1/*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4#include "cache.h"
5#include "tempfile.h"
6#include "quote.h"
7#include "diff.h"
8#include "diffcore.h"
9#include "delta.h"
10#include "xdiff-interface.h"
11#include "color.h"
12#include "attr.h"
13#include "run-command.h"
14#include "utf8.h"
15#include "userdiff.h"
16#include "submodule-config.h"
17#include "submodule.h"
18#include "ll-merge.h"
19#include "string-list.h"
20#include "argv-array.h"
21#include "graph.h"
22#ifdef NO_FAST_WORKING_DIRECTORY
24#define FAST_WORKING_DIRECTORY 0
25#else
26#define FAST_WORKING_DIRECTORY 1
27#endif
28static int diff_detect_rename_default;
30static int diff_indent_heuristic; /* experimental */
31static int diff_compaction_heuristic; /* experimental */
32static int diff_rename_limit_default = 400;
33static int diff_suppress_blank_empty;
34static int diff_use_color_default = -1;
35static int diff_context_default = 3;
36static const char *diff_word_regex_cfg;
37static const char *external_diff_cmd_cfg;
38static const char *diff_order_file_cfg;
39int diff_auto_refresh_index = 1;
40static int diff_mnemonic_prefix;
41static int diff_no_prefix;
42static int diff_stat_graph_width;
43static int diff_dirstat_permille_default = 30;
44static struct diff_options default_diff_options;
45static long diff_algorithm;
46static char diff_colors[][COLOR_MAXLEN] = {
48 GIT_COLOR_RESET,
49 GIT_COLOR_NORMAL, /* CONTEXT */
50 GIT_COLOR_BOLD, /* METAINFO */
51 GIT_COLOR_CYAN, /* FRAGINFO */
52 GIT_COLOR_RED, /* OLD */
53 GIT_COLOR_GREEN, /* NEW */
54 GIT_COLOR_YELLOW, /* COMMIT */
55 GIT_COLOR_BG_RED, /* WHITESPACE */
56 GIT_COLOR_NORMAL, /* FUNCINFO */
57};
58static int parse_diff_color_slot(const char *var)
60{
61 if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
62 return DIFF_CONTEXT;
63 if (!strcasecmp(var, "meta"))
64 return DIFF_METAINFO;
65 if (!strcasecmp(var, "frag"))
66 return DIFF_FRAGINFO;
67 if (!strcasecmp(var, "old"))
68 return DIFF_FILE_OLD;
69 if (!strcasecmp(var, "new"))
70 return DIFF_FILE_NEW;
71 if (!strcasecmp(var, "commit"))
72 return DIFF_COMMIT;
73 if (!strcasecmp(var, "whitespace"))
74 return DIFF_WHITESPACE;
75 if (!strcasecmp(var, "func"))
76 return DIFF_FUNCINFO;
77 return -1;
78}
79static int parse_dirstat_params(struct diff_options *options, const char *params_string,
81 struct strbuf *errmsg)
82{
83 char *params_copy = xstrdup(params_string);
84 struct string_list params = STRING_LIST_INIT_NODUP;
85 int ret = 0;
86 int i;
87 if (*params_copy)
89 string_list_split_in_place(¶ms, params_copy, ',', -1);
90 for (i = 0; i < params.nr; i++) {
91 const char *p = params.items[i].string;
92 if (!strcmp(p, "changes")) {
93 DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
94 DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
95 } else if (!strcmp(p, "lines")) {
96 DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
97 DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
98 } else if (!strcmp(p, "files")) {
99 DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
100 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
101 } else if (!strcmp(p, "noncumulative")) {
102 DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
103 } else if (!strcmp(p, "cumulative")) {
104 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
105 } else if (isdigit(*p)) {
106 char *end;
107 int permille = strtoul(p, &end, 10) * 10;
108 if (*end == '.' && isdigit(*++end)) {
109 /* only use first digit */
110 permille += *end - '0';
111 /* .. and ignore any further digits */
112 while (isdigit(*++end))
113 ; /* nothing */
114 }
115 if (!*end)
116 options->dirstat_permille = permille;
117 else {
118 strbuf_addf(errmsg, _(" Failed to parse dirstat cut-off percentage '%s'\n"),
119 p);
120 ret++;
121 }
122 } else {
123 strbuf_addf(errmsg, _(" Unknown dirstat parameter '%s'\n"), p);
124 ret++;
125 }
126 }
128 string_list_clear(¶ms, 0);
129 free(params_copy);
130 return ret;
131}
132static int parse_submodule_params(struct diff_options *options, const char *value)
134{
135 if (!strcmp(value, "log"))
136 options->submodule_format = DIFF_SUBMODULE_LOG;
137 else if (!strcmp(value, "short"))
138 options->submodule_format = DIFF_SUBMODULE_SHORT;
139 else if (!strcmp(value, "diff"))
140 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
141 else
142 return -1;
143 return 0;
144}
145static int git_config_rename(const char *var, const char *value)
147{
148 if (!value)
149 return DIFF_DETECT_RENAME;
150 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
151 return DIFF_DETECT_COPY;
152 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
153}
154long parse_algorithm_value(const char *value)
156{
157 if (!value)
158 return -1;
159 else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
160 return 0;
161 else if (!strcasecmp(value, "minimal"))
162 return XDF_NEED_MINIMAL;
163 else if (!strcasecmp(value, "patience"))
164 return XDF_PATIENCE_DIFF;
165 else if (!strcasecmp(value, "histogram"))
166 return XDF_HISTOGRAM_DIFF;
167 return -1;
168}
169/*
171 * These are to give UI layer defaults.
172 * The core-level commands such as git-diff-files should
173 * never be affected by the setting of diff.renames
174 * the user happens to have in the configuration file.
175 */
176void init_diff_ui_defaults(void)
177{
178 diff_detect_rename_default = 1;
179}
180int git_diff_heuristic_config(const char *var, const char *value, void *cb)
182{
183 if (!strcmp(var, "diff.indentheuristic")) {
184 diff_indent_heuristic = git_config_bool(var, value);
185 if (diff_indent_heuristic)
186 diff_compaction_heuristic = 0;
187 }
188 if (!strcmp(var, "diff.compactionheuristic")) {
189 diff_compaction_heuristic = git_config_bool(var, value);
190 if (diff_compaction_heuristic)
191 diff_indent_heuristic = 0;
192 }
193 return 0;
194}
195int git_diff_ui_config(const char *var, const char *value, void *cb)
197{
198 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
199 diff_use_color_default = git_config_colorbool(var, value);
200 return 0;
201 }
202 if (!strcmp(var, "diff.context")) {
203 diff_context_default = git_config_int(var, value);
204 if (diff_context_default < 0)
205 return -1;
206 return 0;
207 }
208 if (!strcmp(var, "diff.renames")) {
209 diff_detect_rename_default = git_config_rename(var, value);
210 return 0;
211 }
212 if (!strcmp(var, "diff.autorefreshindex")) {
213 diff_auto_refresh_index = git_config_bool(var, value);
214 return 0;
215 }
216 if (!strcmp(var, "diff.mnemonicprefix")) {
217 diff_mnemonic_prefix = git_config_bool(var, value);
218 return 0;
219 }
220 if (!strcmp(var, "diff.noprefix")) {
221 diff_no_prefix = git_config_bool(var, value);
222 return 0;
223 }
224 if (!strcmp(var, "diff.statgraphwidth")) {
225 diff_stat_graph_width = git_config_int(var, value);
226 return 0;
227 }
228 if (!strcmp(var, "diff.external"))
229 return git_config_string(&external_diff_cmd_cfg, var, value);
230 if (!strcmp(var, "diff.wordregex"))
231 return git_config_string(&diff_word_regex_cfg, var, value);
232 if (!strcmp(var, "diff.orderfile"))
233 return git_config_pathname(&diff_order_file_cfg, var, value);
234 if (!strcmp(var, "diff.ignoresubmodules"))
236 handle_ignore_submodules_arg(&default_diff_options, value);
237 if (!strcmp(var, "diff.submodule")) {
239 if (parse_submodule_params(&default_diff_options, value))
240 warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
241 value);
242 return 0;
243 }
244 if (!strcmp(var, "diff.algorithm")) {
246 diff_algorithm = parse_algorithm_value(value);
247 if (diff_algorithm < 0)
248 return -1;
249 return 0;
250 }
251 if (git_diff_heuristic_config(var, value, cb) < 0)
253 return -1;
254 if (git_color_config(var, value, cb) < 0)
255 return -1;
256 return git_diff_basic_config(var, value, cb);
258}
259int git_diff_basic_config(const char *var, const char *value, void *cb)
261{
262 const char *name;
263 if (!strcmp(var, "diff.renamelimit")) {
265 diff_rename_limit_default = git_config_int(var, value);
266 return 0;
267 }
268 if (userdiff_config(var, value) < 0)
270 return -1;
271 if (skip_prefix(var, "diff.color.", &name) ||
273 skip_prefix(var, "color.diff.", &name)) {
274 int slot = parse_diff_color_slot(name);
275 if (slot < 0)
276 return 0;
277 if (!value)
278 return config_error_nonbool(var);
279 return color_parse(value, diff_colors[slot]);
280 }
281 /* like GNU diff's --suppress-blank-empty option */
283 if (!strcmp(var, "diff.suppressblankempty") ||
284 /* for backwards compatibility */
285 !strcmp(var, "diff.suppress-blank-empty")) {
286 diff_suppress_blank_empty = git_config_bool(var, value);
287 return 0;
288 }
289 if (!strcmp(var, "diff.dirstat")) {
291 struct strbuf errmsg = STRBUF_INIT;
292 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
293 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
294 warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
295 errmsg.buf);
296 strbuf_release(&errmsg);
297 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
298 return 0;
299 }
300 if (starts_with(var, "submodule."))
302 return parse_submodule_config_option(var, value);
303 return git_default_config(var, value, cb);
305}
306static char *quote_two(const char *one, const char *two)
308{
309 int need_one = quote_c_style(one, NULL, NULL, 1);
310 int need_two = quote_c_style(two, NULL, NULL, 1);
311 struct strbuf res = STRBUF_INIT;
312 if (need_one + need_two) {
314 strbuf_addch(&res, '"');
315 quote_c_style(one, &res, NULL, 1);
316 quote_c_style(two, &res, NULL, 1);
317 strbuf_addch(&res, '"');
318 } else {
319 strbuf_addstr(&res, one);
320 strbuf_addstr(&res, two);
321 }
322 return strbuf_detach(&res, NULL);
323}
324static const char *external_diff(void)
326{
327 static const char *external_diff_cmd = NULL;
328 static int done_preparing = 0;
329 if (done_preparing)
331 return external_diff_cmd;
332 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
333 if (!external_diff_cmd)
334 external_diff_cmd = external_diff_cmd_cfg;
335 done_preparing = 1;
336 return external_diff_cmd;
337}
338/*
340 * Keep track of files used for diffing. Sometimes such an entry
341 * refers to a temporary file, sometimes to an existing file, and
342 * sometimes to "/dev/null".
343 */
344static struct diff_tempfile {
345 /*
346 * filename external diff should read from, or NULL if this
347 * entry is currently not in use:
348 */
349 const char *name;
350 char hex[GIT_SHA1_HEXSZ + 1];
352 char mode[10];
353 /*
355 * If this diff_tempfile instance refers to a temporary file,
356 * this tempfile object is used to manage its lifetime.
357 */
358 struct tempfile tempfile;
359} diff_temp[2];
360typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
362struct emit_callback {
364 int color_diff;
365 unsigned ws_rule;
366 int blank_at_eof_in_preimage;
367 int blank_at_eof_in_postimage;
368 int lno_in_preimage;
369 int lno_in_postimage;
370 sane_truncate_fn truncate;
371 const char **label_path;
372 struct diff_words_data *diff_words;
373 struct diff_options *opt;
374 struct strbuf *header;
375};
376static int count_lines(const char *data, int size)
378{
379 int count, ch, completely_empty = 1, nl_just_seen = 0;
380 count = 0;
381 while (0 < size--) {
382 ch = *data++;
383 if (ch == '\n') {
384 count++;
385 nl_just_seen = 1;
386 completely_empty = 0;
387 }
388 else {
389 nl_just_seen = 0;
390 completely_empty = 0;
391 }
392 }
393 if (completely_empty)
394 return 0;
395 if (!nl_just_seen)
396 count++; /* no trailing newline */
397 return count;
398}
399static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
401{
402 if (!DIFF_FILE_VALID(one)) {
403 mf->ptr = (char *)""; /* does not matter */
404 mf->size = 0;
405 return 0;
406 }
407 else if (diff_populate_filespec(one, 0))
408 return -1;
409 mf->ptr = one->data;
411 mf->size = one->size;
412 return 0;
413}
414/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
416static unsigned long diff_filespec_size(struct diff_filespec *one)
417{
418 if (!DIFF_FILE_VALID(one))
419 return 0;
420 diff_populate_filespec(one, CHECK_SIZE_ONLY);
421 return one->size;
422}
423static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
425{
426 char *ptr = mf->ptr;
427 long size = mf->size;
428 int cnt = 0;
429 if (!size)
431 return cnt;
432 ptr += size - 1; /* pointing at the very end */
433 if (*ptr != '\n')
434 ; /* incomplete line */
435 else
436 ptr--; /* skip the last LF */
437 while (mf->ptr < ptr) {
438 char *prev_eol;
439 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
440 if (*prev_eol == '\n')
441 break;
442 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
443 break;
444 cnt++;
445 ptr = prev_eol - 1;
446 }
447 return cnt;
448}
449static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
451 struct emit_callback *ecbdata)
452{
453 int l1, l2, at;
454 unsigned ws_rule = ecbdata->ws_rule;
455 l1 = count_trailing_blank(mf1, ws_rule);
456 l2 = count_trailing_blank(mf2, ws_rule);
457 if (l2 <= l1) {
458 ecbdata->blank_at_eof_in_preimage = 0;
459 ecbdata->blank_at_eof_in_postimage = 0;
460 return;
461 }
462 at = count_lines(mf1->ptr, mf1->size);
463 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
464 at = count_lines(mf2->ptr, mf2->size);
466 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
467}
468static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
470 int first, const char *line, int len)
471{
472 int has_trailing_newline, has_trailing_carriage_return;
473 int nofirst;
474 FILE *file = o->file;
475 fputs(diff_line_prefix(o), file);
477 if (len == 0) {
479 has_trailing_newline = (first == '\n');
480 has_trailing_carriage_return = (!has_trailing_newline &&
481 (first == '\r'));
482 nofirst = has_trailing_newline || has_trailing_carriage_return;
483 } else {
484 has_trailing_newline = (len > 0 && line[len-1] == '\n');
485 if (has_trailing_newline)
486 len--;
487 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
488 if (has_trailing_carriage_return)
489 len--;
490 nofirst = 0;
491 }
492 if (len || !nofirst) {
494 fputs(set, file);
495 if (!nofirst)
496 fputc(first, file);
497 fwrite(line, len, 1, file);
498 fputs(reset, file);
499 }
500 if (has_trailing_carriage_return)
501 fputc('\r', file);
502 if (has_trailing_newline)
503 fputc('\n', file);
504}
505static void emit_line(struct diff_options *o, const char *set, const char *reset,
507 const char *line, int len)
508{
509 emit_line_0(o, set, reset, line[0], line+1, len-1);
510}
511static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
513{
514 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
515 ecbdata->blank_at_eof_in_preimage &&
516 ecbdata->blank_at_eof_in_postimage &&
517 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
518 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
519 return 0;
520 return ws_blank_line(line, len, ecbdata->ws_rule);
521}
522static void emit_line_checked(const char *reset,
524 struct emit_callback *ecbdata,
525 const char *line, int len,
526 enum color_diff color,
527 unsigned ws_error_highlight,
528 char sign)
529{
530 const char *set = diff_get_color(ecbdata->color_diff, color);
531 const char *ws = NULL;
532 if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
534 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
535 if (!*ws)
536 ws = NULL;
537 }
538 if (!ws)
540 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
541 else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
542 /* Blank line at EOF - paint '+' as well */
543 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
544 else {
545 /* Emit just the prefix, then the rest. */
546 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
547 ws_check_emit(line, len, ecbdata->ws_rule,
548 ecbdata->opt->file, set, reset, ws);
549 }
550}
551static void emit_add_line(const char *reset,
553 struct emit_callback *ecbdata,
554 const char *line, int len)
555{
556 emit_line_checked(reset, ecbdata, line, len,
557 DIFF_FILE_NEW, WSEH_NEW, '+');
558}
559static void emit_del_line(const char *reset,
561 struct emit_callback *ecbdata,
562 const char *line, int len)
563{
564 emit_line_checked(reset, ecbdata, line, len,
565 DIFF_FILE_OLD, WSEH_OLD, '-');
566}
567static void emit_context_line(const char *reset,
569 struct emit_callback *ecbdata,
570 const char *line, int len)
571{
572 emit_line_checked(reset, ecbdata, line, len,
573 DIFF_CONTEXT, WSEH_CONTEXT, ' ');
574}
575static void emit_hunk_header(struct emit_callback *ecbdata,
577 const char *line, int len)
578{
579 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
580 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
581 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
582 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
583 static const char atat[2] = { '@', '@' };
584 const char *cp, *ep;
585 struct strbuf msgbuf = STRBUF_INIT;
586 int org_len = len;
587 int i = 1;
588 /*
590 * As a hunk header must begin with "@@ -<old>, +<new> @@",
591 * it always is at least 10 bytes long.
592 */
593 if (len < 10 ||
594 memcmp(line, atat, 2) ||
595 !(ep = memmem(line + 2, len - 2, atat, 2))) {
596 emit_line(ecbdata->opt, context, reset, line, len);
597 return;
598 }
599 ep += 2; /* skip over @@ */
600 /* The hunk header in fraginfo color */
602 strbuf_addstr(&msgbuf, frag);
603 strbuf_add(&msgbuf, line, ep - line);
604 strbuf_addstr(&msgbuf, reset);
605 /*
607 * trailing "\r\n"
608 */
609 for ( ; i < 3; i++)
610 if (line[len - i] == '\r' || line[len - i] == '\n')
611 len--;
612 /* blank before the func header */
614 for (cp = ep; ep - line < len; ep++)
615 if (*ep != ' ' && *ep != '\t')
616 break;
617 if (ep != cp) {
618 strbuf_addstr(&msgbuf, context);
619 strbuf_add(&msgbuf, cp, ep - cp);
620 strbuf_addstr(&msgbuf, reset);
621 }
622 if (ep < line + len) {
624 strbuf_addstr(&msgbuf, func);
625 strbuf_add(&msgbuf, ep, line + len - ep);
626 strbuf_addstr(&msgbuf, reset);
627 }
628 strbuf_add(&msgbuf, line + len, org_len - len);
630 emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
631 strbuf_release(&msgbuf);
632}
633static struct diff_tempfile *claim_diff_tempfile(void) {
635 int i;
636 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
637 if (!diff_temp[i].name)
638 return diff_temp + i;
639 die("BUG: diff is failing to clean up its tempfiles");
640}
641static void remove_tempfile(void)
643{
644 int i;
645 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
646 if (is_tempfile_active(&diff_temp[i].tempfile))
647 delete_tempfile(&diff_temp[i].tempfile);
648 diff_temp[i].name = NULL;
649 }
650}
651static void print_line_count(FILE *file, int count)
653{
654 switch (count) {
655 case 0:
656 fprintf(file, "0,0");
657 break;
658 case 1:
659 fprintf(file, "1");
660 break;
661 default:
662 fprintf(file, "1,%d", count);
663 break;
664 }
665}
666static void emit_rewrite_lines(struct emit_callback *ecb,
668 int prefix, const char *data, int size)
669{
670 const char *endp = NULL;
671 static const char *nneof = " No newline at end of file\n";
672 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
673 while (0 < size) {
675 int len;
676 endp = memchr(data, '\n', size);
678 len = endp ? (endp - data + 1) : size;
679 if (prefix != '+') {
680 ecb->lno_in_preimage++;
681 emit_del_line(reset, ecb, data, len);
682 } else {
683 ecb->lno_in_postimage++;
684 emit_add_line(reset, ecb, data, len);
685 }
686 size -= len;
687 data += len;
688 }
689 if (!endp) {
690 const char *context = diff_get_color(ecb->color_diff,
691 DIFF_CONTEXT);
692 putc('\n', ecb->opt->file);
693 emit_line_0(ecb->opt, context, reset, '\\',
694 nneof, strlen(nneof));
695 }
696}
697static void emit_rewrite_diff(const char *name_a,
699 const char *name_b,
700 struct diff_filespec *one,
701 struct diff_filespec *two,
702 struct userdiff_driver *textconv_one,
703 struct userdiff_driver *textconv_two,
704 struct diff_options *o)
705{
706 int lc_a, lc_b;
707 const char *name_a_tab, *name_b_tab;
708 const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
709 const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
710 const char *reset = diff_get_color(o->use_color, DIFF_RESET);
711 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
712 const char *a_prefix, *b_prefix;
713 char *data_one, *data_two;
714 size_t size_one, size_two;
715 struct emit_callback ecbdata;
716 const char *line_prefix = diff_line_prefix(o);
717 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
719 a_prefix = o->b_prefix;
720 b_prefix = o->a_prefix;
721 } else {
722 a_prefix = o->a_prefix;
723 b_prefix = o->b_prefix;
724 }
725 name_a += (*name_a == '/');
727 name_b += (*name_b == '/');
728 name_a_tab = strchr(name_a, ' ') ? "\t" : "";
729 name_b_tab = strchr(name_b, ' ') ? "\t" : "";
730 strbuf_reset(&a_name);
732 strbuf_reset(&b_name);
733 quote_two_c_style(&a_name, a_prefix, name_a, 0);
734 quote_two_c_style(&b_name, b_prefix, name_b, 0);
735 size_one = fill_textconv(textconv_one, one, &data_one);
737 size_two = fill_textconv(textconv_two, two, &data_two);
738 memset(&ecbdata, 0, sizeof(ecbdata));
740 ecbdata.color_diff = want_color(o->use_color);
741 ecbdata.ws_rule = whitespace_rule(name_b);
742 ecbdata.opt = o;
743 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
744 mmfile_t mf1, mf2;
745 mf1.ptr = (char *)data_one;
746 mf2.ptr = (char *)data_two;
747 mf1.size = size_one;
748 mf2.size = size_two;
749 check_blank_at_eof(&mf1, &mf2, &ecbdata);
750 }
751 ecbdata.lno_in_preimage = 1;
752 ecbdata.lno_in_postimage = 1;
753 lc_a = count_lines(data_one, size_one);
755 lc_b = count_lines(data_two, size_two);
756 fprintf(o->file,
757 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
758 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
759 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
760 line_prefix, fraginfo);
761 if (!o->irreversible_delete)
762 print_line_count(o->file, lc_a);
763 else
764 fprintf(o->file, "?,?");
765 fprintf(o->file, " +");
766 print_line_count(o->file, lc_b);
767 fprintf(o->file, " @@%s\n", reset);
768 if (lc_a && !o->irreversible_delete)
769 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
770 if (lc_b)
771 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
772 if (textconv_one)
773 free((char *)data_one);
774 if (textconv_two)
775 free((char *)data_two);
776}
777struct diff_words_buffer {
779 mmfile_t text;
780 long alloc;
781 struct diff_words_orig {
782 const char *begin, *end;
783 } *orig;
784 int orig_nr, orig_alloc;
785};
786static void diff_words_append(char *line, unsigned long len,
788 struct diff_words_buffer *buffer)
789{
790 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
791 line++;
792 len--;
793 memcpy(buffer->text.ptr + buffer->text.size, line, len);
794 buffer->text.size += len;
795 buffer->text.ptr[buffer->text.size] = '\0';
796}
797struct diff_words_style_elem {
799 const char *prefix;
800 const char *suffix;
801 const char *color; /* NULL; filled in by the setup code if
802 * color is enabled */
803};
804struct diff_words_style {
806 enum diff_words_type type;
807 struct diff_words_style_elem new, old, ctx;
808 const char *newline;
809};
810static struct diff_words_style diff_words_styles[] = {
812 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
813 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
814 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
815};
816struct diff_words_data {
818 struct diff_words_buffer minus, plus;
819 const char *current_plus;
820 int last_minus;
821 struct diff_options *opt;
822 regex_t *word_regex;
823 enum diff_words_type type;
824 struct diff_words_style *style;
825};
826static int fn_out_diff_words_write_helper(FILE *fp,
828 struct diff_words_style_elem *st_el,
829 const char *newline,
830 size_t count, const char *buf,
831 const char *line_prefix)
832{
833 int print = 0;
834 while (count) {
836 char *p = memchr(buf, '\n', count);
837 if (print)
838 fputs(line_prefix, fp);
839 if (p != buf) {
840 if (st_el->color && fputs(st_el->color, fp) < 0)
841 return -1;
842 if (fputs(st_el->prefix, fp) < 0 ||
843 fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
844 fputs(st_el->suffix, fp) < 0)
845 return -1;
846 if (st_el->color && *st_el->color
847 && fputs(GIT_COLOR_RESET, fp) < 0)
848 return -1;
849 }
850 if (!p)
851 return 0;
852 if (fputs(newline, fp) < 0)
853 return -1;
854 count -= p + 1 - buf;
855 buf = p + 1;
856 print = 1;
857 }
858 return 0;
859}
860/*
862 * '--color-words' algorithm can be described as:
863 *
864 * 1. collect a the minus/plus lines of a diff hunk, divided into
865 * minus-lines and plus-lines;
866 *
867 * 2. break both minus-lines and plus-lines into words and
868 * place them into two mmfile_t with one word for each line;
869 *
870 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
871 *
872 * And for the common parts of the both file, we output the plus side text.
873 * diff_words->current_plus is used to trace the current position of the plus file
874 * which printed. diff_words->last_minus is used to trace the last minus word
875 * printed.
876 *
877 * For '--graph' to work with '--color-words', we need to output the graph prefix
878 * on each line of color words output. Generally, there are two conditions on
879 * which we should output the prefix.
880 *
881 * 1. diff_words->last_minus == 0 &&
882 * diff_words->current_plus == diff_words->plus.text.ptr
883 *
884 * that is: the plus text must start as a new line, and if there is no minus
885 * word printed, a graph prefix must be printed.
886 *
887 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
888 * *(diff_words->current_plus - 1) == '\n'
889 *
890 * that is: a graph prefix must be printed following a '\n'
891 */
892static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
893{
894 if ((diff_words->last_minus == 0 &&
895 diff_words->current_plus == diff_words->plus.text.ptr) ||
896 (diff_words->current_plus > diff_words->plus.text.ptr &&
897 *(diff_words->current_plus - 1) == '\n')) {
898 return 1;
899 } else {
900 return 0;
901 }
902}
903static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
905{
906 struct diff_words_data *diff_words = priv;
907 struct diff_words_style *style = diff_words->style;
908 int minus_first, minus_len, plus_first, plus_len;
909 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
910 struct diff_options *opt = diff_words->opt;
911 const char *line_prefix;
912 if (line[0] != '@' || parse_hunk_header(line, len,
914 &minus_first, &minus_len, &plus_first, &plus_len))
915 return;
916 assert(opt);
918 line_prefix = diff_line_prefix(opt);
919 /* POSIX requires that first be decremented by one if len == 0... */
921 if (minus_len) {
922 minus_begin = diff_words->minus.orig[minus_first].begin;
923 minus_end =
924 diff_words->minus.orig[minus_first + minus_len - 1].end;
925 } else
926 minus_begin = minus_end =
927 diff_words->minus.orig[minus_first].end;
928 if (plus_len) {
930 plus_begin = diff_words->plus.orig[plus_first].begin;
931 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
932 } else
933 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
934 if (color_words_output_graph_prefix(diff_words)) {
936 fputs(line_prefix, diff_words->opt->file);
937 }
938 if (diff_words->current_plus != plus_begin) {
939 fn_out_diff_words_write_helper(diff_words->opt->file,
940 &style->ctx, style->newline,
941 plus_begin - diff_words->current_plus,
942 diff_words->current_plus, line_prefix);
943 if (*(plus_begin - 1) == '\n')
944 fputs(line_prefix, diff_words->opt->file);
945 }
946 if (minus_begin != minus_end) {
947 fn_out_diff_words_write_helper(diff_words->opt->file,
948 &style->old, style->newline,
949 minus_end - minus_begin, minus_begin,
950 line_prefix);
951 }
952 if (plus_begin != plus_end) {
953 fn_out_diff_words_write_helper(diff_words->opt->file,
954 &style->new, style->newline,
955 plus_end - plus_begin, plus_begin,
956 line_prefix);
957 }
958 diff_words->current_plus = plus_end;
960 diff_words->last_minus = minus_first;
961}
962/* This function starts looking at *begin, and returns 0 iff a word was found. */
964static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
965 int *begin, int *end)
966{
967 if (word_regex && *begin < buffer->size) {
968 regmatch_t match[1];
969 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
970 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
971 '\n', match[0].rm_eo - match[0].rm_so);
972 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
973 *begin += match[0].rm_so;
974 return *begin >= *end;
975 }
976 return -1;
977 }
978 /* find the next word */
980 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
981 (*begin)++;
982 if (*begin >= buffer->size)
983 return -1;
984 /* find the end of the word */
986 *end = *begin + 1;
987 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
988 (*end)++;
989 return 0;
991}
992/*
994 * This function splits the words in buffer->text, stores the list with
995 * newline separator into out, and saves the offsets of the original words
996 * in buffer->orig.
997 */
998static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
999 regex_t *word_regex)
1000{
1001 int i, j;
1002 long alloc = 0;
1003 out->size = 0;
1005 out->ptr = NULL;
1006 /* fake an empty "0th" word */
1008 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1009 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1010 buffer->orig_nr = 1;
1011 for (i = 0; i < buffer->text.size; i++) {
1013 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1014 return;
1015 /* store original boundaries */
1017 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1018 buffer->orig_alloc);
1019 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1020 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1021 buffer->orig_nr++;
1022 /* store one word */
1024 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1025 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1026 out->ptr[out->size + j - i] = '\n';
1027 out->size += j - i + 1;
1028 i = j - 1;
1030 }
1031}
1032/* this executes the word diff on the accumulated buffers */
1034static void diff_words_show(struct diff_words_data *diff_words)
1035{
1036 xpparam_t xpp;
1037 xdemitconf_t xecfg;
1038 mmfile_t minus, plus;
1039 struct diff_words_style *style = diff_words->style;
1040 struct diff_options *opt = diff_words->opt;
1042 const char *line_prefix;
1043 assert(opt);
1045 line_prefix = diff_line_prefix(opt);
1046 /* special case: only removal */
1048 if (!diff_words->plus.text.size) {
1049 fputs(line_prefix, diff_words->opt->file);
1050 fn_out_diff_words_write_helper(diff_words->opt->file,
1051 &style->old, style->newline,
1052 diff_words->minus.text.size,
1053 diff_words->minus.text.ptr, line_prefix);
1054 diff_words->minus.text.size = 0;
1055 return;
1056 }
1057 diff_words->current_plus = diff_words->plus.text.ptr;
1059 diff_words->last_minus = 0;
1060 memset(&xpp, 0, sizeof(xpp));
1062 memset(&xecfg, 0, sizeof(xecfg));
1063 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1064 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1065 xpp.flags = 0;
1066 /* as only the hunk header will be parsed, we need a 0-context */
1067 xecfg.ctxlen = 0;
1068 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1069 &xpp, &xecfg))
1070 die("unable to generate word diff");
1071 free(minus.ptr);
1072 free(plus.ptr);
1073 if (diff_words->current_plus != diff_words->plus.text.ptr +
1074 diff_words->plus.text.size) {
1075 if (color_words_output_graph_prefix(diff_words))
1076 fputs(line_prefix, diff_words->opt->file);
1077 fn_out_diff_words_write_helper(diff_words->opt->file,
1078 &style->ctx, style->newline,
1079 diff_words->plus.text.ptr + diff_words->plus.text.size
1080 - diff_words->current_plus, diff_words->current_plus,
1081 line_prefix);
1082 }
1083 diff_words->minus.text.size = diff_words->plus.text.size = 0;
1084}
1085/* In "color-words" mode, show word-diff of words accumulated in the buffer */
1087static void diff_words_flush(struct emit_callback *ecbdata)
1088{
1089 if (ecbdata->diff_words->minus.text.size ||
1090 ecbdata->diff_words->plus.text.size)
1091 diff_words_show(ecbdata->diff_words);
1092}
1093static void diff_filespec_load_driver(struct diff_filespec *one)
1095{
1096 /* Use already-loaded driver */
1097 if (one->driver)
1098 return;
1099 if (S_ISREG(one->mode))
1101 one->driver = userdiff_find_by_path(one->path);
1102 /* Fallback to default settings */
1104 if (!one->driver)
1105 one->driver = userdiff_find_by_name("default");
1106}
1107static const char *userdiff_word_regex(struct diff_filespec *one)
1109{
1110 diff_filespec_load_driver(one);
1111 return one->driver->word_regex;
1112}
1113static void init_diff_words_data(struct emit_callback *ecbdata,
1115 struct diff_options *orig_opts,
1116 struct diff_filespec *one,
1117 struct diff_filespec *two)
1118{
1119 int i;
1120 struct diff_options *o = xmalloc(sizeof(struct diff_options));
1121 memcpy(o, orig_opts, sizeof(struct diff_options));
1122 ecbdata->diff_words =
1124 xcalloc(1, sizeof(struct diff_words_data));
1125 ecbdata->diff_words->type = o->word_diff;
1126 ecbdata->diff_words->opt = o;
1127 if (!o->word_regex)
1128 o->word_regex = userdiff_word_regex(one);
1129 if (!o->word_regex)
1130 o->word_regex = userdiff_word_regex(two);
1131 if (!o->word_regex)
1132 o->word_regex = diff_word_regex_cfg;
1133 if (o->word_regex) {
1134 ecbdata->diff_words->word_regex = (regex_t *)
1135 xmalloc(sizeof(regex_t));
1136 if (regcomp(ecbdata->diff_words->word_regex,
1137 o->word_regex,
1138 REG_EXTENDED | REG_NEWLINE))
1139 die ("Invalid regular expression: %s",
1140 o->word_regex);
1141 }
1142 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1143 if (o->word_diff == diff_words_styles[i].type) {
1144 ecbdata->diff_words->style =
1145 &diff_words_styles[i];
1146 break;
1147 }
1148 }
1149 if (want_color(o->use_color)) {
1150 struct diff_words_style *st = ecbdata->diff_words->style;
1151 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1152 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1153 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1154 }
1155}
1156static void free_diff_words_data(struct emit_callback *ecbdata)
1158{
1159 if (ecbdata->diff_words) {
1160 diff_words_flush(ecbdata);
1161 free (ecbdata->diff_words->opt);
1162 free (ecbdata->diff_words->minus.text.ptr);
1163 free (ecbdata->diff_words->minus.orig);
1164 free (ecbdata->diff_words->plus.text.ptr);
1165 free (ecbdata->diff_words->plus.orig);
1166 if (ecbdata->diff_words->word_regex) {
1167 regfree(ecbdata->diff_words->word_regex);
1168 free(ecbdata->diff_words->word_regex);
1169 }
1170 free(ecbdata->diff_words);
1171 ecbdata->diff_words = NULL;
1172 }
1173}
1174const char *diff_get_color(int diff_use_color, enum color_diff ix)
1176{
1177 if (want_color(diff_use_color))
1178 return diff_colors[ix];
1179 return "";
1180}
1181const char *diff_line_prefix(struct diff_options *opt)
1183{
1184 struct strbuf *msgbuf;
1185 if (!opt->output_prefix)
1186 return "";
1187 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1189 return msgbuf->buf;
1190}
1191static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1193{
1194 const char *cp;
1195 unsigned long allot;
1196 size_t l = len;
1197 if (ecb->truncate)
1199 return ecb->truncate(line, len);
1200 cp = line;
1201 allot = l;
1202 while (0 < l) {
1203 (void) utf8_width(&cp, &l);
1204 if (!cp)
1205 break; /* truncated in the middle? */
1206 }
1207 return allot - l;
1208}
1209static void find_lno(const char *line, struct emit_callback *ecbdata)
1211{
1212 const char *p;
1213 ecbdata->lno_in_preimage = 0;
1214 ecbdata->lno_in_postimage = 0;
1215 p = strchr(line, '-');
1216 if (!p)
1217 return; /* cannot happen */
1218 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1219 p = strchr(p, '+');
1220 if (!p)
1221 return; /* cannot happen */
1222 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1223}
1224static void fn_out_consume(void *priv, char *line, unsigned long len)
1226{
1227 struct emit_callback *ecbdata = priv;
1228 const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1229 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1230 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1231 struct diff_options *o = ecbdata->opt;
1232 const char *line_prefix = diff_line_prefix(o);
1233 o->found_changes = 1;
1235 if (ecbdata->header) {
1237 fprintf(o->file, "%s", ecbdata->header->buf);
1238 strbuf_reset(ecbdata->header);
1239 ecbdata->header = NULL;
1240 }
1241 if (ecbdata->label_path[0]) {
1243 const char *name_a_tab, *name_b_tab;
1244 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1246 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1247 fprintf(o->file, "%s%s--- %s%s%s\n",
1249 line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1250 fprintf(o->file, "%s%s+++ %s%s%s\n",
1251 line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1252 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1253 }
1254 if (diff_suppress_blank_empty
1256 && len == 2 && line[0] == ' ' && line[1] == '\n') {
1257 line[0] = '\n';
1258 len = 1;
1259 }
1260 if (line[0] == '@') {
1262 if (ecbdata->diff_words)
1263 diff_words_flush(ecbdata);
1264 len = sane_truncate_line(ecbdata, line, len);
1265 find_lno(line, ecbdata);
1266 emit_hunk_header(ecbdata, line, len);
1267 if (line[len-1] != '\n')
1268 putc('\n', o->file);
1269 return;
1270 }
1271 if (ecbdata->diff_words) {
1273 if (line[0] == '-') {
1274 diff_words_append(line, len,
1275 &ecbdata->diff_words->minus);
1276 return;
1277 } else if (line[0] == '+') {
1278 diff_words_append(line, len,
1279 &ecbdata->diff_words->plus);
1280 return;
1281 } else if (starts_with(line, "\\ ")) {
1282 /*
1283 * Eat the "no newline at eof" marker as if we
1284 * saw a "+" or "-" line with nothing on it,
1285 * and return without diff_words_flush() to
1286 * defer processing. If this is the end of
1287 * preimage, more "+" lines may come after it.
1288 */
1289 return;
1290 }
1291 diff_words_flush(ecbdata);
1292 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1293 emit_line(o, context, reset, line, len);
1294 fputs("~\n", o->file);
1295 } else {
1296 /*
1297 * Skip the prefix character, if any. With
1298 * diff_suppress_blank_empty, there may be
1299 * none.
1300 */
1301 if (line[0] != '\n') {
1302 line++;
1303 len--;
1304 }
1305 emit_line(o, context, reset, line, len);
1306 }
1307 return;
1308 }
1309 switch (line[0]) {
1311 case '+':
1312 ecbdata->lno_in_postimage++;
1313 emit_add_line(reset, ecbdata, line + 1, len - 1);
1314 break;
1315 case '-':
1316 ecbdata->lno_in_preimage++;
1317 emit_del_line(reset, ecbdata, line + 1, len - 1);
1318 break;
1319 case ' ':
1320 ecbdata->lno_in_postimage++;
1321 ecbdata->lno_in_preimage++;
1322 emit_context_line(reset, ecbdata, line + 1, len - 1);
1323 break;
1324 default:
1325 /* incomplete line at the end */
1326 ecbdata->lno_in_preimage++;
1327 emit_line(o, diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1328 reset, line, len);
1329 break;
1330 }
1331}
1332static char *pprint_rename(const char *a, const char *b)
1334{
1335 const char *old = a;
1336 const char *new = b;
1337 struct strbuf name = STRBUF_INIT;
1338 int pfx_length, sfx_length;
1339 int pfx_adjust_for_slash;
1340 int len_a = strlen(a);
1341 int len_b = strlen(b);
1342 int a_midlen, b_midlen;
1343 int qlen_a = quote_c_style(a, NULL, NULL, 0);
1344 int qlen_b = quote_c_style(b, NULL, NULL, 0);
1345 if (qlen_a || qlen_b) {
1347 quote_c_style(a, &name, NULL, 0);
1348 strbuf_addstr(&name, " => ");
1349 quote_c_style(b, &name, NULL, 0);
1350 return strbuf_detach(&name, NULL);
1351 }
1352 /* Find common prefix */
1354 pfx_length = 0;
1355 while (*old && *new && *old == *new) {
1356 if (*old == '/')
1357 pfx_length = old - a + 1;
1358 old++;
1359 new++;
1360 }
1361 /* Find common suffix */
1363 old = a + len_a;
1364 new = b + len_b;
1365 sfx_length = 0;
1366 /*
1367 * If there is a common prefix, it must end in a slash. In
1368 * that case we let this loop run 1 into the prefix to see the
1369 * same slash.
1370 *
1371 * If there is no common prefix, we cannot do this as it would
1372 * underrun the input strings.
1373 */
1374 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1375 while (a + pfx_length - pfx_adjust_for_slash <= old &&
1376 b + pfx_length - pfx_adjust_for_slash <= new &&
1377 *old == *new) {
1378 if (*old == '/')
1379 sfx_length = len_a - (old - a);
1380 old--;
1381 new--;
1382 }
1383 /*
1385 * pfx{mid-a => mid-b}sfx
1386 * {pfx-a => pfx-b}sfx
1387 * pfx{sfx-a => sfx-b}
1388 * name-a => name-b
1389 */
1390 a_midlen = len_a - pfx_length - sfx_length;
1391 b_midlen = len_b - pfx_length - sfx_length;
1392 if (a_midlen < 0)
1393 a_midlen = 0;
1394 if (b_midlen < 0)
1395 b_midlen = 0;
1396 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1398 if (pfx_length + sfx_length) {
1399 strbuf_add(&name, a, pfx_length);
1400 strbuf_addch(&name, '{');
1401 }
1402 strbuf_add(&name, a + pfx_length, a_midlen);
1403 strbuf_addstr(&name, " => ");
1404 strbuf_add(&name, b + pfx_length, b_midlen);
1405 if (pfx_length + sfx_length) {
1406 strbuf_addch(&name, '}');
1407 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1408 }
1409 return strbuf_detach(&name, NULL);
1410}
1411struct diffstat_t {
1413 int nr;
1414 int alloc;
1415 struct diffstat_file {
1416 char *from_name;
1417 char *name;
1418 char *print_name;
1419 unsigned is_unmerged:1;
1420 unsigned is_binary:1;
1421 unsigned is_renamed:1;
1422 unsigned is_interesting:1;
1423 uintmax_t added, deleted;
1424 } **files;
1425};
1426static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1428 const char *name_a,
1429 const char *name_b)
1430{
1431 struct diffstat_file *x;
1432 x = xcalloc(1, sizeof(*x));
1433 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1434 diffstat->files[diffstat->nr++] = x;
1435 if (name_b) {
1436 x->from_name = xstrdup(name_a);
1437 x->name = xstrdup(name_b);
1438 x->is_renamed = 1;
1439 }
1440 else {
1441 x->from_name = NULL;
1442 x->name = xstrdup(name_a);
1443 }
1444 return x;
1445}
1446static void diffstat_consume(void *priv, char *line, unsigned long len)
1448{
1449 struct diffstat_t *diffstat = priv;
1450 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1451 if (line[0] == '+')
1453 x->added++;
1454 else if (line[0] == '-')
1455 x->deleted++;
1456}
1457const char mime_boundary_leader[] = "------------";
1459static int scale_linear(int it, int width, int max_change)
1461{
1462 if (!it)
1463 return 0;
1464 /*
1465 * make sure that at least one '-' or '+' is printed if
1466 * there is any change to this path. The easiest way is to
1467 * scale linearly as if the alloted width is one column shorter
1468 * than it is, and then add 1 to the result.
1469 */
1470 return 1 + (it * (width - 1) / max_change);
1471}
1472static void show_name(FILE *file,
1474 const char *prefix, const char *name, int len)
1475{
1476 fprintf(file, " %s%-*s |", prefix, len, name);
1477}
1478static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1480{
1481 if (cnt <= 0)
1482 return;
1483 fprintf(file, "%s", set);
1484 while (cnt--)
1485 putc(ch, file);
1486 fprintf(file, "%s", reset);
1487}
1488static void fill_print_name(struct diffstat_file *file)
1490{
1491 char *pname;
1492 if (file->print_name)
1494 return;
1495 if (!file->is_renamed) {
1497 struct strbuf buf = STRBUF_INIT;
1498 if (quote_c_style(file->name, &buf, NULL, 0)) {
1499 pname = strbuf_detach(&buf, NULL);
1500 } else {
1501 pname = file->name;
1502 strbuf_release(&buf);
1503 }
1504 } else {
1505 pname = pprint_rename(file->from_name, file->name);
1506 }
1507 file->print_name = pname;
1508}
1509int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1511{
1512 struct strbuf sb = STRBUF_INIT;
1513 int ret;
1514 if (!files) {
1516 assert(insertions == 0 && deletions == 0);
1517 return fprintf(fp, "%s\n", " 0 files changed");
1518 }
1519 strbuf_addf(&sb,
1521 (files == 1) ? " %d file changed" : " %d files changed",
1522 files);
1523 /*
1525 * For binary diff, the caller may want to print "x files
1526 * changed" with insertions == 0 && deletions == 0.
1527 *
1528 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1529 * is probably less confusing (i.e skip over "2 files changed
1530 * but nothing about added/removed lines? Is this a bug in Git?").
1531 */
1532 if (insertions || deletions == 0) {
1533 strbuf_addf(&sb,
1534 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1535 insertions);
1536 }
1537 if (deletions || insertions == 0) {
1539 strbuf_addf(&sb,
1540 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1541 deletions);
1542 }
1543 strbuf_addch(&sb, '\n');
1544 ret = fputs(sb.buf, fp);
1545 strbuf_release(&sb);
1546 return ret;
1547}
1548static void show_stats(struct diffstat_t *data, struct diff_options *options)
1550{
1551 int i, len, add, del, adds = 0, dels = 0;
1552 uintmax_t max_change = 0, max_len = 0;
1553 int total_files = data->nr, count;
1554 int width, name_width, graph_width, number_width = 0, bin_width = 0;
1555 const char *reset, *add_c, *del_c;
1556 const char *line_prefix = "";
1557 int extra_shown = 0;
1558 if (data->nr == 0)
1560 return;
1561 line_prefix = diff_line_prefix(options);
1563 count = options->stat_count ? options->stat_count : data->nr;
1564 reset = diff_get_color_opt(options, DIFF_RESET);
1566 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1567 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1568 /*
1570 * Find the longest filename and max number of changes
1571 */
1572 for (i = 0; (i < count) && (i < data->nr); i++) {
1573 struct diffstat_file *file = data->files[i];
1574 uintmax_t change = file->added + file->deleted;
1575 if (!file->is_interesting && (change == 0)) {
1577 count++; /* not shown == room for one more */
1578 continue;
1579 }
1580 fill_print_name(file);
1581 len = strlen(file->print_name);
1582 if (max_len < len)
1583 max_len = len;
1584 if (file->is_unmerged) {
1586 /* "Unmerged" is 8 characters */
1587 bin_width = bin_width < 8 ? 8 : bin_width;
1588 continue;
1589 }
1590 if (file->is_binary) {
1591 /* "Bin XXX -> YYY bytes" */
1592 int w = 14 + decimal_width(file->added)
1593 + decimal_width(file->deleted);
1594 bin_width = bin_width < w ? w : bin_width;
1595 /* Display change counts aligned with "Bin" */
1596 number_width = 3;
1597 continue;
1598 }
1599 if (max_change < change)
1601 max_change = change;
1602 }
1603 count = i; /* where we can stop scanning in data->files[] */
1604 /*
1606 * We have width = stat_width or term_columns() columns total.
1607 * We want a maximum of min(max_len, stat_name_width) for the name part.
1608 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1609 * We also need 1 for " " and 4 + decimal_width(max_change)
1610 * for " | NNNN " and one the empty column at the end, altogether
1611 * 6 + decimal_width(max_change).
1612 *
1613 * If there's not enough space, we will use the smaller of
1614 * stat_name_width (if set) and 5/8*width for the filename,
1615 * and the rest for constant elements + graph part, but no more
1616 * than stat_graph_width for the graph part.
1617 * (5/8 gives 50 for filename and 30 for the constant parts + graph
1618 * for the standard terminal size).
1619 *
1620 * In other words: stat_width limits the maximum width, and
1621 * stat_name_width fixes the maximum width of the filename,
1622 * and is also used to divide available columns if there
1623 * aren't enough.
1624 *
1625 * Binary files are displayed with "Bin XXX -> YYY bytes"
1626 * instead of the change count and graph. This part is treated
1627 * similarly to the graph part, except that it is not
1628 * "scaled". If total width is too small to accommodate the
1629 * guaranteed minimum width of the filename part and the
1630 * separators and this message, this message will "overflow"
1631 * making the line longer than the maximum width.
1632 */
1633 if (options->stat_width == -1)
1635 width = term_columns() - strlen(line_prefix);
1636 else
1637 width = options->stat_width ? options->stat_width : 80;
1638 number_width = decimal_width(max_change) > number_width ?
1639 decimal_width(max_change) : number_width;
1640 if (options->stat_graph_width == -1)
1642 options->stat_graph_width = diff_stat_graph_width;
1643 /*
1645 * Guarantee 3/8*16==6 for the graph part
1646 * and 5/8*16==10 for the filename part
1647 */
1648 if (width < 16 + 6 + number_width)
1649 width = 16 + 6 + number_width;
1650 /*
1652 * First assign sizes that are wanted, ignoring available width.
1653 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1654 * starting from "XXX" should fit in graph_width.
1655 */
1656 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1657 if (options->stat_graph_width &&
1658 options->stat_graph_width < graph_width)
1659 graph_width = options->stat_graph_width;
1660 name_width = (options->stat_name_width > 0 &&
1662 options->stat_name_width < max_len) ?
1663 options->stat_name_width : max_len;
1664 /*
1666 * Adjust adjustable widths not to exceed maximum width
1667 */
1668 if (name_width + number_width + 6 + graph_width > width) {
1669 if (graph_width > width * 3/8 - number_width - 6) {
1670 graph_width = width * 3/8 - number_width - 6;
1671 if (graph_width < 6)
1672 graph_width = 6;
1673 }
1674 if (options->stat_graph_width &&
1676 graph_width > options->stat_graph_width)
1677 graph_width = options->stat_graph_width;
1678 if (name_width > width - number_width - 6 - graph_width)
1679 name_width = width - number_width - 6 - graph_width;
1680 else
1681 graph_width = width - number_width - 6 - name_width;
1682 }
1683 /*
1685 * From here name_width is the width of the name area,
1686 * and graph_width is the width of the graph area.
1687 * max_change is used to scale graph properly.
1688 */
1689 for (i = 0; i < count; i++) {
1690 const char *prefix = "";
1691 struct diffstat_file *file = data->files[i];
1692 char *name = file->print_name;
1693 uintmax_t added = file->added;
1694 uintmax_t deleted = file->deleted;
1695 int name_len;
1696 if (!file->is_interesting && (added + deleted == 0))
1698 continue;
1699 /*
1701 * "scale" the filename
1702 */
1703 len = name_width;
1704 name_len = strlen(name);
1705 if (name_width < name_len) {
1706 char *slash;
1707 prefix = "...";
1708 len -= 3;
1709 name += name_len - len;
1710 slash = strchr(name, '/');
1711 if (slash)
1712 name = slash;
1713 }
1714 if (file->is_binary) {
1716 fprintf(options->file, "%s", line_prefix);
1717 show_name(options->file, prefix, name, len);
1718 fprintf(options->file, " %*s", number_width, "Bin");
1719 if (!added && !deleted) {
1720 putc('\n', options->file);
1721 continue;
1722 }
1723 fprintf(options->file, " %s%"PRIuMAX"%s",
1724 del_c, deleted, reset);
1725 fprintf(options->file, " -> ");
1726 fprintf(options->file, "%s%"PRIuMAX"%s",
1727 add_c, added, reset);
1728 fprintf(options->file, " bytes");
1729 fprintf(options->file, "\n");
1730 continue;
1731 }
1732 else if (file->is_unmerged) {
1733 fprintf(options->file, "%s", line_prefix);
1734 show_name(options->file, prefix, name, len);
1735 fprintf(options->file, " Unmerged\n");
1736 continue;
1737 }
1738 /*
1740 * scale the add/delete
1741 */
1742 add = added;
1743 del = deleted;
1744 if (graph_width <= max_change) {
1746 int total = scale_linear(add + del, graph_width, max_change);
1747 if (total < 2 && add && del)
1748 /* width >= 2 due to the sanity check */
1749 total = 2;
1750 if (add < del) {
1751 add = scale_linear(add, graph_width, max_change);
1752 del = total - add;
1753 } else {
1754 del = scale_linear(del, graph_width, max_change);
1755 add = total - del;
1756 }
1757 }
1758 fprintf(options->file, "%s", line_prefix);
1759 show_name(options->file, prefix, name, len);
1760 fprintf(options->file, " %*"PRIuMAX"%s",
1761 number_width, added + deleted,
1762 added + deleted ? " " : "");
1763 show_graph(options->file, '+', add, add_c, reset);
1764 show_graph(options->file, '-', del, del_c, reset);
1765 fprintf(options->file, "\n");
1766 }
1767 for (i = 0; i < data->nr; i++) {
1769 struct diffstat_file *file = data->files[i];
1770 uintmax_t added = file->added;
1771 uintmax_t deleted = file->deleted;
1772 if (file->is_unmerged ||
1774 (!file->is_interesting && (added + deleted == 0))) {
1775 total_files--;
1776 continue;
1777 }
1778 if (!file->is_binary) {
1780 adds += added;
1781 dels += deleted;
1782 }
1783 if (i < count)
1784 continue;
1785 if (!extra_shown)
1786 fprintf(options->file, "%s ...\n", line_prefix);
1787 extra_shown = 1;
1788 }
1789 fprintf(options->file, "%s", line_prefix);
1790 print_stat_summary(options->file, total_files, adds, dels);
1791}
1792static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1794{
1795 int i, adds = 0, dels = 0, total_files = data->nr;
1796 if (data->nr == 0)
1798 return;
1799 for (i = 0; i < data->nr; i++) {
1801 int added = data->files[i]->added;
1802 int deleted= data->files[i]->deleted;
1803 if (data->files[i]->is_unmerged ||
1805 (!data->files[i]->is_interesting && (added + deleted == 0))) {
1806 total_files--;
1807 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1808 adds += added;
1809 dels += deleted;
1810 }
1811 }
1812 fprintf(options->file, "%s", diff_line_prefix(options));
1813 print_stat_summary(options->file, total_files, adds, dels);
1814}
1815static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1817{
1818 int i;
1819 if (data->nr == 0)
1821 return;
1822 for (i = 0; i < data->nr; i++) {
1824 struct diffstat_file *file = data->files[i];
1825 fprintf(options->file, "%s", diff_line_prefix(options));
1827 if (file->is_binary)
1829 fprintf(options->file, "-\t-\t");
1830 else
1831 fprintf(options->file,
1832 "%"PRIuMAX"\t%"PRIuMAX"\t",
1833 file->added, file->deleted);
1834 if (options->line_termination) {
1835 fill_print_name(file);
1836 if (!file->is_renamed)
1837 write_name_quoted(file->name, options->file,
1838 options->line_termination);
1839 else {
1840 fputs(file->print_name, options->file);
1841 putc(options->line_termination, options->file);
1842 }
1843 } else {
1844 if (file->is_renamed) {
1845 putc('\0', options->file);
1846 write_name_quoted(file->from_name, options->file, '\0');
1847 }
1848 write_name_quoted(file->name, options->file, '\0');
1849 }
1850 }
1851}
1852struct dirstat_file {
1854 const char *name;
1855 unsigned long changed;
1856};
1857struct dirstat_dir {
1859 struct dirstat_file *files;
1860 int alloc, nr, permille, cumulative;
1861};
1862static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1864 unsigned long changed, const char *base, int baselen)
1865{
1866 unsigned long this_dir = 0;
1867 unsigned int sources = 0;
1868 const char *line_prefix = diff_line_prefix(opt);
1869 while (dir->nr) {
1871 struct dirstat_file *f = dir->files;
1872 int namelen = strlen(f->name);
1873 unsigned long this;
1874 char *slash;
1875 if (namelen < baselen)
1877 break;
1878 if (memcmp(f->name, base, baselen))
1879 break;
1880 slash = strchr(f->name + baselen, '/');
1881 if (slash) {
1882 int newbaselen = slash + 1 - f->name;
1883 this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1884 sources++;
1885 } else {
1886 this = f->changed;
1887 dir->files++;
1888 dir->nr--;
1889 sources += 2;
1890 }
1891 this_dir += this;
1892 }
1893 /*
1895 * We don't report dirstat's for
1896 * - the top level
1897 * - or cases where everything came from a single directory
1898 * under this directory (sources == 1).
1899 */
1900 if (baselen && sources != 1) {
1901 if (this_dir) {
1902 int permille = this_dir * 1000 / changed;
1903 if (permille >= dir->permille) {
1904 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1905 permille / 10, permille % 10, baselen, base);
1906 if (!dir->cumulative)
1907 return 0;
1908 }
1909 }
1910 }
1911 return this_dir;
1912}
1913static int dirstat_compare(const void *_a, const void *_b)
1915{
1916 const struct dirstat_file *a = _a;
1917 const struct dirstat_file *b = _b;
1918 return strcmp(a->name, b->name);
1919}
1920static void show_dirstat(struct diff_options *options)
1922{
1923 int i;
1924 unsigned long changed;
1925 struct dirstat_dir dir;
1926 struct diff_queue_struct *q = &diff_queued_diff;
1927 dir.files = NULL;
1929 dir.alloc = 0;
1930 dir.nr = 0;
1931 dir.permille = options->dirstat_permille;
1932 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1933 changed = 0;
1935 for (i = 0; i < q->nr; i++) {
1936 struct diff_filepair *p = q->queue[i];
1937 const char *name;
1938 unsigned long copied, added, damage;
1939 int content_changed;
1940 name = p->two->path ? p->two->path : p->one->path;
1942 if (p->one->oid_valid && p->two->oid_valid)
1944 content_changed = oidcmp(&p->one->oid, &p->two->oid);
1945 else
1946 content_changed = 1;
1947 if (!content_changed) {
1949 /*
1950 * The SHA1 has not changed, so pre-/post-content is
1951 * identical. We can therefore skip looking at the
1952 * file contents altogether.
1953 */
1954 damage = 0;
1955 goto found_damage;
1956 }
1957 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1959 /*
1960 * In --dirstat-by-file mode, we don't really need to
1961 * look at the actual file contents at all.
1962 * The fact that the SHA1 changed is enough for us to
1963 * add this file to the list of results
1964 * (with each file contributing equal damage).
1965 */
1966 damage = 1;
1967 goto found_damage;
1968 }
1969 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1971 diff_populate_filespec(p->one, 0);
1972 diff_populate_filespec(p->two, 0);
1973 diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1974 &copied, &added);
1975 diff_free_filespec_data(p->one);
1976 diff_free_filespec_data(p->two);
1977 } else if (DIFF_FILE_VALID(p->one)) {
1978 diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
1979 copied = added = 0;
1980 diff_free_filespec_data(p->one);
1981 } else if (DIFF_FILE_VALID(p->two)) {
1982 diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
1983 copied = 0;
1984 added = p->two->size;
1985 diff_free_filespec_data(p->two);
1986 } else
1987 continue;
1988 /*
1990 * Original minus copied is the removed material,
1991 * added is the new material. They are both damages
1992 * made to the preimage.
1993 * If the resulting damage is zero, we know that
1994 * diffcore_count_changes() considers the two entries to
1995 * be identical, but since content_changed is true, we
1996 * know that there must have been _some_ kind of change,
1997 * so we force all entries to have damage > 0.
1998 */
1999 damage = (p->one->size - copied) + added;
2000 if (!damage)
2001 damage = 1;
2002found_damage:
2004 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2005 dir.files[dir.nr].name = name;
2006 dir.files[dir.nr].changed = damage;
2007 changed += damage;
2008 dir.nr++;
2009 }
2010 /* This can happen even with many files, if everything was renames */
2012 if (!changed)
2013 return;
2014 /* Show all directories with more than x% of the changes */
2016 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2017 gather_dirstat(options, &dir, changed, "", 0);
2018}
2019static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2021{
2022 int i;
2023 unsigned long changed;
2024 struct dirstat_dir dir;
2025 if (data->nr == 0)
2027 return;
2028 dir.files = NULL;
2030 dir.alloc = 0;
2031 dir.nr = 0;
2032 dir.permille = options->dirstat_permille;
2033 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2034 changed = 0;
2036 for (i = 0; i < data->nr; i++) {
2037 struct diffstat_file *file = data->files[i];
2038 unsigned long damage = file->added + file->deleted;
2039 if (file->is_binary)
2040 /*
2041 * binary files counts bytes, not lines. Must find some
2042 * way to normalize binary bytes vs. textual lines.
2043 * The following heuristic assumes that there are 64
2044 * bytes per "line".
2045 * This is stupid and ugly, but very cheap...
2046 */
2047 damage = (damage + 63) / 64;
2048 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2049 dir.files[dir.nr].name = file->name;
2050 dir.files[dir.nr].changed = damage;
2051 changed += damage;
2052 dir.nr++;
2053 }
2054 /* This can happen even with many files, if everything was renames */
2056 if (!changed)
2057 return;
2058 /* Show all directories with more than x% of the changes */
2060 qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2061 gather_dirstat(options, &dir, changed, "", 0);
2062}
2063static void free_diffstat_info(struct diffstat_t *diffstat)
2065{
2066 int i;
2067 for (i = 0; i < diffstat->nr; i++) {
2068 struct diffstat_file *f = diffstat->files[i];
2069 if (f->name != f->print_name)
2070 free(f->print_name);
2071 free(f->name);
2072 free(f->from_name);
2073 free(f);
2074 }
2075 free(diffstat->files);
2076}
2077struct checkdiff_t {
2079 const char *filename;
2080 int lineno;
2081 int conflict_marker_size;
2082 struct diff_options *o;
2083 unsigned ws_rule;
2084 unsigned status;
2085};
2086static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2088{
2089 char firstchar;
2090 int cnt;
2091 if (len < marker_size + 1)
2093 return 0;
2094 firstchar = line[0];
2095 switch (firstchar) {
2096 case '=': case '>': case '<': case '|':
2097 break;
2098 default:
2099 return 0;
2100 }
2101 for (cnt = 1; cnt < marker_size; cnt++)
2102 if (line[cnt] != firstchar)
2103 return 0;
2104 /* line[1] thru line[marker_size-1] are same as firstchar */
2105 if (len < marker_size + 1 || !isspace(line[marker_size]))
2106 return 0;
2107 return 1;
2108}
2109static void checkdiff_consume(void *priv, char *line, unsigned long len)
2111{
2112 struct checkdiff_t *data = priv;
2113 int marker_size = data->conflict_marker_size;
2114 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2115 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2116 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2117 char *err;
2118 const char *line_prefix;
2119 assert(data->o);
2121 line_prefix = diff_line_prefix(data->o);
2122 if (line[0] == '+') {
2124 unsigned bad;
2125 data->lineno++;
2126 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2127 data->status |= 1;
2128 fprintf(data->o->file,
2129 "%s%s:%d: leftover conflict marker\n",
2130 line_prefix, data->filename, data->lineno);
2131 }
2132 bad = ws_check(line + 1, len - 1, data->ws_rule);
2133 if (!bad)
2134 return;
2135 data->status |= bad;
2136 err = whitespace_error_string(bad);
2137 fprintf(data->o->file, "%s%s:%d: %s.\n",
2138 line_prefix, data->filename, data->lineno, err);
2139 free(err);
2140 emit_line(data->o, set, reset, line, 1);
2141 ws_check_emit(line + 1, len - 1, data->ws_rule,
2142 data->o->file, set, reset, ws);
2143 } else if (line[0] == ' ') {
2144 data->lineno++;
2145 } else if (line[0] == '@') {
2146 char *plus = strchr(line, '+');
2147 if (plus)
2148 data->lineno = strtol(plus, NULL, 10) - 1;
2149 else
2150 die("invalid diff");
2151 }
2152}
2153static unsigned char *deflate_it(char *data,
2155 unsigned long size,
2156 unsigned long *result_size)
2157{
2158 int bound;
2159 unsigned char *deflated;
2160 git_zstream stream;
2161 git_deflate_init(&stream, zlib_compression_level);
2163 bound = git_deflate_bound(&stream, size);
2164 deflated = xmalloc(bound);
2165 stream.next_out = deflated;
2166 stream.avail_out = bound;
2167 stream.next_in = (unsigned char *)data;
2169 stream.avail_in = size;
2170 while (git_deflate(&stream, Z_FINISH) == Z_OK)
2171 ; /* nothing */
2172 git_deflate_end(&stream);
2173 *result_size = stream.total_out;
2174 return deflated;
2175}
2176static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2178 const char *prefix)
2179{
2180 void *cp;
2181 void *delta;
2182 void *deflated;
2183 void *data;
2184 unsigned long orig_size;
2185 unsigned long delta_size;
2186 unsigned long deflate_size;
2187 unsigned long data_size;
2188 /* We could do deflated delta, or we could do just deflated two,
2190 * whichever is smaller.
2191 */
2192 delta = NULL;
2193 deflated = deflate_it(two->ptr, two->size, &deflate_size);
2194 if (one->size && two->size) {
2195 delta = diff_delta(one->ptr, one->size,
2196 two->ptr, two->size,
2197 &delta_size, deflate_size);
2198 if (delta) {
2199 void *to_free = delta;
2200 orig_size = delta_size;
2201 delta = deflate_it(delta, delta_size, &delta_size);
2202 free(to_free);
2203 }
2204 }
2205 if (delta && delta_size < deflate_size) {
2207 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2208 free(deflated);
2209 data = delta;
2210 data_size = delta_size;
2211 }
2212 else {
2213 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2214 free(delta);
2215 data = deflated;
2216 data_size = deflate_size;
2217 }
2218 /* emit data encoded in base85 */
2220 cp = data;
2221 while (data_size) {
2222 int bytes = (52 < data_size) ? 52 : data_size;
2223 char line[70];
2224 data_size -= bytes;
2225 if (bytes <= 26)
2226 line[0] = bytes + 'A' - 1;
2227 else
2228 line[0] = bytes - 26 + 'a' - 1;
2229 encode_85(line + 1, cp, bytes);
2230 cp = (char *) cp + bytes;
2231 fprintf(file, "%s", prefix);
2232 fputs(line, file);
2233 fputc('\n', file);
2234 }
2235 fprintf(file, "%s\n", prefix);
2236 free(data);
2237}
2238static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2240 const char *prefix)
2241{
2242 fprintf(file, "%sGIT binary patch\n", prefix);
2243 emit_binary_diff_body(file, one, two, prefix);
2244 emit_binary_diff_body(file, two, one, prefix);
2245}
2246int diff_filespec_is_binary(struct diff_filespec *one)
2248{
2249 if (one->is_binary == -1) {
2250 diff_filespec_load_driver(one);
2251 if (one->driver->binary != -1)
2252 one->is_binary = one->driver->binary;
2253 else {
2254 if (!one->data && DIFF_FILE_VALID(one))
2255 diff_populate_filespec(one, CHECK_BINARY);
2256 if (one->is_binary == -1 && one->data)
2257 one->is_binary = buffer_is_binary(one->data,
2258 one->size);
2259 if (one->is_binary == -1)
2260 one->is_binary = 0;
2261 }
2262 }
2263 return one->is_binary;
2264}
2265static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2267{
2268 diff_filespec_load_driver(one);
2269 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2270}
2271void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2273{
2274 if (!options->a_prefix)
2275 options->a_prefix = a;
2276 if (!options->b_prefix)
2277 options->b_prefix = b;
2278}
2279struct userdiff_driver *get_textconv(struct diff_filespec *one)
2281{
2282 if (!DIFF_FILE_VALID(one))
2283 return NULL;
2284 diff_filespec_load_driver(one);
2286 return userdiff_get_textconv(one->driver);
2287}
2288static void builtin_diff(const char *name_a,
2290 const char *name_b,
2291 struct diff_filespec *one,
2292 struct diff_filespec *two,
2293 const char *xfrm_msg,
2294 int must_show_header,
2295 struct diff_options *o,
2296 int complete_rewrite)
2297{
2298 mmfile_t mf1, mf2;
2299 const char *lbl[2];
2300 char *a_one, *b_two;
2301 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2302 const char *reset = diff_get_color_opt(o, DIFF_RESET);
2303 const char *a_prefix, *b_prefix;
2304 struct userdiff_driver *textconv_one = NULL;
2305 struct userdiff_driver *textconv_two = NULL;
2306 struct strbuf header = STRBUF_INIT;
2307 const char *line_prefix = diff_line_prefix(o);
2308 diff_set_mnemonic_prefix(o, "a/", "b/");
2310 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2311 a_prefix = o->b_prefix;
2312 b_prefix = o->a_prefix;
2313 } else {
2314 a_prefix = o->a_prefix;
2315 b_prefix = o->b_prefix;
2316 }
2317 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
2319 (!one->mode || S_ISGITLINK(one->mode)) &&
2320 (!two->mode || S_ISGITLINK(two->mode))) {
2321 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2322 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2323 show_submodule_summary(o->file, one->path ? one->path : two->path,
2324 line_prefix,
2325 &one->oid, &two->oid,
2326 two->dirty_submodule,
2327 meta, del, add, reset);
2328 return;
2329 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
2330 (!one->mode || S_ISGITLINK(one->mode)) &&
2331 (!two->mode || S_ISGITLINK(two->mode))) {
2332 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2333 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2334 show_submodule_inline_diff(o->file, one->path ? one->path : two->path,
2335 line_prefix,
2336 &one->oid, &two->oid,
2337 two->dirty_submodule,
2338 meta, del, add, reset, o);
2339 return;
2340 }
2341 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2343 textconv_one = get_textconv(one);
2344 textconv_two = get_textconv(two);
2345 }
2346 /* Never use a non-valid filename anywhere if at all possible */
2348 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2349 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2350 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2352 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2353 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2354 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2355 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2356 if (lbl[0][0] == '/') {
2357 /* /dev/null */
2358 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2359 if (xfrm_msg)
2360 strbuf_addstr(&header, xfrm_msg);
2361 must_show_header = 1;
2362 }
2363 else if (lbl[1][0] == '/') {
2364 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2365 if (xfrm_msg)
2366 strbuf_addstr(&header, xfrm_msg);
2367 must_show_header = 1;
2368 }
2369 else {
2370 if (one->mode != two->mode) {
2371 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2372 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2373 must_show_header = 1;
2374 }
2375 if (xfrm_msg)
2376 strbuf_addstr(&header, xfrm_msg);
2377 /*
2379 * we do not run diff between different kind
2380 * of objects.
2381 */
2382 if ((one->mode ^ two->mode) & S_IFMT)
2383 goto free_ab_and_return;
2384 if (complete_rewrite &&
2385 (textconv_one || !diff_filespec_is_binary(one)) &&
2386 (textconv_two || !diff_filespec_is_binary(two))) {
2387 fprintf(o->file, "%s", header.buf);
2388 strbuf_reset(&header);
2389 emit_rewrite_diff(name_a, name_b, one, two,
2390 textconv_one, textconv_two, o);
2391 o->found_changes = 1;
2392 goto free_ab_and_return;
2393 }
2394 }
2395 if (o->irreversible_delete && lbl[1][0] == '/') {
2397 fprintf(o->file, "%s", header.buf);
2398 strbuf_reset(&header);
2399 goto free_ab_and_return;
2400 } else if (!DIFF_OPT_TST(o, TEXT) &&
2401 ( (!textconv_one && diff_filespec_is_binary(one)) ||
2402 (!textconv_two && diff_filespec_is_binary(two)) )) {
2403 if (!one->data && !two->data &&
2404 S_ISREG(one->mode) && S_ISREG(two->mode) &&
2405 !DIFF_OPT_TST(o, BINARY)) {
2406 if (!oidcmp(&one->oid, &two->oid)) {
2407 if (must_show_header)
2408 fprintf(o->file, "%s", header.buf);
2409 goto free_ab_and_return;
2410 }
2411 fprintf(o->file, "%s", header.buf);
2412 fprintf(o->file, "%sBinary files %s and %s differ\n",
2413 line_prefix, lbl[0], lbl[1]);
2414 goto free_ab_and_return;
2415 }
2416 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2417 die("unable to read files to diff");
2418 /* Quite common confusing case */
2419 if (mf1.size == mf2.size &&
2420 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2421 if (must_show_header)
2422 fprintf(o->file, "%s", header.buf);
2423 goto free_ab_and_return;
2424 }
2425 fprintf(o->file, "%s", header.buf);
2426 strbuf_reset(&header);
2427 if (DIFF_OPT_TST(o, BINARY))
2428 emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2429 else
2430 fprintf(o->file, "%sBinary files %s and %s differ\n",
2431 line_prefix, lbl[0], lbl[1]);
2432 o->found_changes = 1;
2433 } else {
2434 /* Crazy xdl interfaces.. */
2435 const char *diffopts = getenv("GIT_DIFF_OPTS");
2436 const char *v;
2437 xpparam_t xpp;
2438 xdemitconf_t xecfg;
2439 struct emit_callback ecbdata;
2440 const struct userdiff_funcname *pe;
2441 if (must_show_header) {
2443 fprintf(o->file, "%s", header.buf);
2444 strbuf_reset(&header);
2445 }
2446 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2448 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2449 pe = diff_funcname_pattern(one);
2451 if (!pe)
2452 pe = diff_funcname_pattern(two);
2453 memset(&xpp, 0, sizeof(xpp));
2455 memset(&xecfg, 0, sizeof(xecfg));
2456 memset(&ecbdata, 0, sizeof(ecbdata));
2457 ecbdata.label_path = lbl;
2458 ecbdata.color_diff = want_color(o->use_color);
2459 ecbdata.ws_rule = whitespace_rule(name_b);
2460 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2461 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2462 ecbdata.opt = o;
2463 ecbdata.header = header.len ? &header : NULL;
2464 xpp.flags = o->xdl_opts;
2465 xecfg.ctxlen = o->context;
2466 xecfg.interhunkctxlen = o->interhunkcontext;
2467 xecfg.flags = XDL_EMIT_FUNCNAMES;
2468 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2469 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2470 if (pe)
2471 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2472 if (!diffopts)
2473 ;
2474 else if (skip_prefix(diffopts, "--unified=", &v))
2475 xecfg.ctxlen = strtoul(v, NULL, 10);
2476 else if (skip_prefix(diffopts, "-u", &v))
2477 xecfg.ctxlen = strtoul(v, NULL, 10);
2478 if (o->word_diff)
2479 init_diff_words_data(&ecbdata, o, one, two);
2480 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2481 &xpp, &xecfg))
2482 die("unable to generate diff for %s", one->path);
2483 if (o->word_diff)
2484 free_diff_words_data(&ecbdata);
2485 if (textconv_one)
2486 free(mf1.ptr);
2487 if (textconv_two)
2488 free(mf2.ptr);
2489 xdiff_clear_find_func(&xecfg);
2490 }
2491 free_ab_and_return:
2493 strbuf_release(&header);
2494 diff_free_filespec_data(one);
2495 diff_free_filespec_data(two);
2496 free(a_one);
2497 free(b_two);
2498 return;
2499}
2500static void builtin_diffstat(const char *name_a, const char *name_b,
2502 struct diff_filespec *one,
2503 struct diff_filespec *two,
2504 struct diffstat_t *diffstat,
2505 struct diff_options *o,
2506 struct diff_filepair *p)
2507{
2508 mmfile_t mf1, mf2;
2509 struct diffstat_file *data;
2510 int same_contents;
2511 int complete_rewrite = 0;
2512 if (!DIFF_PAIR_UNMERGED(p)) {
2514 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2515 complete_rewrite = 1;
2516 }
2517 data = diffstat_add(diffstat, name_a, name_b);
2519 data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2520 if (!one || !two) {
2522 data->is_unmerged = 1;
2523 return;
2524 }
2525 same_contents = !oidcmp(&one->oid, &two->oid);
2527 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2529 data->is_binary = 1;
2530 if (same_contents) {
2531 data->added = 0;
2532 data->deleted = 0;
2533 } else {
2534 data->added = diff_filespec_size(two);
2535 data->deleted = diff_filespec_size(one);
2536 }
2537 }
2538 else if (complete_rewrite) {
2540 diff_populate_filespec(one, 0);
2541 diff_populate_filespec(two, 0);
2542 data->deleted = count_lines(one->data, one->size);
2543 data->added = count_lines(two->data, two->size);
2544 }
2545 else if (!same_contents) {
2547 /* Crazy xdl interfaces.. */
2548 xpparam_t xpp;
2549 xdemitconf_t xecfg;
2550 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2552 die("unable to read files to diff");
2553 memset(&xpp, 0, sizeof(xpp));
2555 memset(&xecfg, 0, sizeof(xecfg));
2556 xpp.flags = o->xdl_opts;
2557 xecfg.ctxlen = o->context;
2558 xecfg.interhunkctxlen = o->interhunkcontext;
2559 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2560 &xpp, &xecfg))
2561 die("unable to generate diffstat for %s", one->path);
2562 }
2563 diff_free_filespec_data(one);
2565 diff_free_filespec_data(two);
2566}
2567static void builtin_checkdiff(const char *name_a, const char *name_b,
2569 const char *attr_path,
2570 struct diff_filespec *one,
2571 struct diff_filespec *two,
2572 struct diff_options *o)
2573{
2574 mmfile_t mf1, mf2;
2575 struct checkdiff_t data;
2576 if (!two)
2578 return;
2579 memset(&data, 0, sizeof(data));
2581 data.filename = name_b ? name_b : name_a;
2582 data.lineno = 0;
2583 data.o = o;
2584 data.ws_rule = whitespace_rule(attr_path);
2585 data.conflict_marker_size = ll_merge_marker_size(attr_path);
2586 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2588 die("unable to read files to diff");
2589 /*
2591 * All the other codepaths check both sides, but not checking
2592 * the "old" side here is deliberate. We are checking the newly
2593 * introduced changes, and as long as the "new" side is text, we
2594 * can and should check what it introduces.
2595 */
2596 if (diff_filespec_is_binary(two))
2597 goto free_and_return;
2598 else {
2599 /* Crazy xdl interfaces.. */
2600 xpparam_t xpp;
2601 xdemitconf_t xecfg;
2602 memset(&xpp, 0, sizeof(xpp));
2604 memset(&xecfg, 0, sizeof(xecfg));
2605 xecfg.ctxlen = 1; /* at least one context line */
2606 xpp.flags = 0;
2607 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2608 &xpp, &xecfg))
2609 die("unable to generate checkdiff for %s", one->path);
2610 if (data.ws_rule & WS_BLANK_AT_EOF) {
2612 struct emit_callback ecbdata;
2613 int blank_at_eof;
2614 ecbdata.ws_rule = data.ws_rule;
2616 check_blank_at_eof(&mf1, &mf2, &ecbdata);
2617 blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2618 if (blank_at_eof) {
2620 static char *err;
2621 if (!err)
2622 err = whitespace_error_string(WS_BLANK_AT_EOF);
2623 fprintf(o->file, "%s:%d: %s.\n",
2624 data.filename, blank_at_eof, err);
2625 data.status = 1; /* report errors */
2626 }
2627 }
2628 }
2629 free_and_return:
2630 diff_free_filespec_data(one);
2631 diff_free_filespec_data(two);
2632 if (data.status)
2633 DIFF_OPT_SET(o, CHECK_FAILED);
2634}
2635struct diff_filespec *alloc_filespec(const char *path)
2637{
2638 struct diff_filespec *spec;
2639 FLEXPTR_ALLOC_STR(spec, path, path);
2641 spec->count = 1;
2642 spec->is_binary = -1;
2643 return spec;
2644}
2645void free_filespec(struct diff_filespec *spec)
2647{
2648 if (!--spec->count) {
2649 diff_free_filespec_data(spec);
2650 free(spec);
2651 }
2652}
2653void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2655 int sha1_valid, unsigned short mode)
2656{
2657 if (mode) {
2658 spec->mode = canon_mode(mode);
2659 hashcpy(spec->oid.hash, sha1);
2660 spec->oid_valid = sha1_valid;
2661 }
2662}
2663/*
2665 * Given a name and sha1 pair, if the index tells us the file in
2666 * the work tree has that object contents, return true, so that
2667 * prepare_temp_file() does not have to inflate and extract.
2668 */
2669static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2670{
2671 const struct cache_entry *ce;
2672 struct stat st;
2673 int pos, len;
2674 /*
2676 * We do not read the cache ourselves here, because the
2677 * benchmark with my previous version that always reads cache
2678 * shows that it makes things worse for diff-tree comparing
2679 * two linux-2.6 kernel trees in an already checked out work
2680 * tree. This is because most diff-tree comparisons deal with
2681 * only a small number of files, while reading the cache is
2682 * expensive for a large project, and its cost outweighs the
2683 * savings we get by not inflating the object to a temporary
2684 * file. Practically, this code only helps when we are used
2685 * by diff-cache --cached, which does read the cache before
2686 * calling us.
2687 */
2688 if (!active_cache)
2689 return 0;
2690 /* We want to avoid the working directory if our caller
2692 * doesn't need the data in a normal file, this system
2693 * is rather slow with its stat/open/mmap/close syscalls,
2694 * and the object is contained in a pack file. The pack
2695 * is probably already open and will be faster to obtain
2696 * the data through than the working directory. Loose
2697 * objects however would tend to be slower as they need
2698 * to be individually opened and inflated.
2699 */
2700 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2701 return 0;
2702 /*
2704 * Similarly, if we'd have to convert the file contents anyway, that
2705 * makes the optimization not worthwhile.
2706 */
2707 if (!want_file && would_convert_to_git(name))
2708 return 0;
2709 len = strlen(name);
2711 pos = cache_name_pos(name, len);
2712 if (pos < 0)
2713 return 0;
2714 ce = active_cache[pos];
2715 /*
2717 * This is not the sha1 we are looking for, or
2718 * unreusable because it is not a regular file.
2719 */
2720 if (hashcmp(sha1, ce->oid.hash) || !S_ISREG(ce->ce_mode))
2721 return 0;
2722 /*
2724 * If ce is marked as "assume unchanged", there is no
2725 * guarantee that work tree matches what we are looking for.
2726 */
2727 if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2728 return 0;
2729 /*
2731 * If ce matches the file in the work tree, we can reuse it.
2732 */
2733 if (ce_uptodate(ce) ||
2734 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2735 return 1;
2736 return 0;
2738}
2739static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2741{
2742 struct strbuf buf = STRBUF_INIT;
2743 char *dirty = "";
2744 /* Are we looking at the work tree? */
2746 if (s->dirty_submodule)
2747 dirty = "-dirty";
2748 strbuf_addf(&buf, "Subproject commit %s%s\n",
2750 oid_to_hex(&s->oid), dirty);
2751 s->size = buf.len;
2752 if (size_only) {
2753 s->data = NULL;
2754 strbuf_release(&buf);
2755 } else {
2756 s->data = strbuf_detach(&buf, NULL);
2757 s->should_free = 1;
2758 }
2759 return 0;
2760}
2761/*
2763 * While doing rename detection and pickaxe operation, we may need to
2764 * grab the data for the blob (or file) for our own in-core comparison.
2765 * diff_filespec has data and size fields for this purpose.
2766 */
2767int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2768{
2769 int size_only = flags & CHECK_SIZE_ONLY;
2770 int err = 0;
2771 /*
2772 * demote FAIL to WARN to allow inspecting the situation
2773 * instead of refusing.
2774 */
2775 enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2776 ? SAFE_CRLF_WARN
2777 : safe_crlf);
2778 if (!DIFF_FILE_VALID(s))
2780 die("internal error: asking to populate invalid file.");
2781 if (S_ISDIR(s->mode))
2782 return -1;
2783 if (s->data)
2785 return 0;
2786 if (size_only && 0 < s->size)
2788 return 0;
2789 if (S_ISGITLINK(s->mode))
2791 return diff_populate_gitlink(s, size_only);
2792 if (!s->oid_valid ||
2794 reuse_worktree_file(s->path, s->oid.hash, 0)) {
2795 struct strbuf buf = STRBUF_INIT;
2796 struct stat st;
2797 int fd;
2798 if (lstat(s->path, &st) < 0) {
2800 if (errno == ENOENT) {
2801 err_empty:
2802 err = -1;
2803 empty:
2804 s->data = (char *)"";
2805 s->size = 0;
2806 return err;
2807 }
2808 }
2809 s->size = xsize_t(st.st_size);
2810 if (!s->size)
2811 goto empty;
2812 if (S_ISLNK(st.st_mode)) {
2813 struct strbuf sb = STRBUF_INIT;
2814 if (strbuf_readlink(&sb, s->path, s->size))
2816 goto err_empty;
2817 s->size = sb.len;
2818 s->data = strbuf_detach(&sb, NULL);
2819 s->should_free = 1;
2820 return 0;
2821 }
2822 if (size_only)
2823 return 0;
2824 if ((flags & CHECK_BINARY) &&
2825 s->size > big_file_threshold && s->is_binary == -1) {
2826 s->is_binary = 1;
2827 return 0;
2828 }
2829 fd = open(s->path, O_RDONLY);
2830 if (fd < 0)
2831 goto err_empty;
2832 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2833 close(fd);
2834 s->should_munmap = 1;
2835 /*
2837 * Convert from working tree format to canonical git format
2838 */
2839 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2840 size_t size = 0;
2841 munmap(s->data, s->size);
2842 s->should_munmap = 0;
2843 s->data = strbuf_detach(&buf, &size);
2844 s->size = size;
2845 s->should_free = 1;
2846 }
2847 }
2848 else {
2849 enum object_type type;
2850 if (size_only || (flags & CHECK_BINARY)) {
2851 type = sha1_object_info(s->oid.hash, &s->size);
2852 if (type < 0)
2853 die("unable to read %s",
2854 oid_to_hex(&s->oid));
2855 if (size_only)
2856 return 0;
2857 if (s->size > big_file_threshold && s->is_binary == -1) {
2858 s->is_binary = 1;
2859 return 0;
2860 }
2861 }
2862 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
2863 if (!s->data)
2864 die("unable to read %s", oid_to_hex(&s->oid));
2865 s->should_free = 1;
2866 }
2867 return 0;
2868}
2869void diff_free_filespec_blob(struct diff_filespec *s)
2871{
2872 if (s->should_free)
2873 free(s->data);
2874 else if (s->should_munmap)
2875 munmap(s->data, s->size);
2876 if (s->should_free || s->should_munmap) {
2878 s->should_free = s->should_munmap = 0;
2879 s->data = NULL;
2880 }
2881}
2882void diff_free_filespec_data(struct diff_filespec *s)
2884{
2885 diff_free_filespec_blob(s);
2886 free(s->cnt_data);
2887 s->cnt_data = NULL;
2888}
2889static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2891 void *blob,
2892 unsigned long size,
2893 const struct object_id *oid,
2894 int mode)
2895{
2896 int fd;
2897 struct strbuf buf = STRBUF_INIT;
2898 struct strbuf template = STRBUF_INIT;
2899 char *path_dup = xstrdup(path);
2900 const char *base = basename(path_dup);
2901 /* Generate "XXXXXX_basename.ext" */
2903 strbuf_addstr(&template, "XXXXXX_");
2904 strbuf_addstr(&template, base);
2905 fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2907 if (fd < 0)
2908 die_errno("unable to create temp-file");
2909 if (convert_to_working_tree(path,
2910 (const char *)blob, (size_t)size, &buf)) {
2911 blob = buf.buf;
2912 size = buf.len;
2913 }
2914 if (write_in_full(fd, blob, size) != size)
2915 die_errno("unable to write temp-file");
2916 close_tempfile(&temp->tempfile);
2917 temp->name = get_tempfile_path(&temp->tempfile);
2918 oid_to_hex_r(temp->hex, oid);
2919 xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
2920 strbuf_release(&buf);
2921 strbuf_release(&template);
2922 free(path_dup);
2923}
2924static struct diff_tempfile *prepare_temp_file(const char *name,
2926 struct diff_filespec *one)
2927{
2928 struct diff_tempfile *temp = claim_diff_tempfile();
2929 if (!DIFF_FILE_VALID(one)) {
2931 not_a_valid_file:
2932 /* A '-' entry produces this for file-2, and
2933 * a '+' entry produces this for file-1.
2934 */
2935 temp->name = "/dev/null";
2936 xsnprintf(temp->hex, sizeof(temp->hex), ".");
2937 xsnprintf(temp->mode, sizeof(temp->mode), ".");
2938 return temp;
2939 }
2940 if (!S_ISGITLINK(one->mode) &&
2942 (!one->oid_valid ||
2943 reuse_worktree_file(name, one->oid.hash, 1))) {
2944 struct stat st;
2945 if (lstat(name, &st) < 0) {
2946 if (errno == ENOENT)
2947 goto not_a_valid_file;
2948 die_errno("stat(%s)", name);
2949 }
2950 if (S_ISLNK(st.st_mode)) {
2951 struct strbuf sb = STRBUF_INIT;
2952 if (strbuf_readlink(&sb, name, st.st_size) < 0)
2953 die_errno("readlink(%s)", name);
2954 prep_temp_blob(name, temp, sb.buf, sb.len,
2955 (one->oid_valid ?
2956 &one->oid : &null_oid),
2957 (one->oid_valid ?
2958 one->mode : S_IFLNK));
2959 strbuf_release(&sb);
2960 }
2961 else {
2962 /* we can borrow from the file in the work tree */
2963 temp->name = name;
2964 if (!one->oid_valid)
2965 sha1_to_hex_r(temp->hex, null_sha1);
2966 else
2967 sha1_to_hex_r(temp->hex, one->oid.hash);
2968 /* Even though we may sometimes borrow the
2969 * contents from the work tree, we always want
2970 * one->mode. mode is trustworthy even when
2971 * !(one->sha1_valid), as long as
2972 * DIFF_FILE_VALID(one).
2973 */
2974 xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
2975 }
2976 return temp;
2977 }
2978 else {
2979 if (diff_populate_filespec(one, 0))
2980 die("cannot read data blob for %s", one->path);
2981 prep_temp_blob(name, temp, one->data, one->size,
2982 &one->oid, one->mode);
2983 }
2984 return temp;
2985}
2986static void add_external_diff_name(struct argv_array *argv,
2988 const char *name,
2989 struct diff_filespec *df)
2990{
2991 struct diff_tempfile *temp = prepare_temp_file(name, df);
2992 argv_array_push(argv, temp->name);
2993 argv_array_push(argv, temp->hex);
2994 argv_array_push(argv, temp->mode);
2995}
2996/* An external diff command takes:
2998 *
2999 * diff-cmd name infile1 infile1-sha1 infile1-mode \
3000 * infile2 infile2-sha1 infile2-mode [ rename-to ]
3001 *
3002 */
3003static void run_external_diff(const char *pgm,
3004 const char *name,
3005 const char *other,
3006 struct diff_filespec *one,
3007 struct diff_filespec *two,
3008 const char *xfrm_msg,
3009 int complete_rewrite,
3010 struct diff_options *o)
3011{
3012 struct argv_array argv = ARGV_ARRAY_INIT;
3013 struct argv_array env = ARGV_ARRAY_INIT;
3014 struct diff_queue_struct *q = &diff_queued_diff;
3015 argv_array_push(&argv, pgm);
3017 argv_array_push(&argv, name);
3018 if (one && two) {
3020 add_external_diff_name(&argv, name, one);
3021 if (!other)
3022 add_external_diff_name(&argv, name, two);
3023 else {
3024 add_external_diff_name(&argv, other, two);
3025 argv_array_push(&argv, other);
3026 argv_array_push(&argv, xfrm_msg);
3027 }
3028 }
3029 argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3031 argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3032 if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3034 die(_("external diff died, stopping at %s"), name);
3035 remove_tempfile();
3037 argv_array_clear(&argv);
3038 argv_array_clear(&env);
3039}
3040static int similarity_index(struct diff_filepair *p)
3042{
3043 return p->score * 100 / MAX_SCORE;
3044}
3045static void fill_metainfo(struct strbuf *msg,
3047 const char *name,
3048 const char *other,
3049 struct diff_filespec *one,
3050 struct diff_filespec *two,
3051 struct diff_options *o,
3052 struct diff_filepair *p,
3053 int *must_show_header,
3054 int use_color)
3055{
3056 const char *set = diff_get_color(use_color, DIFF_METAINFO);
3057 const char *reset = diff_get_color(use_color, DIFF_RESET);
3058 const char *line_prefix = diff_line_prefix(o);
3059 *must_show_header = 1;
3061 strbuf_init(msg, PATH_MAX * 2 + 300);
3062 switch (p->status) {
3063 case DIFF_STATUS_COPIED:
3064 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3065 line_prefix, set, similarity_index(p));
3066 strbuf_addf(msg, "%s\n%s%scopy from ",
3067 reset, line_prefix, set);
3068 quote_c_style(name, msg, NULL, 0);
3069 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3070 quote_c_style(other, msg, NULL, 0);
3071 strbuf_addf(msg, "%s\n", reset);
3072 break;
3073 case DIFF_STATUS_RENAMED:
3074 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3075 line_prefix, set, similarity_index(p));
3076 strbuf_addf(msg, "%s\n%s%srename from ",
3077 reset, line_prefix, set);
3078 quote_c_style(name, msg, NULL, 0);
3079 strbuf_addf(msg, "%s\n%s%srename to ",
3080 reset, line_prefix, set);
3081 quote_c_style(other, msg, NULL, 0);
3082 strbuf_addf(msg, "%s\n", reset);
3083 break;
3084 case DIFF_STATUS_MODIFIED:
3085 if (p->score) {
3086 strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3087 line_prefix,
3088 set, similarity_index(p), reset);
3089 break;
3090 }
3091 /* fallthru */
3092 default:
3093 *must_show_header = 0;
3094 }
3095 if (one && two && oidcmp(&one->oid, &two->oid)) {
3096 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3097 if (DIFF_OPT_TST(o, BINARY)) {
3099 mmfile_t mf;
3100 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3101 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3102 abbrev = 40;
3103 }
3104 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3105 find_unique_abbrev(one->oid.hash, abbrev));
3106 strbuf_addstr(msg, find_unique_abbrev(two->oid.hash, abbrev));
3107 if (one->mode == two->mode)
3108 strbuf_addf(msg, " %06o", one->mode);
3109 strbuf_addf(msg, "%s\n", reset);
3110 }
3111}
3112static void run_diff_cmd(const char *pgm,
3114 const char *name,
3115 const char *other,
3116 const char *attr_path,
3117 struct diff_filespec *one,
3118 struct diff_filespec *two,
3119 struct strbuf *msg,
3120 struct diff_options *o,
3121 struct diff_filepair *p)
3122{
3123 const char *xfrm_msg = NULL;
3124 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3125 int must_show_header = 0;
3126 if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3129 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3130 if (drv && drv->external)
3131 pgm = drv->external;
3132 }
3133 if (msg) {
3135 /*
3136 * don't use colors when the header is intended for an
3137 * external diff driver
3138 */
3139 fill_metainfo(msg, name, other, one, two, o, p,
3140 &must_show_header,
3141 want_color(o->use_color) && !pgm);
3142 xfrm_msg = msg->len ? msg->buf : NULL;
3143 }
3144 if (pgm) {
3146 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3147 complete_rewrite, o);
3148 return;
3149 }
3150 if (one && two)
3151 builtin_diff(name, other ? other : name,
3152 one, two, xfrm_msg, must_show_header,
3153 o, complete_rewrite);
3154 else
3155 fprintf(o->file, "* Unmerged path %s\n", name);
3156}
3157static void diff_fill_sha1_info(struct diff_filespec *one)
3159{
3160 if (DIFF_FILE_VALID(one)) {
3161 if (!one->oid_valid) {
3162 struct stat st;
3163 if (one->is_stdin) {
3164 oidclr(&one->oid);
3165 return;
3166 }
3167 if (lstat(one->path, &st) < 0)
3168 die_errno("stat '%s'", one->path);
3169 if (index_path(one->oid.hash, one->path, &st, 0))
3170 die("cannot hash %s", one->path);
3171 }
3172 }
3173 else
3174 oidclr(&one->oid);
3175}
3176static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3178{
3179 /* Strip the prefix but do not molest /dev/null and absolute paths */
3180 if (*namep && **namep != '/') {
3181 *namep += prefix_length;
3182 if (**namep == '/')
3183 ++*namep;
3184 }
3185 if (*otherp && **otherp != '/') {
3186 *otherp += prefix_length;
3187 if (**otherp == '/')
3188 ++*otherp;
3189 }
3190}
3191static void run_diff(struct diff_filepair *p, struct diff_options *o)
3193{
3194 const char *pgm = external_diff();
3195 struct strbuf msg;
3196 struct diff_filespec *one = p->one;
3197 struct diff_filespec *two = p->two;
3198 const char *name;
3199 const char *other;
3200 const char *attr_path;
3201 name = p->one->path;
3203 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3204 attr_path = name;
3205 if (o->prefix_length)
3206 strip_prefix(o->prefix_length, &name, &other);
3207 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3209 pgm = NULL;
3210 if (DIFF_PAIR_UNMERGED(p)) {
3212 run_diff_cmd(pgm, name, NULL, attr_path,
3213 NULL, NULL, NULL, o, p);
3214 return;
3215 }
3216 diff_fill_sha1_info(one);
3218 diff_fill_sha1_info(two);
3219 if (!pgm &&
3221 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3222 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3223 /*
3224 * a filepair that changes between file and symlink
3225 * needs to be split into deletion and creation.
3226 */
3227 struct diff_filespec *null = alloc_filespec(two->path);
3228 run_diff_cmd(NULL, name, other, attr_path,
3229 one, null, &msg, o, p);
3230 free(null);
3231 strbuf_release(&msg);
3232 null = alloc_filespec(one->path);
3234 run_diff_cmd(NULL, name, other, attr_path,
3235 null, two, &msg, o, p);
3236 free(null);
3237 }
3238 else
3239 run_diff_cmd(pgm, name, other, attr_path,
3240 one, two, &msg, o, p);
3241 strbuf_release(&msg);
3243}
3244static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3246 struct diffstat_t *diffstat)
3247{
3248 const char *name;
3249 const char *other;
3250 if (DIFF_PAIR_UNMERGED(p)) {
3252 /* unmerged */
3253 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3254 return;
3255 }
3256 name = p->one->path;
3258 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3259 if (o->prefix_length)
3261 strip_prefix(o->prefix_length, &name, &other);
3262 diff_fill_sha1_info(p->one);
3264 diff_fill_sha1_info(p->two);
3265 builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3267}
3268static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3270{
3271 const char *name;
3272 const char *other;
3273 const char *attr_path;
3274 if (DIFF_PAIR_UNMERGED(p)) {
3276 /* unmerged */
3277 return;
3278 }
3279 name = p->one->path;
3281 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3282 attr_path = other ? other : name;
3283 if (o->prefix_length)
3285 strip_prefix(o->prefix_length, &name, &other);
3286 diff_fill_sha1_info(p->one);
3288 diff_fill_sha1_info(p->two);
3289 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3291}
3292void diff_setup(struct diff_options *options)
3294{
3295 memcpy(options, &default_diff_options, sizeof(*options));
3296 options->file = stdout;
3298 options->line_termination = '\n';
3300 options->break_opt = -1;
3301 options->rename_limit = -1;
3302 options->dirstat_permille = diff_dirstat_permille_default;
3303 options->context = diff_context_default;
3304 options->ws_error_highlight = WSEH_NEW;
3305 DIFF_OPT_SET(options, RENAME_EMPTY);
3306 /* pathchange left =NULL by default */
3308 options->change = diff_change;
3309 options->add_remove = diff_addremove;
3310 options->use_color = diff_use_color_default;
3311 options->detect_rename = diff_detect_rename_default;
3312 options->xdl_opts |= diff_algorithm;
3313 if (diff_indent_heuristic)
3314 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3315 else if (diff_compaction_heuristic)
3316 DIFF_XDL_SET(options, COMPACTION_HEURISTIC);
3317 options->orderfile = diff_order_file_cfg;
3319 if (diff_no_prefix) {
3321 options->a_prefix = options->b_prefix = "";
3322 } else if (!diff_mnemonic_prefix) {
3323 options->a_prefix = "a/";
3324 options->b_prefix = "b/";
3325 }
3326}
3327void diff_setup_done(struct diff_options *options)
3329{
3330 int count = 0;
3331 if (options->set_default)
3333 options->set_default(options);
3334 if (options->output_format & DIFF_FORMAT_NAME)
3336 count++;
3337 if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3338 count++;
3339 if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3340 count++;
3341 if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3342 count++;
3343 if (count > 1)
3344 die("--name-only, --name-status, --check and -s are mutually exclusive");
3345 /*
3347 * Most of the time we can say "there are changes"
3348 * only by checking if there are changed paths, but
3349 * --ignore-whitespace* options force us to look
3350 * inside contents.
3351 */
3352 if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3354 DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3355 DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3356 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3357 else
3358 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3359 if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3361 options->detect_rename = DIFF_DETECT_COPY;
3362 if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3364 options->prefix = NULL;
3365 if (options->prefix)
3366 options->prefix_length = strlen(options->prefix);
3367 else
3368 options->prefix_length = 0;
3369 if (options->output_format & (DIFF_FORMAT_NAME |
3371 DIFF_FORMAT_NAME_STATUS |
3372 DIFF_FORMAT_CHECKDIFF |
3373 DIFF_FORMAT_NO_OUTPUT))
3374 options->output_format &= ~(DIFF_FORMAT_RAW |
3375 DIFF_FORMAT_NUMSTAT |
3376 DIFF_FORMAT_DIFFSTAT |
3377 DIFF_FORMAT_SHORTSTAT |
3378 DIFF_FORMAT_DIRSTAT |
3379 DIFF_FORMAT_SUMMARY |
3380 DIFF_FORMAT_PATCH);
3381 /*
3383 * These cases always need recursive; we do not drop caller-supplied
3384 * recursive bits for other formats here.
3385 */
3386 if (options->output_format & (DIFF_FORMAT_PATCH |
3387 DIFF_FORMAT_NUMSTAT |
3388 DIFF_FORMAT_DIFFSTAT |
3389 DIFF_FORMAT_SHORTSTAT |
3390 DIFF_FORMAT_DIRSTAT |
3391 DIFF_FORMAT_SUMMARY |
3392 DIFF_FORMAT_CHECKDIFF))
3393 DIFF_OPT_SET(options, RECURSIVE);
3394 /*
3395 * Also pickaxe would not work very well if you do not say recursive
3396 */
3397 if (options->pickaxe)
3398 DIFF_OPT_SET(options, RECURSIVE);
3399 /*
3400 * When patches are generated, submodules diffed against the work tree
3401 * must be checked for dirtiness too so it can be shown in the output
3402 */
3403 if (options->output_format & DIFF_FORMAT_PATCH)
3404 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3405 if (options->detect_rename && options->rename_limit < 0)
3407 options->rename_limit = diff_rename_limit_default;
3408 if (options->setup & DIFF_SETUP_USE_CACHE) {
3409 if (!active_cache)
3410 /* read-cache does not die even when it fails
3411 * so it is safe for us to do this here. Also
3412 * it does not smudge active_cache or active_nr
3413 * when it fails, so we do not have to worry about
3414 * cleaning it up ourselves either.
3415 */
3416 read_cache();
3417 }
3418 if (options->abbrev <= 0 || 40 < options->abbrev)
3419 options->abbrev = 40; /* full */
3420 /*
3422 * It does not make sense to show the first hit we happened
3423 * to have found. It does not make sense not to return with
3424 * exit code in such a case either.
3425 */
3426 if (DIFF_OPT_TST(options, QUICK)) {
3427 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3428 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3429 }
3430 options->diff_path_counter = 0;
3432 if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3434 die(_("--follow requires exactly one pathspec"));
3435}
3436static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3438{
3439 char c, *eq;
3440 int len;
3441 if (*arg != '-')
3443 return 0;
3444 c = *++arg;
3445 if (!c)
3446 return 0;
3447 if (c == arg_short) {
3448 c = *++arg;
3449 if (!c)
3450 return 1;
3451 if (val && isdigit(c)) {
3452 char *end;
3453 int n = strtoul(arg, &end, 10);
3454 if (*end)
3455 return 0;
3456 *val = n;
3457 return 1;
3458 }
3459 return 0;
3460 }
3461 if (c != '-')
3462 return 0;
3463 arg++;
3464 eq = strchrnul(arg, '=');
3465 len = eq - arg;
3466 if (!len || strncmp(arg, arg_long, len))
3467 return 0;
3468 if (*eq) {
3469 int n;
3470 char *end;
3471 if (!isdigit(*++eq))
3472 return 0;
3473 n = strtoul(eq, &end, 10);
3474 if (*end)
3475 return 0;
3476 *val = n;
3477 }
3478 return 1;
3479}
3480static int diff_scoreopt_parse(const char *opt);
3482static inline int short_opt(char opt, const char **argv,
3484 const char **optarg)
3485{
3486 const char *arg = argv[0];
3487 if (arg[0] != '-' || arg[1] != opt)
3488 return 0;
3489 if (arg[2] != '\0') {
3490 *optarg = arg + 2;
3491 return 1;
3492 }
3493 if (!argv[1])
3494 die("Option '%c' requires a value", opt);
3495 *optarg = argv[1];
3496 return 2;
3497}
3498int parse_long_opt(const char *opt, const char **argv,
3500 const char **optarg)
3501{
3502 const char *arg = argv[0];
3503 if (!skip_prefix(arg, "--", &arg))
3504 return 0;
3505 if (!skip_prefix(arg, opt, &arg))
3506 return 0;
3507 if (*arg == '=') { /* stuck form: --option=value */
3508 *optarg = arg + 1;
3509 return 1;
3510 }
3511 if (*arg != '\0')
3512 return 0;
3513 /* separate form: --option value */
3514 if (!argv[1])
3515 die("Option '--%s' requires a value", opt);
3516 *optarg = argv[1];
3517 return 2;
3518}
3519static int stat_opt(struct diff_options *options, const char **av)
3521{
3522 const char *arg = av[0];
3523 char *end;
3524 int width = options->stat_width;
3525 int name_width = options->stat_name_width;
3526 int graph_width = options->stat_graph_width;
3527 int count = options->stat_count;
3528 int argcount = 1;
3529 if (!skip_prefix(arg, "--stat", &arg))
3531 die("BUG: stat option does not begin with --stat: %s", arg);
3532 end = (char *)arg;
3533 switch (*arg) {
3535 case '-':
3536 if (skip_prefix(arg, "-width", &arg)) {
3537 if (*arg == '=')
3538 width = strtoul(arg + 1, &end, 10);
3539 else if (!*arg && !av[1])
3540 die("Option '--stat-width' requires a value");
3541 else if (!*arg) {
3542 width = strtoul(av[1], &end, 10);
3543 argcount = 2;
3544 }
3545 } else if (skip_prefix(arg, "-name-width", &arg)) {
3546 if (*arg == '=')
3547 name_width = strtoul(arg + 1, &end, 10);
3548 else if (!*arg && !av[1])
3549 die("Option '--stat-name-width' requires a value");
3550 else if (!*arg) {
3551 name_width = strtoul(av[1], &end, 10);
3552 argcount = 2;
3553 }
3554 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3555 if (*arg == '=')
3556 graph_width = strtoul(arg + 1, &end, 10);
3557 else if (!*arg && !av[1])
3558 die("Option '--stat-graph-width' requires a value");
3559 else if (!*arg) {
3560 graph_width = strtoul(av[1], &end, 10);
3561 argcount = 2;
3562 }
3563 } else if (skip_prefix(arg, "-count", &arg)) {
3564 if (*arg == '=')
3565 count = strtoul(arg + 1, &end, 10);
3566 else if (!*arg && !av[1])
3567 die("Option '--stat-count' requires a value");
3568 else if (!*arg) {
3569 count = strtoul(av[1], &end, 10);
3570 argcount = 2;
3571 }
3572 }
3573 break;
3574 case '=':
3575 width = strtoul(arg+1, &end, 10);
3576 if (*end == ',')
3577 name_width = strtoul(end+1, &end, 10);
3578 if (*end == ',')
3579 count = strtoul(end+1, &end, 10);
3580 }
3581 /* Important! This checks all the error cases! */
3583 if (*end)
3584 return 0;
3585 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3586 options->stat_name_width = name_width;
3587 options->stat_graph_width = graph_width;
3588 options->stat_width = width;
3589 options->stat_count = count;
3590 return argcount;
3591}
3592static int parse_dirstat_opt(struct diff_options *options, const char *params)
3594{
3595 struct strbuf errmsg = STRBUF_INIT;
3596 if (parse_dirstat_params(options, params, &errmsg))
3597 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3598 errmsg.buf);
3599 strbuf_release(&errmsg);
3600 /*
3601 * The caller knows a dirstat-related option is given from the command
3602 * line; allow it to say "return this_function();"
3603 */
3604 options->output_format |= DIFF_FORMAT_DIRSTAT;
3605 return 1;
3606}
3607static int parse_submodule_opt(struct diff_options *options, const char *value)
3609{
3610 if (parse_submodule_params(options, value))
3611 die(_("Failed to parse --submodule option parameter: '%s'"),
3612 value);
3613 return 1;
3614}
3615static const char diff_status_letters[] = {
3617 DIFF_STATUS_ADDED,
3618 DIFF_STATUS_COPIED,
3619 DIFF_STATUS_DELETED,
3620 DIFF_STATUS_MODIFIED,
3621 DIFF_STATUS_RENAMED,
3622 DIFF_STATUS_TYPE_CHANGED,
3623 DIFF_STATUS_UNKNOWN,
3624 DIFF_STATUS_UNMERGED,
3625 DIFF_STATUS_FILTER_AON,
3626 DIFF_STATUS_FILTER_BROKEN,
3627 '\0',
3628};
3629static unsigned int filter_bit['Z' + 1];
3631static void prepare_filter_bits(void)
3633{
3634 int i;
3635 if (!filter_bit[DIFF_STATUS_ADDED]) {
3637 for (i = 0; diff_status_letters[i]; i++)
3638 filter_bit[(int) diff_status_letters[i]] = (1 << i);
3639 }
3640}
3641static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3643{
3644 return opt->filter & filter_bit[(int) status];
3645}
3646static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3648{
3649 int i, optch;
3650 prepare_filter_bits();
3652 /*
3654 * If there is a negation e.g. 'd' in the input, and we haven't
3655 * initialized the filter field with another --diff-filter, start
3656 * from full set of bits, except for AON.
3657 */
3658 if (!opt->filter) {
3659 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3660 if (optch < 'a' || 'z' < optch)
3661 continue;
3662 opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3663 opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3664 break;
3665 }
3666 }
3667 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3669 unsigned int bit;
3670 int negate;
3671 if ('a' <= optch && optch <= 'z') {
3673 negate = 1;
3674 optch = toupper(optch);
3675 } else {
3676 negate = 0;
3677 }
3678 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3680 if (!bit)
3681 return optarg[i];
3682 if (negate)
3683 opt->filter &= ~bit;
3684 else
3685 opt->filter |= bit;
3686 }
3687 return 0;
3688}
3689static void enable_patch_output(int *fmt) {
3691 *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3692 *fmt |= DIFF_FORMAT_PATCH;
3693}
3694static int parse_one_token(const char **arg, const char *token)
3696{
3697 const char *rest;
3698 if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
3699 *arg = rest;
3700 return 1;
3701 }
3702 return 0;
3703}
3704static int parse_ws_error_highlight(struct diff_options *opt, const char *arg)
3706{
3707 const char *orig_arg = arg;
3708 unsigned val = 0;
3709 while (*arg) {
3710 if (parse_one_token(&arg, "none"))
3711 val = 0;
3712 else if (parse_one_token(&arg, "default"))
3713 val = WSEH_NEW;
3714 else if (parse_one_token(&arg, "all"))
3715 val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
3716 else if (parse_one_token(&arg, "new"))
3717 val |= WSEH_NEW;
3718 else if (parse_one_token(&arg, "old"))
3719 val |= WSEH_OLD;
3720 else if (parse_one_token(&arg, "context"))
3721 val |= WSEH_CONTEXT;
3722 else {
3723 error("unknown value after ws-error-highlight=%.*s",
3724 (int)(arg - orig_arg), orig_arg);
3725 return 0;
3726 }
3727 if (*arg)
3728 arg++;
3729 }
3730 opt->ws_error_highlight = val;
3731 return 1;
3732}
3733int diff_opt_parse(struct diff_options *options,
3735 const char **av, int ac, const char *prefix)
3736{
3737 const char *arg = av[0];
3738 const char *optarg;
3739 int argcount;
3740 if (!prefix)
3742 prefix = "";
3743 /* Output format options */
3745 if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3746 || opt_arg(arg, 'U', "unified", &options->context))
3747 enable_patch_output(&options->output_format);
3748 else if (!strcmp(arg, "--raw"))
3749 options->output_format |= DIFF_FORMAT_RAW;
3750 else if (!strcmp(arg, "--patch-with-raw")) {
3751 enable_patch_output(&options->output_format);
3752 options->output_format |= DIFF_FORMAT_RAW;
3753 } else if (!strcmp(arg, "--numstat"))
3754 options->output_format |= DIFF_FORMAT_NUMSTAT;
3755 else if (!strcmp(arg, "--shortstat"))
3756 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3757 else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3758 return parse_dirstat_opt(options, "");
3759 else if (skip_prefix(arg, "-X", &arg))
3760 return parse_dirstat_opt(options, arg);
3761 else if (skip_prefix(arg, "--dirstat=", &arg))
3762 return parse_dirstat_opt(options, arg);
3763 else if (!strcmp(arg, "--cumulative"))
3764 return parse_dirstat_opt(options, "cumulative");
3765 else if (!strcmp(arg, "--dirstat-by-file"))
3766 return parse_dirstat_opt(options, "files");
3767 else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3768 parse_dirstat_opt(options, "files");
3769 return parse_dirstat_opt(options, arg);
3770 }
3771 else if (!strcmp(arg, "--check"))
3772 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3773 else if (!strcmp(arg, "--summary"))
3774 options->output_format |= DIFF_FORMAT_SUMMARY;
3775 else if (!strcmp(arg, "--patch-with-stat")) {
3776 enable_patch_output(&options->output_format);
3777 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3778 } else if (!strcmp(arg, "--name-only"))
3779 options->output_format |= DIFF_FORMAT_NAME;
3780 else if (!strcmp(arg, "--name-status"))
3781 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3782 else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3783 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3784 else if (starts_with(arg, "--stat"))
3785 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3786 return stat_opt(options, av);
3787 /* renames options */
3789 else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3790 !strcmp(arg, "--break-rewrites")) {
3791 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3792 return error("invalid argument to -B: %s", arg+2);
3793 }
3794 else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3795 !strcmp(arg, "--find-renames")) {
3796 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3797 return error("invalid argument to -M: %s", arg+2);
3798 options->detect_rename = DIFF_DETECT_RENAME;
3799 }
3800 else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3801 options->irreversible_delete = 1;
3802 }
3803 else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3804 !strcmp(arg, "--find-copies")) {
3805 if (options->detect_rename == DIFF_DETECT_COPY)
3806 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3807 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3808 return error("invalid argument to -C: %s", arg+2);
3809 options->detect_rename = DIFF_DETECT_COPY;
3810 }
3811 else if (!strcmp(arg, "--no-renames"))
3812 options->detect_rename = 0;
3813 else if (!strcmp(arg, "--rename-empty"))
3814 DIFF_OPT_SET(options, RENAME_EMPTY);
3815 else if (!strcmp(arg, "--no-rename-empty"))
3816 DIFF_OPT_CLR(options, RENAME_EMPTY);
3817 else if (!strcmp(arg, "--relative"))
3818 DIFF_OPT_SET(options, RELATIVE_NAME);
3819 else if (skip_prefix(arg, "--relative=", &arg)) {
3820 DIFF_OPT_SET(options, RELATIVE_NAME);
3821 options->prefix = arg;
3822 }
3823 /* xdiff options */
3825 else if (!strcmp(arg, "--minimal"))
3826 DIFF_XDL_SET(options, NEED_MINIMAL);
3827 else if (!strcmp(arg, "--no-minimal"))
3828 DIFF_XDL_CLR(options, NEED_MINIMAL);
3829 else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3830 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3831 else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3832 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3833 else if (!strcmp(arg, "--ignore-space-at-eol"))
3834 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3835 else if (!strcmp(arg, "--ignore-blank-lines"))
3836 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3837 else if (!strcmp(arg, "--indent-heuristic")) {
3838 DIFF_XDL_SET(options, INDENT_HEURISTIC);
3839 DIFF_XDL_CLR(options, COMPACTION_HEURISTIC);
3840 } else if (!strcmp(arg, "--no-indent-heuristic"))
3841 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
3842 else if (!strcmp(arg, "--compaction-heuristic")) {
3843 DIFF_XDL_SET(options, COMPACTION_HEURISTIC);
3844 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
3845 } else if (!strcmp(arg, "--no-compaction-heuristic"))
3846 DIFF_XDL_CLR(options, COMPACTION_HEURISTIC);
3847 else if (!strcmp(arg, "--patience"))
3848 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3849 else if (!strcmp(arg, "--histogram"))
3850 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3851 else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3852 long value = parse_algorithm_value(optarg);
3853 if (value < 0)
3854 return error("option diff-algorithm accepts \"myers\", "
3855 "\"minimal\", \"patience\" and \"histogram\"");
3856 /* clear out previous settings */
3857 DIFF_XDL_CLR(options, NEED_MINIMAL);
3858 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3859 options->xdl_opts |= value;
3860 return argcount;
3861 }
3862 /* flags options */
3864 else if (!strcmp(arg, "--binary")) {
3865 enable_patch_output(&options->output_format);
3866 DIFF_OPT_SET(options, BINARY);
3867 }
3868 else if (!strcmp(arg, "--full-index"))
3869 DIFF_OPT_SET(options, FULL_INDEX);
3870 else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3871 DIFF_OPT_SET(options, TEXT);
3872 else if (!strcmp(arg, "-R"))
3873 DIFF_OPT_SET(options, REVERSE_DIFF);
3874 else if (!strcmp(arg, "--find-copies-harder"))
3875 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3876 else if (!strcmp(arg, "--follow"))
3877 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3878 else if (!strcmp(arg, "--no-follow")) {
3879 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3880 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3881 } else if (!strcmp(arg, "--color"))
3882 options->use_color = 1;
3883 else if (skip_prefix(arg, "--color=", &arg)) {
3884 int value = git_config_colorbool(NULL, arg);
3885 if (value < 0)
3886 return error("option `color' expects \"always\", \"auto\", or \"never\"");
3887 options->use_color = value;
3888 }
3889 else if (!strcmp(arg, "--no-color"))
3890 options->use_color = 0;
3891 else if (!strcmp(arg, "--color-words")) {
3892 options->use_color = 1;
3893 options->word_diff = DIFF_WORDS_COLOR;
3894 }
3895 else if (skip_prefix(arg, "--color-words=", &arg)) {
3896 options->use_color = 1;
3897 options->word_diff = DIFF_WORDS_COLOR;
3898 options->word_regex = arg;
3899 }
3900 else if (!strcmp(arg, "--word-diff")) {
3901 if (options->word_diff == DIFF_WORDS_NONE)
3902 options->word_diff = DIFF_WORDS_PLAIN;
3903 }
3904 else if (skip_prefix(arg, "--word-diff=", &arg)) {
3905 if (!strcmp(arg, "plain"))
3906 options->word_diff = DIFF_WORDS_PLAIN;
3907 else if (!strcmp(arg, "color")) {
3908 options->use_color = 1;
3909 options->word_diff = DIFF_WORDS_COLOR;
3910 }
3911 else if (!strcmp(arg, "porcelain"))
3912 options->word_diff = DIFF_WORDS_PORCELAIN;
3913 else if (!strcmp(arg, "none"))
3914 options->word_diff = DIFF_WORDS_NONE;
3915 else
3916 die("bad --word-diff argument: %s", arg);
3917 }
3918 else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3919 if (options->word_diff == DIFF_WORDS_NONE)
3920 options->word_diff = DIFF_WORDS_PLAIN;
3921 options->word_regex = optarg;
3922 return argcount;
3923 }
3924 else if (!strcmp(arg, "--exit-code"))
3925 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3926 else if (!strcmp(arg, "--quiet"))
3927 DIFF_OPT_SET(options, QUICK);
3928 else if (!strcmp(arg, "--ext-diff"))
3929 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3930 else if (!strcmp(arg, "--no-ext-diff"))
3931 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3932 else if (!strcmp(arg, "--textconv"))
3933 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3934 else if (!strcmp(arg, "--no-textconv"))
3935 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3936 else if (!strcmp(arg, "--ignore-submodules")) {
3937 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3938 handle_ignore_submodules_arg(options, "all");
3939 } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3940 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3941 handle_ignore_submodules_arg(options, arg);
3942 } else if (!strcmp(arg, "--submodule"))
3943 options->submodule_format = DIFF_SUBMODULE_LOG;
3944 else if (skip_prefix(arg, "--submodule=", &arg))
3945 return parse_submodule_opt(options, arg);
3946 else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3947 return parse_ws_error_highlight(options, arg);
3948 /* misc options */
3950 else if (!strcmp(arg, "-z"))
3951 options->line_termination = 0;
3952 else if ((argcount = short_opt('l', av, &optarg))) {
3953 options->rename_limit = strtoul(optarg, NULL, 10);
3954 return argcount;
3955 }
3956 else if ((argcount = short_opt('S', av, &optarg))) {
3957 options->pickaxe = optarg;
3958 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3959 return argcount;
3960 } else if ((argcount = short_opt('G', av, &optarg))) {
3961 options->pickaxe = optarg;
3962 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3963 return argcount;
3964 }
3965 else if (!strcmp(arg, "--pickaxe-all"))
3966 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3967 else if (!strcmp(arg, "--pickaxe-regex"))
3968 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3969 else if ((argcount = short_opt('O', av, &optarg))) {
3970 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
3971 options->orderfile = xstrdup(path);
3972 return argcount;
3973 }
3974 else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3975 int offending = parse_diff_filter_opt(optarg, options);
3976 if (offending)
3977 die("unknown change class '%c' in --diff-filter=%s",
3978 offending, optarg);
3979 return argcount;
3980 }
3981 else if (!strcmp(arg, "--abbrev"))
3982 options->abbrev = DEFAULT_ABBREV;
3983 else if (skip_prefix(arg, "--abbrev=", &arg)) {
3984 options->abbrev = strtoul(arg, NULL, 10);
3985 if (options->abbrev < MINIMUM_ABBREV)
3986 options->abbrev = MINIMUM_ABBREV;
3987 else if (40 < options->abbrev)
3988 options->abbrev = 40;
3989 }
3990 else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3991 options->a_prefix = optarg;
3992 return argcount;
3993 }
3994 else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
3995 options->line_prefix = optarg;
3996 options->line_prefix_length = strlen(options->line_prefix);
3997 graph_setup_line_prefix(options);
3998 return argcount;
3999 }
4000 else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4001 options->b_prefix = optarg;
4002 return argcount;
4003 }
4004 else if (!strcmp(arg, "--no-prefix"))
4005 options->a_prefix = options->b_prefix = "";
4006 else if (opt_arg(arg, '\0', "inter-hunk-context",
4007 &options->interhunkcontext))
4008 ;
4009 else if (!strcmp(arg, "-W"))
4010 DIFF_OPT_SET(options, FUNCCONTEXT);
4011 else if (!strcmp(arg, "--function-context"))
4012 DIFF_OPT_SET(options, FUNCCONTEXT);
4013 else if (!strcmp(arg, "--no-function-context"))
4014 DIFF_OPT_CLR(options, FUNCCONTEXT);
4015 else if ((argcount = parse_long_opt("output", av, &optarg))) {
4016 const char *path = prefix_filename(prefix, strlen(prefix), optarg);
4017 options->file = fopen(path, "w");
4018 if (!options->file)
4019 die_errno("Could not open '%s'", path);
4020 options->close_file = 1;
4021 if (options->use_color != GIT_COLOR_ALWAYS)
4022 options->use_color = GIT_COLOR_NEVER;
4023 return argcount;
4024 } else
4025 return 0;
4026 return 1;
4027}
4028int parse_rename_score(const char **cp_p)
4030{
4031 unsigned long num, scale;
4032 int ch, dot;
4033 const char *cp = *cp_p;
4034 num = 0;
4036 scale = 1;
4037 dot = 0;
4038 for (;;) {
4039 ch = *cp;
4040 if ( !dot && ch == '.' ) {
4041 scale = 1;
4042 dot = 1;
4043 } else if ( ch == '%' ) {
4044 scale = dot ? scale*100 : 100;
4045 cp++; /* % is always at the end */
4046 break;
4047 } else if ( ch >= '0' && ch <= '9' ) {
4048 if ( scale < 100000 ) {
4049 scale *= 10;
4050 num = (num*10) + (ch-'0');
4051 }
4052 } else {
4053 break;
4054 }
4055 cp++;
4056 }
4057 *cp_p = cp;
4058 /* user says num divided by scale and we say internally that
4060 * is MAX_SCORE * num / scale.
4061 */
4062 return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4063}
4064static int diff_scoreopt_parse(const char *opt)
4066{
4067 int opt1, opt2, cmd;
4068 if (*opt++ != '-')
4070 return -1;
4071 cmd = *opt++;
4072 if (cmd == '-') {
4073 /* convert the long-form arguments into short-form versions */
4074 if (skip_prefix(opt, "break-rewrites", &opt)) {
4075 if (*opt == 0 || *opt++ == '=')
4076 cmd = 'B';
4077 } else if (skip_prefix(opt, "find-copies", &opt)) {
4078 if (*opt == 0 || *opt++ == '=')
4079 cmd = 'C';
4080 } else if (skip_prefix(opt, "find-renames", &opt)) {
4081 if (*opt == 0 || *opt++ == '=')
4082 cmd = 'M';
4083 }
4084 }
4085 if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4086 return -1; /* that is not a -M, -C, or -B option */
4087 opt1 = parse_rename_score(&opt);
4089 if (cmd != 'B')
4090 opt2 = 0;
4091 else {
4092 if (*opt == 0)
4093 opt2 = 0;
4094 else if (*opt != '/')
4095 return -1; /* we expect -B80/99 or -B80 */
4096 else {
4097 opt++;
4098 opt2 = parse_rename_score(&opt);
4099 }
4100 }
4101 if (*opt != 0)
4102 return -1;
4103 return opt1 | (opt2 << 16);
4104}
4105struct diff_queue_struct diff_queued_diff;
4107void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4109{
4110 ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4111 queue->queue[queue->nr++] = dp;
4112}
4113struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4115 struct diff_filespec *one,
4116 struct diff_filespec *two)
4117{
4118 struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4119 dp->one = one;
4120 dp->two = two;
4121 if (queue)
4122 diff_q(queue, dp);
4123 return dp;
4124}
4125void diff_free_filepair(struct diff_filepair *p)
4127{
4128 free_filespec(p->one);
4129 free_filespec(p->two);
4130 free(p);
4131}
4132/* This is different from find_unique_abbrev() in that
4134 * it stuffs the result with dots for alignment.
4135 */
4136const char *diff_unique_abbrev(const unsigned char *sha1, int len)
4137{
4138 int abblen;
4139 const char *abbrev;
4140 if (len == 40)
4141 return sha1_to_hex(sha1);
4142 abbrev = find_unique_abbrev(sha1, len);
4144 abblen = strlen(abbrev);
4145 if (abblen < 37) {
4146 static char hex[41];
4147 if (len < abblen && abblen <= len + 2)
4148 xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4149 else
4150 xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4151 return hex;
4152 }
4153 return sha1_to_hex(sha1);
4154}
4155static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4157{
4158 int line_termination = opt->line_termination;
4159 int inter_name_termination = line_termination ? '\t' : '\0';
4160 fprintf(opt->file, "%s", diff_line_prefix(opt));
4162 if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4163 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4164 diff_unique_abbrev(p->one->oid.hash, opt->abbrev));
4165 fprintf(opt->file, "%s ",
4166 diff_unique_abbrev(p->two->oid.hash, opt->abbrev));
4167 }
4168 if (p->score) {
4169 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4170 inter_name_termination);
4171 } else {
4172 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4173 }
4174 if (p->status == DIFF_STATUS_COPIED ||
4176 p->status == DIFF_STATUS_RENAMED) {
4177 const char *name_a, *name_b;
4178 name_a = p->one->path;
4179 name_b = p->two->path;
4180 strip_prefix(opt->prefix_length, &name_a, &name_b);
4181 write_name_quoted(name_a, opt->file, inter_name_termination);
4182 write_name_quoted(name_b, opt->file, line_termination);
4183 } else {
4184 const char *name_a, *name_b;
4185 name_a = p->one->mode ? p->one->path : p->two->path;
4186 name_b = NULL;
4187 strip_prefix(opt->prefix_length, &name_a, &name_b);
4188 write_name_quoted(name_a, opt->file, line_termination);
4189 }
4190}
4191int diff_unmodified_pair(struct diff_filepair *p)
4193{
4194 /* This function is written stricter than necessary to support
4195 * the currently implemented transformers, but the idea is to
4196 * let transformers to produce diff_filepairs any way they want,
4197 * and filter and clean them up here before producing the output.
4198 */
4199 struct diff_filespec *one = p->one, *two = p->two;
4200 if (DIFF_PAIR_UNMERGED(p))
4202 return 0; /* unmerged is interesting */
4203 /* deletion, addition, mode or type change
4205 * and rename are all interesting.
4206 */
4207 if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4208 DIFF_PAIR_MODE_CHANGED(p) ||
4209 strcmp(one->path, two->path))
4210 return 0;
4211 /* both are valid and point at the same path. that is, we are
4213 * dealing with a change.
4214 */
4215 if (one->oid_valid && two->oid_valid &&
4216 !oidcmp(&one->oid, &two->oid) &&
4217 !one->dirty_submodule && !two->dirty_submodule)
4218 return 1; /* no change */
4219 if (!one->oid_valid && !two->oid_valid)
4220 return 1; /* both look at the same file on the filesystem. */
4221 return 0;
4222}
4223static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4225{
4226 if (diff_unmodified_pair(p))
4227 return;
4228 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4230 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4231 return; /* no tree diffs in patch format */
4232 run_diff(p, o);
4234}
4235static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4237 struct diffstat_t *diffstat)
4238{
4239 if (diff_unmodified_pair(p))
4240 return;
4241 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4243 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4244 return; /* no useful stat for tree diffs */
4245 run_diffstat(p, o, diffstat);
4247}
4248static void diff_flush_checkdiff(struct diff_filepair *p,
4250 struct diff_options *o)
4251{
4252 if (diff_unmodified_pair(p))
4253 return;
4254 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4256 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4257 return; /* nothing to check in tree diffs */
4258 run_checkdiff(p, o);
4260}
4261int diff_queue_is_empty(void)
4263{
4264 struct diff_queue_struct *q = &diff_queued_diff;
4265 int i;
4266 for (i = 0; i < q->nr; i++)
4267 if (!diff_unmodified_pair(q->queue[i]))
4268 return 0;
4269 return 1;
4270}
4271#if DIFF_DEBUG
4273void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4274{
4275 fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4276 x, one ? one : "",
4277 s->path,
4278 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4279 s->mode,
4280 s->oid_valid ? oid_to_hex(&s->oid) : "");
4281 fprintf(stderr, "queue[%d] %s size %lu\n",
4282 x, one ? one : "",
4283 s->size);
4284}
4285void diff_debug_filepair(const struct diff_filepair *p, int i)
4287{
4288 diff_debug_filespec(p->one, i, "one");
4289 diff_debug_filespec(p->two, i, "two");
4290 fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4291 p->score, p->status ? p->status : '?',
4292 p->one->rename_used, p->broken_pair);
4293}
4294void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4296{
4297 int i;
4298 if (msg)
4299 fprintf(stderr, "%s\n", msg);
4300 fprintf(stderr, "q->nr = %d\n", q->nr);
4301 for (i = 0; i < q->nr; i++) {
4302 struct diff_filepair *p = q->queue[i];
4303 diff_debug_filepair(p, i);
4304 }
4305}
4306#endif
4307static void diff_resolve_rename_copy(void)
4309{
4310 int i;
4311 struct diff_filepair *p;
4312 struct diff_queue_struct *q = &diff_queued_diff;
4313 diff_debug_queue("resolve-rename-copy", q);
4315 for (i = 0; i < q->nr; i++) {
4317 p = q->queue[i];
4318 p->status = 0; /* undecided */
4319 if (DIFF_PAIR_UNMERGED(p))
4320 p->status = DIFF_STATUS_UNMERGED;
4321 else if (!DIFF_FILE_VALID(p->one))
4322 p->status = DIFF_STATUS_ADDED;
4323 else if (!DIFF_FILE_VALID(p->two))
4324 p->status = DIFF_STATUS_DELETED;
4325 else if (DIFF_PAIR_TYPE_CHANGED(p))
4326 p->status = DIFF_STATUS_TYPE_CHANGED;
4327 /* from this point on, we are dealing with a pair
4329 * whose both sides are valid and of the same type, i.e.
4330 * either in-place edit or rename/copy edit.
4331 */
4332 else if (DIFF_PAIR_RENAME(p)) {
4333 /*
4334 * A rename might have re-connected a broken
4335 * pair up, causing the pathnames to be the
4336 * same again. If so, that's not a rename at
4337 * all, just a modification..
4338 *
4339 * Otherwise, see if this source was used for
4340 * multiple renames, in which case we decrement
4341 * the count, and call it a copy.
4342 */
4343 if (!strcmp(p->one->path, p->two->path))
4344 p->status = DIFF_STATUS_MODIFIED;
4345 else if (--p->one->rename_used > 0)
4346 p->status = DIFF_STATUS_COPIED;
4347 else
4348 p->status = DIFF_STATUS_RENAMED;
4349 }
4350 else if (oidcmp(&p->one->oid, &p->two->oid) ||
4351 p->one->mode != p->two->mode ||
4352 p->one->dirty_submodule ||
4353 p->two->dirty_submodule ||
4354 is_null_oid(&p->one->oid))
4355 p->status = DIFF_STATUS_MODIFIED;
4356 else {
4357 /* This is a "no-change" entry and should not
4358 * happen anymore, but prepare for broken callers.
4359 */
4360 error("feeding unmodified %s to diffcore",
4361 p->one->path);
4362 p->status = DIFF_STATUS_UNKNOWN;
4363 }
4364 }
4365 diff_debug_queue("resolve-rename-copy done", q);
4366}
4367static int check_pair_status(struct diff_filepair *p)
4369{
4370 switch (p->status) {
4371 case DIFF_STATUS_UNKNOWN:
4372 return 0;
4373 case 0:
4374 die("internal error in diff-resolve-rename-copy");
4375 default:
4376 return 1;
4377 }
4378}
4379static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4381{
4382 int fmt = opt->output_format;
4383 if (fmt & DIFF_FORMAT_CHECKDIFF)
4385 diff_flush_checkdiff(p, opt);
4386 else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4387 diff_flush_raw(p, opt);
4388 else if (fmt & DIFF_FORMAT_NAME) {
4389 const char *name_a, *name_b;
4390 name_a = p->two->path;
4391 name_b = NULL;
4392 strip_prefix(opt->prefix_length, &name_a, &name_b);
4393 write_name_quoted(name_a, opt->file, opt->line_termination);
4394 }
4395}
4396static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4398{
4399 if (fs->mode)
4400 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4401 else
4402 fprintf(file, " %s ", newdelete);
4403 write_name_quoted(fs->path, file, '\n');
4404}
4405static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4408 const char *line_prefix)
4409{
4410 if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4411 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4412 p->two->mode, show_name ? ' ' : '\n');
4413 if (show_name) {
4414 write_name_quoted(p->two->path, file, '\n');
4415 }
4416 }
4417}
4418static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4420 const char *line_prefix)
4421{
4422 char *names = pprint_rename(p->one->path, p->two->path);
4423 fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4425 free(names);
4426 show_mode_change(file, p, 0, line_prefix);
4427}
4428static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4430{
4431 FILE *file = opt->file;
4432 const char *line_prefix = diff_line_prefix(opt);
4433 switch(p->status) {
4435 case DIFF_STATUS_DELETED:
4436 fputs(line_prefix, file);
4437 show_file_mode_name(file, "delete", p->one);
4438 break;
4439 case DIFF_STATUS_ADDED:
4440 fputs(line_prefix, file);
4441 show_file_mode_name(file, "create", p->two);
4442 break;
4443 case DIFF_STATUS_COPIED:
4444 fputs(line_prefix, file);
4445 show_rename_copy(file, "copy", p, line_prefix);
4446 break;
4447 case DIFF_STATUS_RENAMED:
4448 fputs(line_prefix, file);
4449 show_rename_copy(file, "rename", p, line_prefix);
4450 break;
4451 default:
4452 if (p->score) {
4453 fprintf(file, "%s rewrite ", line_prefix);
4454 write_name_quoted(p->two->path, file, ' ');
4455 fprintf(file, "(%d%%)\n", similarity_index(p));
4456 }
4457 show_mode_change(file, p, !p->score, line_prefix);
4458 break;
4459 }
4460}
4461struct patch_id_t {
4463 git_SHA_CTX *ctx;
4464 int patchlen;
4465};
4466static int remove_space(char *line, int len)
4468{
4469 int i;
4470 char *dst = line;
4471 unsigned char c;
4472 for (i = 0; i < len; i++)
4474 if (!isspace((c = line[i])))
4475 *dst++ = c;
4476 return dst - line;
4478}
4479static void patch_id_consume(void *priv, char *line, unsigned long len)
4481{
4482 struct patch_id_t *data = priv;
4483 int new_len;
4484 /* Ignore line numbers when computing the SHA1 of the patch */
4486 if (starts_with(line, "@@ -"))
4487 return;
4488 new_len = remove_space(line, len);
4490 git_SHA1_Update(data->ctx, line, new_len);
4492 data->patchlen += new_len;
4493}
4494/* returns 0 upon success, and writes result into sha1 */
4496static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4497{
4498 struct diff_queue_struct *q = &diff_queued_diff;
4499 int i;
4500 git_SHA_CTX ctx;
4501 struct patch_id_t data;
4502 char buffer[PATH_MAX * 4 + 20];
4503 git_SHA1_Init(&ctx);
4505 memset(&data, 0, sizeof(struct patch_id_t));
4506 data.ctx = &ctx;
4507 for (i = 0; i < q->nr; i++) {
4509 xpparam_t xpp;
4510 xdemitconf_t xecfg;
4511 mmfile_t mf1, mf2;
4512 struct diff_filepair *p = q->queue[i];
4513 int len1, len2;
4514 memset(&xpp, 0, sizeof(xpp));
4516 memset(&xecfg, 0, sizeof(xecfg));
4517 if (p->status == 0)
4518 return error("internal diff status error");
4519 if (p->status == DIFF_STATUS_UNKNOWN)
4520 continue;
4521 if (diff_unmodified_pair(p))
4522 continue;
4523 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4524 (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4525 continue;
4526 if (DIFF_PAIR_UNMERGED(p))
4527 continue;
4528 diff_fill_sha1_info(p->one);
4530 diff_fill_sha1_info(p->two);
4531 len1 = remove_space(p->one->path, strlen(p->one->path));
4533 len2 = remove_space(p->two->path, strlen(p->two->path));
4534 if (p->one->mode == 0)
4535 len1 = snprintf(buffer, sizeof(buffer),
4536 "diff--gita/%.*sb/%.*s"
4537 "newfilemode%06o"
4538 "---/dev/null"
4539 "+++b/%.*s",
4540 len1, p->one->path,
4541 len2, p->two->path,
4542 p->two->mode,
4543 len2, p->two->path);
4544 else if (p->two->mode == 0)
4545 len1 = snprintf(buffer, sizeof(buffer),
4546 "diff--gita/%.*sb/%.*s"
4547 "deletedfilemode%06o"
4548 "---a/%.*s"
4549 "+++/dev/null",
4550 len1, p->one->path,
4551 len2, p->two->path,
4552 p->one->mode,
4553 len1, p->one->path);
4554 else
4555 len1 = snprintf(buffer, sizeof(buffer),
4556 "diff--gita/%.*sb/%.*s"
4557 "---a/%.*s"
4558 "+++b/%.*s",
4559 len1, p->one->path,
4560 len2, p->two->path,
4561 len1, p->one->path,
4562 len2, p->two->path);
4563 git_SHA1_Update(&ctx, buffer, len1);
4564 if (diff_header_only)
4566 continue;
4567 if (fill_mmfile(&mf1, p->one) < 0 ||
4569 fill_mmfile(&mf2, p->two) < 0)
4570 return error("unable to read files to diff");
4571 if (diff_filespec_is_binary(p->one) ||
4573 diff_filespec_is_binary(p->two)) {
4574 git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
4575 40);
4576 git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
4577 40);
4578 continue;
4579 }
4580 xpp.flags = 0;
4582 xecfg.ctxlen = 3;
4583 xecfg.flags = 0;
4584 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4585 &xpp, &xecfg))
4586 return error("unable to generate patch-id diff for %s",
4587 p->one->path);
4588 }
4589 git_SHA1_Final(sha1, &ctx);
4591 return 0;
4592}
4593int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1, int diff_header_only)
4595{
4596 struct diff_queue_struct *q = &diff_queued_diff;
4597 int i;
4598 int result = diff_get_patch_id(options, sha1, diff_header_only);
4599 for (i = 0; i < q->nr; i++)
4601 diff_free_filepair(q->queue[i]);
4602 free(q->queue);
4604 DIFF_QUEUE_CLEAR(q);
4605 return result;
4607}
4608static int is_summary_empty(const struct diff_queue_struct *q)
4610{
4611 int i;
4612 for (i = 0; i < q->nr; i++) {
4614 const struct diff_filepair *p = q->queue[i];
4615 switch (p->status) {
4617 case DIFF_STATUS_DELETED:
4618 case DIFF_STATUS_ADDED:
4619 case DIFF_STATUS_COPIED:
4620 case DIFF_STATUS_RENAMED:
4621 return 0;
4622 default:
4623 if (p->score)
4624 return 0;
4625 if (p->one->mode && p->two->mode &&
4626 p->one->mode != p->two->mode)
4627 return 0;
4628 break;
4629 }
4630 }
4631 return 1;
4632}
4633static const char rename_limit_warning[] =
4635"inexact rename detection was skipped due to too many files.";
4636static const char degrade_cc_to_c_warning[] =
4638"only found copies from modified paths due to too many files.";
4639static const char rename_limit_advice[] =
4641"you may want to set your %s variable to at least "
4642"%d and retry the command.";
4643void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4645{
4646 if (degraded_cc)
4647 warning(degrade_cc_to_c_warning);
4648 else if (needed)
4649 warning(rename_limit_warning);
4650 else
4651 return;
4652 if (0 < needed && needed < 32767)
4653 warning(rename_limit_advice, varname, needed);
4654}
4655void diff_flush(struct diff_options *options)
4657{
4658 struct diff_queue_struct *q = &diff_queued_diff;
4659 int i, output_format = options->output_format;
4660 int separator = 0;
4661 int dirstat_by_line = 0;
4662 /*
4664 * Order: raw, stat, summary, patch
4665 * or: name/name-status/checkdiff (other bits clear)
4666 */
4667 if (!q->nr)
4668 goto free_queue;
4669 if (output_format & (DIFF_FORMAT_RAW |
4671 DIFF_FORMAT_NAME |
4672 DIFF_FORMAT_NAME_STATUS |
4673 DIFF_FORMAT_CHECKDIFF)) {
4674 for (i = 0; i < q->nr; i++) {
4675 struct diff_filepair *p = q->queue[i];
4676 if (check_pair_status(p))
4677 flush_one_pair(p, options);
4678 }
4679 separator++;
4680 }
4681 if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4683 dirstat_by_line = 1;
4684 if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4686 dirstat_by_line) {
4687 struct diffstat_t diffstat;
4688 memset(&diffstat, 0, sizeof(struct diffstat_t));
4690 for (i = 0; i < q->nr; i++) {
4691 struct diff_filepair *p = q->queue[i];
4692 if (check_pair_status(p))
4693 diff_flush_stat(p, options, &diffstat);
4694 }
4695 if (output_format & DIFF_FORMAT_NUMSTAT)
4696 show_numstat(&diffstat, options);
4697 if (output_format & DIFF_FORMAT_DIFFSTAT)
4698 show_stats(&diffstat, options);
4699 if (output_format & DIFF_FORMAT_SHORTSTAT)
4700 show_shortstats(&diffstat, options);
4701 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4702 show_dirstat_by_line(&diffstat, options);
4703 free_diffstat_info(&diffstat);
4704 separator++;
4705 }
4706 if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4707 show_dirstat(options);
4708 if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4710 for (i = 0; i < q->nr; i++) {
4711 diff_summary(options, q->queue[i]);
4712 }
4713 separator++;
4714 }
4715 if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4717 DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4718 DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4719 /*
4720 * run diff_flush_patch for the exit status. setting
4721 * options->file to /dev/null should be safe, because we
4722 * aren't supposed to produce any output anyway.
4723 */
4724 if (options->close_file)
4725 fclose(options->file);
4726 options->file = fopen("/dev/null", "w");
4727 if (!options->file)
4728 die_errno("Could not open /dev/null");
4729 options->close_file = 1;
4730 for (i = 0; i < q->nr; i++) {
4731 struct diff_filepair *p = q->queue[i];
4732 if (check_pair_status(p))
4733 diff_flush_patch(p, options);
4734 if (options->found_changes)
4735 break;
4736 }
4737 }
4738 if (output_format & DIFF_FORMAT_PATCH) {
4740 if (separator) {
4741 fprintf(options->file, "%s%c",
4742 diff_line_prefix(options),
4743 options->line_termination);
4744 if (options->stat_sep) {
4745 /* attach patch instead of inline */
4746 fputs(options->stat_sep, options->file);
4747 }
4748 }
4749 for (i = 0; i < q->nr; i++) {
4751 struct diff_filepair *p = q->queue[i];
4752 if (check_pair_status(p))
4753 diff_flush_patch(p, options);
4754 }
4755 }
4756 if (output_format & DIFF_FORMAT_CALLBACK)
4758 options->format_callback(q, options, options->format_callback_data);
4759 for (i = 0; i < q->nr; i++)
4761 diff_free_filepair(q->queue[i]);
4762free_queue:
4763 free(q->queue);
4764 DIFF_QUEUE_CLEAR(q);
4765 if (options->close_file)
4766 fclose(options->file);
4767 /*
4769 * Report the content-level differences with HAS_CHANGES;
4770 * diff_addremove/diff_change does not set the bit when
4771 * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4772 */
4773 if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4774 if (options->found_changes)
4775 DIFF_OPT_SET(options, HAS_CHANGES);
4776 else
4777 DIFF_OPT_CLR(options, HAS_CHANGES);
4778 }
4779}
4780static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4782{
4783 return (((p->status == DIFF_STATUS_MODIFIED) &&
4784 ((p->score &&
4785 filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4786 (!p->score &&
4787 filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4788 ((p->status != DIFF_STATUS_MODIFIED) &&
4789 filter_bit_tst(p->status, options)));
4790}
4791static void diffcore_apply_filter(struct diff_options *options)
4793{
4794 int i;
4795 struct diff_queue_struct *q = &diff_queued_diff;
4796 struct diff_queue_struct outq;
4797 DIFF_QUEUE_CLEAR(&outq);
4799 if (!options->filter)
4801 return;
4802 if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4804 int found;
4805 for (i = found = 0; !found && i < q->nr; i++) {
4806 if (match_filter(options, q->queue[i]))
4807 found++;
4808 }
4809 if (found)
4810 return;
4811 /* otherwise we will clear the whole queue
4813 * by copying the empty outq at the end of this
4814 * function, but first clear the current entries
4815 * in the queue.
4816 */
4817 for (i = 0; i < q->nr; i++)
4818 diff_free_filepair(q->queue[i]);
4819 }
4820 else {
4821 /* Only the matching ones */
4822 for (i = 0; i < q->nr; i++) {
4823 struct diff_filepair *p = q->queue[i];
4824 if (match_filter(options, p))
4825 diff_q(&outq, p);
4826 else
4827 diff_free_filepair(p);
4828 }
4829 }
4830 free(q->queue);
4831 *q = outq;
4832}
4833/* Check whether two filespecs with the same mode and size are identical */
4835static int diff_filespec_is_identical(struct diff_filespec *one,
4836 struct diff_filespec *two)
4837{
4838 if (S_ISGITLINK(one->mode))
4839 return 0;
4840 if (diff_populate_filespec(one, 0))
4841 return 0;
4842 if (diff_populate_filespec(two, 0))
4843 return 0;
4844 return !memcmp(one->data, two->data, one->size);
4845}
4846static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4848{
4849 if (p->done_skip_stat_unmatch)
4850 return p->skip_stat_unmatch_result;
4851 p->done_skip_stat_unmatch = 1;
4853 p->skip_stat_unmatch_result = 0;
4854 /*
4855 * 1. Entries that come from stat info dirtiness
4856 * always have both sides (iow, not create/delete),
4857 * one side of the object name is unknown, with
4858 * the same mode and size. Keep the ones that
4859 * do not match these criteria. They have real
4860 * differences.
4861 *
4862 * 2. At this point, the file is known to be modified,
4863 * with the same mode and size, and the object
4864 * name of one side is unknown. Need to inspect
4865 * the identical contents.
4866 */
4867 if (!DIFF_FILE_VALID(p->one) || /* (1) */
4868 !DIFF_FILE_VALID(p->two) ||
4869 (p->one->oid_valid && p->two->oid_valid) ||
4870 (p->one->mode != p->two->mode) ||
4871 diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4872 diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4873 (p->one->size != p->two->size) ||
4874 !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4875 p->skip_stat_unmatch_result = 1;
4876 return p->skip_stat_unmatch_result;
4877}
4878static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4880{
4881 int i;
4882 struct diff_queue_struct *q = &diff_queued_diff;
4883 struct diff_queue_struct outq;
4884 DIFF_QUEUE_CLEAR(&outq);
4885 for (i = 0; i < q->nr; i++) {
4887 struct diff_filepair *p = q->queue[i];
4888 if (diff_filespec_check_stat_unmatch(p))
4890 diff_q(&outq, p);
4891 else {
4892 /*
4893 * The caller can subtract 1 from skip_stat_unmatch
4894 * to determine how many paths were dirty only
4895 * due to stat info mismatch.
4896 */
4897 if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4898 diffopt->skip_stat_unmatch++;
4899 diff_free_filepair(p);
4900 }
4901 }
4902 free(q->queue);
4903 *q = outq;
4904}
4905static int diffnamecmp(const void *a_, const void *b_)
4907{
4908 const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4909 const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4910 const char *name_a, *name_b;
4911 name_a = a->one ? a->one->path : a->two->path;
4913 name_b = b->one ? b->one->path : b->two->path;
4914 return strcmp(name_a, name_b);
4915}
4916void diffcore_fix_diff_index(struct diff_options *options)
4918{
4919 struct diff_queue_struct *q = &diff_queued_diff;
4920 qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4921}
4922void diffcore_std(struct diff_options *options)
4924{
4925 /* NOTE please keep the following in sync with diff_tree_combined() */
4926 if (options->skip_stat_unmatch)
4927 diffcore_skip_stat_unmatch(options);
4928 if (!options->found_follow) {
4929 /* See try_to_follow_renames() in tree-diff.c */
4930 if (options->break_opt != -1)
4931 diffcore_break(options->break_opt);
4932 if (options->detect_rename)
4933 diffcore_rename(options);
4934 if (options->break_opt != -1)
4935 diffcore_merge_broken();
4936 }
4937 if (options->pickaxe)
4938 diffcore_pickaxe(options);
4939 if (options->orderfile)
4940 diffcore_order(options->orderfile);
4941 if (!options->found_follow)
4942 /* See try_to_follow_renames() in tree-diff.c */
4943 diff_resolve_rename_copy();
4944 diffcore_apply_filter(options);
4945 if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4947 DIFF_OPT_SET(options, HAS_CHANGES);
4948 else
4949 DIFF_OPT_CLR(options, HAS_CHANGES);
4950 options->found_follow = 0;
4952}
4953int diff_result_code(struct diff_options *opt, int status)
4955{
4956 int result = 0;
4957 diff_warn_rename_limit("diff.renameLimit",
4959 opt->needed_rename_limit,
4960 opt->degraded_cc_to_c);
4961 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4962 !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4963 return status;
4964 if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4965 DIFF_OPT_TST(opt, HAS_CHANGES))
4966 result |= 01;
4967 if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4968 DIFF_OPT_TST(opt, CHECK_FAILED))
4969 result |= 02;
4970 return result;
4971}
4972int diff_can_quit_early(struct diff_options *opt)
4974{
4975 return (DIFF_OPT_TST(opt, QUICK) &&
4976 !opt->filter &&
4977 DIFF_OPT_TST(opt, HAS_CHANGES));
4978}
4979/*
4981 * Shall changes to this submodule be ignored?
4982 *
4983 * Submodule changes can be configured to be ignored separately for each path,
4984 * but that configuration can be overridden from the command line.
4985 */
4986static int is_submodule_ignored(const char *path, struct diff_options *options)
4987{
4988 int ignored = 0;
4989 unsigned orig_flags = options->flags;
4990 if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4991 set_diffopt_flags_from_submodule_config(options, path);
4992 if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4993 ignored = 1;
4994 options->flags = orig_flags;
4995 return ignored;
4996}
4997void diff_addremove(struct diff_options *options,
4999 int addremove, unsigned mode,
5000 const unsigned char *sha1,
5001 int sha1_valid,
5002 const char *concatpath, unsigned dirty_submodule)
5003{
5004 struct diff_filespec *one, *two;
5005 if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5007 return;
5008 /* This may look odd, but it is a preparation for
5010 * feeding "there are unchanged files which should
5011 * not produce diffs, but when you are doing copy
5012 * detection you would need them, so here they are"
5013 * entries to the diff-core. They will be prefixed
5014 * with something like '=' or '*' (I haven't decided
5015 * which but should not make any difference).
5016 * Feeding the same new and old to diff_change()
5017 * also has the same effect.
5018 * Before the final output happens, they are pruned after
5019 * merged into rename/copy pairs as appropriate.
5020 */
5021 if (DIFF_OPT_TST(options, REVERSE_DIFF))
5022 addremove = (addremove == '+' ? '-' :
5023 addremove == '-' ? '+' : addremove);
5024 if (options->prefix &&
5026 strncmp(concatpath, options->prefix, options->prefix_length))
5027 return;
5028 one = alloc_filespec(concatpath);
5030 two = alloc_filespec(concatpath);
5031 if (addremove != '+')
5033 fill_filespec(one, sha1, sha1_valid, mode);
5034 if (addremove != '-') {
5035 fill_filespec(two, sha1, sha1_valid, mode);
5036 two->dirty_submodule = dirty_submodule;
5037 }
5038 diff_queue(&diff_queued_diff, one, two);
5040 if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5041 DIFF_OPT_SET(options, HAS_CHANGES);
5042}
5043void diff_change(struct diff_options *options,
5045 unsigned old_mode, unsigned new_mode,
5046 const unsigned char *old_sha1,
5047 const unsigned char *new_sha1,
5048 int old_sha1_valid, int new_sha1_valid,
5049 const char *concatpath,
5050 unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5051{
5052 struct diff_filespec *one, *two;
5053 struct diff_filepair *p;
5054 if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5056 is_submodule_ignored(concatpath, options))
5057 return;
5058 if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5060 unsigned tmp;
5061 const unsigned char *tmp_c;
5062 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
5063 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
5064 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
5065 new_sha1_valid = tmp;
5066 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
5067 new_dirty_submodule = tmp;
5068 }
5069 if (options->prefix &&
5071 strncmp(concatpath, options->prefix, options->prefix_length))
5072 return;
5073 one = alloc_filespec(concatpath);
5075 two = alloc_filespec(concatpath);
5076 fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5077 fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5078 one->dirty_submodule = old_dirty_submodule;
5079 two->dirty_submodule = new_dirty_submodule;
5080 p = diff_queue(&diff_queued_diff, one, two);
5081 if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5083 return;
5084 if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5086 !diff_filespec_check_stat_unmatch(p))
5087 return;
5088 DIFF_OPT_SET(options, HAS_CHANGES);
5090}
5091struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5093{
5094 struct diff_filepair *pair;
5095 struct diff_filespec *one, *two;
5096 if (options->prefix &&
5098 strncmp(path, options->prefix, options->prefix_length))
5099 return NULL;
5100 one = alloc_filespec(path);
5102 two = alloc_filespec(path);
5103 pair = diff_queue(&diff_queued_diff, one, two);
5104 pair->is_unmerged = 1;
5105 return pair;
5106}
5107static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5109 size_t *outsize)
5110{
5111 struct diff_tempfile *temp;
5112 const char *argv[3];
5113 const char **arg = argv;
5114 struct child_process child = CHILD_PROCESS_INIT;
5115 struct strbuf buf = STRBUF_INIT;
5116 int err = 0;
5117 temp = prepare_temp_file(spec->path, spec);
5119 *arg++ = pgm;
5120 *arg++ = temp->name;
5121 *arg = NULL;
5122 child.use_shell = 1;
5124 child.argv = argv;
5125 child.out = -1;
5126 if (start_command(&child)) {
5127 remove_tempfile();
5128 return NULL;
5129 }
5130 if (strbuf_read(&buf, child.out, 0) < 0)
5132 err = error("error reading from textconv command '%s'", pgm);
5133 close(child.out);
5134 if (finish_command(&child) || err) {
5136 strbuf_release(&buf);
5137 remove_tempfile();
5138 return NULL;
5139 }
5140 remove_tempfile();
5141 return strbuf_detach(&buf, outsize);
5143}
5144size_t fill_textconv(struct userdiff_driver *driver,
5146 struct diff_filespec *df,
5147 char **outbuf)
5148{
5149 size_t size;
5150 if (!driver) {
5152 if (!DIFF_FILE_VALID(df)) {
5153 *outbuf = "";
5154 return 0;
5155 }
5156 if (diff_populate_filespec(df, 0))
5157 die("unable to read files to diff");
5158 *outbuf = df->data;
5159 return df->size;
5160 }
5161 if (!driver->textconv)
5163 die("BUG: fill_textconv called with non-textconv driver");
5164 if (driver->textconv_cache && df->oid_valid) {
5166 *outbuf = notes_cache_get(driver->textconv_cache,
5167 df->oid.hash,
5168 &size);
5169 if (*outbuf)
5170 return size;
5171 }
5172 *outbuf = run_textconv(driver->textconv, df, &size);
5174 if (!*outbuf)
5175 die("unable to read files to diff");
5176 if (driver->textconv_cache && df->oid_valid) {
5178 /* ignore errors, as we might be in a readonly repository */
5179 notes_cache_put(driver->textconv_cache, df->oid.hash, *outbuf,
5180 size);
5181 /*
5182 * we could save up changes and flush them all at the end,
5183 * but we would need an extra call after all diffing is done.
5184 * Since generating a cache entry is the slow path anyway,
5185 * this extra overhead probably isn't a big deal.
5186 */
5187 notes_cache_write(driver->textconv_cache);
5188 }
5189 return size;
5191}
5192void setup_diff_pager(struct diff_options *opt)
5194{
5195 /*
5196 * If the user asked for our exit code, then either they want --quiet
5197 * or --exit-code. We should definitely not bother with a pager in the
5198 * former case, as we will generate no output. Since we still properly
5199 * report our exit code even when a pager is run, we _could_ run a
5200 * pager with --exit-code. But since we have not done so historically,
5201 * and because it is easy to find people oneline advising "git diff
5202 * --exit-code" in hooks and other scripts, we do not do so.
5203 */
5204 if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5205 check_pager_config("diff") != 0)
5206 setup_pager();
5207}