbdata->lno_in_postimage++;
1155 emit_add_line(reset, ecbdata, line + 1, len - 1);
1156 }
1157 }
1158
1159 static char *pprint_rename(const char *a, const char *b)
1160 {
1161 const char *old = a;
1162 const char *new = b;
1163 struct strbuf name = STRBUF_INIT;
1164 int pfx_length, sfx_length;
1165 int len_a = strlen(a);
1166 int len_b = strlen(b);
1167 int a_midlen, b_midlen;
1168 int qlen_a = quote_c_style(a, NULL, NULL, 0);
1169 int qlen_b = quote_c_style(b, NULL, NULL, 0);
1170
1171 if (qlen_a || qlen_b) {
1172 quote_c_style(a, &name, NULL, 0);
1173 strbuf_addstr(&name, " => ");
1174 quote_c_style(b, &name, NULL, 0);
1175 return strbuf_detach(&name, NULL);
1176 }
1177
1178 /* Find common prefix */
1179 pfx_length = 0;
1180 while (*old && *new && *old == *new) {
1181 if (*old == '/')
1182 pfx_length = old - a + 1;
1183 old++;
1184 new++;
1185 }
1186
1187 /* Find common suffix */
1188 old = a + len_a;
1189 new = b + len_b;
1190 sfx_length = 0;
1191 while (a <= old && b <= new && *old == *new) {
1192 if (*old == '/')
1193 sfx_length = len_a - (old - a);
1194 old--;
1195 new--;
1196 }
1197
1198 /*
1199 * pfx{mid-a => mid-b}sfx
1200 * {pfx-a => pfx-b}sfx
1201 * pfx{sfx-a => sfx-b}
1202 * name-a => name-b
1203 */
1204 a_midlen = len_a - pfx_length - sfx_length;
1205 b_midlen = len_b - pfx_length - sfx_length;
1206 if (a_midlen < 0)
1207 a_midlen = 0;
1208 if (b_midlen < 0)
1209 b_midlen = 0;
1210
1211 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1212 if (pfx_length + sfx_length) {
1213 strbuf_add(&name, a, pfx_length);
1214 strbuf_addch(&name, '{');
1215 }
1216 strbuf_add(&name, a + pfx_length, a_midlen);
1217 strbuf_addstr(&name, " => ");
1218 strbuf_add(&name, b + pfx_length, b_midlen);
1219 if (pfx_length + sfx_length) {
1220 strbuf_addch(&name, '}');
1221 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1222 }
1223 return strbuf_detach(&name, NULL);
1224 }
1225
1226 struct diffstat_t {
1227 int nr;
1228 int alloc;
1229 struct diffstat_file {
1230 char *from_name;
1231 char *name;
1232 char *print_name;
1233 unsigned is_unmerged:1;
1234 unsigned is_binary:1;
1235 unsigned is_renamed:1;
1236 uintmax_t added, deleted;
1237 } **files;
1238 };
1239
1240 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1241 const char *name_a,
1242 const char *name_b)
1243 {
1244 struct diffstat_file *x;
1245 x = xcalloc(sizeof (*x), 1);
1246 if (diffstat->nr == diffstat->alloc) {
1247 diffstat->alloc = alloc_nr(diffstat->alloc);
1248 diffstat->files = xrealloc(diffstat->files,
1249 diffstat->alloc * sizeof(x));
1250 }
1251 diffstat->files[diffstat->nr++] = x;
1252 if (name_b) {
1253 x->from_name = xstrdup(name_a);
1254 x->name = xstrdup(name_b);
1255 x->is_renamed = 1;
1256 }
1257 else {
1258 x->from_name = NULL;
1259 x->name = xstrdup(name_a);
1260 }
1261 return x;
1262 }
1263
1264 static void diffstat_consume(void *priv, char *line, unsigned long len)
1265 {
1266 struct diffstat_t *diffstat = priv;
1267 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1268
1269 if (line[0] == '+')
1270 x->added++;
1271 else if (line[0] == '-')
1272 x->deleted++;
1273 }
1274
1275 const char mime_boundary_leader[] = "------------";
1276
1277 static int scale_linear(int it, int width, int max_change)
1278 {
1279 if (!it)
1280 return 0;
1281 /*
1282 * make sure that at least one '-' or '+' is printed if
1283 * there is any change to this path. The easiest way is to
1284 * scale linearly as if the alloted width is one column shorter
1285 * than it is, and then add 1 to the result.
1286 */
1287 return 1 + (it * (width - 1) / max_change);
1288 }
1289
1290 static void show_name(FILE *file,
1291 const char *prefix, const char *name, int len)
1292 {
1293 fprintf(file, " %s%-*s |", prefix, len, name);
1294 }
1295
1296 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1297 {
1298 if (cnt <= 0)
1299 return;
1300 fprintf(file, "%s", set);
1301 while (cnt--)
1302 putc(ch, file);
1303 fprintf(file, "%s", reset);
1304 }
1305
1306 static void fill_print_name(struct diffstat_file *file)
1307 {
1308 char *pname;
1309
1310 if (file->print_name)
1311 return;
1312
1313 if (!file->is_renamed) {
1314 struct strbuf buf = STRBUF_INIT;
1315 if (quote_c_style(file->name, &buf, NULL, 0)) {
1316 pname = strbuf_detach(&buf, NULL);
1317 } else {
1318 pname = file->name;
1319 strbuf_release(&buf);
1320 }
1321 } else {
1322 pname = pprint_rename(file->from_name, file->name);
1323 }
1324 file->print_name = pname;
1325 }
1326
1327 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1328 {
1329 struct strbuf sb = STRBUF_INIT;
1330 int ret;
1331
1332 if (!files) {
1333 assert(insertions == 0 && deletions == 0);
1334 return fputs(_(" 0 files changed\n"), fp);
1335 }
1336
1337 strbuf_addf(&sb,
1338 Q_(" %d file changed", " %d files changed", files),
1339 files);
1340
1341 /*
1342 * For binary diff, the caller may want to print "x files
1343 * changed" with insertions == 0 && deletions == 0.
1344 *
1345 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1346 * is probably less confusing (i.e skip over "2 files changed
1347 * but nothing about added/removed lines? Is this a bug in Git?").
1348 */
1349 if (insertions || deletions == 0) {
1350 /*
1351 * TRANSLATORS: "+" in (+) is a line addition marker;
1352 * do not translate it.
1353 */
1354 strbuf_addf(&sb,
1355 Q_(", %d insertion(+)", ", %d insertions(+)",
1356 insertions),
1357 insertions);
1358 }
1359
1360 if (deletions || insertions == 0) {
1361 /*
1362 * TRANSLATORS: "-" in (-) is a line removal marker;
1363 * do not translate it.
1364 */
1365 strbuf_addf(&sb,
1366 Q_(", %d deletion(-)", ", %d deletions(-)",
1367 deletions),
1368 deletions);
1369 }
1370 strbuf_addch(&sb, '\n');
1371 ret = fputs(sb.buf, fp);
1372 strbuf_release(&sb);
1373 return ret;
1374 }
1375
1376 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1377 {
1378 int i, len, add, del, adds = 0, dels = 0;
1379 uintmax_t max_change = 0, max_len = 0;
1380 int total_files = data->nr;
1381 int width, name_width, count;
1382 const char *reset, *add_c, *del_c;
1383 const char *line_prefix = "";
1384 int extra_shown = 0;
1385 struct strbuf *msg = NULL;
1386
1387 if (data->nr == 0)
1388 return;
1389
1390 if (options->output_prefix) {
1391 msg = options->output_prefix(options, options->output_prefix_data);
1392 line_prefix = msg->buf;
1393 }
1394
1395 width = options->stat_width ? options->stat_width : 80;
1396 name_width = options->stat_name_width ? options->stat_name_width : 50;
1397 count = options->stat_count ? options->stat_count : data->nr;
1398
1399 /* Sanity: give at least 5 columns to the graph,
1400 * but leave at least 10 columns for the name.
1401 */
1402 if (width < 25)
1403 width = 25;
1404 if (name_width < 10)
1405 name_width = 10;
1406 else if (width < name_width + 15)
1407 name_width = width - 15;
1408
1409 /* Find the longest filename and max number of changes */
1410 reset = diff_get_color_opt(options, DIFF_RESET);
1411 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1412 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1413
1414 for (i = 0; (i < count) && (i < data->nr); i++) {
1415 struct diffstat_file *file = data->files[i];
1416 uintmax_t change = file->added + file->deleted;
1417 if (!data->files[i]->is_renamed &&
1418 (change == 0)) {
1419 count++; /* not shown == room for one more */
1420 continue;
1421 }
1422 fill_print_name(file);
1423 len = strlen(file->print_name);
1424 if (max_len < len)
1425 max_len = len;
1426
1427 if (file->is_binary || file->is_unmerged)
1428 continue;
1429 if (max_change < change)
1430 max_change = change;
1431 }
1432 count = i; /* min(count, data->nr) */
1433
1434 /* Compute the width of the graph part;
1435 * 10 is for one blank at the beginning of the line plus
1436 * " | count " between the name and the graph.
1437 *
1438 * From here on, name_width is the width of the name area,
1439 * and width is the width of the graph area.
1440 */
1441 name_width = (name_width < max_len) ? name_width : max_len;
1442 if (width < (name_width + 10) + max_change)
1443 width = width - (name_width + 10);
1444 else
1445 width = max_change;
1446
1447 for (i = 0; i < count; i++) {
1448 const char *prefix = "";
1449 char *name = data->files[i]->print_name;
1450 uintmax_t added = data->files[i]->added;
1451 uintmax_t deleted = data->files[i]->deleted;
1452 int name_len;
1453
1454 if (!data->files[i]->is_renamed &&
1455 (added + deleted == 0)) {
1456 total_files--;
1457 continue;
1458 }
1459 /*
1460 * "scale" the filename
1461 */
1462 len = name_width;
1463 name_len = strlen(name);
1464 if (name_width < name_len) {
1465 char *slash;
1466 prefix = "...";
1467 len -= 3;
1468 name += name_len - len;
1469 slash = strchr(name, '/');
1470 if (slash)
1471 name = slash;
1472 }
1473
1474 if (data->files[i]->is_binary) {
1475 fprintf(options->file, "%s", line_prefix);
1476 show_name(options->file, prefix, name, len);
1477 fprintf(options->file, " Bin ");
1478 fprintf(options->file, "%s%"PRIuMAX"%s",
1479 del_c, deleted, reset);
1480 fprintf(options->file, " -> ");
1481 fprintf(options->file, "%s%"PRIuMAX"%s",
1482 add_c, added, reset);
1483 fprintf(options->file, " bytes");
1484 fprintf(options->file, "\n");
1485 continue;
1486 }
1487 else if (data->files[i]->is_unmerged) {
1488 fprintf(options->file, "%s", line_prefix);
1489 show_name(options->file, prefix, name, len);
1490 fprintf(options->file, " Unmerged\n");
1491 continue;
1492 }
1493
1494 /*
1495 * scale the add/delete
1496 */
1497 add = added;
1498 del = deleted;
1499 adds += add;
1500 dels += del;
1501
1502 if (width <= max_change) {
1503 int total = add + del;
1504
1505 total = scale_linear(add + del, width, max_change);
1506 if (total < 2 && add && del)
1507 /* width >= 2 due to the sanity check */
1508 total = 2;
1509 if (add < del) {
1510 add = scale_linear(add, width, max_change);
1511 del = total - add;
1512 } else {
1513 del = scale_linear(del, width, max_change);
1514 add = total - del;
1515 }
1516 }
1517 fprintf(options->file, "%s", line_prefix);
1518 show_name(options->file, prefix, name, len);
1519 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1520 added + deleted ? " " : "");
1521 show_graph(options->file, '+', add, add_c, reset);
1522 show_graph(options->file, '-', del, del_c, reset);
1523 fprintf(options->file, "\n");
1524 }
1525 for (i = count; i < data->nr; i++) {
1526 uintmax_t added = data->files[i]->added;
1527 uintmax_t deleted = data->files[i]->deleted;
1528 if (!data->files[i]->is_renamed &&
1529 (added + deleted == 0)) {
1530 total_files--;
1531 continue;
1532 }
1533 adds += added;
1534 dels += deleted;
1535 if (!extra_shown)
1536 fprintf(options->file, "%s ...\n", line_prefix);
1537 extra_shown = 1;
1538 }
1539 fprintf(options->file, "%s", line_prefix);
1540 print_stat_summary(options->file, total_files, adds, dels);
1541 }
1542
1543 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1544 {
1545 int i, adds = 0, dels = 0, total_files = data->nr;
1546
1547 if (data->nr == 0)
1548 return;
1549
1550 for (i = 0; i < data->nr; i++) {
1551 if (!data->files[i]->is_binary &&
1552 !data->files[i]->is_unmerged) {
1553 int added = data->files[i]->added;
1554 int deleted= data->files[i]->deleted;
1555 if (!data->files[i]->is_renamed &&
1556 (added + deleted == 0)) {
1557 total_files--;
1558 } else {
1559 adds += added;
1560 dels += deleted;
1561 }
1562 }
1563 }
1564 if (options->output_prefix) {
1565 struct strbuf *msg = NULL;
1566 msg = options->output_prefix(options,
1567 options->output_prefix_data);
1568 fprintf(options->file, "%s", msg->buf);
1569 }
1570 print_stat_summary(options->file, total_files, adds, dels);
1571 }
1572
1573 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1574 {
1575 int i;
1576
1577 if (data->nr == 0)
1578 return;
1579
1580 for (i = 0; i < data->nr; i++) {
1581 struct diffstat_file *file = data->files[i];
1582
1583 if (options->output_prefix) {
1584 struct strbuf *msg = NULL;
1585 msg = options->output_prefix(options,
1586 options->output_prefix_data);
1587 fprintf(options->file, "%s", msg->buf);
1588 }
1589
1590 if (file->is_binary)
1591 fprintf(options->file, "-\t-\t");
1592 else
1593 fprintf(options->file,
1594 "%"PRIuMAX"\t%"PRIuMAX"\t",
1595 file->added, file->deleted);
1596 if (options->line_termination) {
1597 fill_print_name(file);
1598 if (!file->is_renamed)
1599 write_name_quoted(file->name, options->file,
1600 options->line_termination);
1601 else {
1602 fputs(file->print_name, options->file);
1603 putc(options->line_termination, options->file);
1604 }
1605 } else {
1606 if (file->is_renamed) {
1607 putc('\0', options->file);
1608 write_name_quoted(file->from_name, options->file, '\0');
1609 }
1610 write_name_quoted(file->name, options->file, '\0');
1611 }
1612 }
1613 }
1614
1615 struct dirstat_file {
1616 const char *name;
1617 unsigned long changed;
1618 };
1619
1620 struct dirstat_dir {
1621 struct dirstat_file *files;
1622 int alloc, nr, permille, cumulative;
1623 };
1624
1625 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1626 unsigned long changed, const char *base, int baselen)
1627 {
1628 unsigned long this_dir = 0;
1629 unsigned int sources = 0;
1630 const char *line_prefix = "";
1631 struct strbuf *msg = NULL;
1632
1633 if (opt->output_prefix) {
1634 msg = opt->output_prefix(opt, opt->output_prefix_data);
1635 line_prefix = msg->buf;
1636 }
1637
1638 while (dir->nr) {
1639 struct dirstat_file *f = dir->files;
1640 int namelen = strlen(f->name);
1641 unsigned long this;
1642 char *slash;
1643
1644 if (namelen < baselen)
1645 break;
1646 if (memcmp(f->name, base, baselen))
1647 break;
1648 slash = strchr(f->name + baselen, '/');
1649 if (slash) {
1650 int newbaselen = slash + 1 - f->name;
1651 this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1652 sources++;
1653 } else {
1654 this = f->changed;
1655 dir->files++;
1656 dir->nr--;
1657 sources += 2;
1658 }
1659 this_dir += this;
1660 }
1661
1662 /*
1663 * We don't report dirstat's for
1664 * - the top level
1665 * - or cases where everything came from a single directory
1666 * under this directory (sources == 1).
1667 */
1668 if (baselen && sources != 1) {
1669 if (this_dir) {
1670 int permille = this_dir * 1000 / changed;
1671 if (permille >= dir->permille) {
1672 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1673 permille / 10, permille % 10, baselen, base);
1674 if (!dir->cumulative)
1675 return 0;
1676 }
1677 }
1678 }
1679 return this_dir;
1680 }
1681
1682 static int dirstat_compare(const void *_a, const void *_b)
1683 {
1684 const struct dirstat_file *a = _a;
1685 const struct dirstat_file *b = _b;
1686 return strcmp(a->name, b->name);
1687 }
1688
1689 static void show_dirstat(struct diff_options *options)
1690 {
1691 int i;
1692 unsigned long changed;
1693 struct dirstat_dir dir;
1694 struct diff_queue_struct *q = &diff_queued_diff;
1695
1696 dir.files = NULL;
1697 dir.alloc = 0;
1698 dir.nr = 0;
1699 dir.permille = options->dirstat_permille;
1700 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1701
1702 changed = 0;
1703 for (i = 0; i < q->nr; i++) {
1704 struct diff_filepair *p = q->queue[i];
1705 const char *name;
1706 unsigned long copied, added, damage;
1707 int content_changed;
1708
1709 name = p->two->path ? p->two->path : p->one->path;
1710
1711 if (p->one->sha1_valid && p->two->sha1_valid)
1712 content_changed = hashcmp(p->one->sha1, p->two->sha1);
1713 else
1714 content_changed = 1;
1715
1716 if (!content_changed) {
1717 /*
1718 * The SHA1 has not changed, so pre-/post-content is
1719 * identical. We can therefore skip looking at the
1720 * file contents altogether.
1721 */
1722 damage = 0;
1723 goto found_damage;
1724 }
1725
1726 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1727 /*
1728 * In --dirstat-by-file mode, we don't really need to
1729 * look at the actual file contents at all.
1730 * The fact that the SHA1 changed is enough for us to
1731 * add this file to the list of results
1732 * (with each file contributing equal damage).
1733 */
1734 damage = 1;
1735 goto found_damage;
1736 }
1737
1738 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1739 diff_populate_filespec(p->one, 0);
1740 diff_populate_filespec(p->two, 0);
1741 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1742 &copied, &added);
1743 diff_free_filespec_data(p->one);
1744 diff_free_filespec_data(p->two);
1745 } else if (DIFF_FILE_VALID(p->one)) {
1746 diff_populate_filespec(p->one, 1);
1747 copied = added = 0;
1748 diff_free_filespec_data(p->one);
1749 } else if (DIFF_FILE_VALID(p->two)) {
1750 diff_populate_filespec(p->two, 1);
1751 copied = 0;
1752 added = p->two->size;
1753 diff_free_filespec_data(p->two);
1754 } else
1755 continue;
1756
1757 /*
1758 * Original minus copied is the removed material,
1759 * added is the new material. They are both damages
1760 * made to the preimage.
1761 * If the resulting damage is zero, we know that
1762 * diffcore_count_changes() considers the two entries to
1763 * be identical, but since content_changed is true, we
1764 * know that there must have been _some_ kind of change,
1765 * so we force all entries to have damage > 0.
1766 */
1767 damage = (p->one->size - copied) + added;
1768 if (!damage)
1769 damage = 1;
1770
1771 found_damage:
1772 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1773 dir.files[dir.nr].name = name;
1774 dir.files[dir.nr].changed = damage;
1775 changed += damage;
1776 dir.nr++;
1777 }
1778
1779 /* This can happen even with many files, if everything was renames */
1780 if (!changed)
1781 return;
1782
1783 /* Show all directories with more than x% of the changes */
1784 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1785 gather_dirstat(options, &dir, changed, "", 0);
1786 }
1787
1788 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1789 {
1790 int i;
1791 unsigned long changed;
1792 struct dirstat_dir dir;
1793
1794 if (data->nr == 0)
1795 return;
1796
1797 dir.files = NULL;
1798 dir.alloc = 0;
1799 dir.nr = 0;
1800 dir.permille = options->dirstat_permille;
1801 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1802
1803 changed = 0;
1804 for (i = 0; i < data->nr; i++) {
1805 struct diffstat_file *file = data->files[i];
1806 unsigned long damage = file->added + file->deleted;
1807 if (file->is_binary)
1808 /*
1809 * binary files counts bytes, not lines. Must find some
1810 * way to normalize binary bytes vs. textual lines.
1811 * The following heuristic assumes that there are 64
1812 * bytes per "line".
1813 * This is stupid and ugly, but very cheap...
1814 */
1815 damage = (damage + 63) / 64;
1816 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1817 dir.files[dir.nr].name = file->name;
1818 dir.files[dir.nr].changed = damage;
1819 changed += damage;
1820 dir.nr++;
1821 }
1822
1823 /* This can happen even with many files, if everything was renames */
1824 if (!changed)
1825 return;
1826
1827 /* Show all directories with more than x% of the changes */
1828 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1829 gather_dirstat(options, &dir, changed, "", 0);
1830 }
1831
1832 static void free_diffstat_info(struct diffstat_t *diffstat)
1833 {
1834 int i;
1835 for (i = 0; i < diffstat->nr; i++) {
1836 struct diffstat_file *f = diffstat->files[i];
1837 if (f->name != f->print_name)
1838 free(f->print_name);
1839 free(f->name);
1840 free(f->from_name);
1841 free(f);
1842 }
1843 free(diffstat->files);
1844 }
1845
1846 struct checkdiff_t {
1847 const char *filename;
1848 int lineno;
1849 int conflict_marker_size;
1850 struct diff_options *o;
1851 unsigned ws_rule;
1852 unsigned status;
1853 };
1854
1855 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1856 {
1857 char firstchar;
1858 int cnt;
1859
1860 if (len < marker_size + 1)
1861 return 0;
1862 firstchar = line[0];
1863 switch (firstchar) {
1864 case '=': case '>': case '<': case '|':
1865 break;
1866 default:
1867 return 0;
1868 }
1869 for (cnt = 1; cnt < marker_size; cnt++)
1870 if (line[cnt] != firstchar)
1871 return 0;
1872 /* line[1] thru line[marker_size-1] are same as firstchar */
1873 if (len < marker_size + 1 || !isspace(line[marker_size]))
1874 return 0;
1875 return 1;
1876 }
1877
1878 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1879 {
1880 struct checkdiff_t *data = priv;
1881 int marker_size = data->conflict_marker_size;
1882 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
1883 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
1884 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
1885 char *err;
1886 char *line_prefix = "";
1887 struct strbuf *msgbuf;
1888
1889 assert(data->o);
1890 if (data->o->output_prefix) {
1891 msgbuf = data->o->output_prefix(data->o,
1892 data->o->output_prefix_data);
1893 line_prefix = msgbuf->buf;
1894 }
1895
1896 if (line[0] == '+') {
1897 unsigned bad;
1898 data->lineno++;
1899 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1900 data->status |= 1;
1901 fprintf(data->o->file,
1902 "%s%s:%d: leftover conflict marker\n",
1903 line_prefix, data->filename, data->lineno);
1904 }
1905 bad = ws_check(line + 1, len - 1, data->ws_rule);
1906 if (!bad)
1907 return;
1908 data->status |= bad;
1909 err = whitespace_error_string(bad);
1910 fprintf(data->o->file, "%s%s:%d: %s.\n",
1911 line_prefix, data->filename, data->lineno, err);
1912 free(err);
1913 emit_line(data->o, set, reset, line, 1);
1914 ws_check_emit(line + 1, len - 1, data->ws_rule,
1915 data->o->file, set, reset, ws);
1916 } else if (line[0] == ' ') {
1917 data->lineno++;
1918 } else if (line[0] == '@') {
1919 char *plus = strchr(line, '+');
1920 if (plus)
1921 data->lineno = strtol(plus, NULL, 10) - 1;
1922 else
1923 die("invalid diff");
1924 }
1925 }
1926
1927 static unsigned char *deflate_it(char *data,
1928 unsigned long size,
1929 unsigned long *result_size)
1930 {
1931 int bound;
1932 unsigned char *deflated;
1933 git_zstream stream;
1934
1935 memset(&stream, 0, sizeof(stream));
1936 git_deflate_init(&stream, zlib_compression_level);
1937 bound = git_deflate_bound(&stream, size);
1938 deflated = xmalloc(bound);
1939 stream.next_out = deflated;
1940 stream.avail_out = bound;
1941
1942 stream.next_in = (unsigned char *)data;
1943 stream.avail_in = size;
1944 while (git_deflate(&stream, Z_FINISH) == Z_OK)
1945 ; /* nothing */
1946 git_deflate_end(&stream);
1947 *result_size = stream.total_out;
1948 return deflated;
1949 }
1950
1951 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1952 {
1953 void *cp;
1954 void *delta;
1955 void *deflated;
1956 void *data;
1957 unsigned long orig_size;
1958 unsigned long delta_size;
1959 unsigned long deflate_size;
1960 unsigned long data_size;
1961
1962 /* We could do deflated delta, or we could do just deflated two,
1963 * whichever is smaller.
1964 */
1965 delta = NULL;
1966 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1967 if (one->size && two->size) {
1968 delta = diff_delta(one->ptr, one->size,
1969 two->ptr, two->size,
1970 &delta_size, deflate_size);
1971 if (delta) {
1972 void *to_free = delta;
1973 orig_size = delta_size;
1974 delta = deflate_it(delta, delta_size, &delta_size);
1975 free(to_free);
1976 }
1977 }
1978
1979 if (delta && delta_size < deflate_size) {
1980 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
1981 free(deflated);
1982 data = delta;
1983 data_size = delta_size;
1984 }
1985 else {
1986 fprintf(file, "%sliteral %lu\n", prefix, two->size);
1987 free(delta);
1988 data = deflated;
1989 data_size = deflate_size;
1990 }
1991
1992 /* emit data encoded in base85 */
1993 cp = data;
1994 while (data_size) {
1995 int bytes = (52 < data_size) ? 52 : data_size;
1996 char line[70];
1997 data_size -= bytes;
1998 if (bytes <= 26)
1999 line[0] = bytes + 'A' - 1;
2000 else
2001 line[0] = bytes - 26 + 'a' - 1;
2002 encode_85(line + 1, cp, bytes);
2003 cp = (char *) cp + bytes;
2004 fprintf(file, "%s", prefix);
2005 fputs(line, file);
2006 fputc('\n', file);
2007 }
2008 fprintf(file, "%s\n", prefix);
2009 free(data);
2010 }
2011
2012 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
2013 {
2014 fprintf(file, "%sGIT binary patch\n", prefix);
2015 emit_binary_diff_body(file, one, two, prefix);
2016 emit_binary_diff_body(file, two, one, prefix);
2017 }
2018
2019 static void diff_filespec_load_driver(struct diff_filespec *one)
2020 {
2021 /* Use already-loaded driver */
2022 if (one->driver)
2023 return;
2024
2025 if (S_ISREG(one->mode))
2026 one->driver = userdiff_find_by_path(one->path);
2027
2028 /* Fallback to default settings */
2029 if (!one->driver)
2030 one->driver = userdiff_find_by_name("default");
2031 }
2032
2033 int diff_filespec_is_binary(struct diff_filespec *one)
2034 {
2035 if (one->is_binary == -1) {
2036 diff_filespec_load_driver(one);
2037 if (one->driver->binary != -1)
2038 one->is_binary = one->driver->binary;
2039 else {
2040 if (!one->data && DIFF_FILE_VALID(one))
2041 diff_populate_filespec(one, 0);
2042 if (one->data)
2043 one->is_binary = buffer_is_binary(one->data,
2044 one->size);
2045 if (one->is_binary == -1)
2046 one->is_binary = 0;
2047 }
2048 }
2049 return one->is_binary;
2050 }
2051
2052 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2053 {
2054 diff_filespec_load_driver(one);
2055 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2056 }
2057
2058 static const char *userdiff_word_regex(struct diff_filespec *one)
2059 {
2060 diff_filespec_load_driver(one);
2061 return one->driver->word_regex;
2062 }
2063
2064 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2065 {
2066 if (!options->a_prefix)
2067 options->a_prefix = a;
2068 if (!options->b_prefix)
2069 options->b_prefix = b;
2070 }
2071
2072 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2073 {
2074 if (!DIFF_FILE_VALID(one))
2075 return NULL;
2076
2077 diff_filespec_load_driver(one);
2078 return userdiff_get_textconv(one->driver);
2079 }
2080
2081 static void builtin_diff(const char *name_a,
2082 const char *name_b,
2083 struct diff_filespec *one,
2084 struct diff_filespec *two,
2085 const char *xfrm_msg,
2086 int must_show_header,
2087 struct diff_options *o,
2088 int complete_rewrite)
2089 {
2090 mmfile_t mf1, mf2;
2091 const char *lbl[2];
2092 char *a_one, *b_two;
2093 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
2094 const char *reset = diff_get_color_opt(o, DIFF_RESET);
2095 const char *a_prefix, *b_prefix;
2096 struct userdiff_driver *textconv_one = NULL;
2097 struct userdiff_driver *textconv_two = NULL;
2098 struct strbuf header = STRBUF_INIT;
2099 struct strbuf *msgbuf;
2100 char *line_prefix = "";
2101
2102 if (o->output_prefix) {
2103 msgbuf = o->output_prefix(o, o->output_prefix_data);
2104 line_prefix = msgbuf->buf;
2105 }
2106
2107 if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2108 (!one->mode || S_ISGITLINK(one->mode)) &&
2109 (!two->mode || S_ISGITLINK(two->mode))) {
2110 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2111 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2112 show_submodule_summary(o->file, one ? one->path : two->path,
2113 one->sha1, two->sha1, two->dirty_submodule,
2114 del, add, reset);
2115 return;
2116 }
2117
2118 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2119 textconv_one = get_textconv(one);
2120 textconv_two = get_textconv(two);
2121 }
2122
2123 diff_set_mnemonic_prefix(o, "a/", "b/");
2124 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2125 a_prefix = o->b_prefix;
2126 b_prefix = o->a_prefix;
2127 } else {
2128 a_prefix = o->a_prefix;
2129 b_prefix = o->b_prefix;
2130 }
2131
2132 /* Never use a non-valid filename anywhere if at all possible */
2133 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2134 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2135
2136 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2137 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2138 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2139 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2140 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
2141 if (lbl[0][0] == '/') {
2142 /* /dev/null */
2143 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
2144 if (xfrm_msg)
2145 strbuf_addstr(&header, xfrm_msg);
2146 must_show_header = 1;
2147 }
2148 else if (lbl[1][0] == '/') {
2149 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
2150 if (xfrm_msg)
2151 strbuf_addstr(&header, xfrm_msg);
2152 must_show_header = 1;
2153 }
2154 else {
2155 if (one->mode != two->mode) {
2156 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
2157 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
2158 must_show_header = 1;
2159 }
2160 if (xfrm_msg)
2161 strbuf_addstr(&header, xfrm_msg);
2162
2163 /*
2164 * we do not run diff between different kind
2165 * of objects.
2166 */
2167 if ((one->mode ^ two->mode) & S_IFMT)
2168 goto free_ab_and_return;
2169 if (complete_rewrite &&
2170 (textconv_one || !diff_filespec_is_binary(one)) &&
2171 (textconv_two || !diff_filespec_is_binary(two))) {
2172 fprintf(o->file, "%s", header.buf);
2173 strbuf_reset(&header);
2174 emit_rewrite_diff(name_a, name_b, one, two,
2175 textconv_one, textconv_two, o);
2176 o->found_changes = 1;
2177 goto free_ab_and_return;
2178 }
2179 }
2180
2181 if (o->irreversible_delete && lbl[1][0] == '/') {
2182 fprintf(o->file, "%s", header.buf);
2183 strbuf_reset(&header);
2184 goto free_ab_and_return;
2185 } else if (!DIFF_OPT_TST(o, TEXT) &&
2186 ( (!textconv_one && diff_filespec_is_binary(one)) ||
2187 (!textconv_two && diff_filespec_is_binary(two)) )) {
2188 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2189 die("unable to read files to diff");
2190 /* Quite common confusing case */
2191 if (mf1.size == mf2.size &&
2192 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2193 if (must_show_header)
2194 fprintf(o->file, "%s", header.buf);
2195 goto free_ab_and_return;
2196 }
2197 fprintf(o->file, "%s", header.buf);
2198 strbuf_reset(&header);
2199 if (DIFF_OPT_TST(o, BINARY))
2200 emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2201 else
2202 fprintf(o->file, "%sBinary files %s and %s differ\n",
2203 line_prefix, lbl[0], lbl[1]);
2204 o->found_changes = 1;
2205 } else {
2206 /* Crazy xdl interfaces.. */
2207 const char *diffopts = getenv("GIT_DIFF_OPTS");
2208 xpparam_t xpp;
2209 xdemitconf_t xecfg;
2210 struct emit_callback ecbdata;
2211 const struct userdiff_funcname *pe;
2212
2213 if (must_show_header) {
2214 fprintf(o->file, "%s", header.buf);
2215 strbuf_reset(&header);
2216 }
2217
2218 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2219 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2220
2221 pe = diff_funcname_pattern(one);
2222 if (!pe)
2223 pe = diff_funcname_pattern(two);
2224
2225 memset(&xpp, 0, sizeof(xpp));
2226 memset(&xecfg, 0, sizeof(xecfg));
2227 memset(&ecbdata, 0, sizeof(ecbdata));
2228 ecbdata.label_path = lbl;
2229 ecbdata.color_diff = want_color(o->use_color);
2230 ecbdata.found_changesp = &o->found_changes;
2231 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
2232 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2233 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2234 ecbdata.opt = o;
2235 ecbdata.header = header.len ? &header : NULL;
2236 xpp.flags = o->xdl_opts;
2237 xecfg.ctxlen = o->context;
2238 xecfg.interhunkctxlen = o->interhunkcontext;
2239 xecfg.flags = XDL_EMIT_FUNCNAMES;
2240 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2241 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2242 if (pe)
2243 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2244 if (!diffopts)
2245 ;
2246 else if (!prefixcmp(diffopts, "--unified="))
2247 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2248 else if (!prefixcmp(diffopts, "-u"))
2249 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2250 if (o->word_diff) {
2251 int i;
2252
2253 ecbdata.diff_words =
2254 xcalloc(1, sizeof(struct diff_words_data));
2255 ecbdata.diff_words->type = o->word_diff;
2256 ecbdata.diff_words->opt = o;
2257 if (!o->word_regex)
2258 o->word_regex = userdiff_word_regex(one);
2259 if (!o->word_regex)
2260 o->word_regex = userdiff_word_regex(two);
2261 if (!o->word_regex)
2262 o->word_regex = diff_word_regex_cfg;
2263 if (o->word_regex) {
2264 ecbdata.diff_words->word_regex = (regex_t *)
2265 xmalloc(sizeof(regex_t));
2266 if (regcomp(ecbdata.diff_words->word_regex,
2267 o->word_regex,
2268 REG_EXTENDED | REG_NEWLINE))
2269 die ("Invalid regular expression: %s",
2270 o->word_regex);
2271 }
2272 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2273 if (o->word_diff == diff_words_styles[i].type) {
2274 ecbdata.diff_words->style =
2275 &diff_words_styles[i];
2276 break;
2277 }
2278 }
2279 if (want_color(o->use_color)) {
2280 struct diff_words_style *st = ecbdata.diff_words->style;
2281 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2282 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2283 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
2284 }
2285 }
2286 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2287 &xpp, &xecfg);
2288 if (o->word_diff)
2289 free_diff_words_data(&ecbdata);
2290 if (textconv_one)
2291 free(mf1.ptr);
2292 if (textconv_two)
2293 free(mf2.ptr);
2294 xdiff_clear_find_func(&xecfg);
2295 }
2296
2297 free_ab_and_return:
2298 strbuf_release(&header);
2299 diff_free_filespec_data(one);
2300 diff_free_filespec_data(two);
2301 free(a_one);
2302 free(b_two);
2303 return;
2304 }
2305
2306 static void builtin_diffstat(const char *name_a, const char *name_b,
2307 struct diff_filespec *one,
2308 struct diff_filespec *two,
2309 struct diffstat_t *diffstat,
2310 struct diff_options *o,
2311 int complete_rewrite)
2312 {
2313 mmfile_t mf1, mf2;
2314 struct diffstat_file *data;
2315
2316 data = diffstat_add(diffstat, name_a, name_b);
2317
2318 if (!one || !two) {
2319 data->is_unmerged = 1;
2320 return;
2321 }
2322
2323 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2324 data->is_binary = 1;
2325 data->added = diff_filespec_size(two);
2326 data->deleted = diff_filespec_size(one);
2327 }
2328
2329 else if (complete_rewrite) {
2330 diff_populate_filespec(one, 0);
2331 diff_populate_filespec(two, 0);
2332 data->deleted = count_lines(one->data, one->size);
2333 data->added = count_lines(two->data, two->size);
2334 }
2335
2336 else {
2337 /* Crazy xdl interfaces.. */
2338 xpparam_t xpp;
2339 xdemitconf_t xecfg;
2340
2341 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2342 die("unable to read files to diff");
2343
2344 memset(&xpp, 0, sizeof(xpp));
2345 memset(&xecfg, 0, sizeof(xecfg));
2346 xpp.flags = o->xdl_opts;
2347 xecfg.ctxlen = o->context;
2348 xecfg.interhunkctxlen = o->interhunkcontext;
2349 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2350 &xpp, &xecfg);
2351 }
2352
2353 diff_free_filespec_data(one);
2354 diff_free_filespec_data(two);
2355 }
2356
2357 static void builtin_checkdiff(const char *name_a, const char *name_b,
2358 const char *attr_path,
2359 struct diff_filespec *one,
2360 struct diff_filespec *two,
2361 struct diff_options *o)
2362 {
2363 mmfile_t mf1, mf2;
2364 struct checkdiff_t data;
2365
2366 if (!two)
2367 return;
2368
2369 memset(&data, 0, sizeof(data));
2370 data.filename = name_b ? name_b : name_a;
2371 data.lineno = 0;
2372 data.o = o;
2373 data.ws_rule = whitespace_rule(attr_path);
2374 data.conflict_marker_size = ll_merge_marker_size(attr_path);
2375
2376 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2377 die("unable to read files to diff");
2378
2379 /*
2380 * All the other codepaths check both sides, but not checking
2381 * the "old" side here is deliberate. We are checking the newly
2382 * introduced changes, and as long as the "new" side is text, we
2383 * can and should check what it introduces.
2384 */
2385 if (diff_filespec_is_binary(two))
2386 goto free_and_return;
2387 else {
2388 /* Crazy xdl interfaces.. */
2389 xpparam_t xpp;
2390 xdemitconf_t xecfg;
2391
2392 memset(&xpp, 0, sizeof(xpp));
2393 memset(&xecfg, 0, sizeof(xecfg));
2394 xecfg.ctxlen = 1; /* at least one context line */
2395 xpp.flags = 0;
2396 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2397 &xpp, &xecfg);
2398
2399 if (data.ws_rule & WS_BLANK_AT_EOF) {
2400 struct emit_callback ecbdata;
2401 int blank_at_eof;
2402
2403 ecbdata.ws_rule = data.ws_rule;
2404 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2405 blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2406
2407 if (blank_at_eof) {
2408 static char *err;
2409 if (!err)
2410 err = whitespace_error_string(WS_BLANK_AT_EOF);
2411 fprintf(o->file, "%s:%d: %s.\n",
2412 data.filename, blank_at_eof, err);
2413 data.status = 1; /* report errors */
2414 }
2415 }
2416 }
2417 free_and_return:
2418 diff_free_filespec_data(one);
2419 diff_free_filespec_data(two);
2420 if (data.status)
2421 DIFF_OPT_SET(o, CHECK_FAILED);
2422 }
2423
2424 struct diff_filespec *alloc_filespec(const char *path)
2425 {
2426 int namelen = strlen(path);
2427 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2428
2429 memset(spec, 0, sizeof(*spec));
2430 spec->path = (char *)(spec + 1);
2431 memcpy(spec->path, path, namelen+1);
2432 spec->count = 1;
2433 spec->is_binary = -1;
2434 return spec;
2435 }
2436
2437 void free_filespec(struct diff_filespec *spec)
2438 {
2439 if (!--spec->count) {
2440 diff_free_filespec_data(spec);
2441 free(spec);
2442 }
2443 }
2444
2445 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2446 int sha1_valid, unsigned short mode)
2447 {
2448 if (mode) {
2449 spec->mode = canon_mode(mode);
2450 hashcpy(spec->sha1, sha1);
2451 spec->sha1_valid = sha1_valid;
2452 }
2453 }
2454
2455 /*
2456 * Given a name and sha1 pair, if the index tells us the file in
2457 * the work tree has that object contents, return true, so that
2458 * prepare_temp_file() does not have to inflate and extract.
2459 */
2460 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2461 {
2462 struct cache_entry *ce;
2463 struct stat st;
2464 int pos, len;
2465
2466 /*
2467 * We do not read the cache ourselves here, because the
2468 * benchmark with my previous version that always reads cache
2469 * shows that it makes things worse for diff-tree comparing
2470 * two linux-2.6 kernel trees in an already checked out work
2471 * tree. This is because most diff-tree comparisons deal with
2472 * only a small number of files, while reading the cache is
2473 * expensive for a large project, and its cost outweighs the
2474 * savings we get by not inflating the object to a temporary
2475 * file. Practically, this code only helps when we are used
2476 * by diff-cache --cached, which does read the cache before
2477 * calling us.
2478 */
2479 if (!active_cache)
2480 return 0;
2481
2482 /* We want to avoid the working directory if our caller
2483 * doesn't need the data in a normal file, this system
2484 * is rather slow with its stat/open/mmap/close syscalls,
2485 * and the object is contained in a pack file. The pack
2486 * is probably already open and will be faster to obtain
2487 * the data through than the working directory. Loose
2488 * objects however would tend to be slower as they need
2489 * to be individually opened and inflated.
2490 */
2491 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2492 return 0;
2493
2494 len = strlen(name);
2495 pos = cache_name_pos(name, len);
2496 if (pos < 0)
2497 return 0;
2498 ce = active_cache[pos];
2499
2500 /*
2501 * This is not the sha1 we are looking for, or
2502 * unreusable because it is not a regular file.
2503 */
2504 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2505 return 0;
2506
2507 /*
2508 * If ce is marked as "assume unchanged", there is no
2509 * guarantee that work tree matches what we are looking for.
2510 */
2511 if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2512 return 0;
2513
2514 /*
2515 * If ce matches the file in the work tree, we can reuse it.
2516 */
2517 if (ce_uptodate(ce) ||
2518 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2519 return 1;
2520
2521 return 0;
2522 }
2523
2524 static int populate_from_stdin(struct diff_filespec *s)
2525 {
2526 struct strbuf buf = STRBUF_INIT;
2527 size_t size = 0;
2528
2529 if (strbuf_read(&buf, 0, 0) < 0)
2530 return error("error while reading from stdin %s",
2531 strerror(errno));
2532
2533 s->should_munmap = 0;
2534 s->data = strbuf_detach(&buf, &size);
2535 s->size = size;
2536 s->should_free = 1;
2537 return 0;
2538 }
2539
2540 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2541 {
2542 int len;
2543 char *data = xmalloc(100), *dirty = "";
2544
2545 /* Are we looking at the work tree? */
2546 if (s->dirty_submodule)
2547 dirty = "-dirty";
2548
2549 len = snprintf(data, 100,
2550 "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2551 s->data = data;
2552 s->size = len;
2553 s->should_free = 1;
2554 if (size_only) {
2555 s->data = NULL;
2556 free(data);
2557 }
2558 return 0;
2559 }
2560
2561 /*
2562 * While doing rename detection and pickaxe operation, we may need to
2563 * grab the data for the blob (or file) for our own in-core comparison.
2564 * diff_filespec has data and size fields for this purpose.
2565 */
2566 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2567 {
2568 int err = 0;
2569 if (!DIFF_FILE_VALID(s))
2570 die("internal error: asking to populate invalid file.");
2571 if (S_ISDIR(s->mode))
2572 return -1;
2573
2574 if (s->data)
2575 return 0;
2576
2577 if (size_only && 0 < s->size)
2578 return 0;
2579
2580 if (S_ISGITLINK(s->mode))
2581 return diff_populate_gitlink(s, size_only);
2582
2583 if (!s->sha1_valid ||
2584 reuse_worktree_file(s->path, s->sha1, 0)) {
2585 struct strbuf buf = STRBUF_INIT;
2586 struct stat st;
2587 int fd;
2588
2589 if (!strcmp(s->path, "-"))
2590 return populate_from_stdin(s);
2591
2592 if (lstat(s->path, &st) < 0) {
2593 if (errno == ENOENT) {
2594 err_empty:
2595 err = -1;
2596 empty:
2597 s->data = (char *)"";
2598 s->size = 0;
2599 return err;
2600 }
2601 }
2602 s->size = xsize_t(st.st_size);
2603 if (!s->size)
2604 goto empty;
2605 if (S_ISLNK(st.st_mode)) {
2606 struct strbuf sb = STRBUF_INIT;
2607
2608 if (strbuf_readlink(&sb, s->path, s->size))
2609 goto err_empty;
2610 s->size = sb.len;
2611 s->data = strbuf_detach(&sb, NULL);
2612 s->should_free = 1;
2613 return 0;
2614 }
2615 if (size_only)
2616 return 0;
2617 fd = open(s->path, O_RDONLY);
2618 if (fd < 0)
2619 goto err_empty;
2620 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2621 close(fd);
2622 s->should_munmap = 1;
2623
2624 /*
2625 * Convert from working tree format to canonical git format
2626 */
2627 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2628 size_t size = 0;
2629 munmap(s->data, s->size);
2630 s->should_munmap = 0;
2631 s->data = strbuf_detach(&buf, &size);
2632 s->size = size;
2633 s->should_free = 1;
2634 }
2635 }
2636 else {
2637 enum object_type type;
2638 if (size_only) {
2639 type = sha1_object_info(s->sha1, &s->size);
2640 if (type < 0)
2641 die("unable to read %s", sha1_to_hex(s->sha1));
2642 } else {
2643 s->data = read_sha1_file(s->sha1, &type, &s->size);
2644 if (!s->data)
2645 die("unable to read %s", sha1_to_hex(s->sha1));
2646 s->should_free = 1;
2647 }
2648 }
2649 return 0;
2650 }
2651
2652 void diff_free_filespec_blob(struct diff_filespec *s)
2653 {
2654 if (s->should_free)
2655 free(s->data);
2656 else if (s->should_munmap)
2657 munmap(s->data, s->size);
2658
2659 if (s->should_free || s->should_munmap) {
2660 s->should_free = s->should_munmap = 0;
2661 s->data = NULL;
2662 }
2663 }
2664
2665 void diff_free_filespec_data(struct diff_filespec *s)
2666 {
2667 diff_free_filespec_blob(s);
2668 free(s->cnt_data);
2669 s->cnt_data = NULL;
2670 }
2671
2672 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2673 void *blob,
2674 unsigned long size,
2675 const unsigned char *sha1,
2676 int mode)
2677 {
2678 int fd;
2679 struct strbuf buf = STRBUF_INIT;
2680 struct strbuf template = STRBUF_INIT;
2681 char *path_dup = xstrdup(path);
2682 const char *base = basename(path_dup);
2683
2684 /* Generate "XXXXXX_basename.ext" */
2685 strbuf_addstr(&template, "XXXXXX_");
2686 strbuf_addstr(&template, base);
2687
2688 fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2689 strlen(base) + 1);
2690 if (fd < 0)
2691 die_errno("unable to create temp-file");
2692 if (convert_to_working_tree(path,
2693 (const char *)blob, (size_t)size, &buf)) {
2694 blob = buf.buf;
2695 size = buf.len;
2696 }
2697 if (write_in_full(fd, blob, size) != size)
2698 die_errno("unable to write temp-file");
2699 close(fd);
2700 temp->name = temp->tmp_path;
2701 strcpy(temp->hex, sha1_to_hex(sha1));
2702 temp->hex[40] = 0;
2703 sprintf(temp->mode, "%06o", mode);
2704 strbuf_release(&buf);
2705 strbuf_release(&template);
2706 free(path_dup);
2707 }
2708
2709 static struct diff_tempfile *prepare_temp_file(const char *name,
2710 struct diff_filespec *one)
2711 {
2712 struct diff_tempfile *temp = claim_diff_tempfile();
2713
2714 if (!DIFF_FILE_VALID(one)) {
2715 not_a_valid_file:
2716 /* A '-' entry produces this for file-2, and
2717 * a '+' entry produces this for file-1.
2718 */
2719 temp->name = "/dev/null";
2720 strcpy(temp->hex, ".");
2721 strcpy(temp->mode, ".");
2722 return temp;
2723 }
2724
2725 if (!remove_tempfile_installed) {
2726 atexit(remove_tempfile);
2727 sigchain_push_common(remove_tempfile_on_signal);
2728 remove_tempfile_installed = 1;
2729 }
2730
2731 if (!one->sha1_valid ||
2732 reuse_worktree_file(name, one->sha1, 1)) {
2733 struct stat st;
2734 if (lstat(name, &st) < 0) {
2735 if (errno == ENOENT)
2736 goto not_a_valid_file;
2737 die_errno("stat(%s)", name);
2738 }
2739 if (S_ISLNK(st.st_mode)) {
2740 struct strbuf sb = STRBUF_INIT;
2741 if (strbuf_readlink(&sb, name, st.st_size) < 0)
2742 die_errno("readlink(%s)", name);
2743 prep_temp_blob(name, temp, sb.buf, sb.len,
2744 (one->sha1_valid ?
2745 one->sha1 : null_sha1),
2746 (one->sha1_valid ?
2747 one->mode : S_IFLNK));
2748 strbuf_release(&sb);
2749 }
2750 else {
2751 /* we can borrow from the file in the work tree */
2752 temp->name = name;
2753 if (!one->sha1_valid)
2754 strcpy(temp->hex, sha1_to_hex(null_sha1));
2755 else
2756 strcpy(temp->hex, sha1_to_hex(one->sha1));
2757 /* Even though we may sometimes borrow the
2758 * contents from the work tree, we always want
2759 * one->mode. mode is trustworthy even when
2760 * !(one->sha1_valid), as long as
2761 * DIFF_FILE_VALID(one).
2762 */
2763 sprintf(temp->mode, "%06o", one->mode);
2764 }
2765 return temp;
2766 }
2767 else {
2768 if (diff_populate_filespec(one, 0))
2769 die("cannot read data blob for %s", one->path);
2770 prep_temp_blob(name, temp, one->data, one->size,
2771 one->sha1, one->mode);
2772 }
2773 return temp;
2774 }
2775
2776 /* An external diff command takes:
2777 *
2778 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2779 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2780 *
2781 */
2782 static void run_external_diff(const char *pgm,
2783 const char *name,
2784 const char *other,
2785 struct diff_filespec *one,
2786 struct diff_filespec *two,
2787 const char *xfrm_msg,
2788 int complete_rewrite)
2789 {
2790 const char *spawn_arg[10];
2791 int retval;
2792 const char **arg = &spawn_arg[0];
2793
2794 if (one && two) {
2795 struct diff_tempfile *temp_one, *temp_two;
2796 const char *othername = (other ? other : name);
2797 temp_one = prepare_temp_file(name, one);
2798 temp_two = prepare_temp_file(othername, two);
2799 *arg++ = pgm;
2800 *arg++ = name;
2801 *arg++ = temp_one->name;
2802 *arg++ = temp_one->hex;
2803 *arg++ = temp_one->mode;
2804 *arg++ = temp_two->name;
2805 *arg++ = temp_two->hex;
2806 *arg++ = temp_two->mode;
2807 if (other) {
2808 *arg++ = other;
2809 *arg++ = xfrm_msg;
2810 }
2811 } else {
2812 *arg++ = pgm;
2813 *arg++ = name;
2814 }
2815 *arg = NULL;
2816 fflush(NULL);
2817 retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2818 remove_tempfile();
2819 if (retval) {
2820 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2821 exit(1);
2822 }
2823 }
2824
2825 static int similarity_index(struct diff_filepair *p)
2826 {
2827 return p->score * 100 / MAX_SCORE;
2828 }
2829
2830 static void fill_metainfo(struct strbuf *msg,
2831 const char *name,
2832 const char *other,
2833 struct diff_filespec *one,
2834 struct diff_filespec *two,
2835 struct diff_options *o,
2836 struct diff_filepair *p,
2837 int *must_show_header,
2838 int use_color)
2839 {
2840 const char *set = diff_get_color(use_color, DIFF_METAINFO);
2841 const char *reset = diff_get_color(use_color, DIFF_RESET);
2842 struct strbuf *msgbuf;
2843 char *line_prefix = "";
2844
2845 *must_show_header = 1;
2846 if (o->output_prefix) {
2847 msgbuf = o->output_prefix(o, o->output_prefix_data);
2848 line_prefix = msgbuf->buf;
2849 }
2850 strbuf_init(msg, PATH_MAX * 2 + 300);
2851 switch (p->status) {
2852 case DIFF_STATUS_COPIED:
2853 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2854 line_prefix, set, similarity_index(p));
2855 strbuf_addf(msg, "%s\n%s%scopy from ",
2856 reset, line_prefix, set);
2857 quote_c_style(name, msg, NULL, 0);
2858 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2859 quote_c_style(other, msg, NULL, 0);
2860 strbuf_addf(msg, "%s\n", reset);
2861 break;
2862 case DIFF_STATUS_RENAMED:
2863 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2864 line_prefix, set, similarity_index(p));
2865 strbuf_addf(msg, "%s\n%s%srename from ",
2866 reset, line_prefix, set);
2867 quote_c_style(name, msg, NULL, 0);
2868 strbuf_addf(msg, "%s\n%s%srename to ",
2869 reset, line_prefix, set);
2870 quote_c_style(other, msg, NULL, 0);
2871 strbuf_addf(msg, "%s\n", reset);
2872 break;
2873 case DIFF_STATUS_MODIFIED:
2874 if (p->score) {
2875 strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2876 line_prefix,
2877 set, similarity_index(p), reset);
2878 break;
2879 }
2880 /* fallthru */
2881 default:
2882 *must_show_header = 0;
2883 }
2884 if (one && two && hashcmp(one->sha1, two->sha1)) {
2885 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2886
2887 if (DIFF_OPT_TST(o, BINARY)) {
2888 mmfile_t mf;
2889 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2890 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2891 abbrev = 40;
2892 }
2893 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2894 find_unique_abbrev(one->sha1, abbrev));
2895 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2896 if (one->mode == two->mode)
2897 strbuf_addf(msg, " %06o", one->mode);
2898 strbuf_addf(msg, "%s\n", reset);
2899 }
2900 }
2901
2902 static void run_diff_cmd(const char *pgm,
2903 const char *name,
2904 const char *other,
2905 const char *attr_path,
2906 struct diff_filespec *one,
2907 struct diff_filespec *two,
2908 struct strbuf *msg,
2909 struct diff_options *o,
2910 struct diff_filepair *p)
2911 {
2912 const char *xfrm_msg = NULL;
2913 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2914 int must_show_header = 0;
2915
2916 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2917 pgm = NULL;
2918 else {
2919 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2920 if (drv && drv->external)
2921 pgm = drv->external;
2922 }
2923
2924 if (msg) {
2925 /*
2926 * don't use colors when the header is intended for an
2927 * external diff driver
2928 */
2929 fill_metainfo(msg, name, other, one, two, o, p,
2930 &must_show_header,
2931 want_color(o->use_color) && !pgm);
2932 xfrm_msg = msg->len ? msg->buf : NULL;
2933 }
2934
2935 if (pgm) {
2936 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2937 complete_rewrite);
2938 return;
2939 }
2940 if (one && two)
2941 builtin_diff(name, other ? other : name,
2942 one, two, xfrm_msg, must_show_header,
2943 o, complete_rewrite);
2944 else
2945 fprintf(o->file, "* Unmerged path %s\n", name);
2946 }
2947
2948 static void diff_fill_sha1_info(struct diff_filespec *one)
2949 {
2950 if (DIFF_FILE_VALID(one)) {
2951 if (!one->sha1_valid) {
2952 struct stat st;
2953 if (!strcmp(one->path, "-")) {
2954 hashcpy(one->sha1, null_sha1);
2955 return;
2956 }
2957 if (lstat(one->path, &st) < 0)
2958 die_errno("stat '%s'", one->path);
2959 if (index_path(one->sha1, one->path, &st, 0))
2960 die("cannot hash %s", one->path);
2961 }
2962 }
2963 else
2964 hashclr(one->sha1);
2965 }
2966
2967 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2968 {
2969 /* Strip the prefix but do not molest /dev/null and absolute paths */
2970 if (*namep && **namep != '/') {
2971 *namep += prefix_length;
2972 if (**namep == '/')
2973 ++*namep;
2974 }
2975 if (*otherp && **otherp != '/') {
2976 *otherp += prefix_length;
2977 if (**otherp == '/')
2978 ++*otherp;
2979 }
2980 }
2981
2982 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2983 {
2984 const char *pgm = external_diff();
2985 struct strbuf msg;
2986 struct diff_filespec *one = p->one;
2987 struct diff_filespec *two = p->two;
2988 const char *name;
2989 const char *other;
2990 const char *attr_path;
2991
2992 name = p->one->path;
2993 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2994 attr_path = name;
2995 if (o->prefix_length)
2996 strip_prefix(o->prefix_length, &name, &other);
2997
2998 if (DIFF_PAIR_UNMERGED(p)) {
2999 run_diff_cmd(pgm, name, NULL, attr_path,
3000 NULL, NULL, NULL, o, p);
3001 return;
3002 }
3003
3004 diff_fill_sha1_info(one);
3005 diff_fill_sha1_info(two);
3006
3007 if (!pgm &&
3008 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3009 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3010 /*
3011 * a filepair that changes between file and symlink
3012 * needs to be split into deletion and creation.
3013 */
3014 struct diff_filespec *null = alloc_filespec(two->path);
3015 run_diff_cmd(NULL, name, other, attr_path,
3016 one, null, &msg, o, p);
3017 free(null);
3018 strbuf_release(&msg);
3019
3020 null = alloc_filespec(one->path);
3021 run_diff_cmd(NULL, name, other, attr_path,
3022 null, two, &msg, o, p);
3023 free(null);
3024 }
3025 else
3026 run_diff_cmd(pgm, name, other, attr_path,
3027 one, two, &msg, o, p);
3028
3029 strbuf_release(&msg);
3030 }
3031
3032 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3033 struct diffstat_t *diffstat)
3034 {
3035 const char *name;
3036 const char *other;
3037 int complete_rewrite = 0;
3038
3039 if (DIFF_PAIR_UNMERGED(p)) {
3040 /* unmerged */
3041 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
3042 return;
3043 }
3044
3045 name = p->one->path;
3046 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3047
3048 if (o->prefix_length)
3049 strip_prefix(o->prefix_length, &name, &other);
3050
3051 diff_fill_sha1_info(p->one);
3052 diff_fill_sha1_info(p->two);
3053
3054 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3055 complete_rewrite = 1;
3056 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
3057 }
3058
3059 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3060 {
3061 const char *name;
3062 const char *other;
3063 const char *attr_path;
3064
3065 if (DIFF_PAIR_UNMERGED(p)) {
3066 /* unmerged */
3067 return;
3068 }
3069
3070 name = p->one->path;
3071 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3072 attr_path = other ? other : name;
3073
3074 if (o->prefix_length)
3075 strip_prefix(o->prefix_length, &name, &other);
3076
3077 diff_fill_sha1_info(p->one);
3078 diff_fill_sha1_info(p->two);
3079
3080 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3081 }
3082
3083 void diff_setup(struct diff_options *options)
3084 {
3085 memcpy(options, &default_diff_options, sizeof(*options));
3086
3087 options->file = stdout;
3088
3089 options->line_termination = '\n';
3090 options->break_opt = -1;
3091 options->rename_limit = -1;
3092 options->dirstat_permille = diff_dirstat_permille_default;
3093 options->context = 3;
3094
3095 options->change = diff_change;
3096 options->add_remove = diff_addremove;
3097 options->use_color = diff_use_color_default;
3098 options->detect_rename = diff_detect_rename_default;
3099
3100 if (diff_no_prefix) {
3101 options->a_prefix = options->b_prefix = "";
3102 } else if (!diff_mnemonic_prefix) {
3103 options->a_prefix = "a/";
3104 options->b_prefix = "b/";
3105 }
3106 }
3107
3108 int diff_setup_done(struct diff_options *options)
3109 {
3110 int count = 0;
3111
3112 if (options->output_format & DIFF_FORMAT_NAME)
3113 count++;
3114 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3115 count++;
3116 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3117 count++;
3118 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3119 count++;
3120 if (count > 1)
3121 die("--name-only, --name-status, --check and -s are mutually exclusive");
3122
3123 /*
3124 * Most of the time we can say "there are changes"
3125 * only by checking if there are changed paths, but
3126 * --ignore-whitespace* options force us to look
3127 * inside contents.
3128 */
3129
3130 if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3131 DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3132 DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3133 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3134 else
3135 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3136
3137 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3138 options->detect_rename = DIFF_DETECT_COPY;
3139
3140 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3141 options->prefix = NULL;
3142 if (options->prefix)
3143 options->prefix_length = strlen(options->prefix);
3144 else
3145 options->prefix_length = 0;
3146
3147 if (options->output_format & (DIFF_FORMAT_NAME |
3148 DIFF_FORMAT_NAME_STATUS |
3149 DIFF_FORMAT_CHECKDIFF |
3150 DIFF_FORMAT_NO_OUTPUT))
3151 options->output_format &= ~(DIFF_FORMAT_RAW |
3152 DIFF_FORMAT_NUMSTAT |
3153 DIFF_FORMAT_DIFFSTAT |
3154 DIFF_FORMAT_SHORTSTAT |
3155 DIFF_FORMAT_DIRSTAT |
3156 DIFF_FORMAT_SUMMARY |
3157 DIFF_FORMAT_PATCH);
3158
3159 /*
3160 * These cases always need recursive; we do not drop caller-supplied
3161 * recursive bits for other formats here.
3162 */
3163 if (options->output_format & (DIFF_FORMAT_PATCH |
3164 DIFF_FORMAT_NUMSTAT |
3165 DIFF_FORMAT_DIFFSTAT |
3166 DIFF_FORMAT_SHORTSTAT |
3167 DIFF_FORMAT_DIRSTAT |
3168 DIFF_FORMAT_SUMMARY |
3169 DIFF_FORMAT_CHECKDIFF))
3170 DIFF_OPT_SET(options, RECURSIVE);
3171 /*
3172 * Also pickaxe would not work very well if you do not say recursive
3173 */
3174 if (options->pickaxe)
3175 DIFF_OPT_SET(options, RECURSIVE);
3176 /*
3177 * When patches are generated, submodules diffed against the work tree
3178 * must be checked for dirtiness too so it can be shown in the output
3179 */
3180 if (options->output_format & DIFF_FORMAT_PATCH)
3181 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3182
3183 if (options->detect_rename && options->rename_limit < 0)
3184 options->rename_limit = diff_rename_limit_default;
3185 if (options->setup & DIFF_SETUP_USE_CACHE) {
3186 if (!active_cache)
3187 /* read-cache does not die even when it fails
3188 * so it is safe for us to do this here. Also
3189 * it does not smudge active_cache or active_nr
3190 * when it fails, so we do not have to worry about
3191 * cleaning it up ourselves either.
3192 */
3193 read_cache();
3194 }
3195 if (options->abbrev <= 0 || 40 < options->abbrev)
3196 options->abbrev = 40; /* full */
3197
3198 /*
3199 * It does not make sense to show the first hit we happened
3200 * to have found. It does not make sense not to return with
3201 * exit code in such a case either.
3202 */
3203 if (DIFF_OPT_TST(options, QUICK)) {
3204 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3205 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3206 }
3207
3208 return 0;
3209 }
3210
3211 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3212 {
3213 char c, *eq;
3214 int len;
3215
3216 if (*arg != '-')
3217 return 0;
3218 c = *++arg;
3219 if (!c)
3220 return 0;
3221 if (c == arg_short) {
3222 c = *++arg;
3223 if (!c)
3224 return 1;
3225 if (val && isdigit(c)) {
3226 char *end;
3227 int n = strtoul(arg, &end, 10);
3228 if (*end)
3229 return 0;
3230 *val = n;
3231 return 1;
3232 }
3233 return 0;
3234 }
3235 if (c != '-')
3236 return 0;
3237 arg++;
3238 eq = strchr(arg, '=');
3239 if (eq)
3240 len = eq - arg;
3241 else
3242 len = strlen(arg);
3243 if (!len || strncmp(arg, arg_long, len))
3244 return 0;
3245 if (eq) {
3246 int n;
3247 char *end;
3248 if (!isdigit(*++eq))
3249 return 0;
3250 n = strtoul(eq, &end, 10);
3251 if (*end)
3252 return 0;
3253 *val = n;
3254 }
3255 return 1;
3256 }
3257
3258 static int diff_scoreopt_parse(const char *opt);
3259
3260 static inline int short_opt(char opt, const char **argv,
3261 const char **optarg)
3262 {
3263 const char *arg = argv[0];
3264 if (arg[0] != '-' || arg[1] != opt)
3265 return 0;
3266 if (arg[2] != '\0') {
3267 *optarg = arg + 2;
3268 return 1;
3269 }
3270 if (!argv[1])
3271 die("Option '%c' requires a value", opt);
3272 *optarg = argv[1];
3273 return 2;
3274 }
3275
3276 int parse_long_opt(const char *opt, const char **argv,
3277 const char **optarg)
3278 {
3279 const char *arg = argv[0];
3280 if (arg[0] != '-' || arg[1] != '-')
3281 return 0;
3282 arg += strlen("--");
3283 if (prefixcmp(arg, opt))
3284 return 0;
3285 arg += strlen(opt);
3286 if (*arg == '=') { /* sticked form: --option=value */
3287 *optarg = arg + 1;
3288 return 1;
3289 }
3290 if (*arg != '\0')
3291 return 0;
3292 /* separate form: --option value */
3293 if (!argv[1])
3294 die("Option '--%s' requires a value", opt);
3295 *optarg = argv[1];
3296 return 2;
3297 }
3298
3299 static int stat_opt(struct diff_options *options, const char **av)
3300 {
3301 const char *arg = av[0];
3302 char *end;
3303 int width = options->stat_width;
3304 int name_width = options->stat_name_width;
3305 int count = options->stat_count;
3306 int argcount = 1;
3307
3308 arg += strlen("--stat");
3309 end = (char *)arg;
3310
3311 switch (*arg) {
3312 case '-':
3313 if (!prefixcmp(arg, "-width")) {
3314 arg += strlen("-width");
3315 if (*arg == '=')
3316 width = strtoul(arg + 1, &end, 10);
3317 else if (!*arg && !av[1])
3318 die("Option '--stat-width' requires a value");
3319 else if (!*arg) {
3320 width = strtoul(av[1], &end, 10);
3321 argcount = 2;
3322 }
3323 } else if (!prefixcmp(arg, "-name-width")) {
3324 arg += strlen("-name-width");
3325 if (*arg == '=')
3326 name_width = strtoul(arg + 1, &end, 10);
3327 else if (!*arg && !av[1])
3328 die("Option '--stat-name-width' requires a value");
3329 else if (!*arg) {
3330 name_width = strtoul(av[1], &end, 10);
3331 argcount = 2;
3332 }
3333 } else if (!prefixcmp(arg, "-count")) {
3334 arg += strlen("-count");
3335 if (*arg == '=')
3336 count = strtoul(arg + 1, &end, 10);
3337 else if (!*arg && !av[1])
3338 die("Option '--stat-count' requires a value");
3339 else if (!*arg) {
3340 count = strtoul(av[1], &end, 10);
3341 argcount = 2;
3342 }
3343 }
3344 break;
3345 case '=':
3346 width = strtoul(arg+1, &end, 10);
3347 if (*end == ',')
3348 name_width = strtoul(end+1, &end, 10);
3349 if (*end == ',')
3350 count = strtoul(end+1, &end, 10);
3351 }
3352
3353 /* Important! This checks all the error cases! */
3354 if (*end)
3355 return 0;
3356 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3357 options->stat_name_width = name_width;
3358 options->stat_width = width;
3359 options->stat_count = count;
3360 return argcount;
3361 }
3362
3363 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3364 {
3365 struct strbuf errmsg = STRBUF_INIT;
3366 if (parse_dirstat_params(options, params, &errmsg))
3367 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3368 errmsg.buf);
3369 strbuf_release(&errmsg);
3370 /*
3371 * The caller knows a dirstat-related option is given from the command
3372 * line; allow it to say "return this_function();"
3373 */
3374 options->output_format |= DIFF_FORMAT_DIRSTAT;
3375 return 1;
3376 }
3377
3378 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3379 {
3380 const char *arg = av[0];
3381 const char *optarg;
3382 int argcount;
3383
3384 /* Output format options */
3385 if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3386 options->output_format |= DIFF_FORMAT_PATCH;
3387 else if (opt_arg(arg, 'U', "unified", &options->context))
3388 options->output_format |= DIFF_FORMAT_PATCH;
3389 else if (!strcmp(arg, "--raw"))
3390 options->output_format |= DIFF_FORMAT_RAW;
3391 else if (!strcmp(arg, "--patch-with-raw"))
3392 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3393 else if (!strcmp(arg, "--numstat"))
3394 options->output_format |= DIFF_FORMAT_NUMSTAT;
3395 else if (!strcmp(arg, "--shortstat"))
3396 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3397 else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3398 return parse_dirstat_opt(options, "");
3399 else if (!prefixcmp(arg, "-X"))
3400 return parse_dirstat_opt(options, arg + 2);
3401 else if (!prefixcmp(arg, "--dirstat="))
3402 return parse_dirstat_opt(options, arg + 10);
3403 else if (!strcmp(arg, "--cumulative"))
3404 return parse_dirstat_opt(options, "cumulative");
3405 else if (!strcmp(arg, "--dirstat-by-file"))
3406 return parse_dirstat_opt(options, "files");
3407 else if (!prefixcmp(arg, "--dirstat-by-file=")) {
3408 parse_dirstat_opt(options, "files");
3409 return parse_dirstat_opt(options, arg + 18);
3410 }
3411 else if (!strcmp(arg, "--check"))
3412 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3413 else if (!strcmp(arg, "--summary"))
3414 options->output_format |= DIFF_FORMAT_SUMMARY;
3415 else if (!strcmp(arg, "--patch-with-stat"))
3416 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3417 else if (!strcmp(arg, "--name-only"))
3418 options->output_format |= DIFF_FORMAT_NAME;
3419 else if (!strcmp(arg, "--name-status"))
3420 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3421 else if (!strcmp(arg, "-s"))
3422 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3423 else if (!prefixcmp(arg, "--stat"))
3424 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3425 return stat_opt(options, av);
3426
3427 /* renames options */
3428 else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3429 !strcmp(arg, "--break-rewrites")) {
3430 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3431 return error("invalid argument to -B: %s", arg+2);
3432 }
3433 else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3434 !strcmp(arg, "--find-renames")) {
3435 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3436 return error("invalid argument to -M: %s", arg+2);
3437 options->detect_rename = DIFF_DETECT_RENAME;
3438 }
3439 else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3440 options->irreversible_delete = 1;
3441 }
3442 else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3443 !strcmp(arg, "--find-copies")) {
3444 if (options->detect_rename == DIFF_DETECT_COPY)
3445 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3446 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3447 return error("invalid argument to -C: %s", arg+2);
3448 options->detect_rename = DIFF_DETECT_COPY;
3449 }
3450 else if (!strcmp(arg, "--no-renames"))
3451 options->detect_rename = 0;
3452 else if (!strcmp(arg, "--relative"))
3453 DIFF_OPT_SET(options, RELATIVE_NAME);
3454 else if (!prefixcmp(arg, "--relative=")) {
3455 DIFF_OPT_SET(options, RELATIVE_NAME);
3456 options->prefix = arg + 11;
3457 }
3458
3459 /* xdiff options */
3460 else if (!strcmp(arg, "--minimal"))
3461 DIFF_XDL_SET(options, NEED_MINIMAL);
3462 else if (!strcmp(arg, "--no-minimal"))
3463 DIFF_XDL_CLR(options, NEED_MINIMAL);
3464 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3465 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3466 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3467 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3468 else if (!strcmp(arg, "--ignore-space-at-eol"))
3469 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3470 else if (!strcmp(arg, "--patience"))
3471 DIFF_XDL_SET(options, PATIENCE_DIFF);
3472 else if (!strcmp(arg, "--histogram"))
3473 DIFF_XDL_SET(options, HISTOGRAM_DIFF);
3474
3475 /* flags options */
3476 else if (!strcmp(arg, "--binary")) {
3477 options->output_format |= DIFF_FORMAT_PATCH;
3478 DIFF_OPT_SET(options, BINARY);
3479 }
3480 else if (!strcmp(arg, "--full-index"))
3481 DIFF_OPT_SET(options, FULL_INDEX);
3482 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3483 DIFF_OPT_SET(options, TEXT);
3484 else if (!strcmp(arg, "-R"))
3485 DIFF_OPT_SET(options, REVERSE_DIFF);
3486 else if (!strcmp(arg, "--find-copies-harder"))
3487 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3488 else if (!strcmp(arg, "--follow"))
3489 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3490 else if (!strcmp(arg, "--color"))
3491 options->use_color = 1;
3492 else if (!prefixcmp(arg, "--color=")) {
3493 int value = git_config_colorbool(NULL, arg+8);
3494 if (value < 0)
3495 return error("option `color' expects \"always\", \"auto\", or \"never\"");
3496 options->use_color = value;
3497 }
3498 else if (!strcmp(arg, "--no-color"))
3499 options->use_color = 0;
3500 else if (!strcmp(arg, "--color-words")) {
3501 options->use_color = 1;
3502 options->word_diff = DIFF_WORDS_COLOR;
3503 }
3504 else if (!prefixcmp(arg, "--color-words=")) {
3505 options->use_color = 1;
3506 options->word_diff = DIFF_WORDS_COLOR;
3507 options->word_regex = arg + 14;
3508 }
3509 else if (!strcmp(arg, "--word-diff")) {
3510 if (options->word_diff == DIFF_WORDS_NONE)
3511 options->word_diff = DIFF_WORDS_PLAIN;
3512 }
3513 else if (!prefixcmp(arg, "--word-diff=")) {
3514 const char *type = arg + 12;
3515 if (!strcmp(type, "plain"))
3516 options->word_diff = DIFF_WORDS_PLAIN;
3517 else if (!strcmp(type, "color")) {
3518 options->use_color = 1;
3519 options->word_diff = DIFF_WORDS_COLOR;
3520 }
3521 else if (!strcmp(type, "porcelain"))
3522 options->word_diff = DIFF_WORDS_PORCELAIN;
3523 else if (!strcmp(type, "none"))
3524 options->word_diff = DIFF_WORDS_NONE;
3525 else
3526 die("bad --word-diff argument: %s", type);
3527 }
3528 else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3529 if (options->word_diff == DIFF_WORDS_NONE)
3530 options->word_diff = DIFF_WORDS_PLAIN;
3531 options->word_regex = optarg;
3532 return argcount;
3533 }
3534 else if (!strcmp(arg, "--exit-code"))
3535 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3536 else if (!strcmp(arg, "--quiet"))
3537 DIFF_OPT_SET(options, QUICK);
3538 else if (!strcmp(arg, "--ext-diff"))
3539 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3540 else if (!strcmp(arg, "--no-ext-diff"))
3541 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3542 else if (!strcmp(arg, "--textconv"))
3543 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3544 else if (!strcmp(arg, "--no-textconv"))
3545 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3546 else if (!strcmp(arg, "--ignore-submodules")) {
3547 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3548 handle_ignore_submodules_arg(options, "all");
3549 } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3550 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3551 handle_ignore_submodules_arg(options, arg + 20);
3552 } else if (!strcmp(arg, "--submodule"))
3553 DIFF_OPT_SET(options, SUBMODULE_LOG);
3554 else if (!prefixcmp(arg, "--submodule=")) {
3555 if (!strcmp(arg + 12, "log"))
3556 DIFF_OPT_SET(options, SUBMODULE_LOG);
3557 }
3558
3559 /* misc options */
3560 else if (!strcmp(arg, "-z"))
3561 options->line_termination = 0;
3562 else if ((argcount = short_opt('l', av, &optarg))) {
3563 options->rename_limit = strtoul(optarg, NULL, 10);
3564 return argcount;
3565 }
3566 else if ((argcount = short_opt('S', av, &optarg))) {
3567 options->pickaxe = optarg;
3568 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3569 return argcount;
3570 } else if ((argcount = short_opt('G', av, &optarg))) {
3571 options->pickaxe = optarg;
3572 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3573 return argcount;
3574 }
3575 else if (!strcmp(arg, "--pickaxe-all"))
3576 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3577 else if (!strcmp(arg, "--pickaxe-regex"))
3578 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3579 else if ((argcount = short_opt('O', av, &optarg))) {
3580 options->orderfile = optarg;
3581 return argcount;
3582 }
3583 else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3584 options->filter = optarg;
3585 return argcount;
3586 }
3587 else if (!strcmp(arg, "--abbrev"))
3588 options->abbrev = DEFAULT_ABBREV;
3589 else if (!prefixcmp(arg, "--abbrev=")) {
3590 options->abbrev = strtoul(arg + 9, NULL, 10);
3591 if (options->abbrev < MINIMUM_ABBREV)
3592 options->abbrev = MINIMUM_ABBREV;
3593 else if (40 < options->abbrev)
3594 options->abbrev = 40;
3595 }
3596 else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3597 options->a_prefix = optarg;
3598 return argcount;
3599 }
3600 else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3601 options->b_prefix = optarg;
3602 return argcount;
3603 }
3604 else if (!strcmp(arg, "--no-prefix"))
3605 options->a_prefix = options->b_prefix = "";
3606 else if (opt_arg(arg, '\0', "inter-hunk-context",
3607 &options->interhunkcontext))
3608 ;
3609 else if (!strcmp(arg, "-W"))
3610 DIFF_OPT_SET(options, FUNCCONTEXT);
3611 else if (!strcmp(arg, "--function-context"))
3612 DIFF_OPT_SET(options, FUNCCONTEXT);
3613 else if (!strcmp(arg, "--no-function-context"))
3614 DIFF_OPT_CLR(options, FUNCCONTEXT);
3615 else if ((argcount = parse_long_opt("output", av, &optarg))) {
3616 options->file = fopen(optarg, "w");
3617 if (!options->file)
3618 die_errno("Could not open '%s'", optarg);
3619 options->close_file = 1;
3620 return argcount;
3621 } else
3622 return 0;
3623 return 1;
3624 }
3625
3626 int parse_rename_score(const char **cp_p)
3627 {
3628 unsigned long num, scale;
3629 int ch, dot;
3630 const char *cp = *cp_p;
3631
3632 num = 0;