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
15#ifdef NO_FAST_WORKING_DIRECTORY
16#define FAST_WORKING_DIRECTORY 0
17#else
18#define FAST_WORKING_DIRECTORY 1
19#endif
20
21static int diff_detect_rename_default;
22static int diff_rename_limit_default = 200;
23int diff_use_color_default = -1;
24static const char *external_diff_cmd_cfg;
25int diff_auto_refresh_index = 1;
26
27static char diff_colors[][COLOR_MAXLEN] = {
28 "\033[m", /* reset */
29 "", /* PLAIN (normal) */
30 "\033[1m", /* METAINFO (bold) */
31 "\033[36m", /* FRAGINFO (cyan) */
32 "\033[31m", /* OLD (red) */
33 "\033[32m", /* NEW (green) */
34 "\033[33m", /* COMMIT (yellow) */
35 "\033[41m", /* WHITESPACE (red background) */
36};
37
38static int parse_diff_color_slot(const char *var, int ofs)
39{
40 if (!strcasecmp(var+ofs, "plain"))
41 return DIFF_PLAIN;
42 if (!strcasecmp(var+ofs, "meta"))
43 return DIFF_METAINFO;
44 if (!strcasecmp(var+ofs, "frag"))
45 return DIFF_FRAGINFO;
46 if (!strcasecmp(var+ofs, "old"))
47 return DIFF_FILE_OLD;
48 if (!strcasecmp(var+ofs, "new"))
49 return DIFF_FILE_NEW;
50 if (!strcasecmp(var+ofs, "commit"))
51 return DIFF_COMMIT;
52 if (!strcasecmp(var+ofs, "whitespace"))
53 return DIFF_WHITESPACE;
54 die("bad config variable '%s'", var);
55}
56
57static struct ll_diff_driver {
58 const char *name;
59 struct ll_diff_driver *next;
60 const char *cmd;
61} *user_diff, **user_diff_tail;
62
63/*
64 * Currently there is only "diff.<drivername>.command" variable;
65 * because there are "diff.color.<slot>" variables, we are parsing
66 * this in a bit convoluted way to allow low level diff driver
67 * called "color".
68 */
69static int parse_lldiff_command(const char *var, const char *ep, const char *value)
70{
71 const char *name;
72 int namelen;
73 struct ll_diff_driver *drv;
74
75 name = var + 5;
76 namelen = ep - name;
77 for (drv = user_diff; drv; drv = drv->next)
78 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
79 break;
80 if (!drv) {
81 drv = xcalloc(1, sizeof(struct ll_diff_driver));
82 drv->name = xmemdupz(name, namelen);
83 if (!user_diff_tail)
84 user_diff_tail = &user_diff;
85 *user_diff_tail = drv;
86 user_diff_tail = &(drv->next);
87 }
88
89 return git_config_string(&(drv->cmd), var, value);
90}
91
92/*
93 * 'diff.<what>.funcname' attribute can be specified in the configuration
94 * to define a customized regexp to find the beginning of a function to
95 * be used for hunk header lines of "diff -p" style output.
96 */
97static struct funcname_pattern {
98 char *name;
99 char *pattern;
100 struct funcname_pattern *next;
101} *funcname_pattern_list;
102
103static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
104{
105 const char *name;
106 int namelen;
107 struct funcname_pattern *pp;
108
109 name = var + 5; /* "diff." */
110 namelen = ep - name;
111
112 for (pp = funcname_pattern_list; pp; pp = pp->next)
113 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
114 break;
115 if (!pp) {
116 pp = xcalloc(1, sizeof(*pp));
117 pp->name = xmemdupz(name, namelen);
118 pp->next = funcname_pattern_list;
119 funcname_pattern_list = pp;
120 }
121 free(pp->pattern);
122 pp->pattern = xstrdup(value);
123 return 0;
124}
125
126/*
127 * These are to give UI layer defaults.
128 * The core-level commands such as git-diff-files should
129 * never be affected by the setting of diff.renames
130 * the user happens to have in the configuration file.
131 */
132int git_diff_ui_config(const char *var, const char *value, void *cb)
133{
134 if (!strcmp(var, "diff.renamelimit")) {
135 diff_rename_limit_default = git_config_int(var, value);
136 return 0;
137 }
138 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
139 diff_use_color_default = git_config_colorbool(var, value, -1);
140 return 0;
141 }
142 if (!strcmp(var, "diff.renames")) {
143 if (!value)
144 diff_detect_rename_default = DIFF_DETECT_RENAME;
145 else if (!strcasecmp(value, "copies") ||
146 !strcasecmp(value, "copy"))
147 diff_detect_rename_default = DIFF_DETECT_COPY;
148 else if (git_config_bool(var,value))
149 diff_detect_rename_default = DIFF_DETECT_RENAME;
150 return 0;
151 }
152 if (!strcmp(var, "diff.autorefreshindex")) {
153 diff_auto_refresh_index = git_config_bool(var, value);
154 return 0;
155 }
156 if (!strcmp(var, "diff.external")) {
157 if (!value)
158 return config_error_nonbool(var);
159 external_diff_cmd_cfg = xstrdup(value);
160 return 0;
161 }
162 if (!prefixcmp(var, "diff.")) {
163 const char *ep = strrchr(var, '.');
164
165 if (ep != var + 4 && !strcmp(ep, ".command"))
166 return parse_lldiff_command(var, ep, value);
167 }
168
169 return git_diff_basic_config(var, value, cb);
170}
171
172int git_diff_basic_config(const char *var, const char *value, void *cb)
173{
174 if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
175 int slot = parse_diff_color_slot(var, 11);
176 if (!value)
177 return config_error_nonbool(var);
178 color_parse(value, var, diff_colors[slot]);
179 return 0;
180 }
181
182 if (!prefixcmp(var, "diff.")) {
183 const char *ep = strrchr(var, '.');
184 if (ep != var + 4) {
185 if (!strcmp(ep, ".funcname")) {
186 if (!value)
187 return config_error_nonbool(var);
188 return parse_funcname_pattern(var, ep, value);
189 }
190 }
191 }
192
193 return git_color_default_config(var, value, cb);
194}
195
196static char *quote_two(const char *one, const char *two)
197{
198 int need_one = quote_c_style(one, NULL, NULL, 1);
199 int need_two = quote_c_style(two, NULL, NULL, 1);
200 struct strbuf res;
201
202 strbuf_init(&res, 0);
203 if (need_one + need_two) {
204 strbuf_addch(&res, '"');
205 quote_c_style(one, &res, NULL, 1);
206 quote_c_style(two, &res, NULL, 1);
207 strbuf_addch(&res, '"');
208 } else {
209 strbuf_addstr(&res, one);
210 strbuf_addstr(&res, two);
211 }
212 return strbuf_detach(&res, NULL);
213}
214
215static const char *external_diff(void)
216{
217 static const char *external_diff_cmd = NULL;
218 static int done_preparing = 0;
219
220 if (done_preparing)
221 return external_diff_cmd;
222 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
223 if (!external_diff_cmd)
224 external_diff_cmd = external_diff_cmd_cfg;
225 done_preparing = 1;
226 return external_diff_cmd;
227}
228
229static struct diff_tempfile {
230 const char *name; /* filename external diff should read from */
231 char hex[41];
232 char mode[10];
233 char tmp_path[PATH_MAX];
234} diff_temp[2];
235
236static int count_lines(const char *data, int size)
237{
238 int count, ch, completely_empty = 1, nl_just_seen = 0;
239 count = 0;
240 while (0 < size--) {
241 ch = *data++;
242 if (ch == '\n') {
243 count++;
244 nl_just_seen = 1;
245 completely_empty = 0;
246 }
247 else {
248 nl_just_seen = 0;
249 completely_empty = 0;
250 }
251 }
252 if (completely_empty)
253 return 0;
254 if (!nl_just_seen)
255 count++; /* no trailing newline */
256 return count;
257}
258
259static void print_line_count(FILE *file, int count)
260{
261 switch (count) {
262 case 0:
263 fprintf(file, "0,0");
264 break;
265 case 1:
266 fprintf(file, "1");
267 break;
268 default:
269 fprintf(file, "1,%d", count);
270 break;
271 }
272}
273
274static void copy_file_with_prefix(FILE *file,
275 int prefix, const char *data, int size,
276 const char *set, const char *reset)
277{
278 int ch, nl_just_seen = 1;
279 while (0 < size--) {
280 ch = *data++;
281 if (nl_just_seen) {
282 fputs(set, file);
283 putc(prefix, file);
284 }
285 if (ch == '\n') {
286 nl_just_seen = 1;
287 fputs(reset, file);
288 } else
289 nl_just_seen = 0;
290 putc(ch, file);
291 }
292 if (!nl_just_seen)
293 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
294}
295
296static void emit_rewrite_diff(const char *name_a,
297 const char *name_b,
298 struct diff_filespec *one,
299 struct diff_filespec *two,
300 struct diff_options *o)
301{
302 int lc_a, lc_b;
303 int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
304 const char *name_a_tab, *name_b_tab;
305 const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
306 const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
307 const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
308 const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
309 const char *reset = diff_get_color(color_diff, DIFF_RESET);
310 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
311
312 name_a += (*name_a == '/');
313 name_b += (*name_b == '/');
314 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
315 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
316
317 strbuf_reset(&a_name);
318 strbuf_reset(&b_name);
319 quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
320 quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
321
322 diff_populate_filespec(one, 0);
323 diff_populate_filespec(two, 0);
324 lc_a = count_lines(one->data, one->size);
325 lc_b = count_lines(two->data, two->size);
326 fprintf(o->file,
327 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
328 metainfo, a_name.buf, name_a_tab, reset,
329 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
330 print_line_count(o->file, lc_a);
331 fprintf(o->file, " +");
332 print_line_count(o->file, lc_b);
333 fprintf(o->file, " @@%s\n", reset);
334 if (lc_a)
335 copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
336 if (lc_b)
337 copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
338}
339
340static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
341{
342 if (!DIFF_FILE_VALID(one)) {
343 mf->ptr = (char *)""; /* does not matter */
344 mf->size = 0;
345 return 0;
346 }
347 else if (diff_populate_filespec(one, 0))
348 return -1;
349 mf->ptr = one->data;
350 mf->size = one->size;
351 return 0;
352}
353
354struct diff_words_buffer {
355 mmfile_t text;
356 long alloc;
357 long current; /* output pointer */
358 int suppressed_newline;
359};
360
361static void diff_words_append(char *line, unsigned long len,
362 struct diff_words_buffer *buffer)
363{
364 if (buffer->text.size + len > buffer->alloc) {
365 buffer->alloc = (buffer->text.size + len) * 3 / 2;
366 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
367 }
368 line++;
369 len--;
370 memcpy(buffer->text.ptr + buffer->text.size, line, len);
371 buffer->text.size += len;
372}
373
374struct diff_words_data {
375 struct xdiff_emit_state xm;
376 struct diff_words_buffer minus, plus;
377 FILE *file;
378};
379
380static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
381 int suppress_newline)
382{
383 const char *ptr;
384 int eol = 0;
385
386 if (len == 0)
387 return;
388
389 ptr = buffer->text.ptr + buffer->current;
390 buffer->current += len;
391
392 if (ptr[len - 1] == '\n') {
393 eol = 1;
394 len--;
395 }
396
397 fputs(diff_get_color(1, color), file);
398 fwrite(ptr, len, 1, file);
399 fputs(diff_get_color(1, DIFF_RESET), file);
400
401 if (eol) {
402 if (suppress_newline)
403 buffer->suppressed_newline = 1;
404 else
405 putc('\n', file);
406 }
407}
408
409static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
410{
411 struct diff_words_data *diff_words = priv;
412
413 if (diff_words->minus.suppressed_newline) {
414 if (line[0] != '+')
415 putc('\n', diff_words->file);
416 diff_words->minus.suppressed_newline = 0;
417 }
418
419 len--;
420 switch (line[0]) {
421 case '-':
422 print_word(diff_words->file,
423 &diff_words->minus, len, DIFF_FILE_OLD, 1);
424 break;
425 case '+':
426 print_word(diff_words->file,
427 &diff_words->plus, len, DIFF_FILE_NEW, 0);
428 break;
429 case ' ':
430 print_word(diff_words->file,
431 &diff_words->plus, len, DIFF_PLAIN, 0);
432 diff_words->minus.current += len;
433 break;
434 }
435}
436
437/* this executes the word diff on the accumulated buffers */
438static void diff_words_show(struct diff_words_data *diff_words)
439{
440 xpparam_t xpp;
441 xdemitconf_t xecfg;
442 xdemitcb_t ecb;
443 mmfile_t minus, plus;
444 int i;
445
446 memset(&xecfg, 0, sizeof(xecfg));
447 minus.size = diff_words->minus.text.size;
448 minus.ptr = xmalloc(minus.size);
449 memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
450 for (i = 0; i < minus.size; i++)
451 if (isspace(minus.ptr[i]))
452 minus.ptr[i] = '\n';
453 diff_words->minus.current = 0;
454
455 plus.size = diff_words->plus.text.size;
456 plus.ptr = xmalloc(plus.size);
457 memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
458 for (i = 0; i < plus.size; i++)
459 if (isspace(plus.ptr[i]))
460 plus.ptr[i] = '\n';
461 diff_words->plus.current = 0;
462
463 xpp.flags = XDF_NEED_MINIMAL;
464 xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
465 ecb.outf = xdiff_outf;
466 ecb.priv = diff_words;
467 diff_words->xm.consume = fn_out_diff_words_aux;
468 xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
469
470 free(minus.ptr);
471 free(plus.ptr);
472 diff_words->minus.text.size = diff_words->plus.text.size = 0;
473
474 if (diff_words->minus.suppressed_newline) {
475 putc('\n', diff_words->file);
476 diff_words->minus.suppressed_newline = 0;
477 }
478}
479
480typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
481
482struct emit_callback {
483 struct xdiff_emit_state xm;
484 int nparents, color_diff;
485 unsigned ws_rule;
486 sane_truncate_fn truncate;
487 const char **label_path;
488 struct diff_words_data *diff_words;
489 int *found_changesp;
490 FILE *file;
491};
492
493static void free_diff_words_data(struct emit_callback *ecbdata)
494{
495 if (ecbdata->diff_words) {
496 /* flush buffers */
497 if (ecbdata->diff_words->minus.text.size ||
498 ecbdata->diff_words->plus.text.size)
499 diff_words_show(ecbdata->diff_words);
500
501 free (ecbdata->diff_words->minus.text.ptr);
502 free (ecbdata->diff_words->plus.text.ptr);
503 free(ecbdata->diff_words);
504 ecbdata->diff_words = NULL;
505 }
506}
507
508const char *diff_get_color(int diff_use_color, enum color_diff ix)
509{
510 if (diff_use_color)
511 return diff_colors[ix];
512 return "";
513}
514
515static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
516{
517 int has_trailing_newline = (len > 0 && line[len-1] == '\n');
518 if (has_trailing_newline)
519 len--;
520
521 fputs(set, file);
522 fwrite(line, len, 1, file);
523 fputs(reset, file);
524 if (has_trailing_newline)
525 fputc('\n', file);
526}
527
528static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
529{
530 const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
531 const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
532
533 if (!*ws)
534 emit_line(ecbdata->file, set, reset, line, len);
535 else {
536 /* Emit just the prefix, then the rest. */
537 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
538 (void)check_and_emit_line(line + ecbdata->nparents,
539 len - ecbdata->nparents, ecbdata->ws_rule,
540 ecbdata->file, set, reset, ws);
541 }
542}
543
544static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
545{
546 const char *cp;
547 unsigned long allot;
548 size_t l = len;
549
550 if (ecb->truncate)
551 return ecb->truncate(line, len);
552 cp = line;
553 allot = l;
554 while (0 < l) {
555 (void) utf8_width(&cp, &l);
556 if (!cp)
557 break; /* truncated in the middle? */
558 }
559 return allot - l;
560}
561
562static void fn_out_consume(void *priv, char *line, unsigned long len)
563{
564 int i;
565 int color;
566 struct emit_callback *ecbdata = priv;
567 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
568 const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
569 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
570
571 *(ecbdata->found_changesp) = 1;
572
573 if (ecbdata->label_path[0]) {
574 const char *name_a_tab, *name_b_tab;
575
576 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
577 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
578
579 fprintf(ecbdata->file, "%s--- %s%s%s\n",
580 meta, ecbdata->label_path[0], reset, name_a_tab);
581 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
582 meta, ecbdata->label_path[1], reset, name_b_tab);
583 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
584 }
585
586 /* This is not really necessary for now because
587 * this codepath only deals with two-way diffs.
588 */
589 for (i = 0; i < len && line[i] == '@'; i++)
590 ;
591 if (2 <= i && i < len && line[i] == ' ') {
592 ecbdata->nparents = i - 1;
593 len = sane_truncate_line(ecbdata, line, len);
594 emit_line(ecbdata->file,
595 diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
596 reset, line, len);
597 if (line[len-1] != '\n')
598 putc('\n', ecbdata->file);
599 return;
600 }
601
602 if (len < ecbdata->nparents) {
603 emit_line(ecbdata->file, reset, reset, line, len);
604 return;
605 }
606
607 color = DIFF_PLAIN;
608 if (ecbdata->diff_words && ecbdata->nparents != 1)
609 /* fall back to normal diff */
610 free_diff_words_data(ecbdata);
611 if (ecbdata->diff_words) {
612 if (line[0] == '-') {
613 diff_words_append(line, len,
614 &ecbdata->diff_words->minus);
615 return;
616 } else if (line[0] == '+') {
617 diff_words_append(line, len,
618 &ecbdata->diff_words->plus);
619 return;
620 }
621 if (ecbdata->diff_words->minus.text.size ||
622 ecbdata->diff_words->plus.text.size)
623 diff_words_show(ecbdata->diff_words);
624 line++;
625 len--;
626 emit_line(ecbdata->file, plain, reset, line, len);
627 return;
628 }
629 for (i = 0; i < ecbdata->nparents && len; i++) {
630 if (line[i] == '-')
631 color = DIFF_FILE_OLD;
632 else if (line[i] == '+')
633 color = DIFF_FILE_NEW;
634 }
635
636 if (color != DIFF_FILE_NEW) {
637 emit_line(ecbdata->file,
638 diff_get_color(ecbdata->color_diff, color),
639 reset, line, len);
640 return;
641 }
642 emit_add_line(reset, ecbdata, line, len);
643}
644
645static char *pprint_rename(const char *a, const char *b)
646{
647 const char *old = a;
648 const char *new = b;
649 struct strbuf name;
650 int pfx_length, sfx_length;
651 int len_a = strlen(a);
652 int len_b = strlen(b);
653 int a_midlen, b_midlen;
654 int qlen_a = quote_c_style(a, NULL, NULL, 0);
655 int qlen_b = quote_c_style(b, NULL, NULL, 0);
656
657 strbuf_init(&name, 0);
658 if (qlen_a || qlen_b) {
659 quote_c_style(a, &name, NULL, 0);
660 strbuf_addstr(&name, " => ");
661 quote_c_style(b, &name, NULL, 0);
662 return strbuf_detach(&name, NULL);
663 }
664
665 /* Find common prefix */
666 pfx_length = 0;
667 while (*old && *new && *old == *new) {
668 if (*old == '/')
669 pfx_length = old - a + 1;
670 old++;
671 new++;
672 }
673
674 /* Find common suffix */
675 old = a + len_a;
676 new = b + len_b;
677 sfx_length = 0;
678 while (a <= old && b <= new && *old == *new) {
679 if (*old == '/')
680 sfx_length = len_a - (old - a);
681 old--;
682 new--;
683 }
684
685 /*
686 * pfx{mid-a => mid-b}sfx
687 * {pfx-a => pfx-b}sfx
688 * pfx{sfx-a => sfx-b}
689 * name-a => name-b
690 */
691 a_midlen = len_a - pfx_length - sfx_length;
692 b_midlen = len_b - pfx_length - sfx_length;
693 if (a_midlen < 0)
694 a_midlen = 0;
695 if (b_midlen < 0)
696 b_midlen = 0;
697
698 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
699 if (pfx_length + sfx_length) {
700 strbuf_add(&name, a, pfx_length);
701 strbuf_addch(&name, '{');
702 }
703 strbuf_add(&name, a + pfx_length, a_midlen);
704 strbuf_addstr(&name, " => ");
705 strbuf_add(&name, b + pfx_length, b_midlen);
706 if (pfx_length + sfx_length) {
707 strbuf_addch(&name, '}');
708 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
709 }
710 return strbuf_detach(&name, NULL);
711}
712
713struct diffstat_t {
714 struct xdiff_emit_state xm;
715
716 int nr;
717 int alloc;
718 struct diffstat_file {
719 char *from_name;
720 char *name;
721 char *print_name;
722 unsigned is_unmerged:1;
723 unsigned is_binary:1;
724 unsigned is_renamed:1;
725 unsigned int added, deleted;
726 } **files;
727};
728
729static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
730 const char *name_a,
731 const char *name_b)
732{
733 struct diffstat_file *x;
734 x = xcalloc(sizeof (*x), 1);
735 if (diffstat->nr == diffstat->alloc) {
736 diffstat->alloc = alloc_nr(diffstat->alloc);
737 diffstat->files = xrealloc(diffstat->files,
738 diffstat->alloc * sizeof(x));
739 }
740 diffstat->files[diffstat->nr++] = x;
741 if (name_b) {
742 x->from_name = xstrdup(name_a);
743 x->name = xstrdup(name_b);
744 x->is_renamed = 1;
745 }
746 else {
747 x->from_name = NULL;
748 x->name = xstrdup(name_a);
749 }
750 return x;
751}
752
753static void diffstat_consume(void *priv, char *line, unsigned long len)
754{
755 struct diffstat_t *diffstat = priv;
756 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
757
758 if (line[0] == '+')
759 x->added++;
760 else if (line[0] == '-')
761 x->deleted++;
762}
763
764const char mime_boundary_leader[] = "------------";
765
766static int scale_linear(int it, int width, int max_change)
767{
768 /*
769 * make sure that at least one '-' is printed if there were deletions,
770 * and likewise for '+'.
771 */
772 if (max_change < 2)
773 return it;
774 return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
775}
776
777static void show_name(FILE *file,
778 const char *prefix, const char *name, int len,
779 const char *reset, const char *set)
780{
781 fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
782}
783
784static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
785{
786 if (cnt <= 0)
787 return;
788 fprintf(file, "%s", set);
789 while (cnt--)
790 putc(ch, file);
791 fprintf(file, "%s", reset);
792}
793
794static void fill_print_name(struct diffstat_file *file)
795{
796 char *pname;
797
798 if (file->print_name)
799 return;
800
801 if (!file->is_renamed) {
802 struct strbuf buf;
803 strbuf_init(&buf, 0);
804 if (quote_c_style(file->name, &buf, NULL, 0)) {
805 pname = strbuf_detach(&buf, NULL);
806 } else {
807 pname = file->name;
808 strbuf_release(&buf);
809 }
810 } else {
811 pname = pprint_rename(file->from_name, file->name);
812 }
813 file->print_name = pname;
814}
815
816static void show_stats(struct diffstat_t* data, struct diff_options *options)
817{
818 int i, len, add, del, total, adds = 0, dels = 0;
819 int max_change = 0, max_len = 0;
820 int total_files = data->nr;
821 int width, name_width;
822 const char *reset, *set, *add_c, *del_c;
823
824 if (data->nr == 0)
825 return;
826
827 width = options->stat_width ? options->stat_width : 80;
828 name_width = options->stat_name_width ? options->stat_name_width : 50;
829
830 /* Sanity: give at least 5 columns to the graph,
831 * but leave at least 10 columns for the name.
832 */
833 if (width < 25)
834 width = 25;
835 if (name_width < 10)
836 name_width = 10;
837 else if (width < name_width + 15)
838 name_width = width - 15;
839
840 /* Find the longest filename and max number of changes */
841 reset = diff_get_color_opt(options, DIFF_RESET);
842 set = diff_get_color_opt(options, DIFF_PLAIN);
843 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
844 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
845
846 for (i = 0; i < data->nr; i++) {
847 struct diffstat_file *file = data->files[i];
848 int change = file->added + file->deleted;
849 fill_print_name(file);
850 len = strlen(file->print_name);
851 if (max_len < len)
852 max_len = len;
853
854 if (file->is_binary || file->is_unmerged)
855 continue;
856 if (max_change < change)
857 max_change = change;
858 }
859
860 /* Compute the width of the graph part;
861 * 10 is for one blank at the beginning of the line plus
862 * " | count " between the name and the graph.
863 *
864 * From here on, name_width is the width of the name area,
865 * and width is the width of the graph area.
866 */
867 name_width = (name_width < max_len) ? name_width : max_len;
868 if (width < (name_width + 10) + max_change)
869 width = width - (name_width + 10);
870 else
871 width = max_change;
872
873 for (i = 0; i < data->nr; i++) {
874 const char *prefix = "";
875 char *name = data->files[i]->print_name;
876 int added = data->files[i]->added;
877 int deleted = data->files[i]->deleted;
878 int name_len;
879
880 /*
881 * "scale" the filename
882 */
883 len = name_width;
884 name_len = strlen(name);
885 if (name_width < name_len) {
886 char *slash;
887 prefix = "...";
888 len -= 3;
889 name += name_len - len;
890 slash = strchr(name, '/');
891 if (slash)
892 name = slash;
893 }
894
895 if (data->files[i]->is_binary) {
896 show_name(options->file, prefix, name, len, reset, set);
897 fprintf(options->file, " Bin ");
898 fprintf(options->file, "%s%d%s", del_c, deleted, reset);
899 fprintf(options->file, " -> ");
900 fprintf(options->file, "%s%d%s", add_c, added, reset);
901 fprintf(options->file, " bytes");
902 fprintf(options->file, "\n");
903 continue;
904 }
905 else if (data->files[i]->is_unmerged) {
906 show_name(options->file, prefix, name, len, reset, set);
907 fprintf(options->file, " Unmerged\n");
908 continue;
909 }
910 else if (!data->files[i]->is_renamed &&
911 (added + deleted == 0)) {
912 total_files--;
913 continue;
914 }
915
916 /*
917 * scale the add/delete
918 */
919 add = added;
920 del = deleted;
921 total = add + del;
922 adds += add;
923 dels += del;
924
925 if (width <= max_change) {
926 add = scale_linear(add, width, max_change);
927 del = scale_linear(del, width, max_change);
928 total = add + del;
929 }
930 show_name(options->file, prefix, name, len, reset, set);
931 fprintf(options->file, "%5d%s", added + deleted,
932 added + deleted ? " " : "");
933 show_graph(options->file, '+', add, add_c, reset);
934 show_graph(options->file, '-', del, del_c, reset);
935 fprintf(options->file, "\n");
936 }
937 fprintf(options->file,
938 "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
939 set, total_files, adds, dels, reset);
940}
941
942static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
943{
944 int i, adds = 0, dels = 0, total_files = data->nr;
945
946 if (data->nr == 0)
947 return;
948
949 for (i = 0; i < data->nr; i++) {
950 if (!data->files[i]->is_binary &&
951 !data->files[i]->is_unmerged) {
952 int added = data->files[i]->added;
953 int deleted= data->files[i]->deleted;
954 if (!data->files[i]->is_renamed &&
955 (added + deleted == 0)) {
956 total_files--;
957 } else {
958 adds += added;
959 dels += deleted;
960 }
961 }
962 }
963 fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
964 total_files, adds, dels);
965}
966
967static void show_numstat(struct diffstat_t* data, struct diff_options *options)
968{
969 int i;
970
971 if (data->nr == 0)
972 return;
973
974 for (i = 0; i < data->nr; i++) {
975 struct diffstat_file *file = data->files[i];
976
977 if (file->is_binary)
978 fprintf(options->file, "-\t-\t");
979 else
980 fprintf(options->file,
981 "%d\t%d\t", file->added, file->deleted);
982 if (options->line_termination) {
983 fill_print_name(file);
984 if (!file->is_renamed)
985 write_name_quoted(file->name, options->file,
986 options->line_termination);
987 else {
988 fputs(file->print_name, options->file);
989 putc(options->line_termination, options->file);
990 }
991 } else {
992 if (file->is_renamed) {
993 putc('\0', options->file);
994 write_name_quoted(file->from_name, options->file, '\0');
995 }
996 write_name_quoted(file->name, options->file, '\0');
997 }
998 }
999}
1000
1001struct dirstat_file {
1002 const char *name;
1003 unsigned long changed;
1004};
1005
1006struct dirstat_dir {
1007 struct dirstat_file *files;
1008 int alloc, nr, percent, cumulative;
1009};
1010
1011static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
1012{
1013 unsigned long this_dir = 0;
1014 unsigned int sources = 0;
1015
1016 while (dir->nr) {
1017 struct dirstat_file *f = dir->files;
1018 int namelen = strlen(f->name);
1019 unsigned long this;
1020 char *slash;
1021
1022 if (namelen < baselen)
1023 break;
1024 if (memcmp(f->name, base, baselen))
1025 break;
1026 slash = strchr(f->name + baselen, '/');
1027 if (slash) {
1028 int newbaselen = slash + 1 - f->name;
1029 this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1030 sources++;
1031 } else {
1032 this = f->changed;
1033 dir->files++;
1034 dir->nr--;
1035 sources += 2;
1036 }
1037 this_dir += this;
1038 }
1039
1040 /*
1041 * We don't report dirstat's for
1042 * - the top level
1043 * - or cases where everything came from a single directory
1044 * under this directory (sources == 1).
1045 */
1046 if (baselen && sources != 1) {
1047 int permille = this_dir * 1000 / changed;
1048 if (permille) {
1049 int percent = permille / 10;
1050 if (percent >= dir->percent) {
1051 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1052 if (!dir->cumulative)
1053 return 0;
1054 }
1055 }
1056 }
1057 return this_dir;
1058}
1059
1060static void show_dirstat(struct diff_options *options)
1061{
1062 int i;
1063 unsigned long changed;
1064 struct dirstat_dir dir;
1065 struct diff_queue_struct *q = &diff_queued_diff;
1066
1067 dir.files = NULL;
1068 dir.alloc = 0;
1069 dir.nr = 0;
1070 dir.percent = options->dirstat_percent;
1071 dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
1072
1073 changed = 0;
1074 for (i = 0; i < q->nr; i++) {
1075 struct diff_filepair *p = q->queue[i];
1076 const char *name;
1077 unsigned long copied, added, damage;
1078
1079 name = p->one->path ? p->one->path : p->two->path;
1080
1081 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1082 diff_populate_filespec(p->one, 0);
1083 diff_populate_filespec(p->two, 0);
1084 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1085 &copied, &added);
1086 diff_free_filespec_data(p->one);
1087 diff_free_filespec_data(p->two);
1088 } else if (DIFF_FILE_VALID(p->one)) {
1089 diff_populate_filespec(p->one, 1);
1090 copied = added = 0;
1091 diff_free_filespec_data(p->one);
1092 } else if (DIFF_FILE_VALID(p->two)) {
1093 diff_populate_filespec(p->two, 1);
1094 copied = 0;
1095 added = p->two->size;
1096 diff_free_filespec_data(p->two);
1097 } else
1098 continue;
1099
1100 /*
1101 * Original minus copied is the removed material,
1102 * added is the new material. They are both damages
1103 * made to the preimage.
1104 */
1105 damage = (p->one->size - copied) + added;
1106
1107 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1108 dir.files[dir.nr].name = name;
1109 dir.files[dir.nr].changed = damage;
1110 changed += damage;
1111 dir.nr++;
1112 }
1113
1114 /* This can happen even with many files, if everything was renames */
1115 if (!changed)
1116 return;
1117
1118 /* Show all directories with more than x% of the changes */
1119 gather_dirstat(options->file, &dir, changed, "", 0);
1120}
1121
1122static void free_diffstat_info(struct diffstat_t *diffstat)
1123{
1124 int i;
1125 for (i = 0; i < diffstat->nr; i++) {
1126 struct diffstat_file *f = diffstat->files[i];
1127 if (f->name != f->print_name)
1128 free(f->print_name);
1129 free(f->name);
1130 free(f->from_name);
1131 free(f);
1132 }
1133 free(diffstat->files);
1134}
1135
1136struct checkdiff_t {
1137 struct xdiff_emit_state xm;
1138 const char *filename;
1139 int lineno, color_diff;
1140 unsigned ws_rule;
1141 unsigned status;
1142 FILE *file;
1143};
1144
1145static void checkdiff_consume(void *priv, char *line, unsigned long len)
1146{
1147 struct checkdiff_t *data = priv;
1148 const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1149 const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1150 const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1151 char *err;
1152
1153 if (line[0] == '+') {
1154 unsigned bad;
1155 data->lineno++;
1156 bad = check_and_emit_line(line + 1, len - 1,
1157 data->ws_rule, NULL, NULL, NULL, NULL);
1158 if (!bad)
1159 return;
1160 data->status |= bad;
1161 err = whitespace_error_string(bad);
1162 fprintf(data->file, "%s:%d: %s.\n", data->filename, data->lineno, err);
1163 free(err);
1164 emit_line(data->file, set, reset, line, 1);
1165 (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
1166 data->file, set, reset, ws);
1167 } else if (line[0] == ' ')
1168 data->lineno++;
1169 else if (line[0] == '@') {
1170 char *plus = strchr(line, '+');
1171 if (plus)
1172 data->lineno = strtol(plus, NULL, 10) - 1;
1173 else
1174 die("invalid diff");
1175 }
1176}
1177
1178static unsigned char *deflate_it(char *data,
1179 unsigned long size,
1180 unsigned long *result_size)
1181{
1182 int bound;
1183 unsigned char *deflated;
1184 z_stream stream;
1185
1186 memset(&stream, 0, sizeof(stream));
1187 deflateInit(&stream, zlib_compression_level);
1188 bound = deflateBound(&stream, size);
1189 deflated = xmalloc(bound);
1190 stream.next_out = deflated;
1191 stream.avail_out = bound;
1192
1193 stream.next_in = (unsigned char *)data;
1194 stream.avail_in = size;
1195 while (deflate(&stream, Z_FINISH) == Z_OK)
1196 ; /* nothing */
1197 deflateEnd(&stream);
1198 *result_size = stream.total_out;
1199 return deflated;
1200}
1201
1202static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1203{
1204 void *cp;
1205 void *delta;
1206 void *deflated;
1207 void *data;
1208 unsigned long orig_size;
1209 unsigned long delta_size;
1210 unsigned long deflate_size;
1211 unsigned long data_size;
1212
1213 /* We could do deflated delta, or we could do just deflated two,
1214 * whichever is smaller.
1215 */
1216 delta = NULL;
1217 deflated = deflate_it(two->ptr, two->size, &deflate_size);
1218 if (one->size && two->size) {
1219 delta = diff_delta(one->ptr, one->size,
1220 two->ptr, two->size,
1221 &delta_size, deflate_size);
1222 if (delta) {
1223 void *to_free = delta;
1224 orig_size = delta_size;
1225 delta = deflate_it(delta, delta_size, &delta_size);
1226 free(to_free);
1227 }
1228 }
1229
1230 if (delta && delta_size < deflate_size) {
1231 fprintf(file, "delta %lu\n", orig_size);
1232 free(deflated);
1233 data = delta;
1234 data_size = delta_size;
1235 }
1236 else {
1237 fprintf(file, "literal %lu\n", two->size);
1238 free(delta);
1239 data = deflated;
1240 data_size = deflate_size;
1241 }
1242
1243 /* emit data encoded in base85 */
1244 cp = data;
1245 while (data_size) {
1246 int bytes = (52 < data_size) ? 52 : data_size;
1247 char line[70];
1248 data_size -= bytes;
1249 if (bytes <= 26)
1250 line[0] = bytes + 'A' - 1;
1251 else
1252 line[0] = bytes - 26 + 'a' - 1;
1253 encode_85(line + 1, cp, bytes);
1254 cp = (char *) cp + bytes;
1255 fputs(line, file);
1256 fputc('\n', file);
1257 }
1258 fprintf(file, "\n");
1259 free(data);
1260}
1261
1262static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1263{
1264 fprintf(file, "GIT binary patch\n");
1265 emit_binary_diff_body(file, one, two);
1266 emit_binary_diff_body(file, two, one);
1267}
1268
1269static void setup_diff_attr_check(struct git_attr_check *check)
1270{
1271 static struct git_attr *attr_diff;
1272
1273 if (!attr_diff) {
1274 attr_diff = git_attr("diff", 4);
1275 }
1276 check[0].attr = attr_diff;
1277}
1278
1279static void diff_filespec_check_attr(struct diff_filespec *one)
1280{
1281 struct git_attr_check attr_diff_check;
1282 int check_from_data = 0;
1283
1284 if (one->checked_attr)
1285 return;
1286
1287 setup_diff_attr_check(&attr_diff_check);
1288 one->is_binary = 0;
1289 one->funcname_pattern_ident = NULL;
1290
1291 if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1292 const char *value;
1293
1294 /* binaryness */
1295 value = attr_diff_check.value;
1296 if (ATTR_TRUE(value))
1297 ;
1298 else if (ATTR_FALSE(value))
1299 one->is_binary = 1;
1300 else
1301 check_from_data = 1;
1302
1303 /* funcname pattern ident */
1304 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1305 ;
1306 else
1307 one->funcname_pattern_ident = value;
1308 }
1309
1310 if (check_from_data) {
1311 if (!one->data && DIFF_FILE_VALID(one))
1312 diff_populate_filespec(one, 0);
1313
1314 if (one->data)
1315 one->is_binary = buffer_is_binary(one->data, one->size);
1316 }
1317}
1318
1319int diff_filespec_is_binary(struct diff_filespec *one)
1320{
1321 diff_filespec_check_attr(one);
1322 return one->is_binary;
1323}
1324
1325static const char *funcname_pattern(const char *ident)
1326{
1327 struct funcname_pattern *pp;
1328
1329 for (pp = funcname_pattern_list; pp; pp = pp->next)
1330 if (!strcmp(ident, pp->name))
1331 return pp->pattern;
1332 return NULL;
1333}
1334
1335static struct builtin_funcname_pattern {
1336 const char *name;
1337 const char *pattern;
1338} builtin_funcname_pattern[] = {
1339 { "java", "!^[ ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1340 "new\\|return\\|switch\\|throw\\|while\\)\n"
1341 "^[ ]*\\(\\([ ]*"
1342 "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1343 "[ ]*([^;]*\\)$" },
1344 { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1345};
1346
1347static const char *diff_funcname_pattern(struct diff_filespec *one)
1348{
1349 const char *ident, *pattern;
1350 int i;
1351
1352 diff_filespec_check_attr(one);
1353 ident = one->funcname_pattern_ident;
1354
1355 if (!ident)
1356 /*
1357 * If the config file has "funcname.default" defined, that
1358 * regexp is used; otherwise NULL is returned and xemit uses
1359 * the built-in default.
1360 */
1361 return funcname_pattern("default");
1362
1363 /* Look up custom "funcname.$ident" regexp from config. */
1364 pattern = funcname_pattern(ident);
1365 if (pattern)
1366 return pattern;
1367
1368 /*
1369 * And define built-in fallback patterns here. Note that
1370 * these can be overridden by the user's config settings.
1371 */
1372 for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1373 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1374 return builtin_funcname_pattern[i].pattern;
1375
1376 return NULL;
1377}
1378
1379static void builtin_diff(const char *name_a,
1380 const char *name_b,
1381 struct diff_filespec *one,
1382 struct diff_filespec *two,
1383 const char *xfrm_msg,
1384 struct diff_options *o,
1385 int complete_rewrite)
1386{
1387 mmfile_t mf1, mf2;
1388 const char *lbl[2];
1389 char *a_one, *b_two;
1390 const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1391 const char *reset = diff_get_color_opt(o, DIFF_RESET);
1392
1393 a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
1394 b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
1395 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1396 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1397 fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1398 if (lbl[0][0] == '/') {
1399 /* /dev/null */
1400 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1401 if (xfrm_msg && xfrm_msg[0])
1402 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1403 }
1404 else if (lbl[1][0] == '/') {
1405 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1406 if (xfrm_msg && xfrm_msg[0])
1407 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1408 }
1409 else {
1410 if (one->mode != two->mode) {
1411 fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1412 fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1413 }
1414 if (xfrm_msg && xfrm_msg[0])
1415 fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1416 /*
1417 * we do not run diff between different kind
1418 * of objects.
1419 */
1420 if ((one->mode ^ two->mode) & S_IFMT)
1421 goto free_ab_and_return;
1422 if (complete_rewrite) {
1423 emit_rewrite_diff(name_a, name_b, one, two, o);
1424 o->found_changes = 1;
1425 goto free_ab_and_return;
1426 }
1427 }
1428
1429 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1430 die("unable to read files to diff");
1431
1432 if (!DIFF_OPT_TST(o, TEXT) &&
1433 (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1434 /* Quite common confusing case */
1435 if (mf1.size == mf2.size &&
1436 !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1437 goto free_ab_and_return;
1438 if (DIFF_OPT_TST(o, BINARY))
1439 emit_binary_diff(o->file, &mf1, &mf2);
1440 else
1441 fprintf(o->file, "Binary files %s and %s differ\n",
1442 lbl[0], lbl[1]);
1443 o->found_changes = 1;
1444 }
1445 else {
1446 /* Crazy xdl interfaces.. */
1447 const char *diffopts = getenv("GIT_DIFF_OPTS");
1448 xpparam_t xpp;
1449 xdemitconf_t xecfg;
1450 xdemitcb_t ecb;
1451 struct emit_callback ecbdata;
1452 const char *funcname_pattern;
1453
1454 funcname_pattern = diff_funcname_pattern(one);
1455 if (!funcname_pattern)
1456 funcname_pattern = diff_funcname_pattern(two);
1457
1458 memset(&xecfg, 0, sizeof(xecfg));
1459 memset(&ecbdata, 0, sizeof(ecbdata));
1460 ecbdata.label_path = lbl;
1461 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1462 ecbdata.found_changesp = &o->found_changes;
1463 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1464 ecbdata.file = o->file;
1465 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1466 xecfg.ctxlen = o->context;
1467 xecfg.flags = XDL_EMIT_FUNCNAMES;
1468 if (funcname_pattern)
1469 xdiff_set_find_func(&xecfg, funcname_pattern);
1470 if (!diffopts)
1471 ;
1472 else if (!prefixcmp(diffopts, "--unified="))
1473 xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1474 else if (!prefixcmp(diffopts, "-u"))
1475 xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1476 ecb.outf = xdiff_outf;
1477 ecb.priv = &ecbdata;
1478 ecbdata.xm.consume = fn_out_consume;
1479 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1480 ecbdata.diff_words =
1481 xcalloc(1, sizeof(struct diff_words_data));
1482 ecbdata.diff_words->file = o->file;
1483 }
1484 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1485 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1486 free_diff_words_data(&ecbdata);
1487 }
1488
1489 free_ab_and_return:
1490 diff_free_filespec_data(one);
1491 diff_free_filespec_data(two);
1492 free(a_one);
1493 free(b_two);
1494 return;
1495}
1496
1497static void builtin_diffstat(const char *name_a, const char *name_b,
1498 struct diff_filespec *one,
1499 struct diff_filespec *two,
1500 struct diffstat_t *diffstat,
1501 struct diff_options *o,
1502 int complete_rewrite)
1503{
1504 mmfile_t mf1, mf2;
1505 struct diffstat_file *data;
1506
1507 data = diffstat_add(diffstat, name_a, name_b);
1508
1509 if (!one || !two) {
1510 data->is_unmerged = 1;
1511 return;
1512 }
1513 if (complete_rewrite) {
1514 diff_populate_filespec(one, 0);
1515 diff_populate_filespec(two, 0);
1516 data->deleted = count_lines(one->data, one->size);
1517 data->added = count_lines(two->data, two->size);
1518 goto free_and_return;
1519 }
1520 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1521 die("unable to read files to diff");
1522
1523 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1524 data->is_binary = 1;
1525 data->added = mf2.size;
1526 data->deleted = mf1.size;
1527 } else {
1528 /* Crazy xdl interfaces.. */
1529 xpparam_t xpp;
1530 xdemitconf_t xecfg;
1531 xdemitcb_t ecb;
1532
1533 memset(&xecfg, 0, sizeof(xecfg));
1534 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1535 ecb.outf = xdiff_outf;
1536 ecb.priv = diffstat;
1537 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1538 }
1539
1540 free_and_return:
1541 diff_free_filespec_data(one);
1542 diff_free_filespec_data(two);
1543}
1544
1545static void builtin_checkdiff(const char *name_a, const char *name_b,
1546 const char *attr_path,
1547 struct diff_filespec *one,
1548 struct diff_filespec *two, struct diff_options *o)
1549{
1550 mmfile_t mf1, mf2;
1551 struct checkdiff_t data;
1552
1553 if (!two)
1554 return;
1555
1556 memset(&data, 0, sizeof(data));
1557 data.xm.consume = checkdiff_consume;
1558 data.filename = name_b ? name_b : name_a;
1559 data.lineno = 0;
1560 data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1561 data.ws_rule = whitespace_rule(attr_path);
1562 data.file = o->file;
1563
1564 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1565 die("unable to read files to diff");
1566
1567 if (diff_filespec_is_binary(two))
1568 goto free_and_return;
1569 else {
1570 /* Crazy xdl interfaces.. */
1571 xpparam_t xpp;
1572 xdemitconf_t xecfg;
1573 xdemitcb_t ecb;
1574
1575 memset(&xecfg, 0, sizeof(xecfg));
1576 xpp.flags = XDF_NEED_MINIMAL;
1577 ecb.outf = xdiff_outf;
1578 ecb.priv = &data;
1579 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1580 }
1581 free_and_return:
1582 diff_free_filespec_data(one);
1583 diff_free_filespec_data(two);
1584 if (data.status)
1585 DIFF_OPT_SET(o, CHECK_FAILED);
1586}
1587
1588struct diff_filespec *alloc_filespec(const char *path)
1589{
1590 int namelen = strlen(path);
1591 struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1592
1593 memset(spec, 0, sizeof(*spec));
1594 spec->path = (char *)(spec + 1);
1595 memcpy(spec->path, path, namelen+1);
1596 spec->count = 1;
1597 return spec;
1598}
1599
1600void free_filespec(struct diff_filespec *spec)
1601{
1602 if (!--spec->count) {
1603 diff_free_filespec_data(spec);
1604 free(spec);
1605 }
1606}
1607
1608void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1609 unsigned short mode)
1610{
1611 if (mode) {
1612 spec->mode = canon_mode(mode);
1613 hashcpy(spec->sha1, sha1);
1614 spec->sha1_valid = !is_null_sha1(sha1);
1615 }
1616}
1617
1618/*
1619 * Given a name and sha1 pair, if the index tells us the file in
1620 * the work tree has that object contents, return true, so that
1621 * prepare_temp_file() does not have to inflate and extract.
1622 */
1623static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1624{
1625 struct cache_entry *ce;
1626 struct stat st;
1627 int pos, len;
1628
1629 /* We do not read the cache ourselves here, because the
1630 * benchmark with my previous version that always reads cache
1631 * shows that it makes things worse for diff-tree comparing
1632 * two linux-2.6 kernel trees in an already checked out work
1633 * tree. This is because most diff-tree comparisons deal with
1634 * only a small number of files, while reading the cache is
1635 * expensive for a large project, and its cost outweighs the
1636 * savings we get by not inflating the object to a temporary
1637 * file. Practically, this code only helps when we are used
1638 * by diff-cache --cached, which does read the cache before
1639 * calling us.
1640 */
1641 if (!active_cache)
1642 return 0;
1643
1644 /* We want to avoid the working directory if our caller
1645 * doesn't need the data in a normal file, this system
1646 * is rather slow with its stat/open/mmap/close syscalls,
1647 * and the object is contained in a pack file. The pack
1648 * is probably already open and will be faster to obtain
1649 * the data through than the working directory. Loose
1650 * objects however would tend to be slower as they need
1651 * to be individually opened and inflated.
1652 */
1653 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1654 return 0;
1655
1656 len = strlen(name);
1657 pos = cache_name_pos(name, len);
1658 if (pos < 0)
1659 return 0;
1660 ce = active_cache[pos];
1661
1662 /*
1663 * This is not the sha1 we are looking for, or
1664 * unreusable because it is not a regular file.
1665 */
1666 if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1667 return 0;
1668
1669 /*
1670 * If ce matches the file in the work tree, we can reuse it.
1671 */
1672 if (ce_uptodate(ce) ||
1673 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1674 return 1;
1675
1676 return 0;
1677}
1678
1679static int populate_from_stdin(struct diff_filespec *s)
1680{
1681 struct strbuf buf;
1682 size_t size = 0;
1683
1684 strbuf_init(&buf, 0);
1685 if (strbuf_read(&buf, 0, 0) < 0)
1686 return error("error while reading from stdin %s",
1687 strerror(errno));
1688
1689 s->should_munmap = 0;
1690 s->data = strbuf_detach(&buf, &size);
1691 s->size = size;
1692 s->should_free = 1;
1693 return 0;
1694}
1695
1696static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1697{
1698 int len;
1699 char *data = xmalloc(100);
1700 len = snprintf(data, 100,
1701 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1702 s->data = data;
1703 s->size = len;
1704 s->should_free = 1;
1705 if (size_only) {
1706 s->data = NULL;
1707 free(data);
1708 }
1709 return 0;
1710}
1711
1712/*
1713 * While doing rename detection and pickaxe operation, we may need to
1714 * grab the data for the blob (or file) for our own in-core comparison.
1715 * diff_filespec has data and size fields for this purpose.
1716 */
1717int diff_populate_filespec(struct diff_filespec *s, int size_only)
1718{
1719 int err = 0;
1720 if (!DIFF_FILE_VALID(s))
1721 die("internal error: asking to populate invalid file.");
1722 if (S_ISDIR(s->mode))
1723 return -1;
1724
1725 if (s->data)
1726 return 0;
1727
1728 if (size_only && 0 < s->size)
1729 return 0;
1730
1731 if (S_ISGITLINK(s->mode))
1732 return diff_populate_gitlink(s, size_only);
1733
1734 if (!s->sha1_valid ||
1735 reuse_worktree_file(s->path, s->sha1, 0)) {
1736 struct strbuf buf;
1737 struct stat st;
1738 int fd;
1739
1740 if (!strcmp(s->path, "-"))
1741 return populate_from_stdin(s);
1742
1743 if (lstat(s->path, &st) < 0) {
1744 if (errno == ENOENT) {
1745 err_empty:
1746 err = -1;
1747 empty:
1748 s->data = (char *)"";
1749 s->size = 0;
1750 return err;
1751 }
1752 }
1753 s->size = xsize_t(st.st_size);
1754 if (!s->size)
1755 goto empty;
1756 if (size_only)
1757 return 0;
1758 if (S_ISLNK(st.st_mode)) {
1759 int ret;
1760 s->data = xmalloc(s->size);
1761 s->should_free = 1;
1762 ret = readlink(s->path, s->data, s->size);
1763 if (ret < 0) {
1764 free(s->data);
1765 goto err_empty;
1766 }
1767 return 0;
1768 }
1769 fd = open(s->path, O_RDONLY);
1770 if (fd < 0)
1771 goto err_empty;
1772 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1773 close(fd);
1774 s->should_munmap = 1;
1775
1776 /*
1777 * Convert from working tree format to canonical git format
1778 */
1779 strbuf_init(&buf, 0);
1780 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1781 size_t size = 0;
1782 munmap(s->data, s->size);
1783 s->should_munmap = 0;
1784 s->data = strbuf_detach(&buf, &size);
1785 s->size = size;
1786 s->should_free = 1;
1787 }
1788 }
1789 else {
1790 enum object_type type;
1791 if (size_only)
1792 type = sha1_object_info(s->sha1, &s->size);
1793 else {
1794 s->data = read_sha1_file(s->sha1, &type, &s->size);
1795 s->should_free = 1;
1796 }
1797 }
1798 return 0;
1799}
1800
1801void diff_free_filespec_blob(struct diff_filespec *s)
1802{
1803 if (s->should_free)
1804 free(s->data);
1805 else if (s->should_munmap)
1806 munmap(s->data, s->size);
1807
1808 if (s->should_free || s->should_munmap) {
1809 s->should_free = s->should_munmap = 0;
1810 s->data = NULL;
1811 }
1812}
1813
1814void diff_free_filespec_data(struct diff_filespec *s)
1815{
1816 diff_free_filespec_blob(s);
1817 free(s->cnt_data);
1818 s->cnt_data = NULL;
1819}
1820
1821static void prep_temp_blob(struct diff_tempfile *temp,
1822 void *blob,
1823 unsigned long size,
1824 const unsigned char *sha1,
1825 int mode)
1826{
1827 int fd;
1828
1829 fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1830 if (fd < 0)
1831 die("unable to create temp-file: %s", strerror(errno));
1832 if (write_in_full(fd, blob, size) != size)
1833 die("unable to write temp-file");
1834 close(fd);
1835 temp->name = temp->tmp_path;
1836 strcpy(temp->hex, sha1_to_hex(sha1));
1837 temp->hex[40] = 0;
1838 sprintf(temp->mode, "%06o", mode);
1839}
1840
1841static void prepare_temp_file(const char *name,
1842 struct diff_tempfile *temp,
1843 struct diff_filespec *one)
1844{
1845 if (!DIFF_FILE_VALID(one)) {
1846 not_a_valid_file:
1847 /* A '-' entry produces this for file-2, and
1848 * a '+' entry produces this for file-1.
1849 */
1850 temp->name = "/dev/null";
1851 strcpy(temp->hex, ".");
1852 strcpy(temp->mode, ".");
1853 return;
1854 }
1855
1856 if (!one->sha1_valid ||
1857 reuse_worktree_file(name, one->sha1, 1)) {
1858 struct stat st;
1859 if (lstat(name, &st) < 0) {
1860 if (errno == ENOENT)
1861 goto not_a_valid_file;
1862 die("stat(%s): %s", name, strerror(errno));
1863 }
1864 if (S_ISLNK(st.st_mode)) {
1865 int ret;
1866 char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1867 size_t sz = xsize_t(st.st_size);
1868 if (sizeof(buf) <= st.st_size)
1869 die("symlink too long: %s", name);
1870 ret = readlink(name, buf, sz);
1871 if (ret < 0)
1872 die("readlink(%s)", name);
1873 prep_temp_blob(temp, buf, sz,
1874 (one->sha1_valid ?
1875 one->sha1 : null_sha1),
1876 (one->sha1_valid ?
1877 one->mode : S_IFLNK));
1878 }
1879 else {
1880 /* we can borrow from the file in the work tree */
1881 temp->name = name;
1882 if (!one->sha1_valid)
1883 strcpy(temp->hex, sha1_to_hex(null_sha1));
1884 else
1885 strcpy(temp->hex, sha1_to_hex(one->sha1));
1886 /* Even though we may sometimes borrow the
1887 * contents from the work tree, we always want
1888 * one->mode. mode is trustworthy even when
1889 * !(one->sha1_valid), as long as
1890 * DIFF_FILE_VALID(one).
1891 */
1892 sprintf(temp->mode, "%06o", one->mode);
1893 }
1894 return;
1895 }
1896 else {
1897 if (diff_populate_filespec(one, 0))
1898 die("cannot read data blob for %s", one->path);
1899 prep_temp_blob(temp, one->data, one->size,
1900 one->sha1, one->mode);
1901 }
1902}
1903
1904static void remove_tempfile(void)
1905{
1906 int i;
1907
1908 for (i = 0; i < 2; i++)
1909 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1910 unlink(diff_temp[i].name);
1911 diff_temp[i].name = NULL;
1912 }
1913}
1914
1915static void remove_tempfile_on_signal(int signo)
1916{
1917 remove_tempfile();
1918 signal(SIGINT, SIG_DFL);
1919 raise(signo);
1920}
1921
1922/* An external diff command takes:
1923 *
1924 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1925 * infile2 infile2-sha1 infile2-mode [ rename-to ]
1926 *
1927 */
1928static void run_external_diff(const char *pgm,
1929 const char *name,
1930 const char *other,
1931 struct diff_filespec *one,
1932 struct diff_filespec *two,
1933 const char *xfrm_msg,
1934 int complete_rewrite)
1935{
1936 const char *spawn_arg[10];
1937 struct diff_tempfile *temp = diff_temp;
1938 int retval;
1939 static int atexit_asked = 0;
1940 const char *othername;
1941 const char **arg = &spawn_arg[0];
1942
1943 othername = (other? other : name);
1944 if (one && two) {
1945 prepare_temp_file(name, &temp[0], one);
1946 prepare_temp_file(othername, &temp[1], two);
1947 if (! atexit_asked &&
1948 (temp[0].name == temp[0].tmp_path ||
1949 temp[1].name == temp[1].tmp_path)) {
1950 atexit_asked = 1;
1951 atexit(remove_tempfile);
1952 }
1953 signal(SIGINT, remove_tempfile_on_signal);
1954 }
1955
1956 if (one && two) {
1957 *arg++ = pgm;
1958 *arg++ = name;
1959 *arg++ = temp[0].name;
1960 *arg++ = temp[0].hex;
1961 *arg++ = temp[0].mode;
1962 *arg++ = temp[1].name;
1963 *arg++ = temp[1].hex;
1964 *arg++ = temp[1].mode;
1965 if (other) {
1966 *arg++ = other;
1967 *arg++ = xfrm_msg;
1968 }
1969 } else {
1970 *arg++ = pgm;
1971 *arg++ = name;
1972 }
1973 *arg = NULL;
1974 fflush(NULL);
1975 retval = run_command_v_opt(spawn_arg, 0);
1976 remove_tempfile();
1977 if (retval) {
1978 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1979 exit(1);
1980 }
1981}
1982
1983static const char *external_diff_attr(const char *name)
1984{
1985 struct git_attr_check attr_diff_check;
1986
1987 if (!name)
1988 return NULL;
1989
1990 setup_diff_attr_check(&attr_diff_check);
1991 if (!git_checkattr(name, 1, &attr_diff_check)) {
1992 const char *value = attr_diff_check.value;
1993 if (!ATTR_TRUE(value) &&
1994 !ATTR_FALSE(value) &&
1995 !ATTR_UNSET(value)) {
1996 struct ll_diff_driver *drv;
1997
1998 for (drv = user_diff; drv; drv = drv->next)
1999 if (!strcmp(drv->name, value))
2000 return drv->cmd;
2001 }
2002 }
2003 return NULL;
2004}
2005
2006static void run_diff_cmd(const char *pgm,
2007 const char *name,
2008 const char *other,
2009 const char *attr_path,
2010 struct diff_filespec *one,
2011 struct diff_filespec *two,
2012 const char *xfrm_msg,
2013 struct diff_options *o,
2014 int complete_rewrite)
2015{
2016 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2017 pgm = NULL;
2018 else {
2019 const char *cmd = external_diff_attr(attr_path);
2020 if (cmd)
2021 pgm = cmd;
2022 }
2023
2024 if (pgm) {
2025 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2026 complete_rewrite);
2027 return;
2028 }
2029 if (one && two)
2030 builtin_diff(name, other ? other : name,
2031 one, two, xfrm_msg, o, complete_rewrite);
2032 else
2033 fprintf(o->file, "* Unmerged path %s\n", name);
2034}
2035
2036static void diff_fill_sha1_info(struct diff_filespec *one)
2037{
2038 if (DIFF_FILE_VALID(one)) {
2039 if (!one->sha1_valid) {
2040 struct stat st;
2041 if (!strcmp(one->path, "-")) {
2042 hashcpy(one->sha1, null_sha1);
2043 return;
2044 }
2045 if (lstat(one->path, &st) < 0)
2046 die("stat %s", one->path);
2047 if (index_path(one->sha1, one->path, &st, 0))
2048 die("cannot hash %s\n", one->path);
2049 }
2050 }
2051 else
2052 hashclr(one->sha1);
2053}
2054
2055static int similarity_index(struct diff_filepair *p)
2056{
2057 return p->score * 100 / MAX_SCORE;
2058}
2059
2060static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2061{
2062 /* Strip the prefix but do not molest /dev/null and absolute paths */
2063 if (*namep && **namep != '/')
2064 *namep += prefix_length;
2065 if (*otherp && **otherp != '/')
2066 *otherp += prefix_length;
2067}
2068
2069static void run_diff(struct diff_filepair *p, struct diff_options *o)
2070{
2071 const char *pgm = external_diff();
2072 struct strbuf msg;
2073 char *xfrm_msg;
2074 struct diff_filespec *one = p->one;
2075 struct diff_filespec *two = p->two;
2076 const char *name;
2077 const char *other;
2078 const char *attr_path;
2079 int complete_rewrite = 0;
2080
2081 name = p->one->path;
2082 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2083 attr_path = name;
2084 if (o->prefix_length)
2085 strip_prefix(o->prefix_length, &name, &other);
2086
2087 if (DIFF_PAIR_UNMERGED(p)) {
2088 run_diff_cmd(pgm, name, NULL, attr_path,
2089 NULL, NULL, NULL, o, 0);
2090 return;
2091 }
2092
2093 diff_fill_sha1_info(one);
2094 diff_fill_sha1_info(two);
2095
2096 strbuf_init(&msg, PATH_MAX * 2 + 300);
2097 switch (p->status) {
2098 case DIFF_STATUS_COPIED:
2099 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2100 strbuf_addstr(&msg, "\ncopy from ");
2101 quote_c_style(name, &msg, NULL, 0);
2102 strbuf_addstr(&msg, "\ncopy to ");
2103 quote_c_style(other, &msg, NULL, 0);
2104 strbuf_addch(&msg, '\n');
2105 break;
2106 case DIFF_STATUS_RENAMED:
2107 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
2108 strbuf_addstr(&msg, "\nrename from ");
2109 quote_c_style(name, &msg, NULL, 0);
2110 strbuf_addstr(&msg, "\nrename to ");
2111 quote_c_style(other, &msg, NULL, 0);
2112 strbuf_addch(&msg, '\n');
2113 break;
2114 case DIFF_STATUS_MODIFIED:
2115 if (p->score) {
2116 strbuf_addf(&msg, "dissimilarity index %d%%\n",
2117 similarity_index(p));
2118 complete_rewrite = 1;
2119 break;
2120 }
2121 /* fallthru */
2122 default:
2123 /* nothing */
2124 ;
2125 }
2126
2127 if (hashcmp(one->sha1, two->sha1)) {
2128 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2129
2130 if (DIFF_OPT_TST(o, BINARY)) {
2131 mmfile_t mf;
2132 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2133 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2134 abbrev = 40;
2135 }
2136 strbuf_addf(&msg, "index %.*s..%.*s",
2137 abbrev, sha1_to_hex(one->sha1),
2138 abbrev, sha1_to_hex(two->sha1));
2139 if (one->mode == two->mode)
2140 strbuf_addf(&msg, " %06o", one->mode);
2141 strbuf_addch(&msg, '\n');
2142 }
2143
2144 if (msg.len)
2145 strbuf_setlen(&msg, msg.len - 1);
2146 xfrm_msg = msg.len ? msg.buf : NULL;
2147
2148 if (!pgm &&
2149 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2150 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2151 /* a filepair that changes between file and symlink
2152 * needs to be split into deletion and creation.
2153 */
2154 struct diff_filespec *null = alloc_filespec(two->path);
2155 run_diff_cmd(NULL, name, other, attr_path,
2156 one, null, xfrm_msg, o, 0);
2157 free(null);
2158 null = alloc_filespec(one->path);
2159 run_diff_cmd(NULL, name, other, attr_path,
2160 null, two, xfrm_msg, o, 0);
2161 free(null);
2162 }
2163 else
2164 run_diff_cmd(pgm, name, other, attr_path,
2165 one, two, xfrm_msg, o, complete_rewrite);
2166
2167 strbuf_release(&msg);
2168}
2169
2170static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2171 struct diffstat_t *diffstat)
2172{
2173 const char *name;
2174 const char *other;
2175 int complete_rewrite = 0;
2176
2177 if (DIFF_PAIR_UNMERGED(p)) {
2178 /* unmerged */
2179 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2180 return;
2181 }
2182
2183 name = p->one->path;
2184 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2185
2186 if (o->prefix_length)
2187 strip_prefix(o->prefix_length, &name, &other);
2188
2189 diff_fill_sha1_info(p->one);
2190 diff_fill_sha1_info(p->two);
2191
2192 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2193 complete_rewrite = 1;
2194 builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2195}
2196
2197static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2198{
2199 const char *name;
2200 const char *other;
2201 const char *attr_path;
2202
2203 if (DIFF_PAIR_UNMERGED(p)) {
2204 /* unmerged */
2205 return;
2206 }
2207
2208 name = p->one->path;
2209 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2210 attr_path = other ? other : name;
2211
2212 if (o->prefix_length)
2213 strip_prefix(o->prefix_length, &name, &other);
2214
2215 diff_fill_sha1_info(p->one);
2216 diff_fill_sha1_info(p->two);
2217
2218 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2219}
2220
2221void diff_setup(struct diff_options *options)
2222{
2223 memset(options, 0, sizeof(*options));
2224
2225 options->file = stdout;
2226
2227 options->line_termination = '\n';
2228 options->break_opt = -1;
2229 options->rename_limit = -1;
2230 options->dirstat_percent = 3;
2231 options->context = 3;
2232
2233 options->change = diff_change;
2234 options->add_remove = diff_addremove;
2235 if (diff_use_color_default > 0)
2236 DIFF_OPT_SET(options, COLOR_DIFF);
2237 else
2238 DIFF_OPT_CLR(options, COLOR_DIFF);
2239 options->detect_rename = diff_detect_rename_default;
2240
2241 options->a_prefix = "a/";
2242 options->b_prefix = "b/";
2243}
2244
2245int diff_setup_done(struct diff_options *options)
2246{
2247 int count = 0;
2248
2249 if (options->output_format & DIFF_FORMAT_NAME)
2250 count++;
2251 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2252 count++;
2253 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2254 count++;
2255 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2256 count++;
2257 if (count > 1)
2258 die("--name-only, --name-status, --check and -s are mutually exclusive");
2259
2260 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2261 options->detect_rename = DIFF_DETECT_COPY;
2262
2263 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2264 options->prefix = NULL;
2265 if (options->prefix)
2266 options->prefix_length = strlen(options->prefix);
2267 else
2268 options->prefix_length = 0;
2269
2270 if (options->output_format & (DIFF_FORMAT_NAME |
2271 DIFF_FORMAT_NAME_STATUS |
2272 DIFF_FORMAT_CHECKDIFF |
2273 DIFF_FORMAT_NO_OUTPUT))
2274 options->output_format &= ~(DIFF_FORMAT_RAW |
2275 DIFF_FORMAT_NUMSTAT |
2276 DIFF_FORMAT_DIFFSTAT |
2277 DIFF_FORMAT_SHORTSTAT |
2278 DIFF_FORMAT_DIRSTAT |
2279 DIFF_FORMAT_SUMMARY |
2280 DIFF_FORMAT_PATCH);
2281
2282 /*
2283 * These cases always need recursive; we do not drop caller-supplied
2284 * recursive bits for other formats here.
2285 */
2286 if (options->output_format & (DIFF_FORMAT_PATCH |
2287 DIFF_FORMAT_NUMSTAT |
2288 DIFF_FORMAT_DIFFSTAT |
2289 DIFF_FORMAT_SHORTSTAT |
2290 DIFF_FORMAT_DIRSTAT |
2291 DIFF_FORMAT_SUMMARY |
2292 DIFF_FORMAT_CHECKDIFF))
2293 DIFF_OPT_SET(options, RECURSIVE);
2294 /*
2295 * Also pickaxe would not work very well if you do not say recursive
2296 */
2297 if (options->pickaxe)
2298 DIFF_OPT_SET(options, RECURSIVE);
2299
2300 if (options->detect_rename && options->rename_limit < 0)
2301 options->rename_limit = diff_rename_limit_default;
2302 if (options->setup & DIFF_SETUP_USE_CACHE) {
2303 if (!active_cache)
2304 /* read-cache does not die even when it fails
2305 * so it is safe for us to do this here. Also
2306 * it does not smudge active_cache or active_nr
2307 * when it fails, so we do not have to worry about
2308 * cleaning it up ourselves either.
2309 */
2310 read_cache();
2311 }
2312 if (options->abbrev <= 0 || 40 < options->abbrev)
2313 options->abbrev = 40; /* full */
2314
2315 /*
2316 * It does not make sense to show the first hit we happened
2317 * to have found. It does not make sense not to return with
2318 * exit code in such a case either.
2319 */
2320 if (DIFF_OPT_TST(options, QUIET)) {
2321 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2322 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2323 }
2324
2325 /*
2326 * If we postprocess in diffcore, we cannot simply return
2327 * upon the first hit. We need to run diff as usual.
2328 */
2329 if (options->pickaxe || options->filter)
2330 DIFF_OPT_CLR(options, QUIET);
2331
2332 return 0;
2333}
2334
2335static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2336{
2337 char c, *eq;
2338 int len;
2339
2340 if (*arg != '-')
2341 return 0;
2342 c = *++arg;
2343 if (!c)
2344 return 0;
2345 if (c == arg_short) {
2346 c = *++arg;
2347 if (!c)
2348 return 1;
2349 if (val && isdigit(c)) {
2350 char *end;
2351 int n = strtoul(arg, &end, 10);
2352 if (*end)
2353 return 0;
2354 *val = n;
2355 return 1;
2356 }
2357 return 0;
2358 }
2359 if (c != '-')
2360 return 0;
2361 arg++;
2362 eq = strchr(arg, '=');
2363 if (eq)
2364 len = eq - arg;
2365 else
2366 len = strlen(arg);
2367 if (!len || strncmp(arg, arg_long, len))
2368 return 0;
2369 if (eq) {
2370 int n;
2371 char *end;
2372 if (!isdigit(*++eq))
2373 return 0;
2374 n = strtoul(eq, &end, 10);
2375 if (*end)
2376 return 0;
2377 *val = n;
2378 }
2379 return 1;
2380}
2381
2382static int diff_scoreopt_parse(const char *opt);
2383
2384int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2385{
2386 const char *arg = av[0];
2387
2388 /* Output format options */
2389 if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2390 options->output_format |= DIFF_FORMAT_PATCH;
2391 else if (opt_arg(arg, 'U', "unified", &options->context))
2392 options->output_format |= DIFF_FORMAT_PATCH;
2393 else if (!strcmp(arg, "--raw"))
2394 options->output_format |= DIFF_FORMAT_RAW;
2395 else if (!strcmp(arg, "--patch-with-raw"))
2396 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2397 else if (!strcmp(arg, "--numstat"))
2398 options->output_format |= DIFF_FORMAT_NUMSTAT;
2399 else if (!strcmp(arg, "--shortstat"))
2400 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2401 else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2402 options->output_format |= DIFF_FORMAT_DIRSTAT;
2403 else if (!strcmp(arg, "--cumulative"))
2404 options->output_format |= DIFF_FORMAT_CUMULATIVE;
2405 else if (!strcmp(arg, "--check"))
2406 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2407 else if (!strcmp(arg, "--summary"))
2408 options->output_format |= DIFF_FORMAT_SUMMARY;
2409 else if (!strcmp(arg, "--patch-with-stat"))
2410 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2411 else if (!strcmp(arg, "--name-only"))
2412 options->output_format |= DIFF_FORMAT_NAME;
2413 else if (!strcmp(arg, "--name-status"))
2414 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2415 else if (!strcmp(arg, "-s"))
2416 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2417 else if (!prefixcmp(arg, "--stat")) {
2418 char *end;
2419 int width = options->stat_width;
2420 int name_width = options->stat_name_width;
2421 arg += 6;
2422 end = (char *)arg;
2423
2424 switch (*arg) {
2425 case '-':
2426 if (!prefixcmp(arg, "-width="))
2427 width = strtoul(arg + 7, &end, 10);
2428 else if (!prefixcmp(arg, "-name-width="))
2429 name_width = strtoul(arg + 12, &end, 10);
2430 break;
2431 case '=':
2432 width = strtoul(arg+1, &end, 10);
2433 if (*end == ',')
2434 name_width = strtoul(end+1, &end, 10);
2435 }
2436
2437 /* Important! This checks all the error cases! */
2438 if (*end)
2439 return 0;
2440 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2441 options->stat_name_width = name_width;
2442 options->stat_width = width;
2443 }
2444
2445 /* renames options */
2446 else if (!prefixcmp(arg, "-B")) {
2447 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2448 return -1;
2449 }
2450 else if (!prefixcmp(arg, "-M")) {
2451 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2452 return -1;
2453 options->detect_rename = DIFF_DETECT_RENAME;
2454 }
2455 else if (!prefixcmp(arg, "-C")) {
2456 if (options->detect_rename == DIFF_DETECT_COPY)
2457 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2458 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2459 return -1;
2460 options->detect_rename = DIFF_DETECT_COPY;
2461 }
2462 else if (!strcmp(arg, "--no-renames"))
2463 options->detect_rename = 0;
2464 else if (!strcmp(arg, "--relative"))
2465 DIFF_OPT_SET(options, RELATIVE_NAME);
2466 else if (!prefixcmp(arg, "--relative=")) {
2467 DIFF_OPT_SET(options, RELATIVE_NAME);
2468 options->prefix = arg + 11;
2469 }
2470
2471 /* xdiff options */
2472 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2473 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2474 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2475 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2476 else if (!strcmp(arg, "--ignore-space-at-eol"))
2477 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2478
2479 /* flags options */
2480 else if (!strcmp(arg, "--binary")) {
2481 options->output_format |= DIFF_FORMAT_PATCH;
2482 DIFF_OPT_SET(options, BINARY);
2483 }
2484 else if (!strcmp(arg, "--full-index"))
2485 DIFF_OPT_SET(options, FULL_INDEX);
2486 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2487 DIFF_OPT_SET(options, TEXT);
2488 else if (!strcmp(arg, "-R"))
2489 DIFF_OPT_SET(options, REVERSE_DIFF);
2490 else if (!strcmp(arg, "--find-copies-harder"))
2491 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2492 else if (!strcmp(arg, "--follow"))
2493 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2494 else if (!strcmp(arg, "--color"))
2495 DIFF_OPT_SET(options, COLOR_DIFF);
2496 else if (!strcmp(arg, "--no-color"))
2497 DIFF_OPT_CLR(options, COLOR_DIFF);
2498 else if (!strcmp(arg, "--color-words"))
2499 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2500 else if (!strcmp(arg, "--exit-code"))
2501 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2502 else if (!strcmp(arg, "--quiet"))
2503 DIFF_OPT_SET(options, QUIET);
2504 else if (!strcmp(arg, "--ext-diff"))
2505 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2506 else if (!strcmp(arg, "--no-ext-diff"))
2507 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2508 else if (!strcmp(arg, "--ignore-submodules"))
2509 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2510
2511 /* misc options */
2512 else if (!strcmp(arg, "-z"))
2513 options->line_termination = 0;
2514 else if (!prefixcmp(arg, "-l"))
2515 options->rename_limit = strtoul(arg+2, NULL, 10);
2516 else if (!prefixcmp(arg, "-S"))
2517 options->pickaxe = arg + 2;
2518 else if (!strcmp(arg, "--pickaxe-all"))
2519 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2520 else if (!strcmp(arg, "--pickaxe-regex"))
2521 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2522 else if (!prefixcmp(arg, "-O"))
2523 options->orderfile = arg + 2;
2524 else if (!prefixcmp(arg, "--diff-filter="))
2525 options->filter = arg + 14;
2526 else if (!strcmp(arg, "--abbrev"))
2527 options->abbrev = DEFAULT_ABBREV;
2528 else if (!prefixcmp(arg, "--abbrev=")) {
2529 options->abbrev = strtoul(arg + 9, NULL, 10);
2530 if (options->abbrev < MINIMUM_ABBREV)
2531 options->abbrev = MINIMUM_ABBREV;
2532 else if (40 < options->abbrev)
2533 options->abbrev = 40;
2534 }
2535 else if (!prefixcmp(arg, "--src-prefix="))
2536 options->a_prefix = arg + 13;
2537 else if (!prefixcmp(arg, "--dst-prefix="))
2538 options->b_prefix = arg + 13;
2539 else if (!strcmp(arg, "--no-prefix"))
2540 options->a_prefix = options->b_prefix = "";
2541 else if (!prefixcmp(arg, "--output=")) {
2542 options->file = fopen(arg + strlen("--output="), "w");
2543 options->close_file = 1;
2544 } else
2545 return 0;
2546 return 1;
2547}
2548
2549static int parse_num(const char **cp_p)
2550{
2551 unsigned long num, scale;
2552 int ch, dot;
2553 const char *cp = *cp_p;
2554
2555 num = 0;
2556 scale = 1;
2557 dot = 0;
2558 for(;;) {
2559 ch = *cp;
2560 if ( !dot && ch == '.' ) {
2561 scale = 1;
2562 dot = 1;
2563 } else if ( ch == '%' ) {
2564 scale = dot ? scale*100 : 100;
2565 cp++; /* % is always at the end */
2566 break;
2567 } else if ( ch >= '0' && ch <= '9' ) {
2568 if ( scale < 100000 ) {
2569 scale *= 10;
2570 num = (num*10) + (ch-'0');
2571 }
2572 } else {
2573 break;
2574 }
2575 cp++;
2576 }
2577 *cp_p = cp;
2578
2579 /* user says num divided by scale and we say internally that
2580 * is MAX_SCORE * num / scale.
2581 */
2582 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2583}
2584
2585static int diff_scoreopt_parse(const char *opt)
2586{
2587 int opt1, opt2, cmd;
2588
2589 if (*opt++ != '-')
2590 return -1;
2591 cmd = *opt++;
2592 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2593 return -1; /* that is not a -M, -C nor -B option */
2594
2595 opt1 = parse_num(&opt);
2596 if (cmd != 'B')
2597 opt2 = 0;
2598 else {
2599 if (*opt == 0)
2600 opt2 = 0;
2601 else if (*opt != '/')
2602 return -1; /* we expect -B80/99 or -B80 */
2603 else {
2604 opt++;
2605 opt2 = parse_num(&opt);
2606 }
2607 }
2608 if (*opt != 0)
2609 return -1;
2610 return opt1 | (opt2 << 16);
2611}
2612
2613struct diff_queue_struct diff_queued_diff;
2614
2615void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2616{
2617 if (queue->alloc <= queue->nr) {
2618 queue->alloc = alloc_nr(queue->alloc);
2619 queue->queue = xrealloc(queue->queue,
2620 sizeof(dp) * queue->alloc);
2621 }
2622 queue->queue[queue->nr++] = dp;
2623}
2624
2625struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2626 struct diff_filespec *one,
2627 struct diff_filespec *two)
2628{
2629 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2630 dp->one = one;
2631 dp->two = two;
2632 if (queue)
2633 diff_q(queue, dp);
2634 return dp;
2635}
2636
2637void diff_free_filepair(struct diff_filepair *p)
2638{
2639 free_filespec(p->one);
2640 free_filespec(p->two);
2641 free(p);
2642}
2643
2644/* This is different from find_unique_abbrev() in that
2645 * it stuffs the result with dots for alignment.
2646 */
2647const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2648{
2649 int abblen;
2650 const char *abbrev;
2651 if (len == 40)
2652 return sha1_to_hex(sha1);
2653
2654 abbrev = find_unique_abbrev(sha1, len);
2655 abblen = strlen(abbrev);
2656 if (abblen < 37) {
2657 static char hex[41];
2658 if (len < abblen && abblen <= len + 2)
2659 sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2660 else
2661 sprintf(hex, "%s...", abbrev);
2662 return hex;
2663 }
2664 return sha1_to_hex(sha1);
2665}
2666
2667static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2668{
2669 int line_termination = opt->line_termination;
2670 int inter_name_termination = line_termination ? '\t' : '\0';
2671
2672 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2673 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2674 diff_unique_abbrev(p->one->sha1, opt->abbrev));
2675 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2676 }
2677 if (p->score) {
2678 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2679 inter_name_termination);
2680 } else {
2681 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2682 }
2683
2684 if (p->status == DIFF_STATUS_COPIED ||
2685 p->status == DIFF_STATUS_RENAMED) {
2686 const char *name_a, *name_b;
2687 name_a = p->one->path;
2688 name_b = p->two->path;
2689 strip_prefix(opt->prefix_length, &name_a, &name_b);
2690 write_name_quoted(name_a, opt->file, inter_name_termination);
2691 write_name_quoted(name_b, opt->file, line_termination);
2692 } else {
2693 const char *name_a, *name_b;
2694 name_a = p->one->mode ? p->one->path : p->two->path;
2695 name_b = NULL;
2696 strip_prefix(opt->prefix_length, &name_a, &name_b);
2697 write_name_quoted(name_a, opt->file, line_termination);
2698 }
2699}
2700
2701int diff_unmodified_pair(struct diff_filepair *p)
2702{
2703 /* This function is written stricter than necessary to support
2704 * the currently implemented transformers, but the idea is to
2705 * let transformers to produce diff_filepairs any way they want,
2706 * and filter and clean them up here before producing the output.
2707 */
2708 struct diff_filespec *one = p->one, *two = p->two;
2709
2710 if (DIFF_PAIR_UNMERGED(p))
2711 return 0; /* unmerged is interesting */
2712
2713 /* deletion, addition, mode or type change
2714 * and rename are all interesting.
2715 */
2716 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2717 DIFF_PAIR_MODE_CHANGED(p) ||
2718 strcmp(one->path, two->path))
2719 return 0;
2720
2721 /* both are valid and point at the same path. that is, we are
2722 * dealing with a change.
2723 */
2724 if (one->sha1_valid && two->sha1_valid &&
2725 !hashcmp(one->sha1, two->sha1))
2726 return 1; /* no change */
2727 if (!one->sha1_valid && !two->sha1_valid)
2728 return 1; /* both look at the same file on the filesystem. */
2729 return 0;
2730}
2731
2732static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2733{
2734 if (diff_unmodified_pair(p))
2735 return;
2736
2737 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2738 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2739 return; /* no tree diffs in patch format */
2740
2741 run_diff(p, o);
2742}
2743
2744static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2745 struct diffstat_t *diffstat)
2746{
2747 if (diff_unmodified_pair(p))
2748 return;
2749
2750 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2751 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2752 return; /* no tree diffs in patch format */
2753
2754 run_diffstat(p, o, diffstat);
2755}
2756
2757static void diff_flush_checkdiff(struct diff_filepair *p,
2758 struct diff_options *o)
2759{
2760 if (diff_unmodified_pair(p))
2761 return;
2762
2763 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2764 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2765 return; /* no tree diffs in patch format */
2766
2767 run_checkdiff(p, o);
2768}
2769
2770int diff_queue_is_empty(void)
2771{
2772 struct diff_queue_struct *q = &diff_queued_diff;
2773 int i;
2774 for (i = 0; i < q->nr; i++)
2775 if (!diff_unmodified_pair(q->queue[i]))
2776 return 0;
2777 return 1;
2778}
2779
2780#if DIFF_DEBUG
2781void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2782{
2783 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2784 x, one ? one : "",
2785 s->path,
2786 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2787 s->mode,
2788 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2789 fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2790 x, one ? one : "",
2791 s->size, s->xfrm_flags);
2792}
2793
2794void diff_debug_filepair(const struct diff_filepair *p, int i)
2795{
2796 diff_debug_filespec(p->one, i, "one");
2797 diff_debug_filespec(p->two, i, "two");
2798 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2799 p->score, p->status ? p->status : '?',
2800 p->one->rename_used, p->broken_pair);
2801}
2802
2803void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2804{
2805 int i;
2806 if (msg)
2807 fprintf(stderr, "%s\n", msg);
2808 fprintf(stderr, "q->nr = %d\n", q->nr);
2809 for (i = 0; i < q->nr; i++) {
2810 struct diff_filepair *p = q->queue[i];
2811 diff_debug_filepair(p, i);
2812 }
2813}
2814#endif
2815
2816static void diff_resolve_rename_copy(void)
2817{
2818 int i;
2819 struct diff_filepair *p;
2820 struct diff_queue_struct *q = &diff_queued_diff;
2821
2822 diff_debug_queue("resolve-rename-copy", q);
2823
2824 for (i = 0; i < q->nr; i++) {
2825 p = q->queue[i];
2826 p->status = 0; /* undecided */
2827 if (DIFF_PAIR_UNMERGED(p))
2828 p->status = DIFF_STATUS_UNMERGED;
2829 else if (!DIFF_FILE_VALID(p->one))
2830 p->status = DIFF_STATUS_ADDED;
2831 else if (!DIFF_FILE_VALID(p->two))
2832 p->status = DIFF_STATUS_DELETED;
2833 else if (DIFF_PAIR_TYPE_CHANGED(p))
2834 p->status = DIFF_STATUS_TYPE_CHANGED;
2835
2836 /* from this point on, we are dealing with a pair
2837 * whose both sides are valid and of the same type, i.e.
2838 * either in-place edit or rename/copy edit.
2839 */
2840 else if (DIFF_PAIR_RENAME(p)) {
2841 /*
2842 * A rename might have re-connected a broken
2843 * pair up, causing the pathnames to be the
2844 * same again. If so, that's not a rename at
2845 * all, just a modification..
2846 *
2847 * Otherwise, see if this source was used for
2848 * multiple renames, in which case we decrement
2849 * the count, and call it a copy.
2850 */
2851 if (!strcmp(p->one->path, p->two->path))
2852 p->status = DIFF_STATUS_MODIFIED;
2853 else if (--p->one->rename_used > 0)
2854 p->status = DIFF_STATUS_COPIED;
2855 else
2856 p->status = DIFF_STATUS_RENAMED;
2857 }
2858 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2859 p->one->mode != p->two->mode ||
2860 is_null_sha1(p->one->sha1))
2861 p->status = DIFF_STATUS_MODIFIED;
2862 else {
2863 /* This is a "no-change" entry and should not
2864 * happen anymore, but prepare for broken callers.
2865 */
2866 error("feeding unmodified %s to diffcore",
2867 p->one->path);
2868 p->status = DIFF_STATUS_UNKNOWN;
2869 }
2870 }
2871 diff_debug_queue("resolve-rename-copy done", q);
2872}
2873
2874static int check_pair_status(struct diff_filepair *p)
2875{
2876 switch (p->status) {
2877 case DIFF_STATUS_UNKNOWN:
2878 return 0;
2879 case 0:
2880 die("internal error in diff-resolve-rename-copy");
2881 default:
2882 return 1;
2883 }
2884}
2885
2886static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2887{
2888 int fmt = opt->output_format;
2889
2890 if (fmt & DIFF_FORMAT_CHECKDIFF)
2891 diff_flush_checkdiff(p, opt);
2892 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2893 diff_flush_raw(p, opt);
2894 else if (fmt & DIFF_FORMAT_NAME) {
2895 const char *name_a, *name_b;
2896 name_a = p->two->path;
2897 name_b = NULL;
2898 strip_prefix(opt->prefix_length, &name_a, &name_b);
2899 write_name_quoted(name_a, opt->file, opt->line_termination);
2900 }
2901}
2902
2903static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2904{
2905 if (fs->mode)
2906 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2907 else
2908 fprintf(file, " %s ", newdelete);
2909 write_name_quoted(fs->path, file, '\n');
2910}
2911
2912
2913static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2914{
2915 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2916 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2917 show_name ? ' ' : '\n');
2918 if (show_name) {
2919 write_name_quoted(p->two->path, file, '\n');
2920 }
2921 }
2922}
2923
2924static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2925{
2926 char *names = pprint_rename(p->one->path, p->two->path);
2927
2928 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2929 free(names);
2930 show_mode_change(file, p, 0);
2931}
2932
2933static void diff_summary(FILE *file, struct diff_filepair *p)
2934{
2935 switch(p->status) {
2936 case DIFF_STATUS_DELETED:
2937 show_file_mode_name(file, "delete", p->one);
2938 break;
2939 case DIFF_STATUS_ADDED:
2940 show_file_mode_name(file, "create", p->two);
2941 break;
2942 case DIFF_STATUS_COPIED:
2943 show_rename_copy(file, "copy", p);
2944 break;
2945 case DIFF_STATUS_RENAMED:
2946 show_rename_copy(file, "rename", p);
2947 break;
2948 default:
2949 if (p->score) {
2950 fputs(" rewrite ", file);
2951 write_name_quoted(p->two->path, file, ' ');
2952 fprintf(file, "(%d%%)\n", similarity_index(p));
2953 }
2954 show_mode_change(file, p, !p->score);
2955 break;
2956 }
2957}
2958
2959struct patch_id_t {
2960 struct xdiff_emit_state xm;
2961 SHA_CTX *ctx;
2962 int patchlen;
2963};
2964
2965static int remove_space(char *line, int len)
2966{
2967 int i;
2968 char *dst = line;
2969 unsigned char c;
2970
2971 for (i = 0; i < len; i++)
2972 if (!isspace((c = line[i])))
2973 *dst++ = c;
2974
2975 return dst - line;
2976}
2977
2978static void patch_id_consume(void *priv, char *line, unsigned long len)
2979{
2980 struct patch_id_t *data = priv;
2981 int new_len;
2982
2983 /* Ignore line numbers when computing the SHA1 of the patch */
2984 if (!prefixcmp(line, "@@ -"))
2985 return;
2986
2987 new_len = remove_space(line, len);
2988
2989 SHA1_Update(data->ctx, line, new_len);
2990 data->patchlen += new_len;
2991}
2992
2993/* returns 0 upon success, and writes result into sha1 */
2994static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2995{
2996 struct diff_queue_struct *q = &diff_queued_diff;
2997 int i;
2998 SHA_CTX ctx;
2999 struct patch_id_t data;
3000 char buffer[PATH_MAX * 4 + 20];
3001
3002 SHA1_Init(&ctx);
3003 memset(&data, 0, sizeof(struct patch_id_t));
3004 data.ctx = &ctx;
3005 data.xm.consume = patch_id_consume;
3006
3007 for (i = 0; i < q->nr; i++) {
3008 xpparam_t xpp;
3009 xdemitconf_t xecfg;
3010 xdemitcb_t ecb;
3011 mmfile_t mf1, mf2;
3012 struct diff_filepair *p = q->queue[i];
3013 int len1, len2;
3014
3015 memset(&xecfg, 0, sizeof(xecfg));
3016 if (p->status == 0)
3017 return error("internal diff status error");
3018 if (p->status == DIFF_STATUS_UNKNOWN)
3019 continue;
3020 if (diff_unmodified_pair(p))
3021 continue;
3022 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3023 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3024 continue;
3025 if (DIFF_PAIR_UNMERGED(p))
3026 continue;
3027
3028 diff_fill_sha1_info(p->one);
3029 diff_fill_sha1_info(p->two);
3030 if (fill_mmfile(&mf1, p->one) < 0 ||
3031 fill_mmfile(&mf2, p->two) < 0)
3032 return error("unable to read files to diff");
3033
3034 len1 = remove_space(p->one->path, strlen(p->one->path));
3035 len2 = remove_space(p->two->path, strlen(p->two->path));
3036 if (p->one->mode == 0)
3037 len1 = snprintf(buffer, sizeof(buffer),
3038 "diff--gita/%.*sb/%.*s"
3039 "newfilemode%06o"
3040 "---/dev/null"
3041 "+++b/%.*s",
3042 len1, p->one->path,
3043 len2, p->two->path,
3044 p->two->mode,
3045 len2, p->two->path);
3046 else if (p->two->mode == 0)
3047 len1 = snprintf(buffer, sizeof(buffer),
3048 "diff--gita/%.*sb/%.*s"
3049 "deletedfilemode%06o"
3050 "---a/%.*s"
3051 "+++/dev/null",
3052 len1, p->one->path,
3053 len2, p->two->path,
3054 p->one->mode,
3055 len1, p->one->path);
3056 else
3057 len1 = snprintf(buffer, sizeof(buffer),
3058 "diff--gita/%.*sb/%.*s"
3059 "---a/%.*s"
3060 "+++b/%.*s",
3061 len1, p->one->path,
3062 len2, p->two->path,
3063 len1, p->one->path,
3064 len2, p->two->path);
3065 SHA1_Update(&ctx, buffer, len1);
3066
3067 xpp.flags = XDF_NEED_MINIMAL;
3068 xecfg.ctxlen = 3;
3069 xecfg.flags = XDL_EMIT_FUNCNAMES;
3070 ecb.outf = xdiff_outf;
3071 ecb.priv = &data;
3072 xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
3073 }
3074
3075 SHA1_Final(sha1, &ctx);
3076 return 0;
3077}
3078
3079int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3080{
3081 struct diff_queue_struct *q = &diff_queued_diff;
3082 int i;
3083 int result = diff_get_patch_id(options, sha1);
3084
3085 for (i = 0; i < q->nr; i++)
3086 diff_free_filepair(q->queue[i]);
3087
3088 free(q->queue);
3089 q->queue = NULL;
3090 q->nr = q->alloc = 0;
3091
3092 return result;
3093}
3094
3095static int is_summary_empty(const struct diff_queue_struct *q)
3096{
3097 int i;
3098
3099 for (i = 0; i < q->nr; i++) {
3100 const struct diff_filepair *p = q->queue[i];
3101
3102 switch (p->status) {
3103 case DIFF_STATUS_DELETED:
3104 case DIFF_STATUS_ADDED:
3105 case DIFF_STATUS_COPIED:
3106 case DIFF_STATUS_RENAMED:
3107 return 0;
3108 default:
3109 if (p->score)
3110 return 0;
3111 if (p->one->mode && p->two->mode &&
3112 p->one->mode != p->two->mode)
3113 return 0;
3114 break;
3115 }
3116 }
3117 return 1;
3118}
3119
3120void diff_flush(struct diff_options *options)
3121{
3122 struct diff_queue_struct *q = &diff_queued_diff;
3123 int i, output_format = options->output_format;
3124 int separator = 0;
3125
3126 /*
3127 * Order: raw, stat, summary, patch
3128 * or: name/name-status/checkdiff (other bits clear)
3129 */
3130 if (!q->nr)
3131 goto free_queue;
3132
3133 if (output_format & (DIFF_FORMAT_RAW |
3134 DIFF_FORMAT_NAME |
3135 DIFF_FORMAT_NAME_STATUS |
3136 DIFF_FORMAT_CHECKDIFF)) {
3137 for (i = 0; i < q->nr; i++) {
3138 struct diff_filepair *p = q->queue[i];
3139 if (check_pair_status(p))
3140 flush_one_pair(p, options);
3141 }
3142 separator++;
3143 }
3144
3145 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3146 struct diffstat_t diffstat;
3147
3148 memset(&diffstat, 0, sizeof(struct diffstat_t));
3149 diffstat.xm.consume = diffstat_consume;
3150 for (i = 0; i < q->nr; i++) {
3151 struct diff_filepair *p = q->queue[i];
3152 if (check_pair_status(p))
3153 diff_flush_stat(p, options, &diffstat);
3154 }
3155 if (output_format & DIFF_FORMAT_NUMSTAT)
3156 show_numstat(&diffstat, options);
3157 if (output_format & DIFF_FORMAT_DIFFSTAT)
3158 show_stats(&diffstat, options);
3159 if (output_format & DIFF_FORMAT_SHORTSTAT)
3160 show_shortstats(&diffstat, options);
3161 free_diffstat_info(&diffstat);
3162 separator++;
3163 }
3164 if (output_format & DIFF_FORMAT_DIRSTAT)
3165 show_dirstat(options);
3166
3167 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3168 for (i = 0; i < q->nr; i++)
3169 diff_summary(options->file, q->queue[i]);
3170 separator++;
3171 }
3172
3173 if (output_format & DIFF_FORMAT_PATCH) {
3174 if (separator) {
3175 if (options->stat_sep) {
3176 /* attach patch instead of inline */
3177 fputs(options->stat_sep, options->file);
3178 } else {
3179 putc(options->line_termination, options->file);
3180 }
3181 }
3182
3183 for (i = 0; i < q->nr; i++) {
3184 struct diff_filepair *p = q->queue[i];
3185 if (check_pair_status(p))
3186 diff_flush_patch(p, options);
3187 }
3188 }
3189
3190 if (output_format & DIFF_FORMAT_CALLBACK)
3191 options->format_callback(q, options, options->format_callback_data);
3192
3193 for (i = 0; i < q->nr; i++)
3194 diff_free_filepair(q->queue[i]);
3195free_queue:
3196 free(q->queue);
3197 q->queue = NULL;
3198 q->nr = q->alloc = 0;
3199 if (options->close_file)
3200 fclose(options->file);
3201}
3202
3203static void diffcore_apply_filter(const char *filter)
3204{
3205 int i;
3206 struct diff_queue_struct *q = &diff_queued_diff;
3207 struct diff_queue_struct outq;
3208 outq.queue = NULL;
3209 outq.nr = outq.alloc = 0;
3210
3211 if (!filter)
3212 return;
3213
3214 if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3215 int found;
3216 for (i = found = 0; !found && i < q->nr; i++) {
3217 struct diff_filepair *p = q->queue[i];
3218 if (((p->status == DIFF_STATUS_MODIFIED) &&
3219 ((p->score &&
3220 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3221 (!p->score &&
3222 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3223 ((p->status != DIFF_STATUS_MODIFIED) &&
3224 strchr(filter, p->status)))
3225 found++;
3226 }
3227 if (found)
3228 return;
3229
3230 /* otherwise we will clear the whole queue
3231 * by copying the empty outq at the end of this
3232 * function, but first clear the current entries
3233 * in the queue.
3234 */
3235 for (i = 0; i < q->nr; i++)
3236 diff_free_filepair(q->queue[i]);
3237 }
3238 else {
3239 /* Only the matching ones */
3240 for (i = 0; i < q->nr; i++) {
3241 struct diff_filepair *p = q->queue[i];
3242
3243 if (((p->status == DIFF_STATUS_MODIFIED) &&
3244 ((p->score &&
3245 strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3246 (!p->score &&
3247 strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3248 ((p->status != DIFF_STATUS_MODIFIED) &&
3249 strchr(filter, p->status)))
3250 diff_q(&outq, p);
3251 else
3252 diff_free_filepair(p);
3253 }
3254 }
3255 free(q->queue);
3256 *q = outq;
3257}
3258
3259/* Check whether two filespecs with the same mode and size are identical */
3260static int diff_filespec_is_identical(struct diff_filespec *one,
3261 struct diff_filespec *two)
3262{
3263 if (S_ISGITLINK(one->mode))
3264 return 0;
3265 if (diff_populate_filespec(one, 0))
3266 return 0;
3267 if (diff_populate_filespec(two, 0))
3268 return 0;
3269 return !memcmp(one->data, two->data, one->size);
3270}
3271
3272static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3273{
3274 int i;
3275 struct diff_queue_struct *q = &diff_queued_diff;
3276 struct diff_queue_struct outq;
3277 outq.queue = NULL;
3278 outq.nr = outq.alloc = 0;
3279
3280 for (i = 0; i < q->nr; i++) {
3281 struct diff_filepair *p = q->queue[i];
3282
3283 /*
3284 * 1. Entries that come from stat info dirtyness
3285 * always have both sides (iow, not create/delete),
3286 * one side of the object name is unknown, with
3287 * the same mode and size. Keep the ones that
3288 * do not match these criteria. They have real
3289 * differences.
3290 *
3291 * 2. At this point, the file is known to be modified,
3292 * with the same mode and size, and the object
3293 * name of one side is unknown. Need to inspect
3294 * the identical contents.
3295 */
3296 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3297 !DIFF_FILE_VALID(p->two) ||
3298 (p->one->sha1_valid && p->two->sha1_valid) ||
3299 (p->one->mode != p->two->mode) ||
3300 diff_populate_filespec(p->one, 1) ||
3301 diff_populate_filespec(p->two, 1) ||
3302 (p->one->size != p->two->size) ||
3303 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3304 diff_q(&outq, p);
3305 else {
3306 /*
3307 * The caller can subtract 1 from skip_stat_unmatch
3308 * to determine how many paths were dirty only
3309 * due to stat info mismatch.
3310 */
3311 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3312 diffopt->skip_stat_unmatch++;
3313 diff_free_filepair(p);
3314 }
3315 }
3316 free(q->queue);
3317 *q = outq;
3318}
3319
3320void diffcore_std(struct diff_options *options)
3321{
3322 if (DIFF_OPT_TST(options, QUIET))
3323 return;
3324
3325 if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3326 diffcore_skip_stat_unmatch(options);
3327 if (options->break_opt != -1)
3328 diffcore_break(options->break_opt);
3329 if (options->detect_rename)
3330 diffcore_rename(options);
3331 if (options->break_opt != -1)
3332 diffcore_merge_broken();
3333 if (options->pickaxe)
3334 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3335 if (options->orderfile)
3336 diffcore_order(options->orderfile);
3337 diff_resolve_rename_copy();
3338 diffcore_apply_filter(options->filter);
3339
3340 if (diff_queued_diff.nr)
3341 DIFF_OPT_SET(options, HAS_CHANGES);
3342 else
3343 DIFF_OPT_CLR(options, HAS_CHANGES);
3344}
3345
3346int diff_result_code(struct diff_options *opt, int status)
3347{
3348 int result = 0;
3349 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3350 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3351 return status;
3352 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3353 DIFF_OPT_TST(opt, HAS_CHANGES))
3354 result |= 01;
3355 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3356 DIFF_OPT_TST(opt, CHECK_FAILED))
3357 result |= 02;
3358 return result;
3359}
3360
3361void diff_addremove(struct diff_options *options,
3362 int addremove, unsigned mode,
3363 const unsigned char *sha1,
3364 const char *base, const char *path)
3365{
3366 char concatpath[PATH_MAX];
3367 struct diff_filespec *one, *two;
3368
3369 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3370 return;
3371
3372 /* This may look odd, but it is a preparation for
3373 * feeding "there are unchanged files which should
3374 * not produce diffs, but when you are doing copy
3375 * detection you would need them, so here they are"
3376 * entries to the diff-core. They will be prefixed
3377 * with something like '=' or '*' (I haven't decided
3378 * which but should not make any difference).
3379 * Feeding the same new and old to diff_change()
3380 * also has the same effect.
3381 * Before the final output happens, they are pruned after
3382 * merged into rename/copy pairs as appropriate.
3383 */
3384 if (DIFF_OPT_TST(options, REVERSE_DIFF))
3385 addremove = (addremove == '+' ? '-' :
3386 addremove == '-' ? '+' : addremove);
3387
3388 if (!path) path = "";
3389 sprintf(concatpath, "%s%s", base, path);
3390
3391 if (options->prefix &&
3392 strncmp(concatpath, options->prefix, options->prefix_length))
3393 return;
3394
3395 one = alloc_filespec(concatpath);
3396 two = alloc_filespec(concatpath);
3397
3398 if (addremove != '+')
3399 fill_filespec(one, sha1, mode);
3400 if (addremove != '-')
3401 fill_filespec(two, sha1, mode);
3402
3403 diff_queue(&diff_queued_diff, one, two);
3404 DIFF_OPT_SET(options, HAS_CHANGES);
3405}
3406
3407void diff_change(struct diff_options *options,
3408 unsigned old_mode, unsigned new_mode,
3409 const unsigned char *old_sha1,
3410 const unsigned char *new_sha1,
3411 const char *base, const char *path)
3412{
3413 char concatpath[PATH_MAX];
3414 struct diff_filespec *one, *two;
3415
3416 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3417 && S_ISGITLINK(new_mode))
3418 return;
3419
3420 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3421 unsigned tmp;
3422 const unsigned char *tmp_c;
3423 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3424 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3425 }
3426 if (!path) path = "";
3427 sprintf(concatpath, "%s%s", base, path);
3428
3429 if (options->prefix &&
3430 strncmp(concatpath, options->prefix, options->prefix_length))
3431 return;
3432
3433 one = alloc_filespec(concatpath);
3434 two = alloc_filespec(concatpath);
3435 fill_filespec(one, old_sha1, old_mode);
3436 fill_filespec(two, new_sha1, new_mode);
3437
3438 diff_queue(&diff_queued_diff, one, two);
3439 DIFF_OPT_SET(options, HAS_CHANGES);
3440}
3441
3442void diff_unmerge(struct diff_options *options,
3443 const char *path,
3444 unsigned mode, const unsigned char *sha1)
3445{
3446 struct diff_filespec *one, *two;
3447
3448 if (options->prefix &&
3449 strncmp(path, options->prefix, options->prefix_length))
3450 return;
3451
3452 one = alloc_filespec(path);
3453 two = alloc_filespec(path);
3454 fill_filespec(one, sha1, mode);
3455 diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3456}