?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-diff/git-apply: make diff output a bit friendlier to GNU patch (part 2)
[git.git] / diff.c
blobb026c737b682f9249792600032792f3e36230163
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include <sys/types.h>
5 #include <sys/wait.h>
6 #include <signal.h>
7 #include "cache.h"
8 #include "quote.h"
9 #include "diff.h"
10 #include "diffcore.h"
11 #include "delta.h"
12 #include "xdiff-interface.h"
13 #include "color.h"
14
15 static int use_size_cache;
16
17 static int diff_detect_rename_default;
18 static int diff_rename_limit_default = -1;
19 static int diff_use_color_default;
20
21 static char diff_colors[][COLOR_MAXLEN] = {
22 "\033[m", /* reset */
23 "", /* PLAIN (normal) */
24 "\033[1m", /* METAINFO (bold) */
25 "\033[36m", /* FRAGINFO (cyan) */
26 "\033[31m", /* OLD (red) */
27 "\033[32m", /* NEW (green) */
28 "\033[33m", /* COMMIT (yellow) */
29 "\033[41m", /* WHITESPACE (red background) */
30 };
31
32 static int parse_diff_color_slot(const char *var, int ofs)
33 {
34 if (!strcasecmp(var+ofs, "plain"))
35 return DIFF_PLAIN;
36 if (!strcasecmp(var+ofs, "meta"))
37 return DIFF_METAINFO;
38 if (!strcasecmp(var+ofs, "frag"))
39 return DIFF_FRAGINFO;
40 if (!strcasecmp(var+ofs, "old"))
41 return DIFF_FILE_OLD;
42 if (!strcasecmp(var+ofs, "new"))
43 return DIFF_FILE_NEW;
44 if (!strcasecmp(var+ofs, "commit"))
45 return DIFF_COMMIT;
46 if (!strcasecmp(var+ofs, "whitespace"))
47 return DIFF_WHITESPACE;
48 die("bad config variable '%s'", var);
49 }
50
51 /*
52 * These are to give UI layer defaults.
53 * The core-level commands such as git-diff-files should
54 * never be affected by the setting of diff.renames
55 * the user happens to have in the configuration file.
56 */
57 int git_diff_ui_config(const char *var, const char *value)
58 {
59 if (!strcmp(var, "diff.renamelimit")) {
60 diff_rename_limit_default = git_config_int(var, value);
61 return 0;
62 }
63 if (!strcmp(var, "diff.color")) {
64 diff_use_color_default = git_config_colorbool(var, value);
65 return 0;
66 }
67 if (!strcmp(var, "diff.renames")) {
68 if (!value)
69 diff_detect_rename_default = DIFF_DETECT_RENAME;
70 else if (!strcasecmp(value, "copies") ||
71 !strcasecmp(value, "copy"))
72 diff_detect_rename_default = DIFF_DETECT_COPY;
73 else if (git_config_bool(var,value))
74 diff_detect_rename_default = DIFF_DETECT_RENAME;
75 return 0;
76 }
77 if (!strncmp(var, "diff.color.", 11)) {
78 int slot = parse_diff_color_slot(var, 11);
79 color_parse(value, var, diff_colors[slot]);
80 return 0;
81 }
82 return git_default_config(var, value);
83 }
84
85 static char *quote_one(const char *str)
86 {
87 int needlen;
88 char *xp;
89
90 if (!str)
91 return NULL;
92 needlen = quote_c_style(str, NULL, NULL, 0);
93 if (!needlen)
94 return xstrdup(str);
95 xp = xmalloc(needlen + 1);
96 quote_c_style(str, xp, NULL, 0);
97 return xp;
98 }
99
100 static char *quote_two(const char *one, const char *two)
101 {
102 int need_one = quote_c_style(one, NULL, NULL, 1);
103 int need_two = quote_c_style(two, NULL, NULL, 1);
104 char *xp;
105
106 if (need_one + need_two) {
107 if (!need_one) need_one = strlen(one);
108 if (!need_two) need_one = strlen(two);
109
110 xp = xmalloc(need_one + need_two + 3);
111 xp[0] = '"';
112 quote_c_style(one, xp + 1, NULL, 1);
113 quote_c_style(two, xp + need_one + 1, NULL, 1);
114 strcpy(xp + need_one + need_two + 1, "\"");
115 return xp;
116 }
117 need_one = strlen(one);
118 need_two = strlen(two);
119 xp = xmalloc(need_one + need_two + 1);
120 strcpy(xp, one);
121 strcpy(xp + need_one, two);
122 return xp;
123 }
124
125 static const char *external_diff(void)
126 {
127 static const char *external_diff_cmd = NULL;
128 static int done_preparing = 0;
129
130 if (done_preparing)
131 return external_diff_cmd;
132 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
133 done_preparing = 1;
134 return external_diff_cmd;
135 }
136
137 #define TEMPFILE_PATH_LEN 50
138
139 static struct diff_tempfile {
140 const char *name; /* filename external diff should read from */
141 char hex[41];
142 char mode[10];
143 char tmp_path[TEMPFILE_PATH_LEN];
144 } diff_temp[2];
145
146 static int count_lines(const char *data, int size)
147 {
148 int count, ch, completely_empty = 1, nl_just_seen = 0;
149 count = 0;
150 while (0 < size--) {
151 ch = *data++;
152 if (ch == '\n') {
153 count++;
154 nl_just_seen = 1;
155 completely_empty = 0;
156 }
157 else {
158 nl_just_seen = 0;
159 completely_empty = 0;
160 }
161 }
162 if (completely_empty)
163 return 0;
164 if (!nl_just_seen)
165 count++; /* no trailing newline */
166 return count;
167 }
168
169 static void print_line_count(int count)
170 {
171 switch (count) {
172 case 0:
173 printf("0,0");
174 break;
175 case 1:
176 printf("1");
177 break;
178 default:
179 printf("1,%d", count);
180 break;
181 }
182 }
183
184 static void copy_file(int prefix, const char *data, int size)
185 {
186 int ch, nl_just_seen = 1;
187 while (0 < size--) {
188 ch = *data++;
189 if (nl_just_seen)
190 putchar(prefix);
191 putchar(ch);
192 if (ch == '\n')
193 nl_just_seen = 1;
194 else
195 nl_just_seen = 0;
196 }
197 if (!nl_just_seen)
198 printf("\n\\ No newline at end of file\n");
199 }
200
201 static void emit_rewrite_diff(const char *name_a,
202 const char *name_b,
203 struct diff_filespec *one,
204 struct diff_filespec *two)
205 {
206 int lc_a, lc_b;
207 const char *name_a_tab, *name_b_tab;
208
209 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
210 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
211
212 diff_populate_filespec(one, 0);
213 diff_populate_filespec(two, 0);
214 lc_a = count_lines(one->data, one->size);
215 lc_b = count_lines(two->data, two->size);
216 printf("--- a/%s%s\n+++ b/%s%s\n@@ -",
217 name_a, name_a_tab,
218 name_b, name_b_tab);
219 print_line_count(lc_a);
220 printf(" +");
221 print_line_count(lc_b);
222 printf(" @@\n");
223 if (lc_a)
224 copy_file('-', one->data, one->size);
225 if (lc_b)
226 copy_file('+', two->data, two->size);
227 }
228
229 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
230 {
231 if (!DIFF_FILE_VALID(one)) {
232 mf->ptr = (char *)""; /* does not matter */
233 mf->size = 0;
234 return 0;
235 }
236 else if (diff_populate_filespec(one, 0))
237 return -1;
238 mf->ptr = one->data;
239 mf->size = one->size;
240 return 0;
241 }
242
243 struct diff_words_buffer {
244 mmfile_t text;
245 long alloc;
246 long current; /* output pointer */
247 int suppressed_newline;
248 };
249
250 static void diff_words_append(char *line, unsigned long len,
251 struct diff_words_buffer *buffer)
252 {
253 if (buffer->text.size + len > buffer->alloc) {
254 buffer->alloc = (buffer->text.size + len) * 3 / 2;
255 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
256 }
257 line++;
258 len--;
259 memcpy(buffer->text.ptr + buffer->text.size, line, len);
260 buffer->text.size += len;
261 }
262
263 struct diff_words_data {
264 struct xdiff_emit_state xm;
265 struct diff_words_buffer minus, plus;
266 };
267
268 static void print_word(struct diff_words_buffer *buffer, int len, int color,
269 int suppress_newline)
270 {
271 const char *ptr;
272 int eol = 0;
273
274 if (len == 0)
275 return;
276
277 ptr = buffer->text.ptr + buffer->current;
278 buffer->current += len;
279
280 if (ptr[len - 1] == '\n') {
281 eol = 1;
282 len--;
283 }
284
285 fputs(diff_get_color(1, color), stdout);
286 fwrite(ptr, len, 1, stdout);
287 fputs(diff_get_color(1, DIFF_RESET), stdout);
288
289 if (eol) {
290 if (suppress_newline)
291 buffer->suppressed_newline = 1;
292 else
293 putchar('\n');
294 }
295 }
296
297 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
298 {
299 struct diff_words_data *diff_words = priv;
300
301 if (diff_words->minus.suppressed_newline) {
302 if (line[0] != '+')
303 putchar('\n');
304 diff_words->minus.suppressed_newline = 0;
305 }
306
307 len--;
308 switch (line[0]) {
309 case '-':
310 print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
311 break;
312 case '+':
313 print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
314 break;
315 case ' ':
316 print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
317 diff_words->minus.current += len;
318 break;
319 }
320 }
321
322 /* this executes the word diff on the accumulated buffers */
323 static void diff_words_show(struct diff_words_data *diff_words)
324 {
325 xpparam_t xpp;
326 xdemitconf_t xecfg;
327 xdemitcb_t ecb;
328 mmfile_t minus, plus;
329 int i;
330
331 minus.size = diff_words->minus.text.size;
332 minus.ptr = xmalloc(minus.size);
333 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
334 for (i = 0; i < minus.size; i++)
335 if (isspace(minus.ptr[i]))
336 minus.ptr[i] = '\n';
337 diff_words->minus.current = 0;
338
339 plus.size = diff_words->plus.text.size;
340 plus.ptr = xmalloc(plus.size);
341 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
342 for (i = 0; i < plus.size; i++)
343 if (isspace(plus.ptr[i]))
344 plus.ptr[i] = '\n';
345 diff_words->plus.current = 0;
346
347 xpp.flags = XDF_NEED_MINIMAL;
348 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
349 xecfg.flags = 0;
350 ecb.outf = xdiff_outf;
351 ecb.priv = diff_words;
352 diff_words->xm.consume = fn_out_diff_words_aux;
353 xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
354
355 free(minus.ptr);
356 free(plus.ptr);
357 diff_words->minus.text.size = diff_words->plus.text.size = 0;
358
359 if (diff_words->minus.suppressed_newline) {
360 putchar('\n');
361 diff_words->minus.suppressed_newline = 0;
362 }
363 }
364
365 struct emit_callback {
366 struct xdiff_emit_state xm;
367 int nparents, color_diff;
368 const char **label_path;
369 struct diff_words_data *diff_words;
370 };
371
372 static void free_diff_words_data(struct emit_callback *ecbdata)
373 {
374 if (ecbdata->diff_words) {
375 /* flush buffers */
376 if (ecbdata->diff_words->minus.text.size ||
377 ecbdata->diff_words->plus.text.size)
378 diff_words_show(ecbdata->diff_words);
379
380 if (ecbdata->diff_words->minus.text.ptr)
381 free (ecbdata->diff_words->minus.text.ptr);
382 if (ecbdata->diff_words->plus.text.ptr)
383 free (ecbdata->diff_words->plus.text.ptr);
384 free(ecbdata->diff_words);
385 ecbdata->diff_words = NULL;
386 }
387 }
388
389 const char *diff_get_color(int diff_use_color, enum color_diff ix)
390 {
391 if (diff_use_color)
392 return diff_colors[ix];
393 return "";
394 }
395
396 static void emit_line(const char *set, const char *reset, const char *line, int len)
397 {
398 if (len > 0 && line[len-1] == '\n')
399 len--;
400 fputs(set, stdout);
401 fwrite(line, len, 1, stdout);
402 puts(reset);
403 }
404
405 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
406 {
407 int col0 = ecbdata->nparents;
408 int last_tab_in_indent = -1;
409 int last_space_in_indent = -1;
410 int i;
411 int tail = len;
412 int need_highlight_leading_space = 0;
413 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
414 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
415
416 if (!*ws) {
417 emit_line(set, reset, line, len);
418 return;
419 }
420
421 /* The line is a newly added line. Does it have funny leading
422 * whitespaces? In indent, SP should never precede a TAB.
423 */
424 for (i = col0; i < len; i++) {
425 if (line[i] == '\t') {
426 last_tab_in_indent = i;
427 if (0 <= last_space_in_indent)
428 need_highlight_leading_space = 1;
429 }
430 else if (line[i] == ' ')
431 last_space_in_indent = i;
432 else
433 break;
434 }
435 fputs(set, stdout);
436 fwrite(line, col0, 1, stdout);
437 fputs(reset, stdout);
438 if (((i == len) || line[i] == '\n') && i != col0) {
439 /* The whole line was indent */
440 emit_line(ws, reset, line + col0, len - col0);
441 return;
442 }
443 i = col0;
444 if (need_highlight_leading_space) {
445 while (i < last_tab_in_indent) {
446 if (line[i] == ' ') {
447 fputs(ws, stdout);
448 putchar(' ');
449 fputs(reset, stdout);
450 }
451 else
452 putchar(line[i]);
453 i++;
454 }
455 }
456 tail = len - 1;
457 if (line[tail] == '\n' && i < tail)
458 tail--;
459 while (i < tail) {
460 if (!isspace(line[tail]))
461 break;
462 tail--;
463 }
464 if ((i < tail && line[tail + 1] != '\n')) {
465 /* This has whitespace between tail+1..len */
466 fputs(set, stdout);
467 fwrite(line + i, tail - i + 1, 1, stdout);
468 fputs(reset, stdout);
469 emit_line(ws, reset, line + tail + 1, len - tail - 1);
470 }
471 else
472 emit_line(set, reset, line + i, len - i);
473 }
474
475 static void fn_out_consume(void *priv, char *line, unsigned long len)
476 {
477 int i;
478 int color;
479 struct emit_callback *ecbdata = priv;
480 const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
481 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
482
483 if (ecbdata->label_path[0]) {
484 const char *name_a_tab, *name_b_tab;
485
486 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
487 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
488
489 printf("%s--- %s%s%s\n",
490 set, ecbdata->label_path[0], reset, name_a_tab);
491 printf("%s+++ %s%s%s\n",
492 set, ecbdata->label_path[1], reset, name_b_tab);
493 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
494 }
495
496 /* This is not really necessary for now because
497 * this codepath only deals with two-way diffs.
498 */
499 for (i = 0; i < len && line[i] == '@'; i++)
500 ;
501 if (2 <= i && i < len && line[i] == ' ') {
502 ecbdata->nparents = i - 1;
503 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
504 reset, line, len);
505 return;
506 }
507
508 if (len < ecbdata->nparents) {
509 set = reset;
510 emit_line(reset, reset, line, len);
511 return;
512 }
513
514 color = DIFF_PLAIN;
515 if (ecbdata->diff_words && ecbdata->nparents != 1)
516 /* fall back to normal diff */
517 free_diff_words_data(ecbdata);
518 if (ecbdata->diff_words) {
519 if (line[0] == '-') {
520 diff_words_append(line, len,
521 &ecbdata->diff_words->minus);
522 return;
523 } else if (line[0] == '+') {
524 diff_words_append(line, len,
525 &ecbdata->diff_words->plus);
526 return;
527 }
528 if (ecbdata->diff_words->minus.text.size ||
529 ecbdata->diff_words->plus.text.size)
530 diff_words_show(ecbdata->diff_words);
531 line++;
532 len--;
533 emit_line(set, reset, line, len);
534 return;
535 }
536 for (i = 0; i < ecbdata->nparents && len; i++) {
537 if (line[i] == '-')
538 color = DIFF_FILE_OLD;
539 else if (line[i] == '+')
540 color = DIFF_FILE_NEW;
541 }
542
543 if (color != DIFF_FILE_NEW) {
544 emit_line(diff_get_color(ecbdata->color_diff, color),
545 reset, line, len);
546 return;
547 }
548 emit_add_line(reset, ecbdata, line, len);
549 }
550
551 static char *pprint_rename(const char *a, const char *b)
552 {
553 const char *old = a;
554 const char *new = b;
555 char *name = NULL;
556 int pfx_length, sfx_length;
557 int len_a = strlen(a);
558 int len_b = strlen(b);
559
560 /* Find common prefix */
561 pfx_length = 0;
562 while (*old && *new && *old == *new) {
563 if (*old == '/')
564 pfx_length = old - a + 1;
565 old++;
566 new++;
567 }
568
569 /* Find common suffix */
570 old = a + len_a;
571 new = b + len_b;
572 sfx_length = 0;
573 while (a <= old && b <= new && *old == *new) {
574 if (*old == '/')
575 sfx_length = len_a - (old - a);
576 old--;
577 new--;
578 }
579
580 /*
581 * pfx{mid-a => mid-b}sfx
582 * {pfx-a => pfx-b}sfx
583 * pfx{sfx-a => sfx-b}
584 * name-a => name-b
585 */
586 if (pfx_length + sfx_length) {
587 int a_midlen = len_a - pfx_length - sfx_length;
588 int b_midlen = len_b - pfx_length - sfx_length;
589 if (a_midlen < 0) a_midlen = 0;
590 if (b_midlen < 0) b_midlen = 0;
591
592 name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
593 sprintf(name, "%.*s{%.*s => %.*s}%s",
594 pfx_length, a,
595 a_midlen, a + pfx_length,
596 b_midlen, b + pfx_length,
597 a + len_a - sfx_length);
598 }
599 else {
600 name = xmalloc(len_a + len_b + 5);
601 sprintf(name, "%s => %s", a, b);
602 }
603 return name;
604 }
605
606 struct diffstat_t {
607 struct xdiff_emit_state xm;
608
609 int nr;
610 int alloc;
611 struct diffstat_file {
612 char *name;
613 unsigned is_unmerged:1;
614 unsigned is_binary:1;
615 unsigned is_renamed:1;
616 unsigned int added, deleted;
617 } **files;
618 };
619
620 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
621 const char *name_a,
622 const char *name_b)
623 {
624 struct diffstat_file *x;
625 x = xcalloc(sizeof (*x), 1);
626 if (diffstat->nr == diffstat->alloc) {
627 diffstat->alloc = alloc_nr(diffstat->alloc);
628 diffstat->files = xrealloc(diffstat->files,
629 diffstat->alloc * sizeof(x));
630 }
631 diffstat->files[diffstat->nr++] = x;
632 if (name_b) {
633 x->name = pprint_rename(name_a, name_b);
634 x->is_renamed = 1;
635 }
636 else
637 x->name = xstrdup(name_a);
638 return x;
639 }
640
641 static void diffstat_consume(void *priv, char *line, unsigned long len)
642 {
643 struct diffstat_t *diffstat = priv;
644 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
645
646 if (line[0] == '+')
647 x->added++;
648 else if (line[0] == '-')
649 x->deleted++;
650 }
651
652 const char mime_boundary_leader[] = "------------";
653
654 static int scale_linear(int it, int width, int max_change)
655 {
656 /*
657 * make sure that at least one '-' is printed if there were deletions,
658 * and likewise for '+'.
659 */
660 if (max_change < 2)
661 return it;
662 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
663 }
664
665 static void show_name(const char *prefix, const char *name, int len,
666 const char *reset, const char *set)
667 {
668 printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
669 }
670
671 static void show_graph(char ch, int cnt, const char *set, const char *reset)
672 {
673 if (cnt <= 0)
674 return;
675 printf("%s", set);
676 while (cnt--)
677 putchar(ch);
678 printf("%s", reset);
679 }
680
681 static void show_stats(struct diffstat_t* data, struct diff_options *options)
682 {
683 int i, len, add, del, total, adds = 0, dels = 0;
684 int max_change = 0, max_len = 0;
685 int total_files = data->nr;
686 int width, name_width;
687 const char *reset, *set, *add_c, *del_c;
688
689 if (data->nr == 0)
690 return;
691
692 width = options->stat_width ? options->stat_width : 80;
693 name_width = options->stat_name_width ? options->stat_name_width : 50;
694
695 /* Sanity: give at least 5 columns to the graph,
696 * but leave at least 10 columns for the name.
697 */
698 if (width < name_width + 15) {
699 if (name_width <= 25)
700 width = name_width + 15;
701 else
702 name_width = width - 15;
703 }
704
705 /* Find the longest filename and max number of changes */
706 reset = diff_get_color(options->color_diff, DIFF_RESET);
707 set = diff_get_color(options->color_diff, DIFF_PLAIN);
708 add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
709 del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
710
711 for (i = 0; i < data->nr; i++) {
712 struct diffstat_file *file = data->files[i];
713 int change = file->added + file->deleted;
714
715 len = quote_c_style(file->name, NULL, NULL, 0);
716 if (len) {
717 char *qname = xmalloc(len + 1);
718 quote_c_style(file->name, qname, NULL, 0);
719 free(file->name);
720 file->name = qname;
721 }
722
723 len = strlen(file->name);
724 if (max_len < len)
725 max_len = len;
726
727 if (file->is_binary || file->is_unmerged)
728 continue;
729 if (max_change < change)
730 max_change = change;
731 }
732
733 /* Compute the width of the graph part;
734 * 10 is for one blank at the beginning of the line plus
735 * " | count " between the name and the graph.
736 *
737 * From here on, name_width is the width of the name area,
738 * and width is the width of the graph area.
739 */
740 name_width = (name_width < max_len) ? name_width : max_len;
741 if (width < (name_width + 10) + max_change)
742 width = width - (name_width + 10);
743 else
744 width = max_change;
745
746 for (i = 0; i < data->nr; i++) {
747 const char *prefix = "";
748 char *name = data->files[i]->name;
749 int added = data->files[i]->added;
750 int deleted = data->files[i]->deleted;
751 int name_len;
752
753 /*
754 * "scale" the filename
755 */
756 len = name_width;
757 name_len = strlen(name);
758 if (name_width < name_len) {
759 char *slash;
760 prefix = "...";
761 len -= 3;
762 name += name_len - len;
763 slash = strchr(name, '/');
764 if (slash)
765 name = slash;
766 }
767
768 if (data->files[i]->is_binary) {
769 show_name(prefix, name, len, reset, set);
770 printf(" Bin\n");
771 goto free_diffstat_file;
772 }
773 else if (data->files[i]->is_unmerged) {
774 show_name(prefix, name, len, reset, set);
775 printf(" Unmerged\n");
776 goto free_diffstat_file;
777 }
778 else if (!data->files[i]->is_renamed &&
779 (added + deleted == 0)) {
780 total_files--;
781 goto free_diffstat_file;
782 }
783
784 /*
785 * scale the add/delete
786 */
787 add = added;
788 del = deleted;
789 total = add + del;
790 adds += add;
791 dels += del;
792
793 if (width <= max_change) {
794 add = scale_linear(add, width, max_change);
795 del = scale_linear(del, width, max_change);
796 total = add + del;
797 }
798 show_name(prefix, name, len, reset, set);
799 printf("%5d ", added + deleted);
800 show_graph('+', add, add_c, reset);
801 show_graph('-', del, del_c, reset);
802 putchar('\n');
803 free_diffstat_file:
804 free(data->files[i]->name);
805 free(data->files[i]);
806 }
807 free(data->files);
808 printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
809 set, total_files, adds, dels, reset);
810 }
811
812 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
813 {
814 int i;
815
816 for (i = 0; i < data->nr; i++) {
817 struct diffstat_file *file = data->files[i];
818
819 printf("%d\t%d\t", file->added, file->deleted);
820 if (options->line_termination &&
821 quote_c_style(file->name, NULL, NULL, 0))
822 quote_c_style(file->name, NULL, stdout, 0);
823 else
824 fputs(file->name, stdout);
825 putchar(options->line_termination);
826 }
827 }
828
829 struct checkdiff_t {
830 struct xdiff_emit_state xm;
831 const char *filename;
832 int lineno;
833 };
834
835 static void checkdiff_consume(void *priv, char *line, unsigned long len)
836 {
837 struct checkdiff_t *data = priv;
838
839 if (line[0] == '+') {
840 int i, spaces = 0;
841
842 data->lineno++;
843
844 /* check space before tab */
845 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
846 if (line[i] == ' ')
847 spaces++;
848 if (line[i - 1] == '\t' && spaces)
849 printf("%s:%d: space before tab:%.*s\n",
850 data->filename, data->lineno, (int)len, line);
851
852 /* check white space at line end */
853 if (line[len - 1] == '\n')
854 len--;
855 if (isspace(line[len - 1]))
856 printf("%s:%d: white space at end: %.*s\n",
857 data->filename, data->lineno, (int)len, line);
858 } else if (line[0] == ' ')
859 data->lineno++;
860 else if (line[0] == '@') {
861 char *plus = strchr(line, '+');
862 if (plus)
863 data->lineno = strtol(plus, NULL, 10);
864 else
865 die("invalid diff");
866 }
867 }
868
869 static unsigned char *deflate_it(char *data,
870 unsigned long size,
871 unsigned long *result_size)
872 {
873 int bound;
874 unsigned char *deflated;
875 z_stream stream;
876
877 memset(&stream, 0, sizeof(stream));
878 deflateInit(&stream, zlib_compression_level);
879 bound = deflateBound(&stream, size);
880 deflated = xmalloc(bound);
881 stream.next_out = deflated;
882 stream.avail_out = bound;
883
884 stream.next_in = (unsigned char *)data;
885 stream.avail_in = size;
886 while (deflate(&stream, Z_FINISH) == Z_OK)
887 ; /* nothing */
888 deflateEnd(&stream);
889 *result_size = stream.total_out;
890 return deflated;
891 }
892
893 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
894 {
895 void *cp;
896 void *delta;
897 void *deflated;
898 void *data;
899 unsigned long orig_size;
900 unsigned long delta_size;
901 unsigned long deflate_size;
902 unsigned long data_size;
903
904 /* We could do deflated delta, or we could do just deflated two,
905 * whichever is smaller.
906 */
907 delta = NULL;
908 deflated = deflate_it(two->ptr, two->size, &deflate_size);
909 if (one->size && two->size) {
910 delta = diff_delta(one->ptr, one->size,
911 two->ptr, two->size,
912 &delta_size, deflate_size);
913 if (delta) {
914 void *to_free = delta;
915 orig_size = delta_size;
916 delta = deflate_it(delta, delta_size, &delta_size);
917 free(to_free);
918 }
919 }
920
921 if (delta && delta_size < deflate_size) {
922 printf("delta %lu\n", orig_size);
923 free(deflated);
924 data = delta;
925 data_size = delta_size;
926 }
927 else {
928 printf("literal %lu\n", two->size);
929 free(delta);
930 data = deflated;
931 data_size = deflate_size;
932 }
933
934 /* emit data encoded in base85 */
935 cp = data;
936 while (data_size) {
937 int bytes = (52 < data_size) ? 52 : data_size;
938 char line[70];
939 data_size -= bytes;
940 if (bytes <= 26)
941 line[0] = bytes + 'A' - 1;
942 else
943 line[0] = bytes - 26 + 'a' - 1;
944 encode_85(line + 1, cp, bytes);
945 cp = (char *) cp + bytes;
946 puts(line);
947 }
948 printf("\n");
949 free(data);
950 }
951
952 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
953 {
954 printf("GIT binary patch\n");
955 emit_binary_diff_body(one, two);
956 emit_binary_diff_body(two, one);
957 }
958
959 #define FIRST_FEW_BYTES 8000
960 static int mmfile_is_binary(mmfile_t *mf)
961 {
962 long sz = mf->size;
963 if (FIRST_FEW_BYTES < sz)
964 sz = FIRST_FEW_BYTES;
965 return !!memchr(mf->ptr, 0, sz);
966 }
967
968 static void builtin_diff(const char *name_a,
969 const char *name_b,
970 struct diff_filespec *one,
971 struct diff_filespec *two,
972 const char *xfrm_msg,
973 struct diff_options *o,
974 int complete_rewrite)
975 {
976 mmfile_t mf1, mf2;
977 const char *lbl[2];
978 char *a_one, *b_two;
979 const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
980 const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
981
982 a_one = quote_two("a/", name_a);
983 b_two = quote_two("b/", name_b);
984 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
985 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
986 printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
987 if (lbl[0][0] == '/') {
988 /* /dev/null */
989 printf("%snew file mode %06o%s\n", set, two->mode, reset);
990 if (xfrm_msg && xfrm_msg[0])
991 printf("%s%s%s\n", set, xfrm_msg, reset);
992 }
993 else if (lbl[1][0] == '/') {
994 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
995 if (xfrm_msg && xfrm_msg[0])
996 printf("%s%s%s\n", set, xfrm_msg, reset);
997 }
998 else {
999 if (one->mode != two->mode) {
1000 printf("%sold mode %06o%s\n", set, one->mode, reset);
1001 printf("%snew mode %06o%s\n", set, two->mode, reset);
1002 }
1003 if (xfrm_msg && xfrm_msg[0])
1004 printf("%s%s%s\n", set, xfrm_msg, reset);
1005 /*
1006 * we do not run diff between different kind
1007 * of objects.
1008 */
1009 if ((one->mode ^ two->mode) & S_IFMT)
1010 goto free_ab_and_return;
1011 if (complete_rewrite) {
1012 emit_rewrite_diff(name_a, name_b, one, two);
1013 goto free_ab_and_return;
1014 }
1015 }
1016
1017 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1018 die("unable to read files to diff");
1019
1020 if (!o->text && (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))) {
1021 /* Quite common confusing case */
1022 if (mf1.size == mf2.size &&
1023 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1024 goto free_ab_and_return;
1025 if (o->binary)
1026 emit_binary_diff(&mf1, &mf2);
1027 else
1028 printf("Binary files %s and %s differ\n",
1029 lbl[0], lbl[1]);
1030 }
1031 else {
1032 /* Crazy xdl interfaces.. */
1033 const char *diffopts = getenv("GIT_DIFF_OPTS");
1034 xpparam_t xpp;
1035 xdemitconf_t xecfg;
1036 xdemitcb_t ecb;
1037 struct emit_callback ecbdata;
1038
1039 memset(&ecbdata, 0, sizeof(ecbdata));
1040 ecbdata.label_path = lbl;
1041 ecbdata.color_diff = o->color_diff;
1042 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1043 xecfg.ctxlen = o->context;
1044 xecfg.flags = XDL_EMIT_FUNCNAMES;
1045 if (!diffopts)
1046 ;
1047 else if (!strncmp(diffopts, "--unified=", 10))
1048 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1049 else if (!strncmp(diffopts, "-u", 2))
1050 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1051 ecb.outf = xdiff_outf;
1052 ecb.priv = &ecbdata;
1053 ecbdata.xm.consume = fn_out_consume;
1054 if (o->color_diff_words)
1055 ecbdata.diff_words =
1056 xcalloc(1, sizeof(struct diff_words_data));
1057 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1058 if (o->color_diff_words)
1059 free_diff_words_data(&ecbdata);
1060 }
1061
1062 free_ab_and_return:
1063 free(a_one);
1064 free(b_two);
1065 return;
1066 }
1067
1068 static void builtin_diffstat(const char *name_a, const char *name_b,
1069 struct diff_filespec *one,
1070 struct diff_filespec *two,
1071 struct diffstat_t *diffstat,
1072 struct diff_options *o,
1073 int complete_rewrite)
1074 {
1075 mmfile_t mf1, mf2;
1076 struct diffstat_file *data;
1077
1078 data = diffstat_add(diffstat, name_a, name_b);
1079
1080 if (!one || !two) {
1081 data->is_unmerged = 1;
1082 return;
1083 }
1084 if (complete_rewrite) {
1085 diff_populate_filespec(one, 0);
1086 diff_populate_filespec(two, 0);
1087 data->deleted = count_lines(one->data, one->size);
1088 data->added = count_lines(two->data, two->size);
1089 return;
1090 }
1091 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1092 die("unable to read files to diff");
1093
1094 if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
1095 data->is_binary = 1;
1096 else {
1097 /* Crazy xdl interfaces.. */
1098 xpparam_t xpp;
1099 xdemitconf_t xecfg;
1100 xdemitcb_t ecb;
1101
1102 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1103 xecfg.ctxlen = 0;
1104 xecfg.flags = 0;
1105 ecb.outf = xdiff_outf;
1106 ecb.priv = diffstat;
1107 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1108 }
1109 }
1110
1111 static void builtin_checkdiff(const char *name_a, const char *name_b,
1112 struct diff_filespec *one,
1113 struct diff_filespec *two)
1114 {
1115 mmfile_t mf1, mf2;
1116 struct checkdiff_t data;
1117
1118 if (!two)
1119 return;
1120
1121 memset(&data, 0, sizeof(data));
1122 data.xm.consume = checkdiff_consume;
1123 data.filename = name_b ? name_b : name_a;
1124 data.lineno = 0;
1125
1126 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1127 die("unable to read files to diff");
1128
1129 if (mmfile_is_binary(&mf2))
1130 return;
1131 else {
1132 /* Crazy xdl interfaces.. */
1133 xpparam_t xpp;
1134 xdemitconf_t xecfg;
1135 xdemitcb_t ecb;
1136
1137 xpp.flags = XDF_NEED_MINIMAL;
1138 xecfg.ctxlen = 0;
1139 xecfg.flags = 0;
1140 ecb.outf = xdiff_outf;
1141 ecb.priv = &data;
1142 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1143 }
1144 }
1145
1146 struct diff_filespec *alloc_filespec(const char *path)
1147 {
1148 int namelen = strlen(path);
1149 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1150
1151 memset(spec, 0, sizeof(*spec));
1152 spec->path = (char *)(spec + 1);
1153 memcpy(spec->path, path, namelen+1);
1154 return spec;
1155 }
1156
1157 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1158 unsigned short mode)
1159 {
1160 if (mode) {
1161 spec->mode = canon_mode(mode);
1162 hashcpy(spec->sha1, sha1);
1163 spec->sha1_valid = !is_null_sha1(sha1);
1164 }
1165 }
1166
1167 /*
1168 * Given a name and sha1 pair, if the dircache tells us the file in
1169 * the work tree has that object contents, return true, so that
1170 * prepare_temp_file() does not have to inflate and extract.
1171 */
1172 static int work_tree_matches(const char *name, const unsigned char *sha1)
1173 {
1174 struct cache_entry *ce;
1175 struct stat st;
1176 int pos, len;
1177
1178 /* We do not read the cache ourselves here, because the
1179 * benchmark with my previous version that always reads cache
1180 * shows that it makes things worse for diff-tree comparing
1181 * two linux-2.6 kernel trees in an already checked out work
1182 * tree. This is because most diff-tree comparisons deal with
1183 * only a small number of files, while reading the cache is
1184 * expensive for a large project, and its cost outweighs the
1185 * savings we get by not inflating the object to a temporary
1186 * file. Practically, this code only helps when we are used
1187 * by diff-cache --cached, which does read the cache before
1188 * calling us.
1189 */
1190 if (!active_cache)
1191 return 0;
1192
1193 len = strlen(name);
1194 pos = cache_name_pos(name, len);
1195 if (pos < 0)
1196 return 0;
1197 ce = active_cache[pos];
1198 if ((lstat(name, &st) < 0) ||
1199 !S_ISREG(st.st_mode) || /* careful! */
1200 ce_match_stat(ce, &st, 0) ||
1201 hashcmp(sha1, ce->sha1))
1202 return 0;
1203 /* we return 1 only when we can stat, it is a regular file,
1204 * stat information matches, and sha1 recorded in the cache
1205 * matches. I.e. we know the file in the work tree really is
1206 * the same as the <name, sha1> pair.
1207 */
1208 return 1;
1209 }
1210
1211 static struct sha1_size_cache {
1212 unsigned char sha1[20];
1213 unsigned long size;
1214 } **sha1_size_cache;
1215 static int sha1_size_cache_nr, sha1_size_cache_alloc;
1216
1217 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
1218 int find_only,
1219 unsigned long size)
1220 {
1221 int first, last;
1222 struct sha1_size_cache *e;
1223
1224 first = 0;
1225 last = sha1_size_cache_nr;
1226 while (last > first) {
1227 int cmp, next = (last + first) >> 1;
1228 e = sha1_size_cache[next];
1229 cmp = hashcmp(e->sha1, sha1);
1230 if (!cmp)
1231 return e;
1232 if (cmp < 0) {
1233 last = next;
1234 continue;
1235 }
1236 first = next+1;
1237 }
1238 /* not found */
1239 if (find_only)
1240 return NULL;
1241 /* insert to make it at "first" */
1242 if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
1243 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
1244 sha1_size_cache = xrealloc(sha1_size_cache,
1245 sha1_size_cache_alloc *
1246 sizeof(*sha1_size_cache));
1247 }
1248 sha1_size_cache_nr++;
1249 if (first < sha1_size_cache_nr)
1250 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
1251 (sha1_size_cache_nr - first - 1) *
1252 sizeof(*sha1_size_cache));
1253 e = xmalloc(sizeof(struct sha1_size_cache));
1254 sha1_size_cache[first] = e;
1255 hashcpy(e->sha1, sha1);
1256 e->size = size;
1257 return e;
1258 }
1259
1260 /*
1261 * While doing rename detection and pickaxe operation, we may need to
1262 * grab the data for the blob (or file) for our own in-core comparison.
1263 * diff_filespec has data and size fields for this purpose.
1264 */
1265 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1266 {
1267 int err = 0;
1268 if (!DIFF_FILE_VALID(s))
1269 die("internal error: asking to populate invalid file.");
1270 if (S_ISDIR(s->mode))
1271 return -1;
1272
1273 if (!use_size_cache)
1274 size_only = 0;
1275
1276 if (s->data)
1277 return err;
1278 if (!s->sha1_valid ||
1279 work_tree_matches(s->path, s->sha1)) {
1280 struct stat st;
1281 int fd;
1282 if (lstat(s->path, &st) < 0) {
1283 if (errno == ENOENT) {
1284 err_empty:
1285 err = -1;
1286 empty:
1287 s->data = (char *)"";
1288 s->size = 0;
1289 return err;
1290 }
1291 }
1292 s->size = st.st_size;
1293 if (!s->size)
1294 goto empty;
1295 if (size_only)
1296 return 0;
1297 if (S_ISLNK(st.st_mode)) {
1298 int ret;
1299 s->data = xmalloc(s->size);
1300 s->should_free = 1;
1301 ret = readlink(s->path, s->data, s->size);
1302 if (ret < 0) {
1303 free(s->data);
1304 goto err_empty;
1305 }
1306 return 0;
1307 }
1308 fd = open(s->path, O_RDONLY);
1309 if (fd < 0)
1310 goto err_empty;
1311 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1312 close(fd);
1313 if (s->data == MAP_FAILED)
1314 goto err_empty;
1315 s->should_munmap = 1;
1316 }
1317 else {
1318 char type[20];
1319 struct sha1_size_cache *e;
1320
1321 if (size_only) {
1322 e = locate_size_cache(s->sha1, 1, 0);
1323 if (e) {
1324 s->size = e->size;
1325 return 0;
1326 }
1327 if (!sha1_object_info(s->sha1, type, &s->size))
1328 locate_size_cache(s->sha1, 0, s->size);
1329 }
1330 else {
1331 s->data = read_sha1_file(s->sha1, type, &s->size);
1332 s->should_free = 1;
1333 }
1334 }
1335 return 0;
1336 }
1337
1338 void diff_free_filespec_data(struct diff_filespec *s)
1339 {
1340 if (s->should_free)
1341 free(s->data);
1342 else if (s->should_munmap)
1343 munmap(s->data, s->size);
1344 s->should_free = s->should_munmap = 0;
1345 s->data = NULL;
1346 free(s->cnt_data);
1347 s->cnt_data = NULL;
1348 }
1349
1350 static void prep_temp_blob(struct diff_tempfile *temp,
1351 void *blob,
1352 unsigned long size,
1353 const unsigned char *sha1,
1354 int mode)
1355 {
1356 int fd;
1357
1358 fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
1359 if (fd < 0)
1360 die("unable to create temp-file");
1361 if (write(fd, blob, size) != size)
1362 die("unable to write temp-file");
1363 close(fd);
1364 temp->name = temp->tmp_path;
1365 strcpy(temp->hex, sha1_to_hex(sha1));
1366 temp->hex[40] = 0;
1367 sprintf(temp->mode, "%06o", mode);
1368 }
1369
1370 static void prepare_temp_file(const char *name,
1371 struct diff_tempfile *temp,
1372 struct diff_filespec *one)
1373 {
1374 if (!DIFF_FILE_VALID(one)) {
1375 not_a_valid_file:
1376 /* A '-' entry produces this for file-2, and
1377 * a '+' entry produces this for file-1.
1378 */
1379 temp->name = "/dev/null";
1380 strcpy(temp->hex, ".");
1381 strcpy(temp->mode, ".");
1382 return;
1383 }
1384
1385 if (!one->sha1_valid ||
1386 work_tree_matches(name, one->sha1)) {
1387 struct stat st;
1388 if (lstat(name, &st) < 0) {
1389 if (errno == ENOENT)
1390 goto not_a_valid_file;
1391 die("stat(%s): %s", name, strerror(errno));
1392 }
1393 if (S_ISLNK(st.st_mode)) {
1394 int ret;
1395 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1396 if (sizeof(buf) <= st.st_size)
1397 die("symlink too long: %s", name);
1398 ret = readlink(name, buf, st.st_size);
1399 if (ret < 0)
1400 die("readlink(%s)", name);
1401 prep_temp_blob(temp, buf, st.st_size,
1402 (one->sha1_valid ?
1403 one->sha1 : null_sha1),
1404 (one->sha1_valid ?
1405 one->mode : S_IFLNK));
1406 }
1407 else {
1408 /* we can borrow from the file in the work tree */
1409 temp->name = name;
1410 if (!one->sha1_valid)
1411 strcpy(temp->hex, sha1_to_hex(null_sha1));
1412 else
1413 strcpy(temp->hex, sha1_to_hex(one->sha1));
1414 /* Even though we may sometimes borrow the
1415 * contents from the work tree, we always want
1416 * one->mode. mode is trustworthy even when
1417 * !(one->sha1_valid), as long as
1418 * DIFF_FILE_VALID(one).
1419 */
1420 sprintf(temp->mode, "%06o", one->mode);
1421 }
1422 return;
1423 }
1424 else {
1425 if (diff_populate_filespec(one, 0))
1426 die("cannot read data blob for %s", one->path);
1427 prep_temp_blob(temp, one->data, one->size,
1428 one->sha1, one->mode);
1429 }
1430 }
1431
1432 static void remove_tempfile(void)
1433 {
1434 int i;
1435
1436 for (i = 0; i < 2; i++)
1437 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1438 unlink(diff_temp[i].name);
1439 diff_temp[i].name = NULL;
1440 }
1441 }
1442
1443 static void remove_tempfile_on_signal(int signo)
1444 {
1445 remove_tempfile();
1446 signal(SIGINT, SIG_DFL);
1447 raise(signo);
1448 }
1449
1450 static int spawn_prog(const char *pgm, const char **arg)
1451 {
1452 pid_t pid;
1453 int status;
1454
1455 fflush(NULL);
1456 pid = fork();
1457 if (pid < 0)
1458 die("unable to fork");
1459 if (!pid) {
1460 execvp(pgm, (char *const*) arg);
1461 exit(255);
1462 }
1463
1464 while (waitpid(pid, &status, 0) < 0) {
1465 if (errno == EINTR)
1466 continue;
1467 return -1;
1468 }
1469
1470 /* Earlier we did not check the exit status because
1471 * diff exits non-zero if files are different, and
1472 * we are not interested in knowing that. It was a
1473 * mistake which made it harder to quit a diff-*
1474 * session that uses the git-apply-patch-script as
1475 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
1476 * should also exit non-zero only when it wants to
1477 * abort the entire diff-* session.
1478 */
1479 if (WIFEXITED(status) && !WEXITSTATUS(status))
1480 return 0;
1481 return -1;
1482 }
1483
1484 /* An external diff command takes:
1485 *
1486 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1487 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1488 *
1489 */
1490 static void run_external_diff(const char *pgm,
1491 const char *name,
1492 const char *other,
1493 struct diff_filespec *one,
1494 struct diff_filespec *two,
1495 const char *xfrm_msg,
1496 int complete_rewrite)
1497 {
1498 const char *spawn_arg[10];
1499 struct diff_tempfile *temp = diff_temp;
1500 int retval;
1501 static int atexit_asked = 0;
1502 const char *othername;
1503 const char **arg = &spawn_arg[0];
1504
1505 othername = (other? other : name);
1506 if (one && two) {
1507 prepare_temp_file(name, &temp[0], one);
1508 prepare_temp_file(othername, &temp[1], two);
1509 if (! atexit_asked &&
1510 (temp[0].name == temp[0].tmp_path ||
1511 temp[1].name == temp[1].tmp_path)) {
1512 atexit_asked = 1;
1513 atexit(remove_tempfile);
1514 }
1515 signal(SIGINT, remove_tempfile_on_signal);
1516 }
1517
1518 if (one && two) {
1519 *arg++ = pgm;
1520 *arg++ = name;
1521 *arg++ = temp[0].name;
1522 *arg++ = temp[0].hex;
1523 *arg++ = temp[0].mode;
1524 *arg++ = temp[1].name;
1525 *arg++ = temp[1].hex;
1526 *arg++ = temp[1].mode;
1527 if (other) {
1528 *arg++ = other;
1529 *arg++ = xfrm_msg;
1530 }
1531 } else {
1532 *arg++ = pgm;
1533 *arg++ = name;
1534 }
1535 *arg = NULL;
1536 retval = spawn_prog(pgm, spawn_arg);
1537 remove_tempfile();
1538 if (retval) {
1539 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1540 exit(1);
1541 }
1542 }
1543
1544 static void run_diff_cmd(const char *pgm,
1545 const char *name,
1546 const char *other,
1547 struct diff_filespec *one,
1548 struct diff_filespec *two,
1549 const char *xfrm_msg,
1550 struct diff_options *o,
1551 int complete_rewrite)
1552 {
1553 if (pgm) {
1554 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1555 complete_rewrite);
1556 return;
1557 }
1558 if (one && two)
1559 builtin_diff(name, other ? other : name,
1560 one, two, xfrm_msg, o, complete_rewrite);
1561 else
1562 printf("* Unmerged path %s\n", name);
1563 }
1564
1565 static void diff_fill_sha1_info(struct diff_filespec *one)
1566 {
1567 if (DIFF_FILE_VALID(one)) {
1568 if (!one->sha1_valid) {
1569 struct stat st;
1570 if (lstat(one->path, &st) < 0)
1571 die("stat %s", one->path);
1572 if (index_path(one->sha1, one->path, &st, 0))
1573 die("cannot hash %s\n", one->path);
1574 }
1575 }
1576 else
1577 hashclr(one->sha1);
1578 }
1579
1580 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1581 {
1582 const char *pgm = external_diff();
1583 char msg[PATH_MAX*2+300], *xfrm_msg;
1584 struct diff_filespec *one;
1585 struct diff_filespec *two;
1586 const char *name;
1587 const char *other;
1588 char *name_munged, *other_munged;
1589 int complete_rewrite = 0;
1590 int len;
1591
1592 if (DIFF_PAIR_UNMERGED(p)) {
1593 /* unmerged */
1594 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1595 return;
1596 }
1597
1598 name = p->one->path;
1599 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1600 name_munged = quote_one(name);
1601 other_munged = quote_one(other);
1602 one = p->one; two = p->two;
1603
1604 diff_fill_sha1_info(one);
1605 diff_fill_sha1_info(two);
1606
1607 len = 0;
1608 switch (p->status) {
1609 case DIFF_STATUS_COPIED:
1610 len += snprintf(msg + len, sizeof(msg) - len,
1611 "similarity index %d%%\n"
1612 "copy from %s\n"
1613 "copy to %s\n",
1614 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1615 name_munged, other_munged);
1616 break;
1617 case DIFF_STATUS_RENAMED:
1618 len += snprintf(msg + len, sizeof(msg) - len,
1619 "similarity index %d%%\n"
1620 "rename from %s\n"
1621 "rename to %s\n",
1622 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1623 name_munged, other_munged);
1624 break;
1625 case DIFF_STATUS_MODIFIED:
1626 if (p->score) {
1627 len += snprintf(msg + len, sizeof(msg) - len,
1628 "dissimilarity index %d%%\n",
1629 (int)(0.5 + p->score *
1630 100.0/MAX_SCORE));
1631 complete_rewrite = 1;
1632 break;
1633 }
1634 /* fallthru */
1635 default:
1636 /* nothing */
1637 ;
1638 }
1639
1640 if (hashcmp(one->sha1, two->sha1)) {
1641 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1642
1643 if (o->binary) {
1644 mmfile_t mf;
1645 if ((!fill_mmfile(&mf, one) && mmfile_is_binary(&mf)) ||
1646 (!fill_mmfile(&mf, two) && mmfile_is_binary(&mf)))
1647 abbrev = 40;
1648 }
1649 len += snprintf(msg + len, sizeof(msg) - len,
1650 "index %.*s..%.*s",
1651 abbrev, sha1_to_hex(one->sha1),
1652 abbrev, sha1_to_hex(two->sha1));
1653 if (one->mode == two->mode)
1654 len += snprintf(msg + len, sizeof(msg) - len,
1655 " %06o", one->mode);
1656 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1657 }
1658
1659 if (len)
1660 msg[--len] = 0;
1661 xfrm_msg = len ? msg : NULL;
1662
1663 if (!pgm &&
1664 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1665 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1666 /* a filepair that changes between file and symlink
1667 * needs to be split into deletion and creation.
1668 */
1669 struct diff_filespec *null = alloc_filespec(two->path);
1670 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1671 free(null);
1672 null = alloc_filespec(one->path);
1673 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1674 free(null);
1675 }
1676 else
1677 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1678 complete_rewrite);
1679
1680 free(name_munged);
1681 free(other_munged);
1682 }
1683
1684 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1685 struct diffstat_t *diffstat)
1686 {
1687 const char *name;
1688 const char *other;
1689 int complete_rewrite = 0;
1690
1691 if (DIFF_PAIR_UNMERGED(p)) {
1692 /* unmerged */
1693 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
1694 return;
1695 }
1696
1697 name = p->one->path;
1698 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1699
1700 diff_fill_sha1_info(p->one);
1701 diff_fill_sha1_info(p->two);
1702
1703 if (p->status == DIFF_STATUS_MODIFIED && p->score)
1704 complete_rewrite = 1;
1705 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
1706 }
1707
1708 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
1709 {
1710 const char *name;
1711 const char *other;
1712
1713 if (DIFF_PAIR_UNMERGED(p)) {
1714 /* unmerged */
1715 return;
1716 }
1717
1718 name = p->one->path;
1719 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1720
1721 diff_fill_sha1_info(p->one);
1722 diff_fill_sha1_info(p->two);
1723
1724 builtin_checkdiff(name, other, p->one, p->two);
1725 }
1726
1727 void diff_setup(struct diff_options *options)
1728 {
1729 memset(options, 0, sizeof(*options));
1730 options->line_termination = '\n';
1731 options->break_opt = -1;
1732 options->rename_limit = -1;
1733 options->context = 3;
1734 options->msg_sep = "";
1735
1736 options->change = diff_change;
1737 options->add_remove = diff_addremove;
1738 options->color_diff = diff_use_color_default;
1739 options->detect_rename = diff_detect_rename_default;
1740 }
1741
1742 int diff_setup_done(struct diff_options *options)
1743 {
1744 int count = 0;
1745
1746 if (options->output_format & DIFF_FORMAT_NAME)
1747 count++;
1748 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
1749 count++;
1750 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
1751 count++;
1752 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
1753 count++;
1754 if (count > 1)
1755 die("--name-only, --name-status, --check and -s are mutually exclusive");
1756
1757 if (options->find_copies_harder)
1758 options->detect_rename = DIFF_DETECT_COPY;
1759
1760 if (options->output_format & (DIFF_FORMAT_NAME |
1761 DIFF_FORMAT_NAME_STATUS |
1762 DIFF_FORMAT_CHECKDIFF |
1763 DIFF_FORMAT_NO_OUTPUT))
1764 options->output_format &= ~(DIFF_FORMAT_RAW |
1765 DIFF_FORMAT_NUMSTAT |
1766 DIFF_FORMAT_DIFFSTAT |
1767 DIFF_FORMAT_SUMMARY |
1768 DIFF_FORMAT_PATCH);
1769
1770 /*
1771 * These cases always need recursive; we do not drop caller-supplied
1772 * recursive bits for other formats here.
1773 */
1774 if (options->output_format & (DIFF_FORMAT_PATCH |
1775 DIFF_FORMAT_NUMSTAT |
1776 DIFF_FORMAT_DIFFSTAT |
1777 DIFF_FORMAT_SUMMARY |
1778 DIFF_FORMAT_CHECKDIFF))
1779 options->recursive = 1;
1780 /*
1781 * Also pickaxe would not work very well if you do not say recursive
1782 */
1783 if (options->pickaxe)
1784 options->recursive = 1;
1785
1786 if (options->detect_rename && options->rename_limit < 0)
1787 options->rename_limit = diff_rename_limit_default;
1788 if (options->setup & DIFF_SETUP_USE_CACHE) {
1789 if (!active_cache)
1790 /* read-cache does not die even when it fails
1791 * so it is safe for us to do this here. Also
1792 * it does not smudge active_cache or active_nr
1793 * when it fails, so we do not have to worry about
1794 * cleaning it up ourselves either.
1795 */
1796 read_cache();
1797 }
1798 if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1799 use_size_cache = 1;
1800 if (options->abbrev <= 0 || 40 < options->abbrev)
1801 options->abbrev = 40; /* full */
1802
1803 return 0;
1804 }
1805
1806 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
1807 {
1808 char c, *eq;
1809 int len;
1810
1811 if (*arg != '-')
1812 return 0;
1813 c = *++arg;
1814 if (!c)
1815 return 0;
1816 if (c == arg_short) {
1817 c = *++arg;
1818 if (!c)
1819 return 1;
1820 if (val && isdigit(c)) {
1821 char *end;
1822 int n = strtoul(arg, &end, 10);
1823 if (*end)
1824 return 0;
1825 *val = n;
1826 return 1;
1827 }
1828 return 0;
1829 }
1830 if (c != '-')
1831 return 0;
1832 arg++;
1833 eq = strchr(arg, '=');
1834 if (eq)
1835 len = eq - arg;
1836 else
1837 len = strlen(arg);
1838 if (!len || strncmp(arg, arg_long, len))
1839 return 0;
1840 if (eq) {
1841 int n;
1842 char *end;
1843 if (!isdigit(*++eq))
1844 return 0;
1845 n = strtoul(eq, &end, 10);
1846 if (*end)
1847 return 0;
1848 *val = n;
1849 }
1850 return 1;
1851 }
1852
1853 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1854 {
1855 const char *arg = av[0];
1856 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1857 options->output_format |= DIFF_FORMAT_PATCH;
1858 else if (opt_arg(arg, 'U', "unified", &options->context))
1859 options->output_format |= DIFF_FORMAT_PATCH;
1860 else if (!strcmp(arg, "--raw"))
1861 options->output_format |= DIFF_FORMAT_RAW;
1862 else if (!strcmp(arg, "--patch-with-raw")) {
1863 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
1864 }
1865 else if (!strcmp(arg, "--numstat")) {
1866 options->output_format |= DIFF_FORMAT_NUMSTAT;
1867 }
1868 else if (!strncmp(arg, "--stat", 6)) {
1869 char *end;
1870 int width = options->stat_width;
1871 int name_width = options->stat_name_width;
1872 arg += 6;
1873 end = (char *)arg;
1874
1875 switch (*arg) {
1876 case '-':
1877 if (!strncmp(arg, "-width=", 7))
1878 width = strtoul(arg + 7, &end, 10);
1879 else if (!strncmp(arg, "-name-width=", 12))
1880 name_width = strtoul(arg + 12, &end, 10);
1881 break;
1882 case '=':
1883 width = strtoul(arg+1, &end, 10);
1884 if (*end == ',')
1885 name_width = strtoul(end+1, &end, 10);
1886 }
1887
1888 /* Important! This checks all the error cases! */
1889 if (*end)
1890 return 0;
1891 options->output_format |= DIFF_FORMAT_DIFFSTAT;
1892 options->stat_name_width = name_width;
1893 options->stat_width = width;
1894 }
1895 else if (!strcmp(arg, "--check"))
1896 options->output_format |= DIFF_FORMAT_CHECKDIFF;
1897 else if (!strcmp(arg, "--summary"))
1898 options->output_format |= DIFF_FORMAT_SUMMARY;
1899 else if (!strcmp(arg, "--patch-with-stat")) {
1900 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
1901 }
1902 else if (!strcmp(arg, "-z"))
1903 options->line_termination = 0;
1904 else if (!strncmp(arg, "-l", 2))
1905 options->rename_limit = strtoul(arg+2, NULL, 10);
1906 else if (!strcmp(arg, "--full-index"))
1907 options->full_index = 1;
1908 else if (!strcmp(arg, "--binary")) {
1909 options->output_format |= DIFF_FORMAT_PATCH;
1910 options->binary = 1;
1911 }
1912 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
1913 options->text = 1;
1914 }
1915 else if (!strcmp(arg, "--name-only"))
1916 options->output_format |= DIFF_FORMAT_NAME;
1917 else if (!strcmp(arg, "--name-status"))
1918 options->output_format |= DIFF_FORMAT_NAME_STATUS;
1919 else if (!strcmp(arg, "-R"))
1920 options->reverse_diff = 1;
1921 else if (!strncmp(arg, "-S", 2))
1922 options->pickaxe = arg + 2;
1923 else if (!strcmp(arg, "-s")) {
1924 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
1925 }
1926 else if (!strncmp(arg, "-O", 2))
1927 options->orderfile = arg + 2;
1928 else if (!strncmp(arg, "--diff-filter=", 14))
1929 options->filter = arg + 14;
1930 else if (!strcmp(arg, "--pickaxe-all"))
1931 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1932 else if (!strcmp(arg, "--pickaxe-regex"))
1933 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1934 else if (!strncmp(arg, "-B", 2)) {
1935 if ((options->break_opt =
1936 diff_scoreopt_parse(arg)) == -1)
1937 return -1;
1938 }
1939 else if (!strncmp(arg, "-M", 2)) {
1940 if ((options->rename_score =
1941 diff_scoreopt_parse(arg)) == -1)
1942 return -1;
1943 options->detect_rename = DIFF_DETECT_RENAME;
1944 }
1945 else if (!strncmp(arg, "-C", 2)) {
1946 if ((options->rename_score =
1947 diff_scoreopt_parse(arg)) == -1)
1948 return -1;
1949 options->detect_rename = DIFF_DETECT_COPY;
1950 }
1951 else if (!strcmp(arg, "--find-copies-harder"))
1952 options->find_copies_harder = 1;
1953 else if (!strcmp(arg, "--abbrev"))
1954 options->abbrev = DEFAULT_ABBREV;
1955 else if (!strncmp(arg, "--abbrev=", 9)) {
1956 options->abbrev = strtoul(arg + 9, NULL, 10);
1957 if (options->abbrev < MINIMUM_ABBREV)
1958 options->abbrev = MINIMUM_ABBREV;
1959 else if (40 < options->abbrev)
1960 options->abbrev = 40;
1961 }
1962 else if (!strcmp(arg, "--color"))
1963 options->color_diff = 1;
1964 else if (!strcmp(arg, "--no-color"))
1965 options->color_diff = 0;
1966 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
1967 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
1968 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
1969 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
1970 else if (!strcmp(arg, "--color-words"))
1971 options->color_diff = options->color_diff_words = 1;
1972 else if (!strcmp(arg, "--no-renames"))
1973 options->detect_rename = 0;
1974 else
1975 return 0;
1976 return 1;
1977 }
1978
1979 static int parse_num(const char **cp_p)
1980 {
1981 unsigned long num, scale;
1982 int ch, dot;
1983 const char *cp = *cp_p;
1984
1985 num = 0;
1986 scale = 1;
1987 dot = 0;
1988 for(;;) {
1989 ch = *cp;
1990 if ( !dot && ch == '.' ) {
1991 scale = 1;
1992 dot = 1;
1993 } else if ( ch == '%' ) {
1994 scale = dot ? scale*100 : 100;
1995 cp++; /* % is always at the end */
1996 break;
1997 } else if ( ch >= '0' && ch <= '9' ) {
1998 if ( scale < 100000 ) {
1999 scale *= 10;
2000 num = (num*10) + (ch-'0');
2001 }
2002 } else {
2003 break;
2004 }
2005 cp++;
2006 }
2007 *cp_p = cp;
2008
2009 /* user says num divided by scale and we say internally that
2010 * is MAX_SCORE * num / scale.
2011 */
2012 return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
2013 }
2014
2015 int diff_scoreopt_parse(const char *opt)
2016 {
2017 int opt1, opt2, cmd;
2018
2019 if (*opt++ != '-')
2020 return -1;
2021 cmd = *opt++;
2022 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2023 return -1; /* that is not a -M, -C nor -B option */
2024
2025 opt1 = parse_num(&opt);
2026 if (cmd != 'B')
2027 opt2 = 0;
2028 else {
2029 if (*opt == 0)
2030 opt2 = 0;
2031 else if (*opt != '/')
2032 return -1; /* we expect -B80/99 or -B80 */
2033 else {
2034 opt++;
2035 opt2 = parse_num(&opt);
2036 }
2037 }
2038 if (*opt != 0)
2039 return -1;
2040 return opt1 | (opt2 << 16);
2041 }
2042
2043 struct diff_queue_struct diff_queued_diff;
2044
2045 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2046 {
2047 if (queue->alloc <= queue->nr) {
2048 queue->alloc = alloc_nr(queue->alloc);
2049 queue->queue = xrealloc(queue->queue,
2050 sizeof(dp) * queue->alloc);
2051 }
2052 queue->queue[queue->nr++] = dp;
2053 }
2054
2055 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2056 struct diff_filespec *one,
2057 struct diff_filespec *two)
2058 {
2059 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2060 dp->one = one;
2061 dp->two = two;
2062 if (queue)
2063 diff_q(queue, dp);
2064 return dp;
2065 }
2066
2067 void diff_free_filepair(struct diff_filepair *p)
2068 {
2069 diff_free_filespec_data(p->one);
2070 diff_free_filespec_data(p->two);
2071 free(p->one);
2072 free(p->two);
2073 free(p);
2074 }
2075
2076 /* This is different from find_unique_abbrev() in that
2077 * it stuffs the result with dots for alignment.
2078 */
2079 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2080 {
2081 int abblen;
2082 const char *abbrev;
2083 if (len == 40)
2084 return sha1_to_hex(sha1);
2085
2086 abbrev = find_unique_abbrev(sha1, len);
2087 if (!abbrev)
2088 return sha1_to_hex(sha1);
2089 abblen = strlen(abbrev);
2090 if (abblen < 37) {
2091 static char hex[41];
2092 if (len < abblen && abblen <= len + 2)
2093 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2094 else
2095 sprintf(hex, "%s...", abbrev);
2096 return hex;
2097 }
2098 return sha1_to_hex(sha1);
2099 }
2100
2101 static void diff_flush_raw(struct diff_filepair *p,
2102 struct diff_options *options)
2103 {
2104 int two_paths;
2105 char status[10];
2106 int abbrev = options->abbrev;
2107 const char *path_one, *path_two;
2108 int inter_name_termination = '\t';
2109 int line_termination = options->line_termination;
2110
2111 if (!line_termination)
2112 inter_name_termination = 0;
2113
2114 path_one = p->one->path;
2115 path_two = p->two->path;
2116 if (line_termination) {
2117 path_one = quote_one(path_one);
2118 path_two = quote_one(path_two);
2119 }
2120
2121 if (p->score)
2122 sprintf(status, "%c%03d", p->status,
2123 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2124 else {
2125 status[0] = p->status;
2126 status[1] = 0;
2127 }
2128 switch (p->status) {
2129 case DIFF_STATUS_COPIED:
2130 case DIFF_STATUS_RENAMED:
2131 two_paths = 1;
2132 break;
2133 case DIFF_STATUS_ADDED:
2134 case DIFF_STATUS_DELETED:
2135 two_paths = 0;
2136 break;
2137 default:
2138 two_paths = 0;
2139 break;
2140 }
2141 if (!(options->output_format & DIFF_FORMAT_NAME_STATUS)) {
2142 printf(":%06o %06o %s ",
2143 p->one->mode, p->two->mode,
2144 diff_unique_abbrev(p->one->sha1, abbrev));
2145 printf("%s ",
2146 diff_unique_abbrev(p->two->sha1, abbrev));
2147 }
2148 printf("%s%c%s", status, inter_name_termination, path_one);
2149 if (two_paths)
2150 printf("%c%s", inter_name_termination, path_two);
2151 putchar(line_termination);
2152 if (path_one != p->one->path)
2153 free((void*)path_one);
2154 if (path_two != p->two->path)
2155 free((void*)path_two);
2156 }
2157
2158 static void diff_flush_name(struct diff_filepair *p, int line_termination)
2159 {
2160 char *path = p->two->path;
2161
2162 if (line_termination)
2163 path = quote_one(p->two->path);
2164 printf("%s%c", path, line_termination);
2165 if (p->two->path != path)
2166 free(path);
2167 }
2168
2169 int diff_unmodified_pair(struct diff_filepair *p)
2170 {
2171 /* This function is written stricter than necessary to support
2172 * the currently implemented transformers, but the idea is to
2173 * let transformers to produce diff_filepairs any way they want,
2174 * and filter and clean them up here before producing the output.
2175 */
2176 struct diff_filespec *one, *two;
2177
2178 if (DIFF_PAIR_UNMERGED(p))
2179 return 0; /* unmerged is interesting */
2180
2181 one = p->one;
2182 two = p->two;
2183
2184 /* deletion, addition, mode or type change
2185 * and rename are all interesting.
2186 */
2187 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2188 DIFF_PAIR_MODE_CHANGED(p) ||
2189 strcmp(one->path, two->path))
2190 return 0;
2191
2192 /* both are valid and point at the same path. that is, we are
2193 * dealing with a change.
2194 */
2195 if (one->sha1_valid && two->sha1_valid &&
2196 !hashcmp(one->sha1, two->sha1))
2197 return 1; /* no change */
2198 if (!one->sha1_valid && !two->sha1_valid)
2199 return 1; /* both look at the same file on the filesystem. */
2200 return 0;
2201 }
2202
2203 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2204 {
2205 if (diff_unmodified_pair(p))
2206 return;
2207
2208 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2209 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2210 return; /* no tree diffs in patch format */
2211
2212 run_diff(p, o);
2213 }
2214
2215 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2216 struct diffstat_t *diffstat)
2217 {
2218 if (diff_unmodified_pair(p))
2219 return;
2220
2221 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2222 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2223 return; /* no tree diffs in patch format */
2224
2225 run_diffstat(p, o, diffstat);
2226 }
2227
2228 static void diff_flush_checkdiff(struct diff_filepair *p,
2229 struct diff_options *o)
2230 {
2231 if (diff_unmodified_pair(p))
2232 return;
2233
2234 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2235 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2236 return; /* no tree diffs in patch format */
2237
2238 run_checkdiff(p, o);
2239 }
2240
2241 int diff_queue_is_empty(void)
2242 {
2243 struct diff_queue_struct *q = &diff_queued_diff;
2244 int i;
2245 for (i = 0; i < q->nr; i++)
2246 if (!diff_unmodified_pair(q->queue[i]))
2247 return 0;
2248 return 1;
2249 }
2250
2251 #if DIFF_DEBUG
2252 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2253 {
2254 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2255 x, one ? one : "",
2256 s->path,
2257 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2258 s->mode,
2259 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2260 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2261 x, one ? one : "",
2262 s->size, s->xfrm_flags);
2263 }
2264
2265 void diff_debug_filepair(const struct diff_filepair *p, int i)
2266 {
2267 diff_debug_filespec(p->one, i, "one");
2268 diff_debug_filespec(p->two, i, "two");
2269 fprintf(stderr, "score %d, status %c stays %d broken %d\n",
2270 p->score, p->status ? p->status : '?',
2271 p->source_stays, p->broken_pair);
2272 }
2273
2274 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2275 {
2276 int i;
2277 if (msg)
2278 fprintf(stderr, "%s\n", msg);
2279 fprintf(stderr, "q->nr = %d\n", q->nr);
2280 for (i = 0; i < q->nr; i++) {
2281 struct diff_filepair *p = q->queue[i];
2282 diff_debug_filepair(p, i);
2283 }
2284 }
2285 #endif
2286
2287 static void diff_resolve_rename_copy(void)
2288 {
2289 int i, j;
2290 struct diff_filepair *p, *pp;
2291 struct diff_queue_struct *q = &diff_queued_diff;
2292
2293 diff_debug_queue("resolve-rename-copy", q);
2294
2295 for (i = 0; i < q->nr; i++) {
2296 p = q->queue[i];
2297 p->status = 0; /* undecided */
2298 if (DIFF_PAIR_UNMERGED(p))
2299 p->status = DIFF_STATUS_UNMERGED;
2300 else if (!DIFF_FILE_VALID(p->one))
2301 p->status = DIFF_STATUS_ADDED;
2302 else if (!DIFF_FILE_VALID(p->two))
2303 p->status = DIFF_STATUS_DELETED;
2304 else if (DIFF_PAIR_TYPE_CHANGED(p))
2305 p->status = DIFF_STATUS_TYPE_CHANGED;
2306
2307 /* from this point on, we are dealing with a pair
2308 * whose both sides are valid and of the same type, i.e.
2309 * either in-place edit or rename/copy edit.
2310 */
2311 else if (DIFF_PAIR_RENAME(p)) {
2312 if (p->source_stays) {
2313 p->status = DIFF_STATUS_COPIED;
2314 continue;
2315 }
2316 /* See if there is some other filepair that
2317 * copies from the same source as us. If so
2318 * we are a copy. Otherwise we are either a
2319 * copy if the path stays, or a rename if it
2320 * does not, but we already handled "stays" case.
2321 */
2322 for (j = i + 1; j < q->nr; j++) {
2323 pp = q->queue[j];
2324 if (strcmp(pp->one->path, p->one->path))
2325 continue; /* not us */
2326 if (!DIFF_PAIR_RENAME(pp))
2327 continue; /* not a rename/copy */
2328 /* pp is a rename/copy from the same source */
2329 p->status = DIFF_STATUS_COPIED;
2330 break;
2331 }
2332 if (!p->status)
2333 p->status = DIFF_STATUS_RENAMED;
2334 }
2335 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2336 p->one->mode != p->two->mode)
2337 p->status = DIFF_STATUS_MODIFIED;
2338 else {
2339 /* This is a "no-change" entry and should not
2340 * happen anymore, but prepare for broken callers.
2341 */
2342 error("feeding unmodified %s to diffcore",
2343 p->one->path);
2344 p->status = DIFF_STATUS_UNKNOWN;
2345 }
2346 }
2347 diff_debug_queue("resolve-rename-copy done", q);
2348 }
2349
2350 static int check_pair_status(struct diff_filepair *p)
2351 {
2352 switch (p->status) {
2353 case DIFF_STATUS_UNKNOWN:
2354 return 0;
2355 case 0:
2356 die("internal error in diff-resolve-rename-copy");
2357 default:
2358 return 1;
2359 }
2360 }
2361
2362 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2363 {
2364 int fmt = opt->output_format;
2365
2366 if (fmt & DIFF_FORMAT_CHECKDIFF)
2367 diff_flush_checkdiff(p, opt);
2368 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2369 diff_flush_raw(p, opt);
2370 else if (fmt & DIFF_FORMAT_NAME)
2371 diff_flush_name(p, opt->line_termination);
2372 }
2373
2374 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2375 {
2376 if (fs->mode)
2377 printf(" %s mode %06o %s\n", newdelete, fs->mode, fs->path);
2378 else
2379 printf(" %s %s\n", newdelete, fs->path);
2380 }
2381
2382
2383 static void show_mode_change(struct diff_filepair *p, int show_name)
2384 {
2385 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2386 if (show_name)
2387 printf(" mode change %06o => %06o %s\n",
2388 p->one->mode, p->two->mode, p->two->path);
2389 else
2390 printf(" mode change %06o => %06o\n",
2391 p->one->mode, p->two->mode);
2392 }
2393 }
2394
2395 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2396 {
2397 const char *old, *new;
2398
2399 /* Find common prefix */
2400 old = p->one->path;
2401 new = p->two->path;
2402 while (1) {
2403 const char *slash_old, *slash_new;
2404 slash_old = strchr(old, '/');
2405 slash_new = strchr(new, '/');
2406 if (!slash_old ||
2407 !slash_new ||
2408 slash_old - old != slash_new - new ||
2409 memcmp(old, new, slash_new - new))
2410 break;
2411 old = slash_old + 1;
2412 new = slash_new + 1;
2413 }
2414 /* p->one->path thru old is the common prefix, and old and new
2415 * through the end of names are renames
2416 */
2417 if (old != p->one->path)
2418 printf(" %s %.*s{%s => %s} (%d%%)\n", renamecopy,
2419 (int)(old - p->one->path), p->one->path,
2420 old, new, (int)(0.5 + p->score * 100.0/MAX_SCORE));
2421 else
2422 printf(" %s %s => %s (%d%%)\n", renamecopy,
2423 p->one->path, p->two->path,
2424 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2425 show_mode_change(p, 0);
2426 }
2427
2428 static void diff_summary(struct diff_filepair *p)
2429 {
2430 switch(p->status) {
2431 case DIFF_STATUS_DELETED:
2432 show_file_mode_name("delete", p->one);
2433 break;
2434 case DIFF_STATUS_ADDED:
2435 show_file_mode_name("create", p->two);
2436 break;
2437 case DIFF_STATUS_COPIED:
2438 show_rename_copy("copy", p);
2439 break;
2440 case DIFF_STATUS_RENAMED:
2441 show_rename_copy("rename", p);
2442 break;
2443 default:
2444 if (p->score) {
2445 printf(" rewrite %s (%d%%)\n", p->two->path,
2446 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2447 show_mode_change(p, 0);
2448 } else show_mode_change(p, 1);
2449 break;
2450 }
2451 }
2452
2453 struct patch_id_t {
2454 struct xdiff_emit_state xm;
2455 SHA_CTX *ctx;
2456 int patchlen;
2457 };
2458
2459 static int remove_space(char *line, int len)
2460 {
2461 int i;
2462 char *dst = line;
2463 unsigned char c;
2464
2465 for (i = 0; i < len; i++)
2466 if (!isspace((c = line[i])))
2467 *dst++ = c;
2468
2469 return dst - line;
2470 }
2471
2472 static void patch_id_consume(void *priv, char *line, unsigned long len)
2473 {
2474 struct patch_id_t *data = priv;
2475 int new_len;
2476
2477 /* Ignore line numbers when computing the SHA1 of the patch */
2478 if (!strncmp(line, "@@ -", 4))
2479 return;
2480
2481 new_len = remove_space(line, len);
2482
2483 SHA1_Update(data->ctx, line, new_len);
2484 data->patchlen += new_len;
2485 }
2486
2487 /* returns 0 upon success, and writes result into sha1 */
2488 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2489 {
2490 struct diff_queue_struct *q = &diff_queued_diff;
2491 int i;
2492 SHA_CTX ctx;
2493 struct patch_id_t data;
2494 char buffer[PATH_MAX * 4 + 20];
2495
2496 SHA1_Init(&ctx);
2497 memset(&data, 0, sizeof(struct patch_id_t));
2498 data.ctx = &ctx;
2499 data.xm.consume = patch_id_consume;
2500
2501 for (i = 0; i < q->nr; i++) {
2502 xpparam_t xpp;
2503 xdemitconf_t xecfg;
2504 xdemitcb_t ecb;
2505 mmfile_t mf1, mf2;
2506 struct diff_filepair *p = q->queue[i];
2507 int len1, len2;
2508
2509 if (p->status == 0)
2510 return error("internal diff status error");
2511 if (p->status == DIFF_STATUS_UNKNOWN)
2512 continue;
2513 if (diff_unmodified_pair(p))
2514 continue;
2515 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2516 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2517 continue;
2518 if (DIFF_PAIR_UNMERGED(p))
2519 continue;
2520
2521 diff_fill_sha1_info(p->one);
2522 diff_fill_sha1_info(p->two);
2523 if (fill_mmfile(&mf1, p->one) < 0 ||
2524 fill_mmfile(&mf2, p->two) < 0)
2525 return error("unable to read files to diff");
2526
2527 /* Maybe hash p->two? into the patch id? */
2528 if (mmfile_is_binary(&mf2))
2529 continue;
2530
2531 len1 = remove_space(p->one->path, strlen(p->one->path));
2532 len2 = remove_space(p->two->path, strlen(p->two->path));
2533 if (p->one->mode == 0)
2534 len1 = snprintf(buffer, sizeof(buffer),
2535 "diff--gita/%.*sb/%.*s"
2536 "newfilemode%06o"
2537 "---/dev/null"
2538 "+++b/%.*s",
2539 len1, p->one->path,
2540 len2, p->two->path,
2541 p->two->mode,
2542 len2, p->two->path);
2543 else if (p->two->mode == 0)
2544 len1 = snprintf(buffer, sizeof(buffer),
2545 "diff--gita/%.*sb/%.*s"
2546 "deletedfilemode%06o"
2547 "---a/%.*s"
2548 "+++/dev/null",
2549 len1, p->one->path,
2550 len2, p->two->path,
2551 p->one->mode,
2552 len1, p->one->path);
2553 else
2554 len1 = snprintf(buffer, sizeof(buffer),
2555 "diff--gita/%.*sb/%.*s"
2556 "---a/%.*s"
2557 "+++b/%.*s",
2558 len1, p->one->path,
2559 len2, p->two->path,
2560 len1, p->one->path,
2561 len2, p->two->path);
2562 SHA1_Update(&ctx, buffer, len1);
2563
2564 xpp.flags = XDF_NEED_MINIMAL;
2565 xecfg.ctxlen = 3;
2566 xecfg.flags = XDL_EMIT_FUNCNAMES;
2567 ecb.outf = xdiff_outf;
2568 ecb.priv = &data;
2569 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2570 }
2571
2572 SHA1_Final(sha1, &ctx);
2573 return 0;
2574 }
2575
2576 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2577 {
2578 struct diff_queue_struct *q = &diff_queued_diff;
2579 int i;
2580 int result = diff_get_patch_id(options, sha1);
2581
2582 for (i = 0; i < q->nr; i++)
2583 diff_free_filepair(q->queue[i]);
2584
2585 free(q->queue);
2586 q->queue = NULL;
2587 q->nr = q->alloc = 0;
2588
2589 return result;
2590 }
2591
2592 static int is_summary_empty(const struct diff_queue_struct *q)
2593 {
2594 int i;
2595
2596 for (i = 0; i < q->nr; i++) {
2597 const struct diff_filepair *p = q->queue[i];
2598
2599 switch (p->status) {
2600 case DIFF_STATUS_DELETED:
2601 case DIFF_STATUS_ADDED:
2602 case DIFF_STATUS_COPIED:
2603 case DIFF_STATUS_RENAMED:
2604 return 0;
2605 default:
2606 if (p->score)
2607 return 0;
2608 if (p->one->mode && p->two->mode &&
2609 p->one->mode != p->two->mode)
2610 return 0;
2611 break;
2612 }
2613 }
2614 return 1;
2615 }
2616
2617 void diff_flush(struct diff_options *options)
2618 {
2619 struct diff_queue_struct *q = &diff_queued_diff;
2620 int i, output_format = options->output_format;
2621 int separator = 0;
2622
2623 /*
2624 * Order: raw, stat, summary, patch
2625 * or: name/name-status/checkdiff (other bits clear)
2626 */
2627 if (!q->nr)
2628 goto free_queue;
2629
2630 if (output_format & (DIFF_FORMAT_RAW |
2631 DIFF_FORMAT_NAME |
2632 DIFF_FORMAT_NAME_STATUS |
2633 DIFF_FORMAT_CHECKDIFF)) {
2634 for (i = 0; i < q->nr; i++) {
2635 struct diff_filepair *p = q->queue[i];
2636 if (check_pair_status(p))
2637 flush_one_pair(p, options);
2638 }
2639 separator++;
2640 }
2641
2642 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_NUMSTAT)) {
2643 struct diffstat_t diffstat;
2644
2645 memset(&diffstat, 0, sizeof(struct diffstat_t));
2646 diffstat.xm.consume = diffstat_consume;
2647 for (i = 0; i < q->nr; i++) {
2648 struct diff_filepair *p = q->queue[i];
2649 if (check_pair_status(p))
2650 diff_flush_stat(p, options, &diffstat);
2651 }
2652 if (output_format & DIFF_FORMAT_NUMSTAT)
2653 show_numstat(&diffstat, options);
2654 if (output_format & DIFF_FORMAT_DIFFSTAT)
2655 show_stats(&diffstat, options);
2656 separator++;
2657 }
2658
2659 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2660 for (i = 0; i < q->nr; i++)
2661 diff_summary(q->queue[i]);
2662 separator++;
2663 }
2664
2665 if (output_format & DIFF_FORMAT_PATCH) {
2666 if (separator) {
2667 if (options->stat_sep) {
2668 /* attach patch instead of inline */
2669 fputs(options->stat_sep, stdout);
2670 } else {
2671 putchar(options->line_termination);
2672 }
2673 }
2674
2675 for (i = 0; i < q->nr; i++) {
2676 struct diff_filepair *p = q->queue[i];
2677 if (check_pair_status(p))
2678 diff_flush_patch(p, options);
2679 }
2680 }
2681
2682 if (output_format & DIFF_FORMAT_CALLBACK)
2683 options->format_callback(q, options, options->format_callback_data);
2684
2685 for (i = 0; i < q->nr; i++)
2686 diff_free_filepair(q->queue[i]);
2687 free_queue:
2688 free(q->queue);
2689 q->queue = NULL;
2690 q->nr = q->alloc = 0;
2691 }
2692
2693 static void diffcore_apply_filter(const char *filter)
2694 {
2695 int i;
2696 struct diff_queue_struct *q = &diff_queued_diff;
2697 struct diff_queue_struct outq;
2698 outq.queue = NULL;
2699 outq.nr = outq.alloc = 0;
2700
2701 if (!filter)
2702 return;
2703
2704 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2705 int found;
2706 for (i = found = 0; !found && i < q->nr; i++) {
2707 struct diff_filepair *p = q->queue[i];
2708 if (((p->status == DIFF_STATUS_MODIFIED) &&
2709 ((p->score &&
2710 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2711 (!p->score &&
2712 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2713 ((p->status != DIFF_STATUS_MODIFIED) &&
2714 strchr(filter, p->status)))
2715 found++;
2716 }
2717 if (found)
2718 return;
2719
2720 /* otherwise we will clear the whole queue
2721 * by copying the empty outq at the end of this
2722 * function, but first clear the current entries
2723 * in the queue.
2724 */
2725 for (i = 0; i < q->nr; i++)
2726 diff_free_filepair(q->queue[i]);
2727 }
2728 else {
2729 /* Only the matching ones */
2730 for (i = 0; i < q->nr; i++) {
2731 struct diff_filepair *p = q->queue[i];
2732
2733 if (((p->status == DIFF_STATUS_MODIFIED) &&
2734 ((p->score &&
2735 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2736 (!p->score &&
2737 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2738 ((p->status != DIFF_STATUS_MODIFIED) &&
2739 strchr(filter, p->status)))
2740 diff_q(&outq, p);
2741 else
2742 diff_free_filepair(p);
2743 }
2744 }
2745 free(q->queue);
2746 *q = outq;
2747 }
2748
2749 void diffcore_std(struct diff_options *options)
2750 {
2751 if (options->break_opt != -1)
2752 diffcore_break(options->break_opt);
2753 if (options->detect_rename)
2754 diffcore_rename(options);
2755 if (options->break_opt != -1)
2756 diffcore_merge_broken();
2757 if (options->pickaxe)
2758 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2759 if (options->orderfile)
2760 diffcore_order(options->orderfile);
2761 diff_resolve_rename_copy();
2762 diffcore_apply_filter(options->filter);
2763 }
2764
2765
2766 void diffcore_std_no_resolve(struct diff_options *options)
2767 {
2768 if (options->pickaxe)
2769 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2770 if (options->orderfile)
2771 diffcore_order(options->orderfile);
2772 diffcore_apply_filter(options->filter);
2773 }
2774
2775 void diff_addremove(struct diff_options *options,
2776 int addremove, unsigned mode,
2777 const unsigned char *sha1,
2778 const char *base, const char *path)
2779 {
2780 char concatpath[PATH_MAX];
2781 struct diff_filespec *one, *two;
2782
2783 /* This may look odd, but it is a preparation for
2784 * feeding "there are unchanged files which should
2785 * not produce diffs, but when you are doing copy
2786 * detection you would need them, so here they are"
2787 * entries to the diff-core. They will be prefixed
2788 * with something like '=' or '*' (I haven't decided
2789 * which but should not make any difference).
2790 * Feeding the same new and old to diff_change()
2791 * also has the same effect.
2792 * Before the final output happens, they are pruned after
2793 * merged into rename/copy pairs as appropriate.
2794 */
2795 if (options->reverse_diff)
2796 addremove = (addremove == '+' ? '-' :
2797 addremove == '-' ? '+' : addremove);
2798
2799 if (!path) path = "";
2800 sprintf(concatpath, "%s%s", base, path);
2801 one = alloc_filespec(concatpath);
2802 two = alloc_filespec(concatpath);
2803
2804 if (addremove != '+')
2805 fill_filespec(one, sha1, mode);
2806 if (addremove != '-')
2807 fill_filespec(two, sha1, mode);
2808
2809 diff_queue(&diff_queued_diff, one, two);
2810 }
2811
2812 void diff_change(struct diff_options *options,
2813 unsigned old_mode, unsigned new_mode,
2814 const unsigned char *old_sha1,
2815 const unsigned char *new_sha1,
2816 const char *base, const char *path)
2817 {
2818 char concatpath[PATH_MAX];
2819 struct diff_filespec *one, *two;
2820
2821 if (options->reverse_diff) {
2822 unsigned tmp;
2823 const unsigned char *tmp_c;
2824 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
2825 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
2826 }
2827 if (!path) path = "";
2828 sprintf(concatpath, "%s%s", base, path);
2829 one = alloc_filespec(concatpath);
2830 two = alloc_filespec(concatpath);
2831 fill_filespec(one, old_sha1, old_mode);
2832 fill_filespec(two, new_sha1, new_mode);
2833
2834 diff_queue(&diff_queued_diff, one, two);
2835 }
2836
2837 void diff_unmerge(struct diff_options *options,
2838 const char *path)
2839 {
2840 struct diff_filespec *one, *two;
2841 one = alloc_filespec(path);
2842 two = alloc_filespec(path);
2843 diff_queue(&diff_queued_diff, one, two);
2844 }
The Best VCS Around