?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
cherry: split off function to print output lines
[git/jrn.git] / diff.c
blob2f2ab87d96324df35cb56c07d8c416f029ba9f56
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
14 #include "userdiff.h"
15 #include "sigchain.h"
16 #include "submodule.h"
17 #include "ll-merge.h"
18
19 #ifdef NO_FAST_WORKING_DIRECTORY
20 #define FAST_WORKING_DIRECTORY 0
21 #else
22 #define FAST_WORKING_DIRECTORY 1
23 #endif
24
25 static int diff_detect_rename_default;
26 static int diff_rename_limit_default = 200;
27 static int diff_suppress_blank_empty;
28 int diff_use_color_default = -1;
29 static const char *diff_word_regex_cfg;
30 static const char *external_diff_cmd_cfg;
31 int diff_auto_refresh_index = 1;
32 static int diff_mnemonic_prefix;
33 static int diff_no_prefix;
34 static struct diff_options default_diff_options;
35
36 static char diff_colors[][COLOR_MAXLEN] = {
37 GIT_COLOR_RESET,
38 GIT_COLOR_NORMAL, /* PLAIN */
39 GIT_COLOR_BOLD, /* METAINFO */
40 GIT_COLOR_CYAN, /* FRAGINFO */
41 GIT_COLOR_RED, /* OLD */
42 GIT_COLOR_GREEN, /* NEW */
43 GIT_COLOR_YELLOW, /* COMMIT */
44 GIT_COLOR_BG_RED, /* WHITESPACE */
45 GIT_COLOR_NORMAL, /* FUNCINFO */
46 };
47
48 static int parse_diff_color_slot(const char *var, int ofs)
49 {
50 if (!strcasecmp(var+ofs, "plain"))
51 return DIFF_PLAIN;
52 if (!strcasecmp(var+ofs, "meta"))
53 return DIFF_METAINFO;
54 if (!strcasecmp(var+ofs, "frag"))
55 return DIFF_FRAGINFO;
56 if (!strcasecmp(var+ofs, "old"))
57 return DIFF_FILE_OLD;
58 if (!strcasecmp(var+ofs, "new"))
59 return DIFF_FILE_NEW;
60 if (!strcasecmp(var+ofs, "commit"))
61 return DIFF_COMMIT;
62 if (!strcasecmp(var+ofs, "whitespace"))
63 return DIFF_WHITESPACE;
64 if (!strcasecmp(var+ofs, "func"))
65 return DIFF_FUNCINFO;
66 return -1;
67 }
68
69 static int git_config_rename(const char *var, const char *value)
70 {
71 if (!value)
72 return DIFF_DETECT_RENAME;
73 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
74 return DIFF_DETECT_COPY;
75 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
76 }
77
78 /*
79 * These are to give UI layer defaults.
80 * The core-level commands such as git-diff-files should
81 * never be affected by the setting of diff.renames
82 * the user happens to have in the configuration file.
83 */
84 int git_diff_ui_config(const char *var, const char *value, void *cb)
85 {
86 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
87 diff_use_color_default = git_config_colorbool(var, value, -1);
88 return 0;
89 }
90 if (!strcmp(var, "diff.renames")) {
91 diff_detect_rename_default = git_config_rename(var, value);
92 return 0;
93 }
94 if (!strcmp(var, "diff.autorefreshindex")) {
95 diff_auto_refresh_index = git_config_bool(var, value);
96 return 0;
97 }
98 if (!strcmp(var, "diff.mnemonicprefix")) {
99 diff_mnemonic_prefix = git_config_bool(var, value);
100 return 0;
101 }
102 if (!strcmp(var, "diff.noprefix")) {
103 diff_no_prefix = git_config_bool(var, value);
104 return 0;
105 }
106 if (!strcmp(var, "diff.external"))
107 return git_config_string(&external_diff_cmd_cfg, var, value);
108 if (!strcmp(var, "diff.wordregex"))
109 return git_config_string(&diff_word_regex_cfg, var, value);
110
111 if (!strcmp(var, "diff.ignoresubmodules"))
112 handle_ignore_submodules_arg(&default_diff_options, value);
113
114 return git_diff_basic_config(var, value, cb);
115 }
116
117 int git_diff_basic_config(const char *var, const char *value, void *cb)
118 {
119 if (!strcmp(var, "diff.renamelimit")) {
120 diff_rename_limit_default = git_config_int(var, value);
121 return 0;
122 }
123
124 switch (userdiff_config(var, value)) {
125 case 0: break;
126 case -1: return -1;
127 default: return 0;
128 }
129
130 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
131 int slot = parse_diff_color_slot(var, 11);
132 if (slot < 0)
133 return 0;
134 if (!value)
135 return config_error_nonbool(var);
136 color_parse(value, var, diff_colors[slot]);
137 return 0;
138 }
139
140 /* like GNU diff's --suppress-blank-empty option */
141 if (!strcmp(var, "diff.suppressblankempty") ||
142 /* for backwards compatibility */
143 !strcmp(var, "diff.suppress-blank-empty")) {
144 diff_suppress_blank_empty = git_config_bool(var, value);
145 return 0;
146 }
147
148 if (!prefixcmp(var, "submodule."))
149 return parse_submodule_config_option(var, value);
150
151 return git_color_default_config(var, value, cb);
152 }
153
154 static char *quote_two(const char *one, const char *two)
155 {
156 int need_one = quote_c_style(one, NULL, NULL, 1);
157 int need_two = quote_c_style(two, NULL, NULL, 1);
158 struct strbuf res = STRBUF_INIT;
159
160 if (need_one + need_two) {
161 strbuf_addch(&res, '"');
162 quote_c_style(one, &res, NULL, 1);
163 quote_c_style(two, &res, NULL, 1);
164 strbuf_addch(&res, '"');
165 } else {
166 strbuf_addstr(&res, one);
167 strbuf_addstr(&res, two);
168 }
169 return strbuf_detach(&res, NULL);
170 }
171
172 static const char *external_diff(void)
173 {
174 static const char *external_diff_cmd = NULL;
175 static int done_preparing = 0;
176
177 if (done_preparing)
178 return external_diff_cmd;
179 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
180 if (!external_diff_cmd)
181 external_diff_cmd = external_diff_cmd_cfg;
182 done_preparing = 1;
183 return external_diff_cmd;
184 }
185
186 static struct diff_tempfile {
187 const char *name; /* filename external diff should read from */
188 char hex[41];
189 char mode[10];
190 char tmp_path[PATH_MAX];
191 } diff_temp[2];
192
193 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
194
195 struct emit_callback {
196 int color_diff;
197 unsigned ws_rule;
198 int blank_at_eof_in_preimage;
199 int blank_at_eof_in_postimage;
200 int lno_in_preimage;
201 int lno_in_postimage;
202 sane_truncate_fn truncate;
203 const char **label_path;
204 struct diff_words_data *diff_words;
205 struct diff_options *opt;
206 int *found_changesp;
207 struct strbuf *header;
208 };
209
210 static int count_lines(const char *data, int size)
211 {
212 int count, ch, completely_empty = 1, nl_just_seen = 0;
213 count = 0;
214 while (0 < size--) {
215 ch = *data++;
216 if (ch == '\n') {
217 count++;
218 nl_just_seen = 1;
219 completely_empty = 0;
220 }
221 else {
222 nl_just_seen = 0;
223 completely_empty = 0;
224 }
225 }
226 if (completely_empty)
227 return 0;
228 if (!nl_just_seen)
229 count++; /* no trailing newline */
230 return count;
231 }
232
233 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
234 {
235 if (!DIFF_FILE_VALID(one)) {
236 mf->ptr = (char *)""; /* does not matter */
237 mf->size = 0;
238 return 0;
239 }
240 else if (diff_populate_filespec(one, 0))
241 return -1;
242
243 mf->ptr = one->data;
244 mf->size = one->size;
245 return 0;
246 }
247
248 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
249 {
250 char *ptr = mf->ptr;
251 long size = mf->size;
252 int cnt = 0;
253
254 if (!size)
255 return cnt;
256 ptr += size - 1; /* pointing at the very end */
257 if (*ptr != '\n')
258 ; /* incomplete line */
259 else
260 ptr--; /* skip the last LF */
261 while (mf->ptr < ptr) {
262 char *prev_eol;
263 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
264 if (*prev_eol == '\n')
265 break;
266 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
267 break;
268 cnt++;
269 ptr = prev_eol - 1;
270 }
271 return cnt;
272 }
273
274 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
275 struct emit_callback *ecbdata)
276 {
277 int l1, l2, at;
278 unsigned ws_rule = ecbdata->ws_rule;
279 l1 = count_trailing_blank(mf1, ws_rule);
280 l2 = count_trailing_blank(mf2, ws_rule);
281 if (l2 <= l1) {
282 ecbdata->blank_at_eof_in_preimage = 0;
283 ecbdata->blank_at_eof_in_postimage = 0;
284 return;
285 }
286 at = count_lines(mf1->ptr, mf1->size);
287 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
288
289 at = count_lines(mf2->ptr, mf2->size);
290 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
291 }
292
293 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
294 int first, const char *line, int len)
295 {
296 int has_trailing_newline, has_trailing_carriage_return;
297 int nofirst;
298 FILE *file = o->file;
299
300 if (o->output_prefix) {
301 struct strbuf *msg = NULL;
302 msg = o->output_prefix(o, o->output_prefix_data);
303 assert(msg);
304 fwrite(msg->buf, msg->len, 1, file);
305 }
306
307 if (len == 0) {
308 has_trailing_newline = (first == '\n');
309 has_trailing_carriage_return = (!has_trailing_newline &&
310 (first == '\r'));
311 nofirst = has_trailing_newline || has_trailing_carriage_return;
312 } else {
313 has_trailing_newline = (len > 0 && line[len-1] == '\n');
314 if (has_trailing_newline)
315 len--;
316 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
317 if (has_trailing_carriage_return)
318 len--;
319 nofirst = 0;
320 }
321
322 if (len || !nofirst) {
323 fputs(set, file);
324 if (!nofirst)
325 fputc(first, file);
326 fwrite(line, len, 1, file);
327 fputs(reset, file);
328 }
329 if (has_trailing_carriage_return)
330 fputc('\r', file);
331 if (has_trailing_newline)
332 fputc('\n', file);
333 }
334
335 static void emit_line(struct diff_options *o, const char *set, const char *reset,
336 const char *line, int len)
337 {
338 emit_line_0(o, set, reset, line[0], line+1, len-1);
339 }
340
341 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
342 {
343 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
344 ecbdata->blank_at_eof_in_preimage &&
345 ecbdata->blank_at_eof_in_postimage &&
346 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
347 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
348 return 0;
349 return ws_blank_line(line, len, ecbdata->ws_rule);
350 }
351
352 static void emit_add_line(const char *reset,
353 struct emit_callback *ecbdata,
354 const char *line, int len)
355 {
356 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
357 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
358
359 if (!*ws)
360 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
361 else if (new_blank_line_at_eof(ecbdata, line, len))
362 /* Blank line at EOF - paint '+' as well */
363 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
364 else {
365 /* Emit just the prefix, then the rest. */
366 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
367 ws_check_emit(line, len, ecbdata->ws_rule,
368 ecbdata->opt->file, set, reset, ws);
369 }
370 }
371
372 static void emit_hunk_header(struct emit_callback *ecbdata,
373 const char *line, int len)
374 {
375 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
376 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
377 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
378 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
379 static const char atat[2] = { '@', '@' };
380 const char *cp, *ep;
381 struct strbuf msgbuf = STRBUF_INIT;
382 int org_len = len;
383 int i = 1;
384
385 /*
386 * As a hunk header must begin with "@@ -<old>, +<new> @@",
387 * it always is at least 10 bytes long.
388 */
389 if (len < 10 ||
390 memcmp(line, atat, 2) ||
391 !(ep = memmem(line + 2, len - 2, atat, 2))) {
392 emit_line(ecbdata->opt, plain, reset, line, len);
393 return;
394 }
395 ep += 2; /* skip over @@ */
396
397 /* The hunk header in fraginfo color */
398 strbuf_add(&msgbuf, frag, strlen(frag));
399 strbuf_add(&msgbuf, line, ep - line);
400 strbuf_add(&msgbuf, reset, strlen(reset));
401
402 /*
403 * trailing "\r\n"
404 */
405 for ( ; i < 3; i++)
406 if (line[len - i] == '\r' || line[len - i] == '\n')
407 len--;
408
409 /* blank before the func header */
410 for (cp = ep; ep - line < len; ep++)
411 if (*ep != ' ' && *ep != '\t')
412 break;
413 if (ep != cp) {
414 strbuf_add(&msgbuf, plain, strlen(plain));
415 strbuf_add(&msgbuf, cp, ep - cp);
416 strbuf_add(&msgbuf, reset, strlen(reset));
417 }
418
419 if (ep < line + len) {
420 strbuf_add(&msgbuf, func, strlen(func));
421 strbuf_add(&msgbuf, ep, line + len - ep);
422 strbuf_add(&msgbuf, reset, strlen(reset));
423 }
424
425 strbuf_add(&msgbuf, line + len, org_len - len);
426 emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
427 strbuf_release(&msgbuf);
428 }
429
430 static struct diff_tempfile *claim_diff_tempfile(void) {
431 int i;
432 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
433 if (!diff_temp[i].name)
434 return diff_temp + i;
435 die("BUG: diff is failing to clean up its tempfiles");
436 }
437
438 static int remove_tempfile_installed;
439
440 static void remove_tempfile(void)
441 {
442 int i;
443 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
444 if (diff_temp[i].name == diff_temp[i].tmp_path)
445 unlink_or_warn(diff_temp[i].name);
446 diff_temp[i].name = NULL;
447 }
448 }
449
450 static void remove_tempfile_on_signal(int signo)
451 {
452 remove_tempfile();
453 sigchain_pop(signo);
454 raise(signo);
455 }
456
457 static void print_line_count(FILE *file, int count)
458 {
459 switch (count) {
460 case 0:
461 fprintf(file, "0,0");
462 break;
463 case 1:
464 fprintf(file, "1");
465 break;
466 default:
467 fprintf(file, "1,%d", count);
468 break;
469 }
470 }
471
472 static void emit_rewrite_lines(struct emit_callback *ecb,
473 int prefix, const char *data, int size)
474 {
475 const char *endp = NULL;
476 static const char *nneof = " No newline at end of file\n";
477 const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
478 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
479
480 while (0 < size) {
481 int len;
482
483 endp = memchr(data, '\n', size);
484 len = endp ? (endp - data + 1) : size;
485 if (prefix != '+') {
486 ecb->lno_in_preimage++;
487 emit_line_0(ecb->opt, old, reset, '-',
488 data, len);
489 } else {
490 ecb->lno_in_postimage++;
491 emit_add_line(reset, ecb, data, len);
492 }
493 size -= len;
494 data += len;
495 }
496 if (!endp) {
497 const char *plain = diff_get_color(ecb->color_diff,
498 DIFF_PLAIN);
499 emit_line_0(ecb->opt, plain, reset, '\\',
500 nneof, strlen(nneof));
501 }
502 }
503
504 static void emit_rewrite_diff(const char *name_a,
505 const char *name_b,
506 struct diff_filespec *one,
507 struct diff_filespec *two,
508 struct userdiff_driver *textconv_one,
509 struct userdiff_driver *textconv_two,
510 struct diff_options *o)
511 {
512 int lc_a, lc_b;
513 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
514 const char *name_a_tab, *name_b_tab;
515 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
516 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
517 const char *reset = diff_get_color(color_diff, DIFF_RESET);
518 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
519 const char *a_prefix, *b_prefix;
520 char *data_one, *data_two;
521 size_t size_one, size_two;
522 struct emit_callback ecbdata;
523 char *line_prefix = "";
524 struct strbuf *msgbuf;
525
526 if (o && o->output_prefix) {
527 msgbuf = o->output_prefix(o, o->output_prefix_data);
528 line_prefix = msgbuf->buf;
529 }
530
531 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
532 a_prefix = o->b_prefix;
533 b_prefix = o->a_prefix;
534 } else {
535 a_prefix = o->a_prefix;
536 b_prefix = o->b_prefix;
537 }
538
539 name_a += (*name_a == '/');
540 name_b += (*name_b == '/');
541 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
542 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
543
544 strbuf_reset(&a_name);
545 strbuf_reset(&b_name);
546 quote_two_c_style(&a_name, a_prefix, name_a, 0);
547 quote_two_c_style(&b_name, b_prefix, name_b, 0);
548
549 size_one = fill_textconv(textconv_one, one, &data_one);
550 size_two = fill_textconv(textconv_two, two, &data_two);
551
552 memset(&ecbdata, 0, sizeof(ecbdata));
553 ecbdata.color_diff = color_diff;
554 ecbdata.found_changesp = &o->found_changes;
555 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
556 ecbdata.opt = o;
557 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
558 mmfile_t mf1, mf2;
559 mf1.ptr = (char *)data_one;
560 mf2.ptr = (char *)data_two;
561 mf1.size = size_one;
562 mf2.size = size_two;
563 check_blank_at_eof(&mf1, &mf2, &ecbdata);
564 }
565 ecbdata.lno_in_preimage = 1;
566 ecbdata.lno_in_postimage = 1;
567
568 lc_a = count_lines(data_one, size_one);
569 lc_b = count_lines(data_two, size_two);
570 fprintf(o->file,
571 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
572 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
573 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
574 line_prefix, fraginfo);
575 print_line_count(o->file, lc_a);
576 fprintf(o->file, " +");
577 print_line_count(o->file, lc_b);
578 fprintf(o->file, " @@%s\n", reset);
579 if (lc_a)
580 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
581 if (lc_b)
582 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
583 if (textconv_one)
584 free((char *)data_one);
585 if (textconv_two)
586 free((char *)data_two);
587 }
588
589 struct diff_words_buffer {
590 mmfile_t text;
591 long alloc;
592 struct diff_words_orig {
593 const char *begin, *end;
594 } *orig;
595 int orig_nr, orig_alloc;
596 };
597
598 static void diff_words_append(char *line, unsigned long len,
599 struct diff_words_buffer *buffer)
600 {
601 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
602 line++;
603 len--;
604 memcpy(buffer->text.ptr + buffer->text.size, line, len);
605 buffer->text.size += len;
606 buffer->text.ptr[buffer->text.size] = '\0';
607 }
608
609 struct diff_words_style_elem {
610 const char *prefix;
611 const char *suffix;
612 const char *color; /* NULL; filled in by the setup code if
613 * color is enabled */
614 };
615
616 struct diff_words_style {
617 enum diff_words_type type;
618 struct diff_words_style_elem new, old, ctx;
619 const char *newline;
620 };
621
622 struct diff_words_style diff_words_styles[] = {
623 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
624 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
625 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
626 };
627
628 struct diff_words_data {
629 struct diff_words_buffer minus, plus;
630 const char *current_plus;
631 int last_minus;
632 struct diff_options *opt;
633 regex_t *word_regex;
634 enum diff_words_type type;
635 struct diff_words_style *style;
636 };
637
638 static int fn_out_diff_words_write_helper(FILE *fp,
639 struct diff_words_style_elem *st_el,
640 const char *newline,
641 size_t count, const char *buf,
642 const char *line_prefix)
643 {
644 int print = 0;
645
646 while (count) {
647 char *p = memchr(buf, '\n', count);
648 if (print)
649 fputs(line_prefix, fp);
650 if (p != buf) {
651 if (st_el->color && fputs(st_el->color, fp) < 0)
652 return -1;
653 if (fputs(st_el->prefix, fp) < 0 ||
654 fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
655 fputs(st_el->suffix, fp) < 0)
656 return -1;
657 if (st_el->color && *st_el->color
658 && fputs(GIT_COLOR_RESET, fp) < 0)
659 return -1;
660 }
661 if (!p)
662 return 0;
663 if (fputs(newline, fp) < 0)
664 return -1;
665 count -= p + 1 - buf;
666 buf = p + 1;
667 print = 1;
668 }
669 return 0;
670 }
671
672 /*
673 * '--color-words' algorithm can be described as:
674 *
675 * 1. collect a the minus/plus lines of a diff hunk, divided into
676 * minus-lines and plus-lines;
677 *
678 * 2. break both minus-lines and plus-lines into words and
679 * place them into two mmfile_t with one word for each line;
680 *
681 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
682 *
683 * And for the common parts of the both file, we output the plus side text.
684 * diff_words->current_plus is used to trace the current position of the plus file
685 * which printed. diff_words->last_minus is used to trace the last minus word
686 * printed.
687 *
688 * For '--graph' to work with '--color-words', we need to output the graph prefix
689 * on each line of color words output. Generally, there are two conditions on
690 * which we should output the prefix.
691 *
692 * 1. diff_words->last_minus == 0 &&
693 * diff_words->current_plus == diff_words->plus.text.ptr
694 *
695 * that is: the plus text must start as a new line, and if there is no minus
696 * word printed, a graph prefix must be printed.
697 *
698 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
699 * *(diff_words->current_plus - 1) == '\n'
700 *
701 * that is: a graph prefix must be printed following a '\n'
702 */
703 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
704 {
705 if ((diff_words->last_minus == 0 &&
706 diff_words->current_plus == diff_words->plus.text.ptr) ||
707 (diff_words->current_plus > diff_words->plus.text.ptr &&
708 *(diff_words->current_plus - 1) == '\n')) {
709 return 1;
710 } else {
711 return 0;
712 }
713 }
714
715 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
716 {
717 struct diff_words_data *diff_words = priv;
718 struct diff_words_style *style = diff_words->style;
719 int minus_first, minus_len, plus_first, plus_len;
720 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
721 struct diff_options *opt = diff_words->opt;
722 struct strbuf *msgbuf;
723 char *line_prefix = "";
724
725 if (line[0] != '@' || parse_hunk_header(line, len,
726 &minus_first, &minus_len, &plus_first, &plus_len))
727 return;
728
729 assert(opt);
730 if (opt->output_prefix) {
731 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
732 line_prefix = msgbuf->buf;
733 }
734
735 /* POSIX requires that first be decremented by one if len == 0... */
736 if (minus_len) {
737 minus_begin = diff_words->minus.orig[minus_first].begin;
738 minus_end =
739 diff_words->minus.orig[minus_first + minus_len - 1].end;
740 } else
741 minus_begin = minus_end =
742 diff_words->minus.orig[minus_first].end;
743
744 if (plus_len) {
745 plus_begin = diff_words->plus.orig[plus_first].begin;
746 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
747 } else
748 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
749
750 if (color_words_output_graph_prefix(diff_words)) {
751 fputs(line_prefix, diff_words->opt->file);
752 }
753 if (diff_words->current_plus != plus_begin) {
754 fn_out_diff_words_write_helper(diff_words->opt->file,
755 &style->ctx, style->newline,
756 plus_begin - diff_words->current_plus,
757 diff_words->current_plus, line_prefix);
758 if (*(plus_begin - 1) == '\n')
759 fputs(line_prefix, diff_words->opt->file);
760 }
761 if (minus_begin != minus_end) {
762 fn_out_diff_words_write_helper(diff_words->opt->file,
763 &style->old, style->newline,
764 minus_end - minus_begin, minus_begin,
765 line_prefix);
766 }
767 if (plus_begin != plus_end) {
768 fn_out_diff_words_write_helper(diff_words->opt->file,
769 &style->new, style->newline,
770 plus_end - plus_begin, plus_begin,
771 line_prefix);
772 }
773
774 diff_words->current_plus = plus_end;
775 diff_words->last_minus = minus_first;
776 }
777
778 /* This function starts looking at *begin, and returns 0 iff a word was found. */
779 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
780 int *begin, int *end)
781 {
782 if (word_regex && *begin < buffer->size) {
783 regmatch_t match[1];
784 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
785 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
786 '\n', match[0].rm_eo - match[0].rm_so);
787 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
788 *begin += match[0].rm_so;
789 return *begin >= *end;
790 }
791 return -1;
792 }
793
794 /* find the next word */
795 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
796 (*begin)++;
797 if (*begin >= buffer->size)
798 return -1;
799
800 /* find the end of the word */
801 *end = *begin + 1;
802 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
803 (*end)++;
804
805 return 0;
806 }
807
808 /*
809 * This function splits the words in buffer->text, stores the list with
810 * newline separator into out, and saves the offsets of the original words
811 * in buffer->orig.
812 */
813 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
814 regex_t *word_regex)
815 {
816 int i, j;
817 long alloc = 0;
818
819 out->size = 0;
820 out->ptr = NULL;
821
822 /* fake an empty "0th" word */
823 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
824 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
825 buffer->orig_nr = 1;
826
827 for (i = 0; i < buffer->text.size; i++) {
828 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
829 return;
830
831 /* store original boundaries */
832 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
833 buffer->orig_alloc);
834 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
835 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
836 buffer->orig_nr++;
837
838 /* store one word */
839 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
840 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
841 out->ptr[out->size + j - i] = '\n';
842 out->size += j - i + 1;
843
844 i = j - 1;
845 }
846 }
847
848 /* this executes the word diff on the accumulated buffers */
849 static void diff_words_show(struct diff_words_data *diff_words)
850 {
851 xpparam_t xpp;
852 xdemitconf_t xecfg;
853 mmfile_t minus, plus;
854 struct diff_words_style *style = diff_words->style;
855
856 struct diff_options *opt = diff_words->opt;
857 struct strbuf *msgbuf;
858 char *line_prefix = "";
859
860 assert(opt);
861 if (opt->output_prefix) {
862 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
863 line_prefix = msgbuf->buf;
864 }
865
866 /* special case: only removal */
867 if (!diff_words->plus.text.size) {
868 fputs(line_prefix, diff_words->opt->file);
869 fn_out_diff_words_write_helper(diff_words->opt->file,
870 &style->old, style->newline,
871 diff_words->minus.text.size,
872 diff_words->minus.text.ptr, line_prefix);
873 diff_words->minus.text.size = 0;
874 return;
875 }
876
877 diff_words->current_plus = diff_words->plus.text.ptr;
878 diff_words->last_minus = 0;
879
880 memset(&xpp, 0, sizeof(xpp));
881 memset(&xecfg, 0, sizeof(xecfg));
882 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
883 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
884 xpp.flags = 0;
885 /* as only the hunk header will be parsed, we need a 0-context */
886 xecfg.ctxlen = 0;
887 xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
888 &xpp, &xecfg);
889 free(minus.ptr);
890 free(plus.ptr);
891 if (diff_words->current_plus != diff_words->plus.text.ptr +
892 diff_words->plus.text.size) {
893 if (color_words_output_graph_prefix(diff_words))
894 fputs(line_prefix, diff_words->opt->file);
895 fn_out_diff_words_write_helper(diff_words->opt->file,
896 &style->ctx, style->newline,
897 diff_words->plus.text.ptr + diff_words->plus.text.size
898 - diff_words->current_plus, diff_words->current_plus,
899 line_prefix);
900 }
901 diff_words->minus.text.size = diff_words->plus.text.size = 0;
902 }
903
904 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
905 static void diff_words_flush(struct emit_callback *ecbdata)
906 {
907 if (ecbdata->diff_words->minus.text.size ||
908 ecbdata->diff_words->plus.text.size)
909 diff_words_show(ecbdata->diff_words);
910 }
911
912 static void free_diff_words_data(struct emit_callback *ecbdata)
913 {
914 if (ecbdata->diff_words) {
915 diff_words_flush(ecbdata);
916 free (ecbdata->diff_words->minus.text.ptr);
917 free (ecbdata->diff_words->minus.orig);
918 free (ecbdata->diff_words->plus.text.ptr);
919 free (ecbdata->diff_words->plus.orig);
920 if (ecbdata->diff_words->word_regex) {
921 regfree(ecbdata->diff_words->word_regex);
922 free(ecbdata->diff_words->word_regex);
923 }
924 free(ecbdata->diff_words);
925 ecbdata->diff_words = NULL;
926 }
927 }
928
929 const char *diff_get_color(int diff_use_color, enum color_diff ix)
930 {
931 if (diff_use_color)
932 return diff_colors[ix];
933 return "";
934 }
935
936 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
937 {
938 const char *cp;
939 unsigned long allot;
940 size_t l = len;
941
942 if (ecb->truncate)
943 return ecb->truncate(line, len);
944 cp = line;
945 allot = l;
946 while (0 < l) {
947 (void) utf8_width(&cp, &l);
948 if (!cp)
949 break; /* truncated in the middle? */
950 }
951 return allot - l;
952 }
953
954 static void find_lno(const char *line, struct emit_callback *ecbdata)
955 {
956 const char *p;
957 ecbdata->lno_in_preimage = 0;
958 ecbdata->lno_in_postimage = 0;
959 p = strchr(line, '-');
960 if (!p)
961 return; /* cannot happen */
962 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
963 p = strchr(p, '+');
964 if (!p)
965 return; /* cannot happen */
966 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
967 }
968
969 static void fn_out_consume(void *priv, char *line, unsigned long len)
970 {
971 struct emit_callback *ecbdata = priv;
972 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
973 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
974 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
975 struct diff_options *o = ecbdata->opt;
976 char *line_prefix = "";
977 struct strbuf *msgbuf;
978
979 if (o && o->output_prefix) {
980 msgbuf = o->output_prefix(o, o->output_prefix_data);
981 line_prefix = msgbuf->buf;
982 }
983
984 if (ecbdata->header) {
985 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
986 strbuf_reset(ecbdata->header);
987 ecbdata->header = NULL;
988 }
989 *(ecbdata->found_changesp) = 1;
990
991 if (ecbdata->label_path[0]) {
992 const char *name_a_tab, *name_b_tab;
993
994 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
995 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
996
997 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
998 line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
999 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1000 line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1001 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1002 }
1003
1004 if (diff_suppress_blank_empty
1005 && len == 2 && line[0] == ' ' && line[1] == '\n') {
1006 line[0] = '\n';
1007 len = 1;
1008 }
1009
1010 if (line[0] == '@') {
1011 if (ecbdata->diff_words)
1012 diff_words_flush(ecbdata);
1013 len = sane_truncate_line(ecbdata, line, len);
1014 find_lno(line, ecbdata);
1015 emit_hunk_header(ecbdata, line, len);
1016 if (line[len-1] != '\n')
1017 putc('\n', ecbdata->opt->file);
1018 return;
1019 }
1020
1021 if (len < 1) {
1022 emit_line(ecbdata->opt, reset, reset, line, len);
1023 if (ecbdata->diff_words
1024 && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1025 fputs("~\n", ecbdata->opt->file);
1026 return;
1027 }
1028
1029 if (ecbdata->diff_words) {
1030 if (line[0] == '-') {
1031 diff_words_append(line, len,
1032 &ecbdata->diff_words->minus);
1033 return;
1034 } else if (line[0] == '+') {
1035 diff_words_append(line, len,
1036 &ecbdata->diff_words->plus);
1037 return;
1038 }
1039 diff_words_flush(ecbdata);
1040 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1041 emit_line(ecbdata->opt, plain, reset, line, len);
1042 fputs("~\n", ecbdata->opt->file);
1043 } else {
1044 /* don't print the prefix character */
1045 emit_line(ecbdata->opt, plain, reset, line+1, len-1);
1046 }
1047 return;
1048 }
1049
1050 if (line[0] != '+') {
1051 const char *color =
1052 diff_get_color(ecbdata->color_diff,
1053 line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1054 ecbdata->lno_in_preimage++;
1055 if (line[0] == ' ')
1056 ecbdata->lno_in_postimage++;
1057 emit_line(ecbdata->opt, color, reset, line, len);
1058 } else {
1059 ecbdata->lno_in_postimage++;
1060 emit_add_line(reset, ecbdata, line + 1, len - 1);
1061 }
1062 }
1063
1064 static char *pprint_rename(const char *a, const char *b)
1065 {
1066 const char *old = a;
1067 const char *new = b;
1068 struct strbuf name = STRBUF_INIT;
1069 int pfx_length, sfx_length;
1070 int len_a = strlen(a);
1071 int len_b = strlen(b);
1072 int a_midlen, b_midlen;
1073 int qlen_a = quote_c_style(a, NULL, NULL, 0);
1074 int qlen_b = quote_c_style(b, NULL, NULL, 0);
1075
1076 if (qlen_a || qlen_b) {
1077 quote_c_style(a, &name, NULL, 0);
1078 strbuf_addstr(&name, " => ");
1079 quote_c_style(b, &name, NULL, 0);
1080 return strbuf_detach(&name, NULL);
1081 }
1082
1083 /* Find common prefix */
1084 pfx_length = 0;
1085 while (*old && *new && *old == *new) {
1086 if (*old == '/')
1087 pfx_length = old - a + 1;
1088 old++;
1089 new++;
1090 }
1091
1092 /* Find common suffix */
1093 old = a + len_a;
1094 new = b + len_b;
1095 sfx_length = 0;
1096 while (a <= old && b <= new && *old == *new) {
1097 if (*old == '/')
1098 sfx_length = len_a - (old - a);
1099 old--;
1100 new--;
1101 }
1102
1103 /*
1104 * pfx{mid-a => mid-b}sfx
1105 * {pfx-a => pfx-b}sfx
1106 * pfx{sfx-a => sfx-b}
1107 * name-a => name-b
1108 */
1109 a_midlen = len_a - pfx_length - sfx_length;
1110 b_midlen = len_b - pfx_length - sfx_length;
1111 if (a_midlen < 0)
1112 a_midlen = 0;
1113 if (b_midlen < 0)
1114 b_midlen = 0;
1115
1116 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1117 if (pfx_length + sfx_length) {
1118 strbuf_add(&name, a, pfx_length);
1119 strbuf_addch(&name, '{');
1120 }
1121 strbuf_add(&name, a + pfx_length, a_midlen);
1122 strbuf_addstr(&name, " => ");
1123 strbuf_add(&name, b + pfx_length, b_midlen);
1124 if (pfx_length + sfx_length) {
1125 strbuf_addch(&name, '}');
1126 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1127 }
1128 return strbuf_detach(&name, NULL);
1129 }
1130
1131 struct diffstat_t {
1132 int nr;
1133 int alloc;
1134 struct diffstat_file {
1135 char *from_name;
1136 char *name;
1137 char *print_name;
1138 unsigned is_unmerged:1;
1139 unsigned is_binary:1;
1140 unsigned is_renamed:1;
1141 uintmax_t added, deleted;
1142 } **files;
1143 };
1144
1145 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1146 const char *name_a,
1147 const char *name_b)
1148 {
1149 struct diffstat_file *x;
1150 x = xcalloc(sizeof (*x), 1);
1151 if (diffstat->nr == diffstat->alloc) {
1152 diffstat->alloc = alloc_nr(diffstat->alloc);
1153 diffstat->files = xrealloc(diffstat->files,
1154 diffstat->alloc * sizeof(x));
1155 }
1156 diffstat->files[diffstat->nr++] = x;
1157 if (name_b) {
1158 x->from_name = xstrdup(name_a);
1159 x->name = xstrdup(name_b);
1160 x->is_renamed = 1;
1161 }
1162 else {
1163 x->from_name = NULL;
1164 x->name = xstrdup(name_a);
1165 }
1166 return x;
1167 }
1168
1169 static void diffstat_consume(void *priv, char *line, unsigned long len)
1170 {
1171 struct diffstat_t *diffstat = priv;
1172 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1173
1174 if (line[0] == '+')
1175 x->added++;
1176 else if (line[0] == '-')
1177 x->deleted++;
1178 }
1179
1180 const char mime_boundary_leader[] = "------------";
1181
1182 static int scale_linear(int it, int width, int max_change)
1183 {
1184 /*
1185 * make sure that at least one '-' is printed if there were deletions,
1186 * and likewise for '+'.
1187 */
1188 if (max_change < 2)
1189 return it;
1190 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
1191 }
1192
1193 static void show_name(FILE *file,
1194 const char *prefix, const char *name, int len)
1195 {
1196 fprintf(file, " %s%-*s |", prefix, len, name);
1197 }
1198
1199 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1200 {
1201 if (cnt <= 0)
1202 return;
1203 fprintf(file, "%s", set);
1204 while (cnt--)
1205 putc(ch, file);
1206 fprintf(file, "%s", reset);
1207 }
1208
1209 static void fill_print_name(struct diffstat_file *file)
1210 {
1211 char *pname;
1212
1213 if (file->print_name)
1214 return;
1215
1216 if (!file->is_renamed) {
1217 struct strbuf buf = STRBUF_INIT;
1218 if (quote_c_style(file->name, &buf, NULL, 0)) {
1219 pname = strbuf_detach(&buf, NULL);
1220 } else {
1221 pname = file->name;
1222 strbuf_release(&buf);
1223 }
1224 } else {
1225 pname = pprint_rename(file->from_name, file->name);
1226 }
1227 file->print_name = pname;
1228 }
1229
1230 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1231 {
1232 int i, len, add, del, adds = 0, dels = 0;
1233 uintmax_t max_change = 0, max_len = 0;
1234 int total_files = data->nr;
1235 int width, name_width;
1236 const char *reset, *set, *add_c, *del_c;
1237 const char *line_prefix = "";
1238 struct strbuf *msg = NULL;
1239
1240 if (data->nr == 0)
1241 return;
1242
1243 if (options->output_prefix) {
1244 msg = options->output_prefix(options, options->output_prefix_data);
1245 line_prefix = msg->buf;
1246 }
1247
1248 width = options->stat_width ? options->stat_width : 80;
1249 name_width = options->stat_name_width ? options->stat_name_width : 50;
1250
1251 /* Sanity: give at least 5 columns to the graph,
1252 * but leave at least 10 columns for the name.
1253 */
1254 if (width < 25)
1255 width = 25;
1256 if (name_width < 10)
1257 name_width = 10;
1258 else if (width < name_width + 15)
1259 name_width = width - 15;
1260
1261 /* Find the longest filename and max number of changes */
1262 reset = diff_get_color_opt(options, DIFF_RESET);
1263 set = diff_get_color_opt(options, DIFF_PLAIN);
1264 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1265 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1266
1267 for (i = 0; i < data->nr; i++) {
1268 struct diffstat_file *file = data->files[i];
1269 uintmax_t change = file->added + file->deleted;
1270 fill_print_name(file);
1271 len = strlen(file->print_name);
1272 if (max_len < len)
1273 max_len = len;
1274
1275 if (file->is_binary || file->is_unmerged)
1276 continue;
1277 if (max_change < change)
1278 max_change = change;
1279 }
1280
1281 /* Compute the width of the graph part;
1282 * 10 is for one blank at the beginning of the line plus
1283 * " | count " between the name and the graph.
1284 *
1285 * From here on, name_width is the width of the name area,
1286 * and width is the width of the graph area.
1287 */
1288 name_width = (name_width < max_len) ? name_width : max_len;
1289 if (width < (name_width + 10) + max_change)
1290 width = width - (name_width + 10);
1291 else
1292 width = max_change;
1293
1294 for (i = 0; i < data->nr; i++) {
1295 const char *prefix = "";
1296 char *name = data->files[i]->print_name;
1297 uintmax_t added = data->files[i]->added;
1298 uintmax_t deleted = data->files[i]->deleted;
1299 int name_len;
1300
1301 /*
1302 * "scale" the filename
1303 */
1304 len = name_width;
1305 name_len = strlen(name);
1306 if (name_width < name_len) {
1307 char *slash;
1308 prefix = "...";
1309 len -= 3;
1310 name += name_len - len;
1311 slash = strchr(name, '/');
1312 if (slash)
1313 name = slash;
1314 }
1315
1316 if (data->files[i]->is_binary) {
1317 fprintf(options->file, "%s", line_prefix);
1318 show_name(options->file, prefix, name, len);
1319 fprintf(options->file, " Bin ");
1320 fprintf(options->file, "%s%"PRIuMAX"%s",
1321 del_c, deleted, reset);
1322 fprintf(options->file, " -> ");
1323 fprintf(options->file, "%s%"PRIuMAX"%s",
1324 add_c, added, reset);
1325 fprintf(options->file, " bytes");
1326 fprintf(options->file, "\n");
1327 continue;
1328 }
1329 else if (data->files[i]->is_unmerged) {
1330 fprintf(options->file, "%s", line_prefix);
1331 show_name(options->file, prefix, name, len);
1332 fprintf(options->file, " Unmerged\n");
1333 continue;
1334 }
1335 else if (!data->files[i]->is_renamed &&
1336 (added + deleted == 0)) {
1337 total_files--;
1338 continue;
1339 }
1340
1341 /*
1342 * scale the add/delete
1343 */
1344 add = added;
1345 del = deleted;
1346 adds += add;
1347 dels += del;
1348
1349 if (width <= max_change) {
1350 add = scale_linear(add, width, max_change);
1351 del = scale_linear(del, width, max_change);
1352 }
1353 fprintf(options->file, "%s", line_prefix);
1354 show_name(options->file, prefix, name, len);
1355 fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
1356 added + deleted ? " " : "");
1357 show_graph(options->file, '+', add, add_c, reset);
1358 show_graph(options->file, '-', del, del_c, reset);
1359 fprintf(options->file, "\n");
1360 }
1361 fprintf(options->file, "%s", line_prefix);
1362 fprintf(options->file,
1363 " %d files changed, %d insertions(+), %d deletions(-)\n",
1364 total_files, adds, dels);
1365 }
1366
1367 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1368 {
1369 int i, adds = 0, dels = 0, total_files = data->nr;
1370
1371 if (data->nr == 0)
1372 return;
1373
1374 for (i = 0; i < data->nr; i++) {
1375 if (!data->files[i]->is_binary &&
1376 !data->files[i]->is_unmerged) {
1377 int added = data->files[i]->added;
1378 int deleted= data->files[i]->deleted;
1379 if (!data->files[i]->is_renamed &&
1380 (added + deleted == 0)) {
1381 total_files--;
1382 } else {
1383 adds += added;
1384 dels += deleted;
1385 }
1386 }
1387 }
1388 if (options->output_prefix) {
1389 struct strbuf *msg = NULL;
1390 msg = options->output_prefix(options,
1391 options->output_prefix_data);
1392 fprintf(options->file, "%s", msg->buf);
1393 }
1394 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
1395 total_files, adds, dels);
1396 }
1397
1398 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1399 {
1400 int i;
1401
1402 if (data->nr == 0)
1403 return;
1404
1405 for (i = 0; i < data->nr; i++) {
1406 struct diffstat_file *file = data->files[i];
1407
1408 if (options->output_prefix) {
1409 struct strbuf *msg = NULL;
1410 msg = options->output_prefix(options,
1411 options->output_prefix_data);
1412 fprintf(options->file, "%s", msg->buf);
1413 }
1414
1415 if (file->is_binary)
1416 fprintf(options->file, "-\t-\t");
1417 else
1418 fprintf(options->file,
1419 "%"PRIuMAX"\t%"PRIuMAX"\t",
1420 file->added, file->deleted);
1421 if (options->line_termination) {
1422 fill_print_name(file);
1423 if (!file->is_renamed)
1424 write_name_quoted(file->name, options->file,
1425 options->line_termination);
1426 else {
1427 fputs(file->print_name, options->file);
1428 putc(options->line_termination, options->file);
1429 }
1430 } else {
1431 if (file->is_renamed) {
1432 putc('\0', options->file);
1433 write_name_quoted(file->from_name, options->file, '\0');
1434 }
1435 write_name_quoted(file->name, options->file, '\0');
1436 }
1437 }
1438 }
1439
1440 struct dirstat_file {
1441 const char *name;
1442 unsigned long changed;
1443 };
1444
1445 struct dirstat_dir {
1446 struct dirstat_file *files;
1447 int alloc, nr, percent, cumulative;
1448 };
1449
1450 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1451 unsigned long changed, const char *base, int baselen)
1452 {
1453 unsigned long this_dir = 0;
1454 unsigned int sources = 0;
1455 const char *line_prefix = "";
1456 struct strbuf *msg = NULL;
1457
1458 if (opt->output_prefix) {
1459 msg = opt->output_prefix(opt, opt->output_prefix_data);
1460 line_prefix = msg->buf;
1461 }
1462
1463 while (dir->nr) {
1464 struct dirstat_file *f = dir->files;
1465 int namelen = strlen(f->name);
1466 unsigned long this;
1467 char *slash;
1468
1469 if (namelen < baselen)
1470 break;
1471 if (memcmp(f->name, base, baselen))
1472 break;
1473 slash = strchr(f->name + baselen, '/');
1474 if (slash) {
1475 int newbaselen = slash + 1 - f->name;
1476 this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1477 sources++;
1478 } else {
1479 this = f->changed;
1480 dir->files++;
1481 dir->nr--;
1482 sources += 2;
1483 }
1484 this_dir += this;
1485 }
1486
1487 /*
1488 * We don't report dirstat's for
1489 * - the top level
1490 * - or cases where everything came from a single directory
1491 * under this directory (sources == 1).
1492 */
1493 if (baselen && sources != 1) {
1494 int permille = this_dir * 1000 / changed;
1495 if (permille) {
1496 int percent = permille / 10;
1497 if (percent >= dir->percent) {
1498 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1499 percent, permille % 10, baselen, base);
1500 if (!dir->cumulative)
1501 return 0;
1502 }
1503 }
1504 }
1505 return this_dir;
1506 }
1507
1508 static int dirstat_compare(const void *_a, const void *_b)
1509 {
1510 const struct dirstat_file *a = _a;
1511 const struct dirstat_file *b = _b;
1512 return strcmp(a->name, b->name);
1513 }
1514
1515 static void show_dirstat(struct diff_options *options)
1516 {
1517 int i;
1518 unsigned long changed;
1519 struct dirstat_dir dir;
1520 struct diff_queue_struct *q = &diff_queued_diff;
1521
1522 dir.files = NULL;
1523 dir.alloc = 0;
1524 dir.nr = 0;
1525 dir.percent = options->dirstat_percent;
1526 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1527
1528 changed = 0;
1529 for (i = 0; i < q->nr; i++) {
1530 struct diff_filepair *p = q->queue[i];
1531 const char *name;
1532 unsigned long copied, added, damage;
1533
1534 name = p->one->path ? p->one->path : p->two->path;
1535
1536 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1537 diff_populate_filespec(p->one, 0);
1538 diff_populate_filespec(p->two, 0);
1539 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1540 &copied, &added);
1541 diff_free_filespec_data(p->one);
1542 diff_free_filespec_data(p->two);
1543 } else if (DIFF_FILE_VALID(p->one)) {
1544 diff_populate_filespec(p->one, 1);
1545 copied = added = 0;
1546 diff_free_filespec_data(p->one);
1547 } else if (DIFF_FILE_VALID(p->two)) {
1548 diff_populate_filespec(p->two, 1);
1549 copied = 0;
1550 added = p->two->size;
1551 diff_free_filespec_data(p->two);
1552 } else
1553 continue;
1554
1555 /*
1556 * Original minus copied is the removed material,
1557 * added is the new material. They are both damages
1558 * made to the preimage. In --dirstat-by-file mode, count
1559 * damaged files, not damaged lines. This is done by
1560 * counting only a single damaged line per file.
1561 */
1562 damage = (p->one->size - copied) + added;
1563 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1564 damage = 1;
1565
1566 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1567 dir.files[dir.nr].name = name;
1568 dir.files[dir.nr].changed = damage;
1569 changed += damage;
1570 dir.nr++;
1571 }
1572
1573 /* This can happen even with many files, if everything was renames */
1574 if (!changed)
1575 return;
1576
1577 /* Show all directories with more than x% of the changes */
1578 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1579 gather_dirstat(options, &dir, changed, "", 0);
1580 }
1581
1582 static void free_diffstat_info(struct diffstat_t *diffstat)
1583 {
1584 int i;
1585 for (i = 0; i < diffstat->nr; i++) {
1586 struct diffstat_file *f = diffstat->files[i];
1587 if (f->name != f->print_name)
1588 free(f->print_name);
1589 free(f->name);
1590 free(f->from_name);
1591 free(f);
1592 }
1593 free(diffstat->files);
1594 }
1595
1596 struct checkdiff_t {
1597 const char *filename;
1598 int lineno;
1599 int conflict_marker_size;
1600 struct diff_options *o;
1601 unsigned ws_rule;
1602 unsigned status;
1603 };
1604
1605 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
1606 {
1607 char firstchar;
1608 int cnt;
1609
1610 if (len < marker_size + 1)
1611 return 0;
1612 firstchar = line[0];
1613 switch (firstchar) {
1614 case '=': case '>': case '<': case '|':
1615 break;
1616 default:
1617 return 0;
1618 }
1619 for (cnt = 1; cnt < marker_size; cnt++)
1620 if (line[cnt] != firstchar)
1621 return 0;
1622 /* line[1] thru line[marker_size-1] are same as firstchar */
1623 if (len < marker_size + 1 || !isspace(line[marker_size]))
1624 return 0;
1625 return 1;
1626 }
1627
1628 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1629 {
1630 struct checkdiff_t *data = priv;
1631 int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1632 int marker_size = data->conflict_marker_size;
1633 const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1634 const char *reset = diff_get_color(color_diff, DIFF_RESET);
1635 const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1636 char *err;
1637 char *line_prefix = "";
1638 struct strbuf *msgbuf;
1639
1640 assert(data->o);
1641 if (data->o->output_prefix) {
1642 msgbuf = data->o->output_prefix(data->o,
1643 data->o->output_prefix_data);
1644 line_prefix = msgbuf->buf;
1645 }
1646
1647 if (line[0] == '+') {
1648 unsigned bad;
1649 data->lineno++;
1650 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
1651 data->status |= 1;
1652 fprintf(data->o->file,
1653 "%s%s:%d: leftover conflict marker\n",
1654 line_prefix, data->filename, data->lineno);
1655 }
1656 bad = ws_check(line + 1, len - 1, data->ws_rule);
1657 if (!bad)
1658 return;
1659 data->status |= bad;
1660 err = whitespace_error_string(bad);
1661 fprintf(data->o->file, "%s%s:%d: %s.\n",
1662 line_prefix, data->filename, data->lineno, err);
1663 free(err);
1664 emit_line(data->o, set, reset, line, 1);
1665 ws_check_emit(line + 1, len - 1, data->ws_rule,
1666 data->o->file, set, reset, ws);
1667 } else if (line[0] == ' ') {
1668 data->lineno++;
1669 } else if (line[0] == '@') {
1670 char *plus = strchr(line, '+');
1671 if (plus)
1672 data->lineno = strtol(plus, NULL, 10) - 1;
1673 else
1674 die("invalid diff");
1675 }
1676 }
1677
1678 static unsigned char *deflate_it(char *data,
1679 unsigned long size,
1680 unsigned long *result_size)
1681 {
1682 int bound;
1683 unsigned char *deflated;
1684 z_stream stream;
1685
1686 memset(&stream, 0, sizeof(stream));
1687 deflateInit(&stream, zlib_compression_level);
1688 bound = deflateBound(&stream, size);
1689 deflated = xmalloc(bound);
1690 stream.next_out = deflated;
1691 stream.avail_out = bound;
1692
1693 stream.next_in = (unsigned char *)data;
1694 stream.avail_in = size;
1695 while (deflate(&stream, Z_FINISH) == Z_OK)
1696 ; /* nothing */
1697 deflateEnd(&stream);
1698 *result_size = stream.total_out;
1699 return deflated;
1700 }
1701
1702 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1703 {
1704 void *cp;
1705 void *delta;
1706 void *deflated;
1707 void *data;
1708 unsigned long orig_size;
1709 unsigned long delta_size;
1710 unsigned long deflate_size;
1711 unsigned long data_size;
1712
1713 /* We could do deflated delta, or we could do just deflated two,
1714 * whichever is smaller.
1715 */
1716 delta = NULL;
1717 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1718 if (one->size && two->size) {
1719 delta = diff_delta(one->ptr, one->size,
1720 two->ptr, two->size,
1721 &delta_size, deflate_size);
1722 if (delta) {
1723 void *to_free = delta;
1724 orig_size = delta_size;
1725 delta = deflate_it(delta, delta_size, &delta_size);
1726 free(to_free);
1727 }
1728 }
1729
1730 if (delta && delta_size < deflate_size) {
1731 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
1732 free(deflated);
1733 data = delta;
1734 data_size = delta_size;
1735 }
1736 else {
1737 fprintf(file, "%sliteral %lu\n", prefix, two->size);
1738 free(delta);
1739 data = deflated;
1740 data_size = deflate_size;
1741 }
1742
1743 /* emit data encoded in base85 */
1744 cp = data;
1745 while (data_size) {
1746 int bytes = (52 < data_size) ? 52 : data_size;
1747 char line[70];
1748 data_size -= bytes;
1749 if (bytes <= 26)
1750 line[0] = bytes + 'A' - 1;
1751 else
1752 line[0] = bytes - 26 + 'a' - 1;
1753 encode_85(line + 1, cp, bytes);
1754 cp = (char *) cp + bytes;
1755 fprintf(file, "%s", prefix);
1756 fputs(line, file);
1757 fputc('\n', file);
1758 }
1759 fprintf(file, "%s\n", prefix);
1760 free(data);
1761 }
1762
1763 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
1764 {
1765 fprintf(file, "%sGIT binary patch\n", prefix);
1766 emit_binary_diff_body(file, one, two, prefix);
1767 emit_binary_diff_body(file, two, one, prefix);
1768 }
1769
1770 static void diff_filespec_load_driver(struct diff_filespec *one)
1771 {
1772 /* Use already-loaded driver */
1773 if (one->driver)
1774 return;
1775
1776 if (S_ISREG(one->mode))
1777 one->driver = userdiff_find_by_path(one->path);
1778
1779 /* Fallback to default settings */
1780 if (!one->driver)
1781 one->driver = userdiff_find_by_name("default");
1782 }
1783
1784 int diff_filespec_is_binary(struct diff_filespec *one)
1785 {
1786 if (one->is_binary == -1) {
1787 diff_filespec_load_driver(one);
1788 if (one->driver->binary != -1)
1789 one->is_binary = one->driver->binary;
1790 else {
1791 if (!one->data && DIFF_FILE_VALID(one))
1792 diff_populate_filespec(one, 0);
1793 if (one->data)
1794 one->is_binary = buffer_is_binary(one->data,
1795 one->size);
1796 if (one->is_binary == -1)
1797 one->is_binary = 0;
1798 }
1799 }
1800 return one->is_binary;
1801 }
1802
1803 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1804 {
1805 diff_filespec_load_driver(one);
1806 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1807 }
1808
1809 static const char *userdiff_word_regex(struct diff_filespec *one)
1810 {
1811 diff_filespec_load_driver(one);
1812 return one->driver->word_regex;
1813 }
1814
1815 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1816 {
1817 if (!options->a_prefix)
1818 options->a_prefix = a;
1819 if (!options->b_prefix)
1820 options->b_prefix = b;
1821 }
1822
1823 struct userdiff_driver *get_textconv(struct diff_filespec *one)
1824 {
1825 if (!DIFF_FILE_VALID(one))
1826 return NULL;
1827
1828 diff_filespec_load_driver(one);
1829 if (!one->driver->textconv)
1830 return NULL;
1831
1832 if (one->driver->textconv_want_cache && !one->driver->textconv_cache) {
1833 struct notes_cache *c = xmalloc(sizeof(*c));
1834 struct strbuf name = STRBUF_INIT;
1835
1836 strbuf_addf(&name, "textconv/%s", one->driver->name);
1837 notes_cache_init(c, name.buf, one->driver->textconv);
1838 one->driver->textconv_cache = c;
1839 }
1840
1841 return one->driver;
1842 }
1843
1844 static void builtin_diff(const char *name_a,
1845 const char *name_b,
1846 struct diff_filespec *one,
1847 struct diff_filespec *two,
1848 const char *xfrm_msg,
1849 int must_show_header,
1850 struct diff_options *o,
1851 int complete_rewrite)
1852 {
1853 mmfile_t mf1, mf2;
1854 const char *lbl[2];
1855 char *a_one, *b_two;
1856 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1857 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1858 const char *a_prefix, *b_prefix;
1859 struct userdiff_driver *textconv_one = NULL;
1860 struct userdiff_driver *textconv_two = NULL;
1861 struct strbuf header = STRBUF_INIT;
1862 struct strbuf *msgbuf;
1863 char *line_prefix = "";
1864
1865 if (o->output_prefix) {
1866 msgbuf = o->output_prefix(o, o->output_prefix_data);
1867 line_prefix = msgbuf->buf;
1868 }
1869
1870 if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
1871 (!one->mode || S_ISGITLINK(one->mode)) &&
1872 (!two->mode || S_ISGITLINK(two->mode))) {
1873 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
1874 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
1875 show_submodule_summary(o->file, one ? one->path : two->path,
1876 one->sha1, two->sha1, two->dirty_submodule,
1877 del, add, reset);
1878 return;
1879 }
1880
1881 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1882 textconv_one = get_textconv(one);
1883 textconv_two = get_textconv(two);
1884 }
1885
1886 diff_set_mnemonic_prefix(o, "a/", "b/");
1887 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1888 a_prefix = o->b_prefix;
1889 b_prefix = o->a_prefix;
1890 } else {
1891 a_prefix = o->a_prefix;
1892 b_prefix = o->b_prefix;
1893 }
1894
1895 /* Never use a non-valid filename anywhere if at all possible */
1896 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1897 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1898
1899 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1900 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1901 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1902 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1903 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
1904 if (lbl[0][0] == '/') {
1905 /* /dev/null */
1906 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
1907 if (xfrm_msg)
1908 strbuf_addstr(&header, xfrm_msg);
1909 must_show_header = 1;
1910 }
1911 else if (lbl[1][0] == '/') {
1912 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
1913 if (xfrm_msg)
1914 strbuf_addstr(&header, xfrm_msg);
1915 must_show_header = 1;
1916 }
1917 else {
1918 if (one->mode != two->mode) {
1919 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
1920 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
1921 must_show_header = 1;
1922 }
1923 if (xfrm_msg)
1924 strbuf_addstr(&header, xfrm_msg);
1925
1926 /*
1927 * we do not run diff between different kind
1928 * of objects.
1929 */
1930 if ((one->mode ^ two->mode) & S_IFMT)
1931 goto free_ab_and_return;
1932 if (complete_rewrite &&
1933 (textconv_one || !diff_filespec_is_binary(one)) &&
1934 (textconv_two || !diff_filespec_is_binary(two))) {
1935 fprintf(o->file, "%s", header.buf);
1936 strbuf_reset(&header);
1937 emit_rewrite_diff(name_a, name_b, one, two,
1938 textconv_one, textconv_two, o);
1939 o->found_changes = 1;
1940 goto free_ab_and_return;
1941 }
1942 }
1943
1944 if (!DIFF_OPT_TST(o, TEXT) &&
1945 ( (!textconv_one && diff_filespec_is_binary(one)) ||
1946 (!textconv_two && diff_filespec_is_binary(two)) )) {
1947 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1948 die("unable to read files to diff");
1949 /* Quite common confusing case */
1950 if (mf1.size == mf2.size &&
1951 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
1952 if (must_show_header)
1953 fprintf(o->file, "%s", header.buf);
1954 goto free_ab_and_return;
1955 }
1956 fprintf(o->file, "%s", header.buf);
1957 strbuf_reset(&header);
1958 if (DIFF_OPT_TST(o, BINARY))
1959 emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
1960 else
1961 fprintf(o->file, "%sBinary files %s and %s differ\n",
1962 line_prefix, lbl[0], lbl[1]);
1963 o->found_changes = 1;
1964 }
1965 else {
1966 /* Crazy xdl interfaces.. */
1967 const char *diffopts = getenv("GIT_DIFF_OPTS");
1968 xpparam_t xpp;
1969 xdemitconf_t xecfg;
1970 struct emit_callback ecbdata;
1971 const struct userdiff_funcname *pe;
1972
1973 if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || must_show_header) {
1974 fprintf(o->file, "%s", header.buf);
1975 strbuf_reset(&header);
1976 }
1977
1978 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
1979 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
1980
1981 pe = diff_funcname_pattern(one);
1982 if (!pe)
1983 pe = diff_funcname_pattern(two);
1984
1985 memset(&xpp, 0, sizeof(xpp));
1986 memset(&xecfg, 0, sizeof(xecfg));
1987 memset(&ecbdata, 0, sizeof(ecbdata));
1988 ecbdata.label_path = lbl;
1989 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1990 ecbdata.found_changesp = &o->found_changes;
1991 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1992 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
1993 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1994 ecbdata.opt = o;
1995 ecbdata.header = header.len ? &header : NULL;
1996 xpp.flags = o->xdl_opts;
1997 xecfg.ctxlen = o->context;
1998 xecfg.interhunkctxlen = o->interhunkcontext;
1999 xecfg.flags = XDL_EMIT_FUNCNAMES;
2000 if (pe)
2001 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2002 if (!diffopts)
2003 ;
2004 else if (!prefixcmp(diffopts, "--unified="))
2005 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
2006 else if (!prefixcmp(diffopts, "-u"))
2007 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
2008 if (o->word_diff) {
2009 int i;
2010
2011 ecbdata.diff_words =
2012 xcalloc(1, sizeof(struct diff_words_data));
2013 ecbdata.diff_words->type = o->word_diff;
2014 ecbdata.diff_words->opt = o;
2015 if (!o->word_regex)
2016 o->word_regex = userdiff_word_regex(one);
2017 if (!o->word_regex)
2018 o->word_regex = userdiff_word_regex(two);
2019 if (!o->word_regex)
2020 o->word_regex = diff_word_regex_cfg;
2021 if (o->word_regex) {
2022 ecbdata.diff_words->word_regex = (regex_t *)
2023 xmalloc(sizeof(regex_t));
2024 if (regcomp(ecbdata.diff_words->word_regex,
2025 o->word_regex,
2026 REG_EXTENDED | REG_NEWLINE))
2027 die ("Invalid regular expression: %s",
2028 o->word_regex);
2029 }
2030 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2031 if (o->word_diff == diff_words_styles[i].type) {
2032 ecbdata.diff_words->style =
2033 &diff_words_styles[i];
2034 break;
2035 }
2036 }
2037 if (DIFF_OPT_TST(o, COLOR_DIFF)) {
2038 struct diff_words_style *st = ecbdata.diff_words->style;
2039 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2040 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2041 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
2042 }
2043 }
2044 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2045 &xpp, &xecfg);
2046 if (o->word_diff)
2047 free_diff_words_data(&ecbdata);
2048 if (textconv_one)
2049 free(mf1.ptr);
2050 if (textconv_two)
2051 free(mf2.ptr);
2052 xdiff_clear_find_func(&xecfg);
2053 }
2054
2055 free_ab_and_return:
2056 strbuf_release(&header);
2057 diff_free_filespec_data(one);
2058 diff_free_filespec_data(two);
2059 free(a_one);
2060 free(b_two);
2061 return;
2062 }
2063
2064 static void builtin_diffstat(const char *name_a, const char *name_b,
2065 struct diff_filespec *one,
2066 struct diff_filespec *two,
2067 struct diffstat_t *diffstat,
2068 struct diff_options *o,
2069 int complete_rewrite)
2070 {
2071 mmfile_t mf1, mf2;
2072 struct diffstat_file *data;
2073
2074 data = diffstat_add(diffstat, name_a, name_b);
2075
2076 if (!one || !two) {
2077 data->is_unmerged = 1;
2078 return;
2079 }
2080 if (complete_rewrite) {
2081 diff_populate_filespec(one, 0);
2082 diff_populate_filespec(two, 0);
2083 data->deleted = count_lines(one->data, one->size);
2084 data->added = count_lines(two->data, two->size);
2085 goto free_and_return;
2086 }
2087 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2088 die("unable to read files to diff");
2089
2090 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2091 data->is_binary = 1;
2092 data->added = mf2.size;
2093 data->deleted = mf1.size;
2094 } else {
2095 /* Crazy xdl interfaces.. */
2096 xpparam_t xpp;
2097 xdemitconf_t xecfg;
2098
2099 memset(&xpp, 0, sizeof(xpp));
2100 memset(&xecfg, 0, sizeof(xecfg));
2101 xpp.flags = o->xdl_opts;
2102 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2103 &xpp, &xecfg);
2104 }
2105
2106 free_and_return:
2107 diff_free_filespec_data(one);
2108 diff_free_filespec_data(two);
2109 }
2110
2111 static void builtin_checkdiff(const char *name_a, const char *name_b,
2112 const char *attr_path,
2113 struct diff_filespec *one,
2114 struct diff_filespec *two,
2115 struct diff_options *o)
2116 {
2117 mmfile_t mf1, mf2;
2118 struct checkdiff_t data;
2119
2120 if (!two)
2121 return;
2122
2123 memset(&data, 0, sizeof(data));
2124 data.filename = name_b ? name_b : name_a;
2125 data.lineno = 0;
2126 data.o = o;
2127 data.ws_rule = whitespace_rule(attr_path);
2128 data.conflict_marker_size = ll_merge_marker_size(attr_path);
2129
2130 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2131 die("unable to read files to diff");
2132
2133 /*
2134 * All the other codepaths check both sides, but not checking
2135 * the "old" side here is deliberate. We are checking the newly
2136 * introduced changes, and as long as the "new" side is text, we
2137 * can and should check what it introduces.
2138 */
2139 if (diff_filespec_is_binary(two))
2140 goto free_and_return;
2141 else {
2142 /* Crazy xdl interfaces.. */
2143 xpparam_t xpp;
2144 xdemitconf_t xecfg;
2145
2146 memset(&xpp, 0, sizeof(xpp));
2147 memset(&xecfg, 0, sizeof(xecfg));
2148 xecfg.ctxlen = 1; /* at least one context line */
2149 xpp.flags = 0;
2150 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2151 &xpp, &xecfg);
2152
2153 if (data.ws_rule & WS_BLANK_AT_EOF) {
2154 struct emit_callback ecbdata;
2155 int blank_at_eof;
2156
2157 ecbdata.ws_rule = data.ws_rule;
2158 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2159 blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2160
2161 if (blank_at_eof) {
2162 static char *err;
2163 if (!err)
2164 err = whitespace_error_string(WS_BLANK_AT_EOF);
2165 fprintf(o->file, "%s:%d: %s.\n",
2166 data.filename, blank_at_eof, err);
2167 data.status = 1; /* report errors */
2168 }
2169 }
2170 }
2171 free_and_return:
2172 diff_free_filespec_data(one);
2173 diff_free_filespec_data(two);
2174 if (data.status)
2175 DIFF_OPT_SET(o, CHECK_FAILED);
2176 }
2177
2178 struct diff_filespec *alloc_filespec(const char *path)
2179 {
2180 int namelen = strlen(path);
2181 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2182
2183 memset(spec, 0, sizeof(*spec));
2184 spec->path = (char *)(spec + 1);
2185 memcpy(spec->path, path, namelen+1);
2186 spec->count = 1;
2187 spec->is_binary = -1;
2188 return spec;
2189 }
2190
2191 void free_filespec(struct diff_filespec *spec)
2192 {
2193 if (!--spec->count) {
2194 diff_free_filespec_data(spec);
2195 free(spec);
2196 }
2197 }
2198
2199 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2200 unsigned short mode)
2201 {
2202 if (mode) {
2203 spec->mode = canon_mode(mode);
2204 hashcpy(spec->sha1, sha1);
2205 spec->sha1_valid = !is_null_sha1(sha1);
2206 }
2207 }
2208
2209 /*
2210 * Given a name and sha1 pair, if the index tells us the file in
2211 * the work tree has that object contents, return true, so that
2212 * prepare_temp_file() does not have to inflate and extract.
2213 */
2214 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2215 {
2216 struct cache_entry *ce;
2217 struct stat st;
2218 int pos, len;
2219
2220 /*
2221 * We do not read the cache ourselves here, because the
2222 * benchmark with my previous version that always reads cache
2223 * shows that it makes things worse for diff-tree comparing
2224 * two linux-2.6 kernel trees in an already checked out work
2225 * tree. This is because most diff-tree comparisons deal with
2226 * only a small number of files, while reading the cache is
2227 * expensive for a large project, and its cost outweighs the
2228 * savings we get by not inflating the object to a temporary
2229 * file. Practically, this code only helps when we are used
2230 * by diff-cache --cached, which does read the cache before
2231 * calling us.
2232 */
2233 if (!active_cache)
2234 return 0;
2235
2236 /* We want to avoid the working directory if our caller
2237 * doesn't need the data in a normal file, this system
2238 * is rather slow with its stat/open/mmap/close syscalls,
2239 * and the object is contained in a pack file. The pack
2240 * is probably already open and will be faster to obtain
2241 * the data through than the working directory. Loose
2242 * objects however would tend to be slower as they need
2243 * to be individually opened and inflated.
2244 */
2245 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2246 return 0;
2247
2248 len = strlen(name);
2249 pos = cache_name_pos(name, len);
2250 if (pos < 0)
2251 return 0;
2252 ce = active_cache[pos];
2253
2254 /*
2255 * This is not the sha1 we are looking for, or
2256 * unreusable because it is not a regular file.
2257 */
2258 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2259 return 0;
2260
2261 /*
2262 * If ce is marked as "assume unchanged", there is no
2263 * guarantee that work tree matches what we are looking for.
2264 */
2265 if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2266 return 0;
2267
2268 /*
2269 * If ce matches the file in the work tree, we can reuse it.
2270 */
2271 if (ce_uptodate(ce) ||
2272 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2273 return 1;
2274
2275 return 0;
2276 }
2277
2278 static int populate_from_stdin(struct diff_filespec *s)
2279 {
2280 struct strbuf buf = STRBUF_INIT;
2281 size_t size = 0;
2282
2283 if (strbuf_read(&buf, 0, 0) < 0)
2284 return error("error while reading from stdin %s",
2285 strerror(errno));
2286
2287 s->should_munmap = 0;
2288 s->data = strbuf_detach(&buf, &size);
2289 s->size = size;
2290 s->should_free = 1;
2291 return 0;
2292 }
2293
2294 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2295 {
2296 int len;
2297 char *data = xmalloc(100), *dirty = "";
2298
2299 /* Are we looking at the work tree? */
2300 if (s->dirty_submodule)
2301 dirty = "-dirty";
2302
2303 len = snprintf(data, 100,
2304 "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2305 s->data = data;
2306 s->size = len;
2307 s->should_free = 1;
2308 if (size_only) {
2309 s->data = NULL;
2310 free(data);
2311 }
2312 return 0;
2313 }
2314
2315 /*
2316 * While doing rename detection and pickaxe operation, we may need to
2317 * grab the data for the blob (or file) for our own in-core comparison.
2318 * diff_filespec has data and size fields for this purpose.
2319 */
2320 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2321 {
2322 int err = 0;
2323 if (!DIFF_FILE_VALID(s))
2324 die("internal error: asking to populate invalid file.");
2325 if (S_ISDIR(s->mode))
2326 return -1;
2327
2328 if (s->data)
2329 return 0;
2330
2331 if (size_only && 0 < s->size)
2332 return 0;
2333
2334 if (S_ISGITLINK(s->mode))
2335 return diff_populate_gitlink(s, size_only);
2336
2337 if (!s->sha1_valid ||
2338 reuse_worktree_file(s->path, s->sha1, 0)) {
2339 struct strbuf buf = STRBUF_INIT;
2340 struct stat st;
2341 int fd;
2342
2343 if (!strcmp(s->path, "-"))
2344 return populate_from_stdin(s);
2345
2346 if (lstat(s->path, &st) < 0) {
2347 if (errno == ENOENT) {
2348 err_empty:
2349 err = -1;
2350 empty:
2351 s->data = (char *)"";
2352 s->size = 0;
2353 return err;
2354 }
2355 }
2356 s->size = xsize_t(st.st_size);
2357 if (!s->size)
2358 goto empty;
2359 if (S_ISLNK(st.st_mode)) {
2360 struct strbuf sb = STRBUF_INIT;
2361
2362 if (strbuf_readlink(&sb, s->path, s->size))
2363 goto err_empty;
2364 s->size = sb.len;
2365 s->data = strbuf_detach(&sb, NULL);
2366 s->should_free = 1;
2367 return 0;
2368 }
2369 if (size_only)
2370 return 0;
2371 fd = open(s->path, O_RDONLY);
2372 if (fd < 0)
2373 goto err_empty;
2374 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2375 close(fd);
2376 s->should_munmap = 1;
2377
2378 /*
2379 * Convert from working tree format to canonical git format
2380 */
2381 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
2382 size_t size = 0;
2383 munmap(s->data, s->size);
2384 s->should_munmap = 0;
2385 s->data = strbuf_detach(&buf, &size);
2386 s->size = size;
2387 s->should_free = 1;
2388 }
2389 }
2390 else {
2391 enum object_type type;
2392 if (size_only) {
2393 type = sha1_object_info(s->sha1, &s->size);
2394 if (type < 0)
2395 die("unable to read %s", sha1_to_hex(s->sha1));
2396 } else {
2397 s->data = read_sha1_file(s->sha1, &type, &s->size);
2398 if (!s->data)
2399 die("unable to read %s", sha1_to_hex(s->sha1));
2400 s->should_free = 1;
2401 }
2402 }
2403 return 0;
2404 }
2405
2406 void diff_free_filespec_blob(struct diff_filespec *s)
2407 {
2408 if (s->should_free)
2409 free(s->data);
2410 else if (s->should_munmap)
2411 munmap(s->data, s->size);
2412
2413 if (s->should_free || s->should_munmap) {
2414 s->should_free = s->should_munmap = 0;
2415 s->data = NULL;
2416 }
2417 }
2418
2419 void diff_free_filespec_data(struct diff_filespec *s)
2420 {
2421 diff_free_filespec_blob(s);
2422 free(s->cnt_data);
2423 s->cnt_data = NULL;
2424 }
2425
2426 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2427 void *blob,
2428 unsigned long size,
2429 const unsigned char *sha1,
2430 int mode)
2431 {
2432 int fd;
2433 struct strbuf buf = STRBUF_INIT;
2434 struct strbuf template = STRBUF_INIT;
2435 char *path_dup = xstrdup(path);
2436 const char *base = basename(path_dup);
2437
2438 /* Generate "XXXXXX_basename.ext" */
2439 strbuf_addstr(&template, "XXXXXX_");
2440 strbuf_addstr(&template, base);
2441
2442 fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2443 strlen(base) + 1);
2444 if (fd < 0)
2445 die_errno("unable to create temp-file");
2446 if (convert_to_working_tree(path,
2447 (const char *)blob, (size_t)size, &buf)) {
2448 blob = buf.buf;
2449 size = buf.len;
2450 }
2451 if (write_in_full(fd, blob, size) != size)
2452 die_errno("unable to write temp-file");
2453 close(fd);
2454 temp->name = temp->tmp_path;
2455 strcpy(temp->hex, sha1_to_hex(sha1));
2456 temp->hex[40] = 0;
2457 sprintf(temp->mode, "%06o", mode);
2458 strbuf_release(&buf);
2459 strbuf_release(&template);
2460 free(path_dup);
2461 }
2462
2463 static struct diff_tempfile *prepare_temp_file(const char *name,
2464 struct diff_filespec *one)
2465 {
2466 struct diff_tempfile *temp = claim_diff_tempfile();
2467
2468 if (!DIFF_FILE_VALID(one)) {
2469 not_a_valid_file:
2470 /* A '-' entry produces this for file-2, and
2471 * a '+' entry produces this for file-1.
2472 */
2473 temp->name = "/dev/null";
2474 strcpy(temp->hex, ".");
2475 strcpy(temp->mode, ".");
2476 return temp;
2477 }
2478
2479 if (!remove_tempfile_installed) {
2480 atexit(remove_tempfile);
2481 sigchain_push_common(remove_tempfile_on_signal);
2482 remove_tempfile_installed = 1;
2483 }
2484
2485 if (!one->sha1_valid ||
2486 reuse_worktree_file(name, one->sha1, 1)) {
2487 struct stat st;
2488 if (lstat(name, &st) < 0) {
2489 if (errno == ENOENT)
2490 goto not_a_valid_file;
2491 die_errno("stat(%s)", name);
2492 }
2493 if (S_ISLNK(st.st_mode)) {
2494 struct strbuf sb = STRBUF_INIT;
2495 if (strbuf_readlink(&sb, name, st.st_size) < 0)
2496 die_errno("readlink(%s)", name);
2497 prep_temp_blob(name, temp, sb.buf, sb.len,
2498 (one->sha1_valid ?
2499 one->sha1 : null_sha1),
2500 (one->sha1_valid ?
2501 one->mode : S_IFLNK));
2502 strbuf_release(&sb);
2503 }
2504 else {
2505 /* we can borrow from the file in the work tree */
2506 temp->name = name;
2507 if (!one->sha1_valid)
2508 strcpy(temp->hex, sha1_to_hex(null_sha1));
2509 else
2510 strcpy(temp->hex, sha1_to_hex(one->sha1));
2511 /* Even though we may sometimes borrow the
2512 * contents from the work tree, we always want
2513 * one->mode. mode is trustworthy even when
2514 * !(one->sha1_valid), as long as
2515 * DIFF_FILE_VALID(one).
2516 */
2517 sprintf(temp->mode, "%06o", one->mode);
2518 }
2519 return temp;
2520 }
2521 else {
2522 if (diff_populate_filespec(one, 0))
2523 die("cannot read data blob for %s", one->path);
2524 prep_temp_blob(name, temp, one->data, one->size,
2525 one->sha1, one->mode);
2526 }
2527 return temp;
2528 }
2529
2530 /* An external diff command takes:
2531 *
2532 * diff-cmd name infile1 infile1-sha1 infile1-mode \
2533 * infile2 infile2-sha1 infile2-mode [ rename-to ]
2534 *
2535 */
2536 static void run_external_diff(const char *pgm,
2537 const char *name,
2538 const char *other,
2539 struct diff_filespec *one,
2540 struct diff_filespec *two,
2541 const char *xfrm_msg,
2542 int complete_rewrite)
2543 {
2544 const char *spawn_arg[10];
2545 int retval;
2546 const char **arg = &spawn_arg[0];
2547
2548 if (one && two) {
2549 struct diff_tempfile *temp_one, *temp_two;
2550 const char *othername = (other ? other : name);
2551 temp_one = prepare_temp_file(name, one);
2552 temp_two = prepare_temp_file(othername, two);
2553 *arg++ = pgm;
2554 *arg++ = name;
2555 *arg++ = temp_one->name;
2556 *arg++ = temp_one->hex;
2557 *arg++ = temp_one->mode;
2558 *arg++ = temp_two->name;
2559 *arg++ = temp_two->hex;
2560 *arg++ = temp_two->mode;
2561 if (other) {
2562 *arg++ = other;
2563 *arg++ = xfrm_msg;
2564 }
2565 } else {
2566 *arg++ = pgm;
2567 *arg++ = name;
2568 }
2569 *arg = NULL;
2570 fflush(NULL);
2571 retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
2572 remove_tempfile();
2573 if (retval) {
2574 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2575 exit(1);
2576 }
2577 }
2578
2579 static int similarity_index(struct diff_filepair *p)
2580 {
2581 return p->score * 100 / MAX_SCORE;
2582 }
2583
2584 static void fill_metainfo(struct strbuf *msg,
2585 const char *name,
2586 const char *other,
2587 struct diff_filespec *one,
2588 struct diff_filespec *two,
2589 struct diff_options *o,
2590 struct diff_filepair *p,
2591 int *must_show_header,
2592 int use_color)
2593 {
2594 const char *set = diff_get_color(use_color, DIFF_METAINFO);
2595 const char *reset = diff_get_color(use_color, DIFF_RESET);
2596 struct strbuf *msgbuf;
2597 char *line_prefix = "";
2598
2599 *must_show_header = 1;
2600 if (o->output_prefix) {
2601 msgbuf = o->output_prefix(o, o->output_prefix_data);
2602 line_prefix = msgbuf->buf;
2603 }
2604 strbuf_init(msg, PATH_MAX * 2 + 300);
2605 switch (p->status) {
2606 case DIFF_STATUS_COPIED:
2607 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2608 line_prefix, set, similarity_index(p));
2609 strbuf_addf(msg, "%s\n%s%scopy from ",
2610 reset, line_prefix, set);
2611 quote_c_style(name, msg, NULL, 0);
2612 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2613 quote_c_style(other, msg, NULL, 0);
2614 strbuf_addf(msg, "%s\n", reset);
2615 break;
2616 case DIFF_STATUS_RENAMED:
2617 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2618 line_prefix, set, similarity_index(p));
2619 strbuf_addf(msg, "%s\n%s%srename from ",
2620 reset, line_prefix, set);
2621 quote_c_style(name, msg, NULL, 0);
2622 strbuf_addf(msg, "%s\n%s%srename to ",
2623 reset, line_prefix, set);
2624 quote_c_style(other, msg, NULL, 0);
2625 strbuf_addf(msg, "%s\n", reset);
2626 break;
2627 case DIFF_STATUS_MODIFIED:
2628 if (p->score) {
2629 strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2630 line_prefix,
2631 set, similarity_index(p), reset);
2632 break;
2633 }
2634 /* fallthru */
2635 default:
2636 *must_show_header = 0;
2637 }
2638 if (one && two && hashcmp(one->sha1, two->sha1)) {
2639 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2640
2641 if (DIFF_OPT_TST(o, BINARY)) {
2642 mmfile_t mf;
2643 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2644 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2645 abbrev = 40;
2646 }
2647 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
2648 find_unique_abbrev(one->sha1, abbrev));
2649 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
2650 if (one->mode == two->mode)
2651 strbuf_addf(msg, " %06o", one->mode);
2652 strbuf_addf(msg, "%s\n", reset);
2653 }
2654 }
2655
2656 static void run_diff_cmd(const char *pgm,
2657 const char *name,
2658 const char *other,
2659 const char *attr_path,
2660 struct diff_filespec *one,
2661 struct diff_filespec *two,
2662 struct strbuf *msg,
2663 struct diff_options *o,
2664 struct diff_filepair *p)
2665 {
2666 const char *xfrm_msg = NULL;
2667 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2668 int must_show_header = 0;
2669
2670 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2671 pgm = NULL;
2672 else {
2673 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2674 if (drv && drv->external)
2675 pgm = drv->external;
2676 }
2677
2678 if (msg) {
2679 /*
2680 * don't use colors when the header is intended for an
2681 * external diff driver
2682 */
2683 fill_metainfo(msg, name, other, one, two, o, p,
2684 &must_show_header,
2685 DIFF_OPT_TST(o, COLOR_DIFF) && !pgm);
2686 xfrm_msg = msg->len ? msg->buf : NULL;
2687 }
2688
2689 if (pgm) {
2690 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2691 complete_rewrite);
2692 return;
2693 }
2694 if (one && two)
2695 builtin_diff(name, other ? other : name,
2696 one, two, xfrm_msg, must_show_header,
2697 o, complete_rewrite);
2698 else
2699 fprintf(o->file, "* Unmerged path %s\n", name);
2700 }
2701
2702 static void diff_fill_sha1_info(struct diff_filespec *one)
2703 {
2704 if (DIFF_FILE_VALID(one)) {
2705 if (!one->sha1_valid) {
2706 struct stat st;
2707 if (!strcmp(one->path, "-")) {
2708 hashcpy(one->sha1, null_sha1);
2709 return;
2710 }
2711 if (lstat(one->path, &st) < 0)
2712 die_errno("stat '%s'", one->path);
2713 if (index_path(one->sha1, one->path, &st, 0))
2714 die("cannot hash %s", one->path);
2715 }
2716 }
2717 else
2718 hashclr(one->sha1);
2719 }
2720
2721 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2722 {
2723 /* Strip the prefix but do not molest /dev/null and absolute paths */
2724 if (*namep && **namep != '/') {
2725 *namep += prefix_length;
2726 if (**namep == '/')
2727 ++*namep;
2728 }
2729 if (*otherp && **otherp != '/') {
2730 *otherp += prefix_length;
2731 if (**otherp == '/')
2732 ++*otherp;
2733 }
2734 }
2735
2736 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2737 {
2738 const char *pgm = external_diff();
2739 struct strbuf msg;
2740 struct diff_filespec *one = p->one;
2741 struct diff_filespec *two = p->two;
2742 const char *name;
2743 const char *other;
2744 const char *attr_path;
2745
2746 name = p->one->path;
2747 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2748 attr_path = name;
2749 if (o->prefix_length)
2750 strip_prefix(o->prefix_length, &name, &other);
2751
2752 if (DIFF_PAIR_UNMERGED(p)) {
2753 run_diff_cmd(pgm, name, NULL, attr_path,
2754 NULL, NULL, NULL, o, p);
2755 return;
2756 }
2757
2758 diff_fill_sha1_info(one);
2759 diff_fill_sha1_info(two);
2760
2761 if (!pgm &&
2762 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2763 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2764 /*
2765 * a filepair that changes between file and symlink
2766 * needs to be split into deletion and creation.
2767 */
2768 struct diff_filespec *null = alloc_filespec(two->path);
2769 run_diff_cmd(NULL, name, other, attr_path,
2770 one, null, &msg, o, p);
2771 free(null);
2772 strbuf_release(&msg);
2773
2774 null = alloc_filespec(one->path);
2775 run_diff_cmd(NULL, name, other, attr_path,
2776 null, two, &msg, o, p);
2777 free(null);
2778 }
2779 else
2780 run_diff_cmd(pgm, name, other, attr_path,
2781 one, two, &msg, o, p);
2782
2783 strbuf_release(&msg);
2784 }
2785
2786 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2787 struct diffstat_t *diffstat)
2788 {
2789 const char *name;
2790 const char *other;
2791 int complete_rewrite = 0;
2792
2793 if (DIFF_PAIR_UNMERGED(p)) {
2794 /* unmerged */
2795 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2796 return;
2797 }
2798
2799 name = p->one->path;
2800 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2801
2802 if (o->prefix_length)
2803 strip_prefix(o->prefix_length, &name, &other);
2804
2805 diff_fill_sha1_info(p->one);
2806 diff_fill_sha1_info(p->two);
2807
2808 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2809 complete_rewrite = 1;
2810 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2811 }
2812
2813 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2814 {
2815 const char *name;
2816 const char *other;
2817 const char *attr_path;
2818
2819 if (DIFF_PAIR_UNMERGED(p)) {
2820 /* unmerged */
2821 return;
2822 }
2823
2824 name = p->one->path;
2825 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2826 attr_path = other ? other : name;
2827
2828 if (o->prefix_length)
2829 strip_prefix(o->prefix_length, &name, &other);
2830
2831 diff_fill_sha1_info(p->one);
2832 diff_fill_sha1_info(p->two);
2833
2834 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2835 }
2836
2837 void diff_setup(struct diff_options *options)
2838 {
2839 memcpy(options, &default_diff_options, sizeof(*options));
2840
2841 options->file = stdout;
2842
2843 options->line_termination = '\n';
2844 options->break_opt = -1;
2845 options->rename_limit = -1;
2846 options->dirstat_percent = 3;
2847 options->context = 3;
2848
2849 options->change = diff_change;
2850 options->add_remove = diff_addremove;
2851 if (diff_use_color_default > 0)
2852 DIFF_OPT_SET(options, COLOR_DIFF);
2853 options->detect_rename = diff_detect_rename_default;
2854
2855 if (diff_no_prefix) {
2856 options->a_prefix = options->b_prefix = "";
2857 } else if (!diff_mnemonic_prefix) {
2858 options->a_prefix = "a/";
2859 options->b_prefix = "b/";
2860 }
2861 }
2862
2863 int diff_setup_done(struct diff_options *options)
2864 {
2865 int count = 0;
2866
2867 if (options->output_format & DIFF_FORMAT_NAME)
2868 count++;
2869 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2870 count++;
2871 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2872 count++;
2873 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2874 count++;
2875 if (count > 1)
2876 die("--name-only, --name-status, --check and -s are mutually exclusive");
2877
2878 /*
2879 * Most of the time we can say "there are changes"
2880 * only by checking if there are changed paths, but
2881 * --ignore-whitespace* options force us to look
2882 * inside contents.
2883 */
2884
2885 if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
2886 DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
2887 DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
2888 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
2889 else
2890 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
2891
2892 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2893 options->detect_rename = DIFF_DETECT_COPY;
2894
2895 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2896 options->prefix = NULL;
2897 if (options->prefix)
2898 options->prefix_length = strlen(options->prefix);
2899 else
2900 options->prefix_length = 0;
2901
2902 if (options->output_format & (DIFF_FORMAT_NAME |
2903 DIFF_FORMAT_NAME_STATUS |
2904 DIFF_FORMAT_CHECKDIFF |
2905 DIFF_FORMAT_NO_OUTPUT))
2906 options->output_format &= ~(DIFF_FORMAT_RAW |
2907 DIFF_FORMAT_NUMSTAT |
2908 DIFF_FORMAT_DIFFSTAT |
2909 DIFF_FORMAT_SHORTSTAT |
2910 DIFF_FORMAT_DIRSTAT |
2911 DIFF_FORMAT_SUMMARY |
2912 DIFF_FORMAT_PATCH);
2913
2914 /*
2915 * These cases always need recursive; we do not drop caller-supplied
2916 * recursive bits for other formats here.
2917 */
2918 if (options->output_format & (DIFF_FORMAT_PATCH |
2919 DIFF_FORMAT_NUMSTAT |
2920 DIFF_FORMAT_DIFFSTAT |
2921 DIFF_FORMAT_SHORTSTAT |
2922 DIFF_FORMAT_DIRSTAT |
2923 DIFF_FORMAT_SUMMARY |
2924 DIFF_FORMAT_CHECKDIFF))
2925 DIFF_OPT_SET(options, RECURSIVE);
2926 /*
2927 * Also pickaxe would not work very well if you do not say recursive
2928 */
2929 if (options->pickaxe)
2930 DIFF_OPT_SET(options, RECURSIVE);
2931 /*
2932 * When patches are generated, submodules diffed against the work tree
2933 * must be checked for dirtiness too so it can be shown in the output
2934 */
2935 if (options->output_format & DIFF_FORMAT_PATCH)
2936 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
2937
2938 if (options->detect_rename && options->rename_limit < 0)
2939 options->rename_limit = diff_rename_limit_default;
2940 if (options->setup & DIFF_SETUP_USE_CACHE) {
2941 if (!active_cache)
2942 /* read-cache does not die even when it fails
2943 * so it is safe for us to do this here. Also
2944 * it does not smudge active_cache or active_nr
2945 * when it fails, so we do not have to worry about
2946 * cleaning it up ourselves either.
2947 */
2948 read_cache();
2949 }
2950 if (options->abbrev <= 0 || 40 < options->abbrev)
2951 options->abbrev = 40; /* full */
2952
2953 /*
2954 * It does not make sense to show the first hit we happened
2955 * to have found. It does not make sense not to return with
2956 * exit code in such a case either.
2957 */
2958 if (DIFF_OPT_TST(options, QUICK)) {
2959 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2960 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2961 }
2962
2963 return 0;
2964 }
2965
2966 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2967 {
2968 char c, *eq;
2969 int len;
2970
2971 if (*arg != '-')
2972 return 0;
2973 c = *++arg;
2974 if (!c)
2975 return 0;
2976 if (c == arg_short) {
2977 c = *++arg;
2978 if (!c)
2979 return 1;
2980 if (val && isdigit(c)) {
2981 char *end;
2982 int n = strtoul(arg, &end, 10);
2983 if (*end)
2984 return 0;
2985 *val = n;
2986 return 1;
2987 }
2988 return 0;
2989 }
2990 if (c != '-')
2991 return 0;
2992 arg++;
2993 eq = strchr(arg, '=');
2994 if (eq)
2995 len = eq - arg;
2996 else
2997 len = strlen(arg);
2998 if (!len || strncmp(arg, arg_long, len))
2999 return 0;
3000 if (eq) {
3001 int n;
3002 char *end;
3003 if (!isdigit(*++eq))
3004 return 0;
3005 n = strtoul(eq, &end, 10);
3006 if (*end)
3007 return 0;
3008 *val = n;
3009 }
3010 return 1;
3011 }
3012
3013 static int diff_scoreopt_parse(const char *opt);
3014
3015 static inline int short_opt(char opt, const char **argv,
3016 const char **optarg)
3017 {
3018 const char *arg = argv[0];
3019 if (arg[0] != '-' || arg[1] != opt)
3020 return 0;
3021 if (arg[2] != '\0') {
3022 *optarg = arg + 2;
3023 return 1;
3024 }
3025 if (!argv[1])
3026 die("Option '%c' requires a value", opt);
3027 *optarg = argv[1];
3028 return 2;
3029 }
3030
3031 int parse_long_opt(const char *opt, const char **argv,
3032 const char **optarg)
3033 {
3034 const char *arg = argv[0];
3035 if (arg[0] != '-' || arg[1] != '-')
3036 return 0;
3037 arg += strlen("--");
3038 if (prefixcmp(arg, opt))
3039 return 0;
3040 arg += strlen(opt);
3041 if (*arg == '=') { /* sticked form: --option=value */
3042 *optarg = arg + 1;
3043 return 1;
3044 }
3045 if (*arg != '\0')
3046 return 0;
3047 /* separate form: --option value */
3048 if (!argv[1])
3049 die("Option '--%s' requires a value", opt);
3050 *optarg = argv[1];
3051 return 2;
3052 }
3053
3054 static int stat_opt(struct diff_options *options, const char **av)
3055 {
3056 const char *arg = av[0];
3057 char *end;
3058 int width = options->stat_width;
3059 int name_width = options->stat_name_width;
3060 int argcount = 1;
3061
3062 arg += strlen("--stat");
3063 end = (char *)arg;
3064
3065 switch (*arg) {
3066 case '-':
3067 if (!prefixcmp(arg, "-width")) {
3068 arg += strlen("-width");
3069 if (*arg == '=')
3070 width = strtoul(arg + 1, &end, 10);
3071 else if (!*arg && !av[1])
3072 die("Option '--stat-width' requires a value");
3073 else if (!*arg) {
3074 width = strtoul(av[1], &end, 10);
3075 argcount = 2;
3076 }
3077 } else if (!prefixcmp(arg, "-name-width")) {
3078 arg += strlen("-name-width");
3079 if (*arg == '=')
3080 name_width = strtoul(arg + 1, &end, 10);
3081 else if (!*arg && !av[1])
3082 die("Option '--stat-name-width' requires a value");
3083 else if (!*arg) {
3084 name_width = strtoul(av[1], &end, 10);
3085 argcount = 2;
3086 }
3087 }
3088 break;
3089 case '=':
3090 width = strtoul(arg+1, &end, 10);
3091 if (*end == ',')
3092 name_width = strtoul(end+1, &end, 10);
3093 }
3094
3095 /* Important! This checks all the error cases! */
3096 if (*end)
3097 return 0;
3098 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3099 options->stat_name_width = name_width;
3100 options->stat_width = width;
3101 return argcount;
3102 }
3103
3104 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3105 {
3106 const char *arg = av[0];
3107 const char *optarg;
3108 int argcount;
3109
3110 /* Output format options */
3111 if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
3112 options->output_format |= DIFF_FORMAT_PATCH;
3113 else if (opt_arg(arg, 'U', "unified", &options->context))
3114 options->output_format |= DIFF_FORMAT_PATCH;
3115 else if (!strcmp(arg, "--raw"))
3116 options->output_format |= DIFF_FORMAT_RAW;
3117 else if (!strcmp(arg, "--patch-with-raw"))
3118 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
3119 else if (!strcmp(arg, "--numstat"))
3120 options->output_format |= DIFF_FORMAT_NUMSTAT;
3121 else if (!strcmp(arg, "--shortstat"))
3122 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3123 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
3124 options->output_format |= DIFF_FORMAT_DIRSTAT;
3125 else if (!strcmp(arg, "--cumulative")) {
3126 options->output_format |= DIFF_FORMAT_DIRSTAT;
3127 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
3128 } else if (opt_arg(arg, 0, "dirstat-by-file",
3129 &options->dirstat_percent)) {
3130 options->output_format |= DIFF_FORMAT_DIRSTAT;
3131 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
3132 }
3133 else if (!strcmp(arg, "--check"))
3134 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3135 else if (!strcmp(arg, "--summary"))
3136 options->output_format |= DIFF_FORMAT_SUMMARY;
3137 else if (!strcmp(arg, "--patch-with-stat"))
3138 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
3139 else if (!strcmp(arg, "--name-only"))
3140 options->output_format |= DIFF_FORMAT_NAME;
3141 else if (!strcmp(arg, "--name-status"))
3142 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3143 else if (!strcmp(arg, "-s"))
3144 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3145 else if (!prefixcmp(arg, "--stat"))
3146 /* --stat, --stat-width, or --stat-name-width */
3147 return stat_opt(options, av);
3148
3149 /* renames options */
3150 else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
3151 !strcmp(arg, "--break-rewrites")) {
3152 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3153 return error("invalid argument to -B: %s", arg+2);
3154 }
3155 else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
3156 !strcmp(arg, "--find-renames")) {
3157 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3158 return error("invalid argument to -M: %s", arg+2);
3159 options->detect_rename = DIFF_DETECT_RENAME;
3160 }
3161 else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
3162 !strcmp(arg, "--find-copies")) {
3163 if (options->detect_rename == DIFF_DETECT_COPY)
3164 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3165 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3166 return error("invalid argument to -C: %s", arg+2);
3167 options->detect_rename = DIFF_DETECT_COPY;
3168 }
3169 else if (!strcmp(arg, "--no-renames"))
3170 options->detect_rename = 0;
3171 else if (!strcmp(arg, "--relative"))
3172 DIFF_OPT_SET(options, RELATIVE_NAME);
3173 else if (!prefixcmp(arg, "--relative=")) {
3174 DIFF_OPT_SET(options, RELATIVE_NAME);
3175 options->prefix = arg + 11;
3176 }
3177
3178 /* xdiff options */
3179 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3180 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3181 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3182 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3183 else if (!strcmp(arg, "--ignore-space-at-eol"))
3184 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3185 else if (!strcmp(arg, "--patience"))
3186 DIFF_XDL_SET(options, PATIENCE_DIFF);
3187
3188 /* flags options */
3189 else if (!strcmp(arg, "--binary")) {
3190 options->output_format |= DIFF_FORMAT_PATCH;
3191 DIFF_OPT_SET(options, BINARY);
3192 }
3193 else if (!strcmp(arg, "--full-index"))
3194 DIFF_OPT_SET(options, FULL_INDEX);
3195 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3196 DIFF_OPT_SET(options, TEXT);
3197 else if (!strcmp(arg, "-R"))
3198 DIFF_OPT_SET(options, REVERSE_DIFF);
3199 else if (!strcmp(arg, "--find-copies-harder"))
3200 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3201 else if (!strcmp(arg, "--follow"))
3202 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3203 else if (!strcmp(arg, "--color"))
3204 DIFF_OPT_SET(options, COLOR_DIFF);
3205 else if (!prefixcmp(arg, "--color=")) {
3206 int value = git_config_colorbool(NULL, arg+8, -1);
3207 if (value == 0)
3208 DIFF_OPT_CLR(options, COLOR_DIFF);
3209 else if (value > 0)
3210 DIFF_OPT_SET(options, COLOR_DIFF);
3211 else
3212 return error("option `color' expects \"always\", \"auto\", or \"never\"");
3213 }
3214 else if (!strcmp(arg, "--no-color"))
3215 DIFF_OPT_CLR(options, COLOR_DIFF);
3216 else if (!strcmp(arg, "--color-words")) {
3217 DIFF_OPT_SET(options, COLOR_DIFF);
3218 options->word_diff = DIFF_WORDS_COLOR;
3219 }
3220 else if (!prefixcmp(arg, "--color-words=")) {
3221 DIFF_OPT_SET(options, COLOR_DIFF);
3222 options->word_diff = DIFF_WORDS_COLOR;
3223 options->word_regex = arg + 14;
3224 }
3225 else if (!strcmp(arg, "--word-diff")) {
3226 if (options->word_diff == DIFF_WORDS_NONE)
3227 options->word_diff = DIFF_WORDS_PLAIN;
3228 }
3229 else if (!prefixcmp(arg, "--word-diff=")) {
3230 const char *type = arg + 12;
3231 if (!strcmp(type, "plain"))
3232 options->word_diff = DIFF_WORDS_PLAIN;
3233 else if (!strcmp(type, "color")) {
3234 DIFF_OPT_SET(options, COLOR_DIFF);
3235 options->word_diff = DIFF_WORDS_COLOR;
3236 }
3237 else if (!strcmp(type, "porcelain"))
3238 options->word_diff = DIFF_WORDS_PORCELAIN;
3239 else if (!strcmp(type, "none"))
3240 options->word_diff = DIFF_WORDS_NONE;
3241 else
3242 die("bad --word-diff argument: %s", type);
3243 }
3244 else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3245 if (options->word_diff == DIFF_WORDS_NONE)
3246 options->word_diff = DIFF_WORDS_PLAIN;
3247 options->word_regex = optarg;
3248 return argcount;
3249 }
3250 else if (!strcmp(arg, "--exit-code"))
3251 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3252 else if (!strcmp(arg, "--quiet"))
3253 DIFF_OPT_SET(options, QUICK);
3254 else if (!strcmp(arg, "--ext-diff"))
3255 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3256 else if (!strcmp(arg, "--no-ext-diff"))
3257 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3258 else if (!strcmp(arg, "--textconv"))
3259 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3260 else if (!strcmp(arg, "--no-textconv"))
3261 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3262 else if (!strcmp(arg, "--ignore-submodules")) {
3263 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3264 handle_ignore_submodules_arg(options, "all");
3265 } else if (!prefixcmp(arg, "--ignore-submodules=")) {
3266 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3267 handle_ignore_submodules_arg(options, arg + 20);
3268 } else if (!strcmp(arg, "--submodule"))
3269 DIFF_OPT_SET(options, SUBMODULE_LOG);
3270 else if (!prefixcmp(arg, "--submodule=")) {
3271 if (!strcmp(arg + 12, "log"))
3272 DIFF_OPT_SET(options, SUBMODULE_LOG);
3273 }
3274
3275 /* misc options */
3276 else if (!strcmp(arg, "-z"))
3277 options->line_termination = 0;
3278 else if ((argcount = short_opt('l', av, &optarg))) {
3279 options->rename_limit = strtoul(optarg, NULL, 10);
3280 return argcount;
3281 }
3282 else if ((argcount = short_opt('S', av, &optarg))) {
3283 options->pickaxe = optarg;
3284 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3285 return argcount;
3286 } else if ((argcount = short_opt('G', av, &optarg))) {
3287 options->pickaxe = optarg;
3288 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3289 return argcount;
3290 }
3291 else if (!strcmp(arg, "--pickaxe-all"))
3292 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3293 else if (!strcmp(arg, "--pickaxe-regex"))
3294 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3295 else if ((argcount = short_opt('O', av, &optarg))) {
3296 options->orderfile = optarg;
3297 return argcount;
3298 }
3299 else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3300 options->filter = optarg;
3301 return argcount;
3302 }
3303 else if (!strcmp(arg, "--abbrev"))
3304 options->abbrev = DEFAULT_ABBREV;
3305 else if (!prefixcmp(arg, "--abbrev=")) {
3306 options->abbrev = strtoul(arg + 9, NULL, 10);
3307 if (options->abbrev < MINIMUM_ABBREV)
3308 options->abbrev = MINIMUM_ABBREV;
3309 else if (40 < options->abbrev)
3310 options->abbrev = 40;
3311 }
3312 else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3313 options->a_prefix = optarg;
3314 return argcount;
3315 }
3316 else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3317 options->b_prefix = optarg;
3318 return argcount;
3319 }
3320 else if (!strcmp(arg, "--no-prefix"))
3321 options->a_prefix = options->b_prefix = "";
3322 else if (opt_arg(arg, '\0', "inter-hunk-context",
3323 &options->interhunkcontext))
3324 ;
3325 else if ((argcount = parse_long_opt("output", av, &optarg))) {
3326 options->file = fopen(optarg, "w");
3327 if (!options->file)
3328 die_errno("Could not open '%s'", optarg);
3329 options->close_file = 1;
3330 return argcount;
3331 } else
3332 return 0;
3333 return 1;
3334 }
3335
3336 int parse_rename_score(const char **cp_p)
3337 {
3338 unsigned long num, scale;
3339 int ch, dot;
3340 const char *cp = *cp_p;
3341
3342 num = 0;
3343 scale = 1;
3344 dot = 0;
3345 for (;;) {
3346 ch = *cp;
3347 if ( !dot && ch == '.' ) {
3348 scale = 1;
3349 dot = 1;
3350 } else if ( ch == '%' ) {
3351 scale = dot ? scale*100 : 100;
3352 cp++; /* % is always at the end */
3353 break;
3354 } else if ( ch >= '0' && ch <= '9' ) {
3355 if ( scale < 100000 ) {
3356 scale *= 10;
3357 num = (num*10) + (ch-'0');
3358 }
3359 } else {
3360 break;
3361 }
3362 cp++;
3363 }
3364 *cp_p = cp;
3365
3366 /* user says num divided by scale and we say internally that
3367 * is MAX_SCORE * num / scale.
3368 */
3369 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3370 }
3371
3372 static int diff_scoreopt_parse(const char *opt)
3373 {
3374 int opt1, opt2, cmd;
3375
3376 if (*opt++ != '-')
3377 return -1;
3378 cmd = *opt++;
3379 if (cmd == '-') {
3380 /* convert the long-form arguments into short-form versions */
3381 if (!prefixcmp(opt, "break-rewrites")) {
3382 opt += strlen("break-rewrites");
3383 if (*opt == 0 || *opt++ == '=')
3384 cmd = 'B';
3385 } else if (!prefixcmp(opt, "find-copies")) {
3386 opt += strlen("find-copies");
3387 if (*opt == 0 || *opt++ == '=')
3388 cmd = 'C';
3389 } else if (!prefixcmp(opt, "find-renames")) {
3390 opt += strlen("find-renames");
3391 if (*opt == 0 || *opt++ == '=')
3392 cmd = 'M';
3393 }
3394 }
3395 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3396 return -1; /* that is not a -M, -C nor -B option */
3397
3398 opt1 = parse_rename_score(&opt);
3399 if (cmd != 'B')
3400 opt2 = 0;
3401 else {
3402 if (*opt == 0)
3403 opt2 = 0;
3404 else if (*opt != '/')
3405 return -1; /* we expect -B80/99 or -B80 */
3406 else {
3407 opt++;
3408 opt2 = parse_rename_score(&opt);
3409 }
3410 }
3411 if (*opt != 0)
3412 return -1;
3413 return opt1 | (opt2 << 16);
3414 }
3415
3416 struct diff_queue_struct diff_queued_diff;
3417
3418 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3419 {
3420 if (queue->alloc <= queue->nr) {
3421 queue->alloc = alloc_nr(queue->alloc);
3422 queue->queue = xrealloc(queue->queue,
3423 sizeof(dp) * queue->alloc);
3424 }
3425 queue->queue[queue->nr++] = dp;
3426 }
3427
3428 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3429 struct diff_filespec *one,
3430 struct diff_filespec *two)
3431 {
3432 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3433 dp->one = one;
3434 dp->two = two;
3435 if (queue)
3436 diff_q(queue, dp);
3437 return dp;
3438 }
3439
3440 void diff_free_filepair(struct diff_filepair *p)
3441 {
3442 free_filespec(p->one);
3443 free_filespec(p->two);
3444 free(p);
3445 }
3446
3447 /* This is different from find_unique_abbrev() in that
3448 * it stuffs the result with dots for alignment.
3449 */
3450 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3451 {
3452 int abblen;
3453 const char *abbrev;
3454 if (len == 40)
3455 return sha1_to_hex(sha1);
3456
3457 abbrev = find_unique_abbrev(sha1, len);
3458 abblen = strlen(abbrev);
3459 if (abblen < 37) {
3460 static char hex[41];
3461 if (len < abblen && abblen <= len + 2)
3462 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3463 else
3464 sprintf(hex, "%s...", abbrev);
3465 return hex;
3466 }
3467 return sha1_to_hex(sha1);
3468 }
3469
3470 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3471 {
3472 int line_termination = opt->line_termination;
3473 int inter_name_termination = line_termination ? '\t' : '\0';
3474 if (opt->output_prefix) {
3475 struct strbuf *msg = NULL;
3476 msg = opt->output_prefix(opt, opt->output_prefix_data);
3477 fprintf(opt->file, "%s", msg->buf);
3478 }
3479
3480 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3481 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3482 diff_unique_abbrev(p->one->sha1, opt->abbrev));
3483 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3484 }
3485 if (p->score) {
3486 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3487 inter_name_termination);
3488 } else {
3489 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
3490 }
3491
3492 if (p->status == DIFF_STATUS_COPIED ||
3493 p->status == DIFF_STATUS_RENAMED) {
3494 const char *name_a, *name_b;
3495 name_a = p->one->path;
3496 name_b = p->two->path;
3497 strip_prefix(opt->prefix_length, &name_a, &name_b);
3498 write_name_quoted(name_a, opt->file, inter_name_termination);
3499 write_name_quoted(name_b, opt->file, line_termination);
3500 } else {
3501 const char *name_a, *name_b;
3502 name_a = p->one->mode ? p->one->path : p->two->path;
3503 name_b = NULL;
3504 strip_prefix(opt->prefix_length, &name_a, &name_b);
3505 write_name_quoted(name_a, opt->file, line_termination);
3506 }
3507 }
3508
3509 int diff_unmodified_pair(struct diff_filepair *p)
3510 {
3511 /* This function is written stricter than necessary to support
3512 * the currently implemented transformers, but the idea is to
3513 * let transformers to produce diff_filepairs any way they want,
3514 * and filter and clean them up here before producing the output.
3515 */
3516 struct diff_filespec *one = p->one, *two = p->two;
3517
3518 if (DIFF_PAIR_UNMERGED(p))
3519 return 0; /* unmerged is interesting */
3520
3521 /* deletion, addition, mode or type change
3522 * and rename are all interesting.
3523 */
3524 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
3525 DIFF_PAIR_MODE_CHANGED(p) ||
3526 strcmp(one->path, two->path))
3527 return 0;
3528
3529 /* both are valid and point at the same path. that is, we are
3530 * dealing with a change.
3531 */
3532 if (one->sha1_valid && two->sha1_valid &&
3533 !hashcmp(one->sha1, two->sha1) &&
3534 !one->dirty_submodule && !two->dirty_submodule)
3535 return 1; /* no change */
3536 if (!one->sha1_valid && !two->sha1_valid)
3537 return 1; /* both look at the same file on the filesystem. */
3538 return 0;
3539 }
3540
3541 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
3542 {
3543 if (diff_unmodified_pair(p))
3544 return;
3545
3546 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3547 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3548 return; /* no tree diffs in patch format */
3549
3550 run_diff(p, o);
3551 }
3552
3553 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
3554 struct diffstat_t *diffstat)
3555 {
3556 if (diff_unmodified_pair(p))
3557 return;
3558
3559 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3560 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3561 return; /* no useful stat for tree diffs */
3562
3563 run_diffstat(p, o, diffstat);
3564 }
3565
3566 static void diff_flush_checkdiff(struct diff_filepair *p,
3567 struct diff_options *o)
3568 {
3569 if (diff_unmodified_pair(p))
3570 return;
3571
3572 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3573 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3574 return; /* nothing to check in tree diffs */
3575
3576 run_checkdiff(p, o);
3577 }
3578
3579 int diff_queue_is_empty(void)
3580 {
3581 struct diff_queue_struct *q = &diff_queued_diff;
3582 int i;
3583 for (i = 0; i < q->nr; i++)
3584 if (!diff_unmodified_pair(q->queue[i]))
3585 return 0;
3586 return 1;
3587 }
3588
3589 #if DIFF_DEBUG
3590 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
3591 {
3592 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
3593 x, one ? one : "",
3594 s->path,
3595 DIFF_FILE_VALID(s) ? "valid" : "invalid",
3596 s->mode,
3597 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
3598 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
3599 x, one ? one : "",
3600 s->size, s->xfrm_flags);
3601 }
3602
3603 void diff_debug_filepair(const struct diff_filepair *p, int i)
3604 {
3605 diff_debug_filespec(p->one, i, "one");
3606 diff_debug_filespec(p->two, i, "two");
3607 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
3608 p->score, p->status ? p->status : '?',
3609 p->one->rename_used, p->broken_pair);
3610 }
3611
3612 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
3613 {
3614 int i;
3615 if (msg)
3616 fprintf(stderr, "%s\n", msg);
3617 fprintf(stderr, "q->nr = %d\n", q->nr);
3618 for (i = 0; i < q->nr; i++) {
3619 struct diff_filepair *p = q->queue[i];
3620 diff_debug_filepair(p, i);
3621 }
3622 }
3623 #endif
3624
3625 static void diff_resolve_rename_copy(void)
3626 {
3627 int i;
3628 struct diff_filepair *p;
3629 struct diff_queue_struct *q = &diff_queued_diff;
3630
3631 diff_debug_queue("resolve-rename-copy", q);
3632
3633 for (i = 0; i < q->nr; i++) {
3634 p = q->queue[i];
3635 p->status = 0; /* undecided */
3636 if (DIFF_PAIR_UNMERGED(p))
3637 p->status = DIFF_STATUS_UNMERGED;
3638 else if (!DIFF_FILE_VALID(p->one))
3639 p->status = DIFF_STATUS_ADDED;
3640 else if (!DIFF_FILE_VALID(p->two))
3641 p->status = DIFF_STATUS_DELETED;
3642 else if (DIFF_PAIR_TYPE_CHANGED(p))
3643 p->status = DIFF_STATUS_TYPE_CHANGED;
3644
3645 /* from this point on, we are dealing with a pair
3646 * whose both sides are valid and of the same type, i.e.
3647 * either in-place edit or rename/copy edit.
3648 */
3649 else if (DIFF_PAIR_RENAME(p)) {
3650 /*
3651 * A rename might have re-connected a broken
3652 * pair up, causing the pathnames to be the
3653 * same again. If so, that's not a rename at
3654 * all, just a modification..
3655 *
3656 * Otherwise, see if this source was used for
3657 * multiple renames, in which case we decrement
3658 * the count, and call it a copy.
3659 */
3660 if (!strcmp(p->one->path, p->two->path))
3661 p->status = DIFF_STATUS_MODIFIED;
3662 else if (--p->one->rename_used > 0)
3663 p->status = DIFF_STATUS_COPIED;
3664 else
3665 p->status = DIFF_STATUS_RENAMED;
3666 }
3667 else if (hashcmp(p->one->sha1, p->two->sha1) ||
3668 p->one->mode != p->two->mode ||
3669 p->one->dirty_submodule ||
3670 p->two->dirty_submodule ||
3671 is_null_sha1(p->one->sha1))
3672 p->status = DIFF_STATUS_MODIFIED;
3673 else {
3674 /* This is a "no-change" entry and should not
3675 * happen anymore, but prepare for broken callers.
3676 */
3677 error("feeding unmodified %s to diffcore",
3678 p->one->path);
3679 p->status = DIFF_STATUS_UNKNOWN;
3680 }
3681 }
3682 diff_debug_queue("resolve-rename-copy done", q);
3683 }
3684
3685 static int check_pair_status(struct diff_filepair *p)
3686 {
3687 switch (p->status) {
3688 case DIFF_STATUS_UNKNOWN:
3689 return 0;
3690 case 0:
3691 die("internal error in diff-resolve-rename-copy");
3692 default:
3693 return 1;
3694 }
3695 }
3696
3697 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
3698 {
3699 int fmt = opt->output_format;
3700
3701 if (fmt & DIFF_FORMAT_CHECKDIFF)
3702 diff_flush_checkdiff(p, opt);
3703 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
3704 diff_flush_raw(p, opt);
3705 else if (fmt & DIFF_FORMAT_NAME) {
3706 const char *name_a, *name_b;
3707 name_a = p->two->path;
3708 name_b = NULL;
3709 strip_prefix(opt->prefix_length, &name_a, &name_b);
3710 write_name_quoted(name_a, opt->file, opt->line_termination);
3711 }
3712 }
3713
3714 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
3715 {
3716 if (fs->mode)
3717 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
3718 else
3719 fprintf(file, " %s ", newdelete);
3720 write_name_quoted(fs->path, file, '\n');
3721 }
3722
3723
3724 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
3725 const char *line_prefix)
3726 {
3727 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
3728 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
3729 p->two->mode, show_name ? ' ' : '\n');
3730 if (show_name) {
3731 write_name_quoted(p->two->path, file, '\n');
3732 }
3733 }
3734 }
3735
3736 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
3737 const char *line_prefix)
3738 {
3739 char *names = pprint_rename(p->one->path, p->two->path);
3740
3741 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
3742 free(names);
3743 show_mode_change(file, p, 0, line_prefix);
3744 }
3745
3746 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
3747 {
3748 FILE *file = opt->file;
3749 char *line_prefix = "";
3750
3751 if (opt->output_prefix) {
3752 struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
3753 line_prefix = buf->buf;
3754 }
3755
3756 switch(p->status) {
3757 case DIFF_STATUS_DELETED:
3758 fputs(line_prefix, file);
3759 show_file_mode_name(file, "delete", p->one);
3760 break;
3761 case DIFF_STATUS_ADDED:
3762 fputs(line_prefix, file);
3763 show_file_mode_name(file, "create", p->two);
3764 break;
3765 case DIFF_STATUS_COPIED:
3766 fputs(line_prefix, file);
3767 show_rename_copy(file, "copy", p, line_prefix);
3768 break;
3769 case DIFF_STATUS_RENAMED:
3770 fputs(line_prefix, file);
3771 show_rename_copy(file, "rename", p, line_prefix);
3772 break;
3773 default:
3774 if (p->score) {
3775 fprintf(file, "%s rewrite ", line_prefix);
3776 write_name_quoted(p->two->path, file, ' ');
3777 fprintf(file, "(%d%%)\n", similarity_index(p));
3778 }
3779 show_mode_change(file, p, !p->score, line_prefix);
3780 break;
3781 }
3782 }
3783
3784 struct patch_id_t {
3785 git_SHA_CTX *ctx;
3786 int patchlen;
3787 };
3788
3789 static int remove_space(char *line, int len)
3790 {
3791 int i;
3792 char *dst = line;
3793 unsigned char c;
3794
3795 for (i = 0; i < len; i++)
3796 if (!isspace((c = line[i])))
3797 *dst++ = c;
3798
3799 return dst - line;
3800 }
3801
3802 static void patch_id_consume(void *priv, char *line, unsigned long len)
3803 {
3804 struct patch_id_t *data = priv;
3805 int new_len;
3806
3807 /* Ignore line numbers when computing the SHA1 of the patch */
3808 if (!prefixcmp(line, "@@ -"))
3809 return;
3810
3811 new_len = remove_space(line, len);
3812
3813 git_SHA1_Update(data->ctx, line, new_len);
3814 data->patchlen += new_len;
3815 }
3816
3817 /* returns 0 upon success, and writes result into sha1 */
3818 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3819 {
3820 struct diff_queue_struct *q = &diff_queued_diff;
3821 int i;
3822 git_SHA_CTX ctx;
3823 struct patch_id_t data;
3824 char buffer[PATH_MAX * 4 + 20];
3825
3826 git_SHA1_Init(&ctx);
3827 memset(&data, 0, sizeof(struct patch_id_t));
3828 data.ctx = &ctx;
3829
3830 for (i = 0; i < q->nr; i++) {
3831 xpparam_t xpp;
3832 xdemitconf_t xecfg;
3833 mmfile_t mf1, mf2;
3834 struct diff_filepair *p = q->queue[i];
3835 int len1, len2;
3836
3837 memset(&xpp, 0, sizeof(xpp));
3838 memset(&xecfg, 0, sizeof(xecfg));
3839 if (p->status == 0)
3840 return error("internal diff status error");
3841 if (p->status == DIFF_STATUS_UNKNOWN)
3842 continue;
3843 if (diff_unmodified_pair(p))
3844 continue;
3845 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3846 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3847 continue;
3848 if (DIFF_PAIR_UNMERGED(p))
3849 continue;
3850
3851 diff_fill_sha1_info(p->one);
3852 diff_fill_sha1_info(p->two);
3853 if (fill_mmfile(&mf1, p->one) < 0 ||
3854 fill_mmfile(&mf2, p->two) < 0)
3855 return error("unable to read files to diff");
3856
3857 len1 = remove_space(p->one->path, strlen(p->one->path));
3858 len2 = remove_space(p->two->path, strlen(p->two->path));
3859 if (p->one->mode == 0)
3860 len1 = snprintf(buffer, sizeof(buffer),
3861 "diff--gita/%.*sb/%.*s"
3862 "newfilemode%06o"
3863 "---/dev/null"
3864 "+++b/%.*s",
3865 len1, p->one->path,
3866 len2, p->two->path,
3867 p->two->mode,
3868 len2, p->two->path);
3869 else if (p->two->mode == 0)
3870 len1 = snprintf(buffer, sizeof(buffer),
3871 "diff--gita/%.*sb/%.*s"
3872 "deletedfilemode%06o"
3873 "---a/%.*s"
3874 "+++/dev/null",
3875 len1, p->one->path,
3876 len2, p->two->path,
3877 p->one->mode,
3878 len1, p->one->path);
3879 else
3880 len1 = snprintf(buffer, sizeof(buffer),
3881 "diff--gita/%.*sb/%.*s"
3882 "---a/%.*s"
3883 "+++b/%.*s",
3884 len1, p->one->path,
3885 len2, p->two->path,
3886 len1, p->one->path,
3887 len2, p->two->path);
3888 git_SHA1_Update(&ctx, buffer, len1);
3889
3890 if (diff_filespec_is_binary(p->one) ||
3891 diff_filespec_is_binary(p->two)) {
3892 git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
3893 git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
3894 continue;
3895 }
3896
3897 xpp.flags = 0;
3898 xecfg.ctxlen = 3;
3899 xecfg.flags = 0;
3900 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3901 &xpp, &xecfg);
3902 }
3903
3904 git_SHA1_Final(sha1, &ctx);
3905 return 0;
3906 }
3907
3908 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3909 {
3910 struct diff_queue_struct *q = &diff_queued_diff;
3911 int i;
3912 int result = diff_get_patch_id(options, sha1);
3913
3914 for (i = 0; i < q->nr; i++)
3915 diff_free_filepair(q->queue[i]);
3916
3917 free(q->queue);
3918 DIFF_QUEUE_CLEAR(q);
3919
3920 return result;
3921 }
3922
3923 static int is_summary_empty(const struct diff_queue_struct *q)
3924 {
3925 int i;
3926
3927 for (i = 0; i < q->nr; i++) {
3928 const struct diff_filepair *p = q->queue[i];
3929
3930 switch (p->status) {
3931 case DIFF_STATUS_DELETED:
3932 case DIFF_STATUS_ADDED:
3933 case DIFF_STATUS_COPIED:
3934 case DIFF_STATUS_RENAMED:
3935 return 0;
3936 default:
3937 if (p->score)
3938 return 0;
3939 if (p->one->mode && p->two->mode &&
3940 p->one->mode != p->two->mode)
3941 return 0;
3942 break;
3943 }
3944 }
3945 return 1;
3946 }
3947
3948 void diff_flush(struct diff_options *options)
3949 {
3950 struct diff_queue_struct *q = &diff_queued_diff;
3951 int i, output_format = options->output_format;
3952 int separator = 0;
3953
3954 /*
3955 * Order: raw, stat, summary, patch
3956 * or: name/name-status/checkdiff (other bits clear)
3957 */
3958 if (!q->nr)
3959 goto free_queue;
3960
3961 if (output_format & (DIFF_FORMAT_RAW |
3962 DIFF_FORMAT_NAME |
3963 DIFF_FORMAT_NAME_STATUS |
3964 DIFF_FORMAT_CHECKDIFF)) {
3965 for (i = 0; i < q->nr; i++) {
3966 struct diff_filepair *p = q->queue[i];
3967 if (check_pair_status(p))
3968 flush_one_pair(p, options);
3969 }
3970 separator++;
3971 }
3972
3973 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3974 struct diffstat_t diffstat;
3975
3976 memset(&diffstat, 0, sizeof(struct diffstat_t));
3977 for (i = 0; i < q->nr; i++) {
3978 struct diff_filepair *p = q->queue[i];
3979 if (check_pair_status(p))
3980 diff_flush_stat(p, options, &diffstat);
3981 }
3982 if (output_format & DIFF_FORMAT_NUMSTAT)
3983 show_numstat(&diffstat, options);
3984 if (output_format & DIFF_FORMAT_DIFFSTAT)
3985 show_stats(&diffstat, options);
3986 if (output_format & DIFF_FORMAT_SHORTSTAT)
3987 show_shortstats(&diffstat, options);
3988 free_diffstat_info(&diffstat);
3989 separator++;
3990 }
3991 if (output_format & DIFF_FORMAT_DIRSTAT)
3992 show_dirstat(options);
3993
3994 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3995 for (i = 0; i < q->nr; i++) {
3996 diff_summary(options, q->queue[i]);
3997 }
3998 separator++;
3999 }
4000
4001 if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4002 DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4003 DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4004 /*
4005 * run diff_flush_patch for the exit status. setting
4006 * options->file to /dev/null should be safe, becaue we
4007 * aren't supposed to produce any output anyway.
4008 */
4009 if (options->close_file)
4010 fclose(options->file);
4011 options->file = fopen("/dev/null", "w");
4012 if (!options->file)
4013 die_errno("Could not open /dev/null");
4014 options->close_file = 1;
4015 for (i = 0; i < q->nr; i++) {
4016 struct diff_filepair *p = q->queue[i];
4017 if (check_pair_status(p))
4018 diff_flush_patch(p, options);
4019 if (options->found_changes)
4020 break;
4021 }
4022 }
4023
4024 if (output_format & DIFF_FORMAT_PATCH) {
4025 if (separator) {
4026 putc(options->line_termination, options->file);
4027 if (options->stat_sep) {
4028 /* attach patch instead of inline */
4029 fputs(options->stat_sep, options->file);
4030 }
4031 }
4032
4033 for (i = 0; i < q->nr; i++) {
4034 struct diff_filepair *p = q->queue[i];
4035 if (check_pair_status(p))
4036 diff_flush_patch(p, options);
4037 }
4038 }
4039
4040 if (output_format & DIFF_FORMAT_CALLBACK)
4041 options->format_callback(q, options, options->format_callback_data);
4042
4043 for (i = 0; i < q->nr; i++)
4044 diff_free_filepair(q->queue[i]);
4045 free_queue:
4046 free(q->queue);
4047 DIFF_QUEUE_CLEAR(q);
4048 if (options->close_file)
4049 fclose(options->file);
4050
4051 /*
4052 * Report the content-level differences with HAS_CHANGES;
4053 * diff_addremove/diff_change does not set the bit when
4054 * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4055 */
4056 if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4057 if (options->found_changes)
4058 DIFF_OPT_SET(options, HAS_CHANGES);
4059 else
4060 DIFF_OPT_CLR(options, HAS_CHANGES);
4061 }
4062 }
4063
4064 static void diffcore_apply_filter(const char *filter)
4065 {
4066 int i;
4067 struct diff_queue_struct *q = &diff_queued_diff;
4068 struct diff_queue_struct outq;
4069 DIFF_QUEUE_CLEAR(&outq);
4070
4071 if (!filter)
4072 return;
4073
4074 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
4075 int found;
4076 for (i = found = 0; !found && i < q->nr; i++) {
4077 struct diff_filepair *p = q->queue[i];
4078 if (((p->status == DIFF_STATUS_MODIFIED) &&
4079 ((p->score &&
4080 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4081 (!p->score &&
4082 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4083 ((p->status != DIFF_STATUS_MODIFIED) &&
4084 strchr(filter, p->status)))
4085 found++;
4086 }
4087 if (found)
4088 return;
4089
4090 /* otherwise we will clear the whole queue
4091 * by copying the empty outq at the end of this
4092 * function, but first clear the current entries
4093 * in the queue.
4094 */
4095 for (i = 0; i < q->nr; i++)
4096 diff_free_filepair(q->queue[i]);
4097 }
4098 else {
4099 /* Only the matching ones */
4100 for (i = 0; i < q->nr; i++) {
4101 struct diff_filepair *p = q->queue[i];
4102
4103 if (((p->status == DIFF_STATUS_MODIFIED) &&
4104 ((p->score &&
4105 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
4106 (!p->score &&
4107 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
4108 ((p->status != DIFF_STATUS_MODIFIED) &&
4109 strchr(filter, p->status)))
4110 diff_q(&outq, p);
4111 else
4112 diff_free_filepair(p);
4113 }
4114 }
4115 free(q->queue);
4116 *q = outq;
4117 }
4118
4119 /* Check whether two filespecs with the same mode and size are identical */
4120 static int diff_filespec_is_identical(struct diff_filespec *one,
4121 struct diff_filespec *two)
4122 {
4123 if (S_ISGITLINK(one->mode))
4124 return 0;
4125 if (diff_populate_filespec(one, 0))
4126 return 0;
4127 if (diff_populate_filespec(two, 0))
4128 return 0;
4129 return !memcmp(one->data, two->data, one->size);
4130 }
4131
4132 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4133 {
4134 int i;
4135 struct diff_queue_struct *q = &diff_queued_diff;
4136 struct diff_queue_struct outq;
4137 DIFF_QUEUE_CLEAR(&outq);
4138
4139 for (i = 0; i < q->nr; i++) {
4140 struct diff_filepair *p = q->queue[i];
4141
4142 /*
4143 * 1. Entries that come from stat info dirtiness
4144 * always have both sides (iow, not create/delete),
4145 * one side of the object name is unknown, with
4146 * the same mode and size. Keep the ones that
4147 * do not match these criteria. They have real
4148 * differences.
4149 *
4150 * 2. At this point, the file is known to be modified,
4151 * with the same mode and size, and the object
4152 * name of one side is unknown. Need to inspect
4153 * the identical contents.
4154 */
4155 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4156 !DIFF_FILE_VALID(p->two) ||
4157 (p->one->sha1_valid && p->two->sha1_valid) ||
4158 (p->one->mode != p->two->mode) ||
4159 diff_populate_filespec(p->one, 1) ||
4160 diff_populate_filespec(p->two, 1) ||
4161 (p->one->size != p->two->size) ||
4162 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4163 diff_q(&outq, p);
4164 else {
4165 /*
4166 * The caller can subtract 1 from skip_stat_unmatch
4167 * to determine how many paths were dirty only
4168 * due to stat info mismatch.
4169 */
4170 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4171 diffopt->skip_stat_unmatch++;
4172 diff_free_filepair(p);
4173 }
4174 }
4175 free(q->queue);
4176 *q = outq;
4177 }
4178
4179 static int diffnamecmp(const void *a_, const void *b_)
4180 {
4181 const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4182 const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4183 const char *name_a, *name_b;
4184
4185 name_a = a->one ? a->one->path : a->two->path;
4186 name_b = b->one ? b->one->path : b->two->path;
4187 return strcmp(name_a, name_b);
4188 }
4189
4190 void diffcore_fix_diff_index(struct diff_options *options)
4191 {
4192 struct diff_queue_struct *q = &diff_queued_diff;
4193 qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4194 }
4195
4196 void diffcore_std(struct diff_options *options)
4197 {
4198 if (options->skip_stat_unmatch)
4199 diffcore_skip_stat_unmatch(options);
4200 if (!options->found_follow) {
4201 /* See try_to_follow_renames() in tree-diff.c */
4202 if (options->break_opt != -1)
4203 diffcore_break(options->break_opt);
4204 if (options->detect_rename)
4205 diffcore_rename(options);
4206 if (options->break_opt != -1)
4207 diffcore_merge_broken();
4208 }
4209 if (options->pickaxe)
4210 diffcore_pickaxe(options);
4211 if (options->orderfile)
4212 diffcore_order(options->orderfile);
4213 if (!options->found_follow)
4214 /* See try_to_follow_renames() in tree-diff.c */
4215 diff_resolve_rename_copy();
4216 diffcore_apply_filter(options->filter);
4217
4218 if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4219 DIFF_OPT_SET(options, HAS_CHANGES);
4220 else
4221 DIFF_OPT_CLR(options, HAS_CHANGES);
4222
4223 options->found_follow = 0;
4224 }
4225
4226 int diff_result_code(struct diff_options *opt, int status)
4227 {
4228 int result = 0;
4229 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4230 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4231 return status;
4232 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4233 DIFF_OPT_TST(opt, HAS_CHANGES))
4234 result |= 01;
4235 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4236 DIFF_OPT_TST(opt, CHECK_FAILED))
4237 result |= 02;
4238 return result;
4239 }
4240
4241 /*
4242 * Shall changes to this submodule be ignored?
4243 *
4244 * Submodule changes can be configured to be ignored separately for each path,
4245 * but that configuration can be overridden from the command line.
4246 */
4247 static int is_submodule_ignored(const char *path, struct diff_options *options)
4248 {
4249 int ignored = 0;
4250 unsigned orig_flags = options->flags;
4251 if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4252 set_diffopt_flags_from_submodule_config(options, path);
4253 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4254 ignored = 1;
4255 options->flags = orig_flags;
4256 return ignored;
4257 }
4258
4259 void diff_addremove(struct diff_options *options,
4260 int addremove, unsigned mode,
4261 const unsigned char *sha1,
4262 const char *concatpath, unsigned dirty_submodule)
4263 {
4264 struct diff_filespec *one, *two;
4265
4266 if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4267 return;
4268
4269 /* This may look odd, but it is a preparation for
4270 * feeding "there are unchanged files which should
4271 * not produce diffs, but when you are doing copy
4272 * detection you would need them, so here they are"
4273 * entries to the diff-core. They will be prefixed
4274 * with something like '=' or '*' (I haven't decided
4275 * which but should not make any difference).
4276 * Feeding the same new and old to diff_change()
4277 * also has the same effect.
4278 * Before the final output happens, they are pruned after
4279 * merged into rename/copy pairs as appropriate.
4280 */
4281 if (DIFF_OPT_TST(options, REVERSE_DIFF))
4282 addremove = (addremove == '+' ? '-' :
4283 addremove == '-' ? '+' : addremove);
4284
4285 if (options->prefix &&
4286 strncmp(concatpath, options->prefix, options->prefix_length))
4287 return;
4288
4289 one = alloc_filespec(concatpath);
4290 two = alloc_filespec(concatpath);
4291
4292 if (addremove != '+')
4293 fill_filespec(one, sha1, mode);
4294 if (addremove != '-') {
4295 fill_filespec(two, sha1, mode);
4296 two->dirty_submodule = dirty_submodule;
4297 }
4298
4299 diff_queue(&diff_queued_diff, one, two);
4300 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4301 DIFF_OPT_SET(options, HAS_CHANGES);
4302 }
4303
4304 void diff_change(struct diff_options *options,
4305 unsigned old_mode, unsigned new_mode,
4306 const unsigned char *old_sha1,
4307 const unsigned char *new_sha1,
4308 const char *concatpath,
4309 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4310 {
4311 struct diff_filespec *one, *two;
4312
4313 if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4314 is_submodule_ignored(concatpath, options))
4315 return;
4316
4317 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4318 unsigned tmp;
4319 const unsigned char *tmp_c;
4320 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4321 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4322 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4323 new_dirty_submodule = tmp;
4324 }
4325
4326 if (options->prefix &&
4327 strncmp(concatpath, options->prefix, options->prefix_length))
4328 return;
4329
4330 one = alloc_filespec(concatpath);
4331 two = alloc_filespec(concatpath);
4332 fill_filespec(one, old_sha1, old_mode);
4333 fill_filespec(two, new_sha1, new_mode);
4334 one->dirty_submodule = old_dirty_submodule;
4335 two->dirty_submodule = new_dirty_submodule;
4336
4337 diff_queue(&diff_queued_diff, one, two);
4338 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4339 DIFF_OPT_SET(options, HAS_CHANGES);
4340 }
4341
4342 void diff_unmerge(struct diff_options *options,
4343 const char *path,
4344 unsigned mode, const unsigned char *sha1)
4345 {
4346 struct diff_filespec *one, *two;
4347
4348 if (options->prefix &&
4349 strncmp(path, options->prefix, options->prefix_length))
4350 return;
4351
4352 one = alloc_filespec(path);
4353 two = alloc_filespec(path);
4354 fill_filespec(one, sha1, mode);
4355 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
4356 }
4357
4358 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4359 size_t *outsize)
4360 {
4361 struct diff_tempfile *temp;
4362 const char *argv[3];
4363 const char **arg = argv;
4364 struct child_process child;
4365 struct strbuf buf = STRBUF_INIT;
4366 int err = 0;
4367
4368 temp = prepare_temp_file(spec->path, spec);
4369 *arg++ = pgm;
4370 *arg++ = temp->name;
4371 *arg = NULL;
4372
4373 memset(&child, 0, sizeof(child));
4374 child.use_shell = 1;
4375 child.argv = argv;
4376 child.out = -1;
4377 if (start_command(&child)) {
4378 remove_tempfile();
4379 return NULL;
4380 }
4381
4382 if (strbuf_read(&buf, child.out, 0) < 0)
4383 err = error("error reading from textconv command '%s'", pgm);
4384 close(child.out);
4385
4386 if (finish_command(&child) || err) {
4387 strbuf_release(&buf);
4388 remove_tempfile();
4389 return NULL;
4390 }
4391 remove_tempfile();
4392
4393 return strbuf_detach(&buf, outsize);
4394 }
4395
4396 size_t fill_textconv(struct userdiff_driver *driver,
4397 struct diff_filespec *df,
4398 char **outbuf)
4399 {
4400 size_t size;
4401
4402 if (!driver || !driver->textconv) {
4403 if (!DIFF_FILE_VALID(df)) {
4404 *outbuf = "";
4405 return 0;
4406 }
4407 if (diff_populate_filespec(df, 0))
4408 die("unable to read files to diff");
4409 *outbuf = df->data;
4410 return df->size;
4411 }
4412
4413 if (driver->textconv_cache && df->sha1_valid) {
4414 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4415 &size);
4416 if (*outbuf)
4417 return size;
4418 }
4419
4420 *outbuf = run_textconv(driver->textconv, df, &size);
4421 if (!*outbuf)
4422 die("unable to read files to diff");
4423
4424 if (driver->textconv_cache && df->sha1_valid) {
4425 /* ignore errors, as we might be in a readonly repository */
4426 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4427 size);
4428 /*
4429 * we could save up changes and flush them all at the end,
4430 * but we would need an extra call after all diffing is done.
4431 * Since generating a cache entry is the slow path anyway,
4432 * this extra overhead probably isn't a big deal.
4433 */
4434 notes_cache_write(driver->textconv_cache);
4435 }
4436
4437 return size;
4438 }
Jonathan Nieder’s miscellaneous git branches