Merge branch 'jt/namespaced-ls-refs-fix'
[git/git.git] / diff.c
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "config.h"
6 #include "tempfile.h"
7 #include "quote.h"
8 #include "diff.h"
9 #include "diffcore.h"
10 #include "delta.h"
11 #include "xdiff-interface.h"
12 #include "color.h"
13 #include "attr.h"
14 #include "run-command.h"
15 #include "utf8.h"
16 #include "object-store.h"
17 #include "userdiff.h"
18 #include "submodule-config.h"
19 #include "submodule.h"
20 #include "hashmap.h"
21 #include "ll-merge.h"
22 #include "string-list.h"
23 #include "argv-array.h"
24 #include "graph.h"
25 #include "packfile.h"
26 #include "help.h"
27
28 #ifdef NO_FAST_WORKING_DIRECTORY
29 #define FAST_WORKING_DIRECTORY 0
30 #else
31 #define FAST_WORKING_DIRECTORY 1
32 #endif
33
34 static int diff_detect_rename_default;
35 static int diff_indent_heuristic = 1;
36 static int diff_rename_limit_default = 400;
37 static int diff_suppress_blank_empty;
38 static int diff_use_color_default = -1;
39 static int diff_color_moved_default;
40 static int diff_color_moved_ws_default;
41 static int diff_context_default = 3;
42 static int diff_interhunk_context_default;
43 static const char *diff_word_regex_cfg;
44 static const char *external_diff_cmd_cfg;
45 static const char *diff_order_file_cfg;
46 int diff_auto_refresh_index = 1;
47 static int diff_mnemonic_prefix;
48 static int diff_no_prefix;
49 static int diff_stat_graph_width;
50 static int diff_dirstat_permille_default = 30;
51 static struct diff_options default_diff_options;
52 static long diff_algorithm;
53 static unsigned ws_error_highlight_default = WSEH_NEW;
54
55 static char diff_colors[][COLOR_MAXLEN] = {
56 GIT_COLOR_RESET,
57 GIT_COLOR_NORMAL, /* CONTEXT */
58 GIT_COLOR_BOLD, /* METAINFO */
59 GIT_COLOR_CYAN, /* FRAGINFO */
60 GIT_COLOR_RED, /* OLD */
61 GIT_COLOR_GREEN, /* NEW */
62 GIT_COLOR_YELLOW, /* COMMIT */
63 GIT_COLOR_BG_RED, /* WHITESPACE */
64 GIT_COLOR_NORMAL, /* FUNCINFO */
65 GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
66 GIT_COLOR_BOLD_BLUE, /* OLD_MOVED ALTERNATIVE */
67 GIT_COLOR_FAINT, /* OLD_MOVED_DIM */
68 GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
69 GIT_COLOR_BOLD_CYAN, /* NEW_MOVED */
70 GIT_COLOR_BOLD_YELLOW, /* NEW_MOVED ALTERNATIVE */
71 GIT_COLOR_FAINT, /* NEW_MOVED_DIM */
72 GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
73 GIT_COLOR_FAINT, /* CONTEXT_DIM */
74 GIT_COLOR_FAINT_RED, /* OLD_DIM */
75 GIT_COLOR_FAINT_GREEN, /* NEW_DIM */
76 GIT_COLOR_BOLD, /* CONTEXT_BOLD */
77 GIT_COLOR_BOLD_RED, /* OLD_BOLD */
78 GIT_COLOR_BOLD_GREEN, /* NEW_BOLD */
79 };
80
81 static const char *color_diff_slots[] = {
82 [DIFF_CONTEXT] = "context",
83 [DIFF_METAINFO] = "meta",
84 [DIFF_FRAGINFO] = "frag",
85 [DIFF_FILE_OLD] = "old",
86 [DIFF_FILE_NEW] = "new",
87 [DIFF_COMMIT] = "commit",
88 [DIFF_WHITESPACE] = "whitespace",
89 [DIFF_FUNCINFO] = "func",
90 [DIFF_FILE_OLD_MOVED] = "oldMoved",
91 [DIFF_FILE_OLD_MOVED_ALT] = "oldMovedAlternative",
92 [DIFF_FILE_OLD_MOVED_DIM] = "oldMovedDimmed",
93 [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
94 [DIFF_FILE_NEW_MOVED] = "newMoved",
95 [DIFF_FILE_NEW_MOVED_ALT] = "newMovedAlternative",
96 [DIFF_FILE_NEW_MOVED_DIM] = "newMovedDimmed",
97 [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
98 [DIFF_CONTEXT_DIM] = "contextDimmed",
99 [DIFF_FILE_OLD_DIM] = "oldDimmed",
100 [DIFF_FILE_NEW_DIM] = "newDimmed",
101 [DIFF_CONTEXT_BOLD] = "contextBold",
102 [DIFF_FILE_OLD_BOLD] = "oldBold",
103 [DIFF_FILE_NEW_BOLD] = "newBold",
104 };
105
106 static NORETURN void die_want_option(const char *option_name)
107 {
108 die(_("option '%s' requires a value"), option_name);
109 }
110
111 define_list_config_array_extra(color_diff_slots, {"plain"});
112
113 static int parse_diff_color_slot(const char *var)
114 {
115 if (!strcasecmp(var, "plain"))
116 return DIFF_CONTEXT;
117 return LOOKUP_CONFIG(color_diff_slots, var);
118 }
119
120 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
121 struct strbuf *errmsg)
122 {
123 char *params_copy = xstrdup(params_string);
124 struct string_list params = STRING_LIST_INIT_NODUP;
125 int ret = 0;
126 int i;
127
128 if (*params_copy)
129 string_list_split_in_place(&params, params_copy, ',', -1);
130 for (i = 0; i < params.nr; i++) {
131 const char *p = params.items[i].string;
132 if (!strcmp(p, "changes")) {
133 options->flags.dirstat_by_line = 0;
134 options->flags.dirstat_by_file = 0;
135 } else if (!strcmp(p, "lines")) {
136 options->flags.dirstat_by_line = 1;
137 options->flags.dirstat_by_file = 0;
138 } else if (!strcmp(p, "files")) {
139 options->flags.dirstat_by_line = 0;
140 options->flags.dirstat_by_file = 1;
141 } else if (!strcmp(p, "noncumulative")) {
142 options->flags.dirstat_cumulative = 0;
143 } else if (!strcmp(p, "cumulative")) {
144 options->flags.dirstat_cumulative = 1;
145 } else if (isdigit(*p)) {
146 char *end;
147 int permille = strtoul(p, &end, 10) * 10;
148 if (*end == '.' && isdigit(*++end)) {
149 /* only use first digit */
150 permille += *end - '0';
151 /* .. and ignore any further digits */
152 while (isdigit(*++end))
153 ; /* nothing */
154 }
155 if (!*end)
156 options->dirstat_permille = permille;
157 else {
158 strbuf_addf(errmsg, _(" Failed to parse dirstat cut-off percentage '%s'\n"),
159 p);
160 ret++;
161 }
162 } else {
163 strbuf_addf(errmsg, _(" Unknown dirstat parameter '%s'\n"), p);
164 ret++;
165 }
166
167 }
168 string_list_clear(&params, 0);
169 free(params_copy);
170 return ret;
171 }
172
173 static int parse_submodule_params(struct diff_options *options, const char *value)
174 {
175 if (!strcmp(value, "log"))
176 options->submodule_format = DIFF_SUBMODULE_LOG;
177 else if (!strcmp(value, "short"))
178 options->submodule_format = DIFF_SUBMODULE_SHORT;
179 else if (!strcmp(value, "diff"))
180 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
181 else
182 return -1;
183 return 0;
184 }
185
186 int git_config_rename(const char *var, const char *value)
187 {
188 if (!value)
189 return DIFF_DETECT_RENAME;
190 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
191 return DIFF_DETECT_COPY;
192 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
193 }
194
195 long parse_algorithm_value(const char *value)
196 {
197 if (!value)
198 return -1;
199 else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
200 return 0;
201 else if (!strcasecmp(value, "minimal"))
202 return XDF_NEED_MINIMAL;
203 else if (!strcasecmp(value, "patience"))
204 return XDF_PATIENCE_DIFF;
205 else if (!strcasecmp(value, "histogram"))
206 return XDF_HISTOGRAM_DIFF;
207 return -1;
208 }
209
210 static int parse_one_token(const char **arg, const char *token)
211 {
212 const char *rest;
213 if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
214 *arg = rest;
215 return 1;
216 }
217 return 0;
218 }
219
220 static int parse_ws_error_highlight(const char *arg)
221 {
222 const char *orig_arg = arg;
223 unsigned val = 0;
224
225 while (*arg) {
226 if (parse_one_token(&arg, "none"))
227 val = 0;
228 else if (parse_one_token(&arg, "default"))
229 val = WSEH_NEW;
230 else if (parse_one_token(&arg, "all"))
231 val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
232 else if (parse_one_token(&arg, "new"))
233 val |= WSEH_NEW;
234 else if (parse_one_token(&arg, "old"))
235 val |= WSEH_OLD;
236 else if (parse_one_token(&arg, "context"))
237 val |= WSEH_CONTEXT;
238 else {
239 return -1 - (int)(arg - orig_arg);
240 }
241 if (*arg)
242 arg++;
243 }
244 return val;
245 }
246
247 /*
248 * These are to give UI layer defaults.
249 * The core-level commands such as git-diff-files should
250 * never be affected by the setting of diff.renames
251 * the user happens to have in the configuration file.
252 */
253 void init_diff_ui_defaults(void)
254 {
255 diff_detect_rename_default = DIFF_DETECT_RENAME;
256 }
257
258 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
259 {
260 if (!strcmp(var, "diff.indentheuristic"))
261 diff_indent_heuristic = git_config_bool(var, value);
262 return 0;
263 }
264
265 static int parse_color_moved(const char *arg)
266 {
267 switch (git_parse_maybe_bool(arg)) {
268 case 0:
269 return COLOR_MOVED_NO;
270 case 1:
271 return COLOR_MOVED_DEFAULT;
272 default:
273 break;
274 }
275
276 if (!strcmp(arg, "no"))
277 return COLOR_MOVED_NO;
278 else if (!strcmp(arg, "plain"))
279 return COLOR_MOVED_PLAIN;
280 else if (!strcmp(arg, "blocks"))
281 return COLOR_MOVED_BLOCKS;
282 else if (!strcmp(arg, "zebra"))
283 return COLOR_MOVED_ZEBRA;
284 else if (!strcmp(arg, "default"))
285 return COLOR_MOVED_DEFAULT;
286 else if (!strcmp(arg, "dimmed-zebra"))
287 return COLOR_MOVED_ZEBRA_DIM;
288 else if (!strcmp(arg, "dimmed_zebra"))
289 return COLOR_MOVED_ZEBRA_DIM;
290 else
291 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
292 }
293
294 static unsigned parse_color_moved_ws(const char *arg)
295 {
296 int ret = 0;
297 struct string_list l = STRING_LIST_INIT_DUP;
298 struct string_list_item *i;
299
300 string_list_split(&l, arg, ',', -1);
301
302 for_each_string_list_item(i, &l) {
303 struct strbuf sb = STRBUF_INIT;
304 strbuf_addstr(&sb, i->string);
305 strbuf_trim(&sb);
306
307 if (!strcmp(sb.buf, "no"))
308 ret = 0;
309 else if (!strcmp(sb.buf, "ignore-space-change"))
310 ret |= XDF_IGNORE_WHITESPACE_CHANGE;
311 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
312 ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
313 else if (!strcmp(sb.buf, "ignore-all-space"))
314 ret |= XDF_IGNORE_WHITESPACE;
315 else if (!strcmp(sb.buf, "allow-indentation-change"))
316 ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
317 else {
318 ret |= COLOR_MOVED_WS_ERROR;
319 error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
320 }
321
322 strbuf_release(&sb);
323 }
324
325 if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
326 (ret & XDF_WHITESPACE_FLAGS)) {
327 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
328 ret |= COLOR_MOVED_WS_ERROR;
329 }
330
331 string_list_clear(&l, 0);
332
333 return ret;
334 }
335
336 int git_diff_ui_config(const char *var, const char *value, void *cb)
337 {
338 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
339 diff_use_color_default = git_config_colorbool(var, value);
340 return 0;
341 }
342 if (!strcmp(var, "diff.colormoved")) {
343 int cm = parse_color_moved(value);
344 if (cm < 0)
345 return -1;
346 diff_color_moved_default = cm;
347 return 0;
348 }
349 if (!strcmp(var, "diff.colormovedws")) {
350 unsigned cm = parse_color_moved_ws(value);
351 if (cm & COLOR_MOVED_WS_ERROR)
352 return -1;
353 diff_color_moved_ws_default = cm;
354 return 0;
355 }
356 if (!strcmp(var, "diff.context")) {
357 diff_context_default = git_config_int(var, value);
358 if (diff_context_default < 0)
359 return -1;
360 return 0;
361 }
362 if (!strcmp(var, "diff.interhunkcontext")) {
363 diff_interhunk_context_default = git_config_int(var, value);
364 if (diff_interhunk_context_default < 0)
365 return -1;
366 return 0;
367 }
368 if (!strcmp(var, "diff.renames")) {
369 diff_detect_rename_default = git_config_rename(var, value);
370 return 0;
371 }
372 if (!strcmp(var, "diff.autorefreshindex")) {
373 diff_auto_refresh_index = git_config_bool(var, value);
374 return 0;
375 }
376 if (!strcmp(var, "diff.mnemonicprefix")) {
377 diff_mnemonic_prefix = git_config_bool(var, value);
378 return 0;
379 }
380 if (!strcmp(var, "diff.noprefix")) {
381 diff_no_prefix = git_config_bool(var, value);
382 return 0;
383 }
384 if (!strcmp(var, "diff.statgraphwidth")) {
385 diff_stat_graph_width = git_config_int(var, value);
386 return 0;
387 }
388 if (!strcmp(var, "diff.external"))
389 return git_config_string(&external_diff_cmd_cfg, var, value);
390 if (!strcmp(var, "diff.wordregex"))
391 return git_config_string(&diff_word_regex_cfg, var, value);
392 if (!strcmp(var, "diff.orderfile"))
393 return git_config_pathname(&diff_order_file_cfg, var, value);
394
395 if (!strcmp(var, "diff.ignoresubmodules"))
396 handle_ignore_submodules_arg(&default_diff_options, value);
397
398 if (!strcmp(var, "diff.submodule")) {
399 if (parse_submodule_params(&default_diff_options, value))
400 warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
401 value);
402 return 0;
403 }
404
405 if (!strcmp(var, "diff.algorithm")) {
406 diff_algorithm = parse_algorithm_value(value);
407 if (diff_algorithm < 0)
408 return -1;
409 return 0;
410 }
411
412 if (!strcmp(var, "diff.wserrorhighlight")) {
413 int val = parse_ws_error_highlight(value);
414 if (val < 0)
415 return -1;
416 ws_error_highlight_default = val;
417 return 0;
418 }
419
420 if (git_color_config(var, value, cb) < 0)
421 return -1;
422
423 return git_diff_basic_config(var, value, cb);
424 }
425
426 int git_diff_basic_config(const char *var, const char *value, void *cb)
427 {
428 const char *name;
429
430 if (!strcmp(var, "diff.renamelimit")) {
431 diff_rename_limit_default = git_config_int(var, value);
432 return 0;
433 }
434
435 if (userdiff_config(var, value) < 0)
436 return -1;
437
438 if (skip_prefix(var, "diff.color.", &name) ||
439 skip_prefix(var, "color.diff.", &name)) {
440 int slot = parse_diff_color_slot(name);
441 if (slot < 0)
442 return 0;
443 if (!value)
444 return config_error_nonbool(var);
445 return color_parse(value, diff_colors[slot]);
446 }
447
448 /* like GNU diff's --suppress-blank-empty option */
449 if (!strcmp(var, "diff.suppressblankempty") ||
450 /* for backwards compatibility */
451 !strcmp(var, "diff.suppress-blank-empty")) {
452 diff_suppress_blank_empty = git_config_bool(var, value);
453 return 0;
454 }
455
456 if (!strcmp(var, "diff.dirstat")) {
457 struct strbuf errmsg = STRBUF_INIT;
458 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
459 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
460 warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
461 errmsg.buf);
462 strbuf_release(&errmsg);
463 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
464 return 0;
465 }
466
467 if (git_diff_heuristic_config(var, value, cb) < 0)
468 return -1;
469
470 return git_default_config(var, value, cb);
471 }
472
473 static char *quote_two(const char *one, const char *two)
474 {
475 int need_one = quote_c_style(one, NULL, NULL, 1);
476 int need_two = quote_c_style(two, NULL, NULL, 1);
477 struct strbuf res = STRBUF_INIT;
478
479 if (need_one + need_two) {
480 strbuf_addch(&res, '"');
481 quote_c_style(one, &res, NULL, 1);
482 quote_c_style(two, &res, NULL, 1);
483 strbuf_addch(&res, '"');
484 } else {
485 strbuf_addstr(&res, one);
486 strbuf_addstr(&res, two);
487 }
488 return strbuf_detach(&res, NULL);
489 }
490
491 static const char *external_diff(void)
492 {
493 static const char *external_diff_cmd = NULL;
494 static int done_preparing = 0;
495
496 if (done_preparing)
497 return external_diff_cmd;
498 external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
499 if (!external_diff_cmd)
500 external_diff_cmd = external_diff_cmd_cfg;
501 done_preparing = 1;
502 return external_diff_cmd;
503 }
504
505 /*
506 * Keep track of files used for diffing. Sometimes such an entry
507 * refers to a temporary file, sometimes to an existing file, and
508 * sometimes to "/dev/null".
509 */
510 static struct diff_tempfile {
511 /*
512 * filename external diff should read from, or NULL if this
513 * entry is currently not in use:
514 */
515 const char *name;
516
517 char hex[GIT_MAX_HEXSZ + 1];
518 char mode[10];
519
520 /*
521 * If this diff_tempfile instance refers to a temporary file,
522 * this tempfile object is used to manage its lifetime.
523 */
524 struct tempfile *tempfile;
525 } diff_temp[2];
526
527 struct emit_callback {
528 int color_diff;
529 unsigned ws_rule;
530 int blank_at_eof_in_preimage;
531 int blank_at_eof_in_postimage;
532 int lno_in_preimage;
533 int lno_in_postimage;
534 const char **label_path;
535 struct diff_words_data *diff_words;
536 struct diff_options *opt;
537 struct strbuf *header;
538 };
539
540 static int count_lines(const char *data, int size)
541 {
542 int count, ch, completely_empty = 1, nl_just_seen = 0;
543 count = 0;
544 while (0 < size--) {
545 ch = *data++;
546 if (ch == '\n') {
547 count++;
548 nl_just_seen = 1;
549 completely_empty = 0;
550 }
551 else {
552 nl_just_seen = 0;
553 completely_empty = 0;
554 }
555 }
556 if (completely_empty)
557 return 0;
558 if (!nl_just_seen)
559 count++; /* no trailing newline */
560 return count;
561 }
562
563 static int fill_mmfile(struct repository *r, mmfile_t *mf,
564 struct diff_filespec *one)
565 {
566 if (!DIFF_FILE_VALID(one)) {
567 mf->ptr = (char *)""; /* does not matter */
568 mf->size = 0;
569 return 0;
570 }
571 else if (diff_populate_filespec(r, one, 0))
572 return -1;
573
574 mf->ptr = one->data;
575 mf->size = one->size;
576 return 0;
577 }
578
579 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
580 static unsigned long diff_filespec_size(struct repository *r,
581 struct diff_filespec *one)
582 {
583 if (!DIFF_FILE_VALID(one))
584 return 0;
585 diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
586 return one->size;
587 }
588
589 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
590 {
591 char *ptr = mf->ptr;
592 long size = mf->size;
593 int cnt = 0;
594
595 if (!size)
596 return cnt;
597 ptr += size - 1; /* pointing at the very end */
598 if (*ptr != '\n')
599 ; /* incomplete line */
600 else
601 ptr--; /* skip the last LF */
602 while (mf->ptr < ptr) {
603 char *prev_eol;
604 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
605 if (*prev_eol == '\n')
606 break;
607 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
608 break;
609 cnt++;
610 ptr = prev_eol - 1;
611 }
612 return cnt;
613 }
614
615 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
616 struct emit_callback *ecbdata)
617 {
618 int l1, l2, at;
619 unsigned ws_rule = ecbdata->ws_rule;
620 l1 = count_trailing_blank(mf1, ws_rule);
621 l2 = count_trailing_blank(mf2, ws_rule);
622 if (l2 <= l1) {
623 ecbdata->blank_at_eof_in_preimage = 0;
624 ecbdata->blank_at_eof_in_postimage = 0;
625 return;
626 }
627 at = count_lines(mf1->ptr, mf1->size);
628 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
629
630 at = count_lines(mf2->ptr, mf2->size);
631 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
632 }
633
634 static void emit_line_0(struct diff_options *o,
635 const char *set_sign, const char *set, unsigned reverse, const char *reset,
636 int first, const char *line, int len)
637 {
638 int has_trailing_newline, has_trailing_carriage_return;
639 int needs_reset = 0; /* at the end of the line */
640 FILE *file = o->file;
641
642 fputs(diff_line_prefix(o), file);
643
644 has_trailing_newline = (len > 0 && line[len-1] == '\n');
645 if (has_trailing_newline)
646 len--;
647
648 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
649 if (has_trailing_carriage_return)
650 len--;
651
652 if (!len && !first)
653 goto end_of_line;
654
655 if (reverse && want_color(o->use_color)) {
656 fputs(GIT_COLOR_REVERSE, file);
657 needs_reset = 1;
658 }
659
660 if (set_sign) {
661 fputs(set_sign, file);
662 needs_reset = 1;
663 }
664
665 if (first)
666 fputc(first, file);
667
668 if (!len)
669 goto end_of_line;
670
671 if (set) {
672 if (set_sign && set != set_sign)
673 fputs(reset, file);
674 fputs(set, file);
675 needs_reset = 1;
676 }
677 fwrite(line, len, 1, file);
678 needs_reset = 1; /* 'line' may contain color codes. */
679
680 end_of_line:
681 if (needs_reset)
682 fputs(reset, file);
683 if (has_trailing_carriage_return)
684 fputc('\r', file);
685 if (has_trailing_newline)
686 fputc('\n', file);
687 }
688
689 static void emit_line(struct diff_options *o, const char *set, const char *reset,
690 const char *line, int len)
691 {
692 emit_line_0(o, set, NULL, 0, reset, 0, line, len);
693 }
694
695 enum diff_symbol {
696 DIFF_SYMBOL_BINARY_DIFF_HEADER,
697 DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
698 DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
699 DIFF_SYMBOL_BINARY_DIFF_BODY,
700 DIFF_SYMBOL_BINARY_DIFF_FOOTER,
701 DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
702 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
703 DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
704 DIFF_SYMBOL_STATS_LINE,
705 DIFF_SYMBOL_WORD_DIFF,
706 DIFF_SYMBOL_STAT_SEP,
707 DIFF_SYMBOL_SUMMARY,
708 DIFF_SYMBOL_SUBMODULE_ADD,
709 DIFF_SYMBOL_SUBMODULE_DEL,
710 DIFF_SYMBOL_SUBMODULE_UNTRACKED,
711 DIFF_SYMBOL_SUBMODULE_MODIFIED,
712 DIFF_SYMBOL_SUBMODULE_HEADER,
713 DIFF_SYMBOL_SUBMODULE_ERROR,
714 DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
715 DIFF_SYMBOL_REWRITE_DIFF,
716 DIFF_SYMBOL_BINARY_FILES,
717 DIFF_SYMBOL_HEADER,
718 DIFF_SYMBOL_FILEPAIR_PLUS,
719 DIFF_SYMBOL_FILEPAIR_MINUS,
720 DIFF_SYMBOL_WORDS_PORCELAIN,
721 DIFF_SYMBOL_WORDS,
722 DIFF_SYMBOL_CONTEXT,
723 DIFF_SYMBOL_CONTEXT_INCOMPLETE,
724 DIFF_SYMBOL_PLUS,
725 DIFF_SYMBOL_MINUS,
726 DIFF_SYMBOL_NO_LF_EOF,
727 DIFF_SYMBOL_CONTEXT_FRAGINFO,
728 DIFF_SYMBOL_CONTEXT_MARKER,
729 DIFF_SYMBOL_SEPARATOR
730 };
731 /*
732 * Flags for content lines:
733 * 0..12 are whitespace rules
734 * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
735 * 16 is marking if the line is blank at EOF
736 */
737 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF (1<<16)
738 #define DIFF_SYMBOL_MOVED_LINE (1<<17)
739 #define DIFF_SYMBOL_MOVED_LINE_ALT (1<<18)
740 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING (1<<19)
741 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
742
743 /*
744 * This struct is used when we need to buffer the output of the diff output.
745 *
746 * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
747 * into the pre/post image file. This pointer could be a union with the
748 * line pointer. By storing an offset into the file instead of the literal line,
749 * we can decrease the memory footprint for the buffered output. At first we
750 * may want to only have indirection for the content lines, but we could also
751 * enhance the state for emitting prefabricated lines, e.g. the similarity
752 * score line or hunk/file headers would only need to store a number or path
753 * and then the output can be constructed later on depending on state.
754 */
755 struct emitted_diff_symbol {
756 const char *line;
757 int len;
758 int flags;
759 int indent_off; /* Offset to first non-whitespace character */
760 int indent_width; /* The visual width of the indentation */
761 enum diff_symbol s;
762 };
763 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
764
765 struct emitted_diff_symbols {
766 struct emitted_diff_symbol *buf;
767 int nr, alloc;
768 };
769 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
770
771 static void append_emitted_diff_symbol(struct diff_options *o,
772 struct emitted_diff_symbol *e)
773 {
774 struct emitted_diff_symbol *f;
775
776 ALLOC_GROW(o->emitted_symbols->buf,
777 o->emitted_symbols->nr + 1,
778 o->emitted_symbols->alloc);
779 f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
780
781 memcpy(f, e, sizeof(struct emitted_diff_symbol));
782 f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
783 }
784
785 struct moved_entry {
786 struct hashmap_entry ent;
787 const struct emitted_diff_symbol *es;
788 struct moved_entry *next_line;
789 };
790
791 struct moved_block {
792 struct moved_entry *match;
793 int wsd; /* The whitespace delta of this block */
794 };
795
796 static void moved_block_clear(struct moved_block *b)
797 {
798 memset(b, 0, sizeof(*b));
799 }
800
801 #define INDENT_BLANKLINE INT_MIN
802
803 static void fill_es_indent_data(struct emitted_diff_symbol *es)
804 {
805 unsigned int off = 0, i;
806 int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
807 const char *s = es->line;
808 const int len = es->len;
809
810 /* skip any \v \f \r at start of indentation */
811 while (s[off] == '\f' || s[off] == '\v' ||
812 (s[off] == '\r' && off < len - 1))
813 off++;
814
815 /* calculate the visual width of indentation */
816 while(1) {
817 if (s[off] == ' ') {
818 width++;
819 off++;
820 } else if (s[off] == '\t') {
821 width += tab_width - (width % tab_width);
822 while (s[++off] == '\t')
823 width += tab_width;
824 } else {
825 break;
826 }
827 }
828
829 /* check if this line is blank */
830 for (i = off; i < len; i++)
831 if (!isspace(s[i]))
832 break;
833
834 if (i == len) {
835 es->indent_width = INDENT_BLANKLINE;
836 es->indent_off = len;
837 } else {
838 es->indent_off = off;
839 es->indent_width = width;
840 }
841 }
842
843 static int compute_ws_delta(const struct emitted_diff_symbol *a,
844 const struct emitted_diff_symbol *b,
845 int *out)
846 {
847 int a_len = a->len,
848 b_len = b->len,
849 a_off = a->indent_off,
850 a_width = a->indent_width,
851 b_off = b->indent_off,
852 b_width = b->indent_width;
853 int delta;
854
855 if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
856 *out = INDENT_BLANKLINE;
857 return 1;
858 }
859
860 if (a->s == DIFF_SYMBOL_PLUS)
861 delta = a_width - b_width;
862 else
863 delta = b_width - a_width;
864
865 if (a_len - a_off != b_len - b_off ||
866 memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
867 return 0;
868
869 *out = delta;
870
871 return 1;
872 }
873
874 static int cmp_in_block_with_wsd(const struct diff_options *o,
875 const struct moved_entry *cur,
876 const struct moved_entry *match,
877 struct moved_block *pmb,
878 int n)
879 {
880 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
881 int al = cur->es->len, bl = match->es->len, cl = l->len;
882 const char *a = cur->es->line,
883 *b = match->es->line,
884 *c = l->line;
885 int a_off = cur->es->indent_off,
886 a_width = cur->es->indent_width,
887 c_off = l->indent_off,
888 c_width = l->indent_width;
889 int delta;
890
891 /*
892 * We need to check if 'cur' is equal to 'match'. As those
893 * are from the same (+/-) side, we do not need to adjust for
894 * indent changes. However these were found using fuzzy
895 * matching so we do have to check if they are equal. Here we
896 * just check the lengths. We delay calling memcmp() to check
897 * the contents until later as if the length comparison for a
898 * and c fails we can avoid the call all together.
899 */
900 if (al != bl)
901 return 1;
902
903 /* If 'l' and 'cur' are both blank then they match. */
904 if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
905 return 0;
906
907 /*
908 * The indent changes of the block are known and stored in pmb->wsd;
909 * however we need to check if the indent changes of the current line
910 * match those of the current block and that the text of 'l' and 'cur'
911 * after the indentation match.
912 */
913 if (cur->es->s == DIFF_SYMBOL_PLUS)
914 delta = a_width - c_width;
915 else
916 delta = c_width - a_width;
917
918 /*
919 * If the previous lines of this block were all blank then set its
920 * whitespace delta.
921 */
922 if (pmb->wsd == INDENT_BLANKLINE)
923 pmb->wsd = delta;
924
925 return !(delta == pmb->wsd && al - a_off == cl - c_off &&
926 !memcmp(a, b, al) && !
927 memcmp(a + a_off, c + c_off, al - a_off));
928 }
929
930 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
931 const void *entry,
932 const void *entry_or_key,
933 const void *keydata)
934 {
935 const struct diff_options *diffopt = hashmap_cmp_fn_data;
936 const struct moved_entry *a = entry;
937 const struct moved_entry *b = entry_or_key;
938 unsigned flags = diffopt->color_moved_ws_handling
939 & XDF_WHITESPACE_FLAGS;
940
941 if (diffopt->color_moved_ws_handling &
942 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
943 /*
944 * As there is not specific white space config given,
945 * we'd need to check for a new block, so ignore all
946 * white space. The setup of the white space
947 * configuration for the next block is done else where
948 */
949 flags |= XDF_IGNORE_WHITESPACE;
950
951 return !xdiff_compare_lines(a->es->line, a->es->len,
952 b->es->line, b->es->len,
953 flags);
954 }
955
956 static struct moved_entry *prepare_entry(struct diff_options *o,
957 int line_no)
958 {
959 struct moved_entry *ret = xmalloc(sizeof(*ret));
960 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
961 unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
962
963 ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
964 ret->es = l;
965 ret->next_line = NULL;
966
967 return ret;
968 }
969
970 static void add_lines_to_move_detection(struct diff_options *o,
971 struct hashmap *add_lines,
972 struct hashmap *del_lines)
973 {
974 struct moved_entry *prev_line = NULL;
975
976 int n;
977 for (n = 0; n < o->emitted_symbols->nr; n++) {
978 struct hashmap *hm;
979 struct moved_entry *key;
980
981 switch (o->emitted_symbols->buf[n].s) {
982 case DIFF_SYMBOL_PLUS:
983 hm = add_lines;
984 break;
985 case DIFF_SYMBOL_MINUS:
986 hm = del_lines;
987 break;
988 default:
989 prev_line = NULL;
990 continue;
991 }
992
993 if (o->color_moved_ws_handling &
994 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
995 fill_es_indent_data(&o->emitted_symbols->buf[n]);
996 key = prepare_entry(o, n);
997 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
998 prev_line->next_line = key;
999
1000 hashmap_add(hm, key);
1001 prev_line = key;
1002 }
1003 }
1004
1005 static void pmb_advance_or_null(struct diff_options *o,
1006 struct moved_entry *match,
1007 struct hashmap *hm,
1008 struct moved_block *pmb,
1009 int pmb_nr)
1010 {
1011 int i;
1012 for (i = 0; i < pmb_nr; i++) {
1013 struct moved_entry *prev = pmb[i].match;
1014 struct moved_entry *cur = (prev && prev->next_line) ?
1015 prev->next_line : NULL;
1016 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
1017 pmb[i].match = cur;
1018 } else {
1019 pmb[i].match = NULL;
1020 }
1021 }
1022 }
1023
1024 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1025 struct moved_entry *match,
1026 struct hashmap *hm,
1027 struct moved_block *pmb,
1028 int pmb_nr, int n)
1029 {
1030 int i;
1031 char *got_match = xcalloc(1, pmb_nr);
1032
1033 for (; match; match = hashmap_get_next(hm, match)) {
1034 for (i = 0; i < pmb_nr; i++) {
1035 struct moved_entry *prev = pmb[i].match;
1036 struct moved_entry *cur = (prev && prev->next_line) ?
1037 prev->next_line : NULL;
1038 if (!cur)
1039 continue;
1040 if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1041 got_match[i] |= 1;
1042 }
1043 }
1044
1045 for (i = 0; i < pmb_nr; i++) {
1046 if (got_match[i]) {
1047 /* Advance to the next line */
1048 pmb[i].match = pmb[i].match->next_line;
1049 } else {
1050 moved_block_clear(&pmb[i]);
1051 }
1052 }
1053
1054 free(got_match);
1055 }
1056
1057 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1058 int pmb_nr)
1059 {
1060 int lp, rp;
1061
1062 /* Shrink the set of potential block to the remaining running */
1063 for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1064 while (lp < pmb_nr && pmb[lp].match)
1065 lp++;
1066 /* lp points at the first NULL now */
1067
1068 while (rp > -1 && !pmb[rp].match)
1069 rp--;
1070 /* rp points at the last non-NULL */
1071
1072 if (lp < pmb_nr && rp > -1 && lp < rp) {
1073 pmb[lp] = pmb[rp];
1074 memset(&pmb[rp], 0, sizeof(pmb[rp]));
1075 rp--;
1076 lp++;
1077 }
1078 }
1079
1080 /* Remember the number of running sets */
1081 return rp + 1;
1082 }
1083
1084 /*
1085 * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1086 *
1087 * Otherwise, if the last block has fewer alphanumeric characters than
1088 * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1089 * that block.
1090 *
1091 * The last block consists of the (n - block_length)'th line up to but not
1092 * including the nth line.
1093 *
1094 * Returns 0 if the last block is empty or is unset by this function, non zero
1095 * otherwise.
1096 *
1097 * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1098 * Think of a way to unify them.
1099 */
1100 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1101 {
1102 int i, alnum_count = 0;
1103 if (o->color_moved == COLOR_MOVED_PLAIN)
1104 return block_length;
1105 for (i = 1; i < block_length + 1; i++) {
1106 const char *c = o->emitted_symbols->buf[n - i].line;
1107 for (; *c; c++) {
1108 if (!isalnum(*c))
1109 continue;
1110 alnum_count++;
1111 if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1112 return 1;
1113 }
1114 }
1115 for (i = 1; i < block_length + 1; i++)
1116 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1117 return 0;
1118 }
1119
1120 /* Find blocks of moved code, delegate actual coloring decision to helper */
1121 static void mark_color_as_moved(struct diff_options *o,
1122 struct hashmap *add_lines,
1123 struct hashmap *del_lines)
1124 {
1125 struct moved_block *pmb = NULL; /* potentially moved blocks */
1126 int pmb_nr = 0, pmb_alloc = 0;
1127 int n, flipped_block = 0, block_length = 0;
1128
1129
1130 for (n = 0; n < o->emitted_symbols->nr; n++) {
1131 struct hashmap *hm = NULL;
1132 struct moved_entry *key;
1133 struct moved_entry *match = NULL;
1134 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1135 enum diff_symbol last_symbol = 0;
1136
1137 switch (l->s) {
1138 case DIFF_SYMBOL_PLUS:
1139 hm = del_lines;
1140 key = prepare_entry(o, n);
1141 match = hashmap_get(hm, key, NULL);
1142 free(key);
1143 break;
1144 case DIFF_SYMBOL_MINUS:
1145 hm = add_lines;
1146 key = prepare_entry(o, n);
1147 match = hashmap_get(hm, key, NULL);
1148 free(key);
1149 break;
1150 default:
1151 flipped_block = 0;
1152 }
1153
1154 if (!match) {
1155 int i;
1156
1157 adjust_last_block(o, n, block_length);
1158 for(i = 0; i < pmb_nr; i++)
1159 moved_block_clear(&pmb[i]);
1160 pmb_nr = 0;
1161 block_length = 0;
1162 flipped_block = 0;
1163 last_symbol = l->s;
1164 continue;
1165 }
1166
1167 if (o->color_moved == COLOR_MOVED_PLAIN) {
1168 last_symbol = l->s;
1169 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1170 continue;
1171 }
1172
1173 if (o->color_moved_ws_handling &
1174 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1175 pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1176 else
1177 pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1178
1179 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1180
1181 if (pmb_nr == 0) {
1182 /*
1183 * The current line is the start of a new block.
1184 * Setup the set of potential blocks.
1185 */
1186 for (; match; match = hashmap_get_next(hm, match)) {
1187 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1188 if (o->color_moved_ws_handling &
1189 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1190 if (compute_ws_delta(l, match->es,
1191 &pmb[pmb_nr].wsd))
1192 pmb[pmb_nr++].match = match;
1193 } else {
1194 pmb[pmb_nr].wsd = 0;
1195 pmb[pmb_nr++].match = match;
1196 }
1197 }
1198
1199 if (adjust_last_block(o, n, block_length) &&
1200 pmb_nr && last_symbol != l->s)
1201 flipped_block = (flipped_block + 1) % 2;
1202 else
1203 flipped_block = 0;
1204
1205 block_length = 0;
1206 }
1207
1208 if (pmb_nr) {
1209 block_length++;
1210 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1211 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1212 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1213 }
1214 last_symbol = l->s;
1215 }
1216 adjust_last_block(o, n, block_length);
1217
1218 for(n = 0; n < pmb_nr; n++)
1219 moved_block_clear(&pmb[n]);
1220 free(pmb);
1221 }
1222
1223 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1224 (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1225 static void dim_moved_lines(struct diff_options *o)
1226 {
1227 int n;
1228 for (n = 0; n < o->emitted_symbols->nr; n++) {
1229 struct emitted_diff_symbol *prev = (n != 0) ?
1230 &o->emitted_symbols->buf[n - 1] : NULL;
1231 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1232 struct emitted_diff_symbol *next =
1233 (n < o->emitted_symbols->nr - 1) ?
1234 &o->emitted_symbols->buf[n + 1] : NULL;
1235
1236 /* Not a plus or minus line? */
1237 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1238 continue;
1239
1240 /* Not a moved line? */
1241 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1242 continue;
1243
1244 /*
1245 * If prev or next are not a plus or minus line,
1246 * pretend they don't exist
1247 */
1248 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1249 prev->s != DIFF_SYMBOL_MINUS)
1250 prev = NULL;
1251 if (next && next->s != DIFF_SYMBOL_PLUS &&
1252 next->s != DIFF_SYMBOL_MINUS)
1253 next = NULL;
1254
1255 /* Inside a block? */
1256 if ((prev &&
1257 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1258 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1259 (next &&
1260 (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1261 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1262 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1263 continue;
1264 }
1265
1266 /* Check if we are at an interesting bound: */
1267 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1268 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1269 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1270 continue;
1271 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1272 (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1273 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1274 continue;
1275
1276 /*
1277 * The boundary to prev and next are not interesting,
1278 * so this line is not interesting as a whole
1279 */
1280 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1281 }
1282 }
1283
1284 static void emit_line_ws_markup(struct diff_options *o,
1285 const char *set_sign, const char *set,
1286 const char *reset,
1287 int sign_index, const char *line, int len,
1288 unsigned ws_rule, int blank_at_eof)
1289 {
1290 const char *ws = NULL;
1291 int sign = o->output_indicators[sign_index];
1292
1293 if (o->ws_error_highlight & ws_rule) {
1294 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1295 if (!*ws)
1296 ws = NULL;
1297 }
1298
1299 if (!ws && !set_sign)
1300 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1301 else if (!ws) {
1302 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1303 } else if (blank_at_eof)
1304 /* Blank line at EOF - paint '+' as well */
1305 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1306 else {
1307 /* Emit just the prefix, then the rest. */
1308 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1309 sign, "", 0);
1310 ws_check_emit(line, len, ws_rule,
1311 o->file, set, reset, ws);
1312 }
1313 }
1314
1315 static void emit_diff_symbol_from_struct(struct diff_options *o,
1316 struct emitted_diff_symbol *eds)
1317 {
1318 static const char *nneof = " No newline at end of file\n";
1319 const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1320 struct strbuf sb = STRBUF_INIT;
1321
1322 enum diff_symbol s = eds->s;
1323 const char *line = eds->line;
1324 int len = eds->len;
1325 unsigned flags = eds->flags;
1326
1327 switch (s) {
1328 case DIFF_SYMBOL_NO_LF_EOF:
1329 context = diff_get_color_opt(o, DIFF_CONTEXT);
1330 reset = diff_get_color_opt(o, DIFF_RESET);
1331 putc('\n', o->file);
1332 emit_line_0(o, context, NULL, 0, reset, '\\',
1333 nneof, strlen(nneof));
1334 break;
1335 case DIFF_SYMBOL_SUBMODULE_HEADER:
1336 case DIFF_SYMBOL_SUBMODULE_ERROR:
1337 case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1338 case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1339 case DIFF_SYMBOL_SUMMARY:
1340 case DIFF_SYMBOL_STATS_LINE:
1341 case DIFF_SYMBOL_BINARY_DIFF_BODY:
1342 case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1343 emit_line(o, "", "", line, len);
1344 break;
1345 case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1346 case DIFF_SYMBOL_CONTEXT_MARKER:
1347 context = diff_get_color_opt(o, DIFF_CONTEXT);
1348 reset = diff_get_color_opt(o, DIFF_RESET);
1349 emit_line(o, context, reset, line, len);
1350 break;
1351 case DIFF_SYMBOL_SEPARATOR:
1352 fprintf(o->file, "%s%c",
1353 diff_line_prefix(o),
1354 o->line_termination);
1355 break;
1356 case DIFF_SYMBOL_CONTEXT:
1357 set = diff_get_color_opt(o, DIFF_CONTEXT);
1358 reset = diff_get_color_opt(o, DIFF_RESET);
1359 set_sign = NULL;
1360 if (o->flags.dual_color_diffed_diffs) {
1361 char c = !len ? 0 : line[0];
1362
1363 if (c == '+')
1364 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1365 else if (c == '@')
1366 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1367 else if (c == '-')
1368 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1369 }
1370 emit_line_ws_markup(o, set_sign, set, reset,
1371 OUTPUT_INDICATOR_CONTEXT, line, len,
1372 flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1373 break;
1374 case DIFF_SYMBOL_PLUS:
1375 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1376 DIFF_SYMBOL_MOVED_LINE_ALT |
1377 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1378 case DIFF_SYMBOL_MOVED_LINE |
1379 DIFF_SYMBOL_MOVED_LINE_ALT |
1380 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1381 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1382 break;
1383 case DIFF_SYMBOL_MOVED_LINE |
1384 DIFF_SYMBOL_MOVED_LINE_ALT:
1385 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1386 break;
1387 case DIFF_SYMBOL_MOVED_LINE |
1388 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1389 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1390 break;
1391 case DIFF_SYMBOL_MOVED_LINE:
1392 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1393 break;
1394 default:
1395 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1396 }
1397 reset = diff_get_color_opt(o, DIFF_RESET);
1398 if (!o->flags.dual_color_diffed_diffs)
1399 set_sign = NULL;
1400 else {
1401 char c = !len ? 0 : line[0];
1402
1403 set_sign = set;
1404 if (c == '-')
1405 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1406 else if (c == '@')
1407 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1408 else if (c == '+')
1409 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1410 else
1411 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1412 flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1413 }
1414 emit_line_ws_markup(o, set_sign, set, reset,
1415 OUTPUT_INDICATOR_NEW, line, len,
1416 flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1417 flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1418 break;
1419 case DIFF_SYMBOL_MINUS:
1420 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1421 DIFF_SYMBOL_MOVED_LINE_ALT |
1422 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1423 case DIFF_SYMBOL_MOVED_LINE |
1424 DIFF_SYMBOL_MOVED_LINE_ALT |
1425 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1426 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1427 break;
1428 case DIFF_SYMBOL_MOVED_LINE |
1429 DIFF_SYMBOL_MOVED_LINE_ALT:
1430 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1431 break;
1432 case DIFF_SYMBOL_MOVED_LINE |
1433 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1434 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1435 break;
1436 case DIFF_SYMBOL_MOVED_LINE:
1437 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1438 break;
1439 default:
1440 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1441 }
1442 reset = diff_get_color_opt(o, DIFF_RESET);
1443 if (!o->flags.dual_color_diffed_diffs)
1444 set_sign = NULL;
1445 else {
1446 char c = !len ? 0 : line[0];
1447
1448 set_sign = set;
1449 if (c == '+')
1450 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1451 else if (c == '@')
1452 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1453 else if (c == '-')
1454 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1455 else
1456 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1457 }
1458 emit_line_ws_markup(o, set_sign, set, reset,
1459 OUTPUT_INDICATOR_OLD, line, len,
1460 flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1461 break;
1462 case DIFF_SYMBOL_WORDS_PORCELAIN:
1463 context = diff_get_color_opt(o, DIFF_CONTEXT);
1464 reset = diff_get_color_opt(o, DIFF_RESET);
1465 emit_line(o, context, reset, line, len);
1466 fputs("~\n", o->file);
1467 break;
1468 case DIFF_SYMBOL_WORDS:
1469 context = diff_get_color_opt(o, DIFF_CONTEXT);
1470 reset = diff_get_color_opt(o, DIFF_RESET);
1471 /*
1472 * Skip the prefix character, if any. With
1473 * diff_suppress_blank_empty, there may be
1474 * none.
1475 */
1476 if (line[0] != '\n') {
1477 line++;
1478 len--;
1479 }
1480 emit_line(o, context, reset, line, len);
1481 break;
1482 case DIFF_SYMBOL_FILEPAIR_PLUS:
1483 meta = diff_get_color_opt(o, DIFF_METAINFO);
1484 reset = diff_get_color_opt(o, DIFF_RESET);
1485 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1486 line, reset,
1487 strchr(line, ' ') ? "\t" : "");
1488 break;
1489 case DIFF_SYMBOL_FILEPAIR_MINUS:
1490 meta = diff_get_color_opt(o, DIFF_METAINFO);
1491 reset = diff_get_color_opt(o, DIFF_RESET);
1492 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1493 line, reset,
1494 strchr(line, ' ') ? "\t" : "");
1495 break;
1496 case DIFF_SYMBOL_BINARY_FILES:
1497 case DIFF_SYMBOL_HEADER:
1498 fprintf(o->file, "%s", line);
1499 break;
1500 case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1501 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1502 break;
1503 case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1504 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1505 break;
1506 case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1507 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1508 break;
1509 case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1510 fputs(diff_line_prefix(o), o->file);
1511 fputc('\n', o->file);
1512 break;
1513 case DIFF_SYMBOL_REWRITE_DIFF:
1514 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1515 reset = diff_get_color_opt(o, DIFF_RESET);
1516 emit_line(o, fraginfo, reset, line, len);
1517 break;
1518 case DIFF_SYMBOL_SUBMODULE_ADD:
1519 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1520 reset = diff_get_color_opt(o, DIFF_RESET);
1521 emit_line(o, set, reset, line, len);
1522 break;
1523 case DIFF_SYMBOL_SUBMODULE_DEL:
1524 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1525 reset = diff_get_color_opt(o, DIFF_RESET);
1526 emit_line(o, set, reset, line, len);
1527 break;
1528 case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1529 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1530 diff_line_prefix(o), line);
1531 break;
1532 case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1533 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1534 diff_line_prefix(o), line);
1535 break;
1536 case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1537 emit_line(o, "", "", " 0 files changed\n",
1538 strlen(" 0 files changed\n"));
1539 break;
1540 case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1541 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1542 break;
1543 case DIFF_SYMBOL_WORD_DIFF:
1544 fprintf(o->file, "%.*s", len, line);
1545 break;
1546 case DIFF_SYMBOL_STAT_SEP:
1547 fputs(o->stat_sep, o->file);
1548 break;
1549 default:
1550 BUG("unknown diff symbol");
1551 }
1552 strbuf_release(&sb);
1553 }
1554
1555 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1556 const char *line, int len, unsigned flags)
1557 {
1558 struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1559
1560 if (o->emitted_symbols)
1561 append_emitted_diff_symbol(o, &e);
1562 else
1563 emit_diff_symbol_from_struct(o, &e);
1564 }
1565
1566 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1567 {
1568 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1569 }
1570
1571 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1572 {
1573 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1574 }
1575
1576 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1577 {
1578 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1579 path, strlen(path), 0);
1580 }
1581
1582 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1583 {
1584 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1585 path, strlen(path), 0);
1586 }
1587
1588 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1589 {
1590 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1591 header, strlen(header), 0);
1592 }
1593
1594 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1595 {
1596 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1597 }
1598
1599 void diff_emit_submodule_pipethrough(struct diff_options *o,
1600 const char *line, int len)
1601 {
1602 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1603 }
1604
1605 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1606 {
1607 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1608 ecbdata->blank_at_eof_in_preimage &&
1609 ecbdata->blank_at_eof_in_postimage &&
1610 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1611 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1612 return 0;
1613 return ws_blank_line(line, len, ecbdata->ws_rule);
1614 }
1615
1616 static void emit_add_line(const char *reset,
1617 struct emit_callback *ecbdata,
1618 const char *line, int len)
1619 {
1620 unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1621 if (new_blank_line_at_eof(ecbdata, line, len))
1622 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1623
1624 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1625 }
1626
1627 static void emit_del_line(const char *reset,
1628 struct emit_callback *ecbdata,
1629 const char *line, int len)
1630 {
1631 unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1632 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1633 }
1634
1635 static void emit_context_line(const char *reset,
1636 struct emit_callback *ecbdata,
1637 const char *line, int len)
1638 {
1639 unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1640 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1641 }
1642
1643 static void emit_hunk_header(struct emit_callback *ecbdata,
1644 const char *line, int len)
1645 {
1646 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1647 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1648 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1649 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1650 const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1651 static const char atat[2] = { '@', '@' };
1652 const char *cp, *ep;
1653 struct strbuf msgbuf = STRBUF_INIT;
1654 int org_len = len;
1655 int i = 1;
1656
1657 /*
1658 * As a hunk header must begin with "@@ -<old>, +<new> @@",
1659 * it always is at least 10 bytes long.
1660 */
1661 if (len < 10 ||
1662 memcmp(line, atat, 2) ||
1663 !(ep = memmem(line + 2, len - 2, atat, 2))) {
1664 emit_diff_symbol(ecbdata->opt,
1665 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1666 return;
1667 }
1668 ep += 2; /* skip over @@ */
1669
1670 /* The hunk header in fraginfo color */
1671 if (ecbdata->opt->flags.dual_color_diffed_diffs)
1672 strbuf_addstr(&msgbuf, reverse);
1673 strbuf_addstr(&msgbuf, frag);
1674 strbuf_add(&msgbuf, line, ep - line);
1675 strbuf_addstr(&msgbuf, reset);
1676
1677 /*
1678 * trailing "\r\n"
1679 */
1680 for ( ; i < 3; i++)
1681 if (line[len - i] == '\r' || line[len - i] == '\n')
1682 len--;
1683
1684 /* blank before the func header */
1685 for (cp = ep; ep - line < len; ep++)
1686 if (*ep != ' ' && *ep != '\t')
1687 break;
1688 if (ep != cp) {
1689 strbuf_addstr(&msgbuf, context);
1690 strbuf_add(&msgbuf, cp, ep - cp);
1691 strbuf_addstr(&msgbuf, reset);
1692 }
1693
1694 if (ep < line + len) {
1695 strbuf_addstr(&msgbuf, func);
1696 strbuf_add(&msgbuf, ep, line + len - ep);
1697 strbuf_addstr(&msgbuf, reset);
1698 }
1699
1700 strbuf_add(&msgbuf, line + len, org_len - len);
1701 strbuf_complete_line(&msgbuf);
1702 emit_diff_symbol(ecbdata->opt,
1703 DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1704 strbuf_release(&msgbuf);
1705 }
1706
1707 static struct diff_tempfile *claim_diff_tempfile(void)
1708 {
1709 int i;
1710 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1711 if (!diff_temp[i].name)
1712 return diff_temp + i;
1713 BUG("diff is failing to clean up its tempfiles");
1714 }
1715
1716 static void remove_tempfile(void)
1717 {
1718 int i;
1719 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1720 if (is_tempfile_active(diff_temp[i].tempfile))
1721 delete_tempfile(&diff_temp[i].tempfile);
1722 diff_temp[i].name = NULL;
1723 }
1724 }
1725
1726 static void add_line_count(struct strbuf *out, int count)
1727 {
1728 switch (count) {
1729 case 0:
1730 strbuf_addstr(out, "0,0");
1731 break;
1732 case 1:
1733 strbuf_addstr(out, "1");
1734 break;
1735 default:
1736 strbuf_addf(out, "1,%d", count);
1737 break;
1738 }
1739 }
1740
1741 static void emit_rewrite_lines(struct emit_callback *ecb,
1742 int prefix, const char *data, int size)
1743 {
1744 const char *endp = NULL;
1745 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1746
1747 while (0 < size) {
1748 int len;
1749
1750 endp = memchr(data, '\n', size);
1751 len = endp ? (endp - data + 1) : size;
1752 if (prefix != '+') {
1753 ecb->lno_in_preimage++;
1754 emit_del_line(reset, ecb, data, len);
1755 } else {
1756 ecb->lno_in_postimage++;
1757 emit_add_line(reset, ecb, data, len);
1758 }
1759 size -= len;
1760 data += len;
1761 }
1762 if (!endp)
1763 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1764 }
1765
1766 static void emit_rewrite_diff(const char *name_a,
1767 const char *name_b,
1768 struct diff_filespec *one,
1769 struct diff_filespec *two,
1770 struct userdiff_driver *textconv_one,
1771 struct userdiff_driver *textconv_two,
1772 struct diff_options *o)
1773 {
1774 int lc_a, lc_b;
1775 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1776 const char *a_prefix, *b_prefix;
1777 char *data_one, *data_two;
1778 size_t size_one, size_two;
1779 struct emit_callback ecbdata;
1780 struct strbuf out = STRBUF_INIT;
1781
1782 if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1783 a_prefix = o->b_prefix;
1784 b_prefix = o->a_prefix;
1785 } else {
1786 a_prefix = o->a_prefix;
1787 b_prefix = o->b_prefix;
1788 }
1789
1790 name_a += (*name_a == '/');
1791 name_b += (*name_b == '/');
1792
1793 strbuf_reset(&a_name);
1794 strbuf_reset(&b_name);
1795 quote_two_c_style(&a_name, a_prefix, name_a, 0);
1796 quote_two_c_style(&b_name, b_prefix, name_b, 0);
1797
1798 size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1799 size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1800
1801 memset(&ecbdata, 0, sizeof(ecbdata));
1802 ecbdata.color_diff = want_color(o->use_color);
1803 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1804 ecbdata.opt = o;
1805 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1806 mmfile_t mf1, mf2;
1807 mf1.ptr = (char *)data_one;
1808 mf2.ptr = (char *)data_two;
1809 mf1.size = size_one;
1810 mf2.size = size_two;
1811 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1812 }
1813 ecbdata.lno_in_preimage = 1;
1814 ecbdata.lno_in_postimage = 1;
1815
1816 lc_a = count_lines(data_one, size_one);
1817 lc_b = count_lines(data_two, size_two);
1818
1819 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1820 a_name.buf, a_name.len, 0);
1821 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1822 b_name.buf, b_name.len, 0);
1823
1824 strbuf_addstr(&out, "@@ -");
1825 if (!o->irreversible_delete)
1826 add_line_count(&out, lc_a);
1827 else
1828 strbuf_addstr(&out, "?,?");
1829 strbuf_addstr(&out, " +");
1830 add_line_count(&out, lc_b);
1831 strbuf_addstr(&out, " @@\n");
1832 emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1833 strbuf_release(&out);
1834
1835 if (lc_a && !o->irreversible_delete)
1836 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1837 if (lc_b)
1838 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1839 if (textconv_one)
1840 free((char *)data_one);
1841 if (textconv_two)
1842 free((char *)data_two);
1843 }
1844
1845 struct diff_words_buffer {
1846 mmfile_t text;
1847 unsigned long alloc;
1848 struct diff_words_orig {
1849 const char *begin, *end;
1850 } *orig;
1851 int orig_nr, orig_alloc;
1852 };
1853
1854 static void diff_words_append(char *line, unsigned long len,
1855 struct diff_words_buffer *buffer)
1856 {
1857 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1858 line++;
1859 len--;
1860 memcpy(buffer->text.ptr + buffer->text.size, line, len);
1861 buffer->text.size += len;
1862 buffer->text.ptr[buffer->text.size] = '\0';
1863 }
1864
1865 struct diff_words_style_elem {
1866 const char *prefix;
1867 const char *suffix;
1868 const char *color; /* NULL; filled in by the setup code if
1869 * color is enabled */
1870 };
1871
1872 struct diff_words_style {
1873 enum diff_words_type type;
1874 struct diff_words_style_elem new_word, old_word, ctx;
1875 const char *newline;
1876 };
1877
1878 static struct diff_words_style diff_words_styles[] = {
1879 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1880 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1881 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1882 };
1883
1884 struct diff_words_data {
1885 struct diff_words_buffer minus, plus;
1886 const char *current_plus;
1887 int last_minus;
1888 struct diff_options *opt;
1889 regex_t *word_regex;
1890 enum diff_words_type type;
1891 struct diff_words_style *style;
1892 };
1893
1894 static int fn_out_diff_words_write_helper(struct diff_options *o,
1895 struct diff_words_style_elem *st_el,
1896 const char *newline,
1897 size_t count, const char *buf)
1898 {
1899 int print = 0;
1900 struct strbuf sb = STRBUF_INIT;
1901
1902 while (count) {
1903 char *p = memchr(buf, '\n', count);
1904 if (print)
1905 strbuf_addstr(&sb, diff_line_prefix(o));
1906
1907 if (p != buf) {
1908 const char *reset = st_el->color && *st_el->color ?
1909 GIT_COLOR_RESET : NULL;
1910 if (st_el->color && *st_el->color)
1911 strbuf_addstr(&sb, st_el->color);
1912 strbuf_addstr(&sb, st_el->prefix);
1913 strbuf_add(&sb, buf, p ? p - buf : count);
1914 strbuf_addstr(&sb, st_el->suffix);
1915 if (reset)
1916 strbuf_addstr(&sb, reset);
1917 }
1918 if (!p)
1919 goto out;
1920
1921 strbuf_addstr(&sb, newline);
1922 count -= p + 1 - buf;
1923 buf = p + 1;
1924 print = 1;
1925 if (count) {
1926 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1927 sb.buf, sb.len, 0);
1928 strbuf_reset(&sb);
1929 }
1930 }
1931
1932 out:
1933 if (sb.len)
1934 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1935 sb.buf, sb.len, 0);
1936 strbuf_release(&sb);
1937 return 0;
1938 }
1939
1940 /*
1941 * '--color-words' algorithm can be described as:
1942 *
1943 * 1. collect the minus/plus lines of a diff hunk, divided into
1944 * minus-lines and plus-lines;
1945 *
1946 * 2. break both minus-lines and plus-lines into words and
1947 * place them into two mmfile_t with one word for each line;
1948 *
1949 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1950 *
1951 * And for the common parts of the both file, we output the plus side text.
1952 * diff_words->current_plus is used to trace the current position of the plus file
1953 * which printed. diff_words->last_minus is used to trace the last minus word
1954 * printed.
1955 *
1956 * For '--graph' to work with '--color-words', we need to output the graph prefix
1957 * on each line of color words output. Generally, there are two conditions on
1958 * which we should output the prefix.
1959 *
1960 * 1. diff_words->last_minus == 0 &&
1961 * diff_words->current_plus == diff_words->plus.text.ptr
1962 *
1963 * that is: the plus text must start as a new line, and if there is no minus
1964 * word printed, a graph prefix must be printed.
1965 *
1966 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
1967 * *(diff_words->current_plus - 1) == '\n'
1968 *
1969 * that is: a graph prefix must be printed following a '\n'
1970 */
1971 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1972 {
1973 if ((diff_words->last_minus == 0 &&
1974 diff_words->current_plus == diff_words->plus.text.ptr) ||
1975 (diff_words->current_plus > diff_words->plus.text.ptr &&
1976 *(diff_words->current_plus - 1) == '\n')) {
1977 return 1;
1978 } else {
1979 return 0;
1980 }
1981 }
1982
1983 static void fn_out_diff_words_aux(void *priv,
1984 long minus_first, long minus_len,
1985 long plus_first, long plus_len,
1986 const char *func, long funclen)
1987 {
1988 struct diff_words_data *diff_words = priv;
1989 struct diff_words_style *style = diff_words->style;
1990 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1991 struct diff_options *opt = diff_words->opt;
1992 const char *line_prefix;
1993
1994 assert(opt);
1995 line_prefix = diff_line_prefix(opt);
1996
1997 /* POSIX requires that first be decremented by one if len == 0... */
1998 if (minus_len) {
1999 minus_begin = diff_words->minus.orig[minus_first].begin;
2000 minus_end =
2001 diff_words->minus.orig[minus_first + minus_len - 1].end;
2002 } else
2003 minus_begin = minus_end =
2004 diff_words->minus.orig[minus_first].end;
2005
2006 if (plus_len) {
2007 plus_begin = diff_words->plus.orig[plus_first].begin;
2008 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2009 } else
2010 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2011
2012 if (color_words_output_graph_prefix(diff_words)) {
2013 fputs(line_prefix, diff_words->opt->file);
2014 }
2015 if (diff_words->current_plus != plus_begin) {
2016 fn_out_diff_words_write_helper(diff_words->opt,
2017 &style->ctx, style->newline,
2018 plus_begin - diff_words->current_plus,
2019 diff_words->current_plus);
2020 }
2021 if (minus_begin != minus_end) {
2022 fn_out_diff_words_write_helper(diff_words->opt,
2023 &style->old_word, style->newline,
2024 minus_end - minus_begin, minus_begin);
2025 }
2026 if (plus_begin != plus_end) {
2027 fn_out_diff_words_write_helper(diff_words->opt,
2028 &style->new_word, style->newline,
2029 plus_end - plus_begin, plus_begin);
2030 }
2031
2032 diff_words->current_plus = plus_end;
2033 diff_words->last_minus = minus_first;
2034 }
2035
2036 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2037 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2038 int *begin, int *end)
2039 {
2040 if (word_regex && *begin < buffer->size) {
2041 regmatch_t match[1];
2042 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2043 buffer->size - *begin, 1, match, 0)) {
2044 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2045 '\n', match[0].rm_eo - match[0].rm_so);
2046 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2047 *begin += match[0].rm_so;
2048 return *begin >= *end;
2049 }
2050 return -1;
2051 }
2052
2053 /* find the next word */
2054 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2055 (*begin)++;
2056 if (*begin >= buffer->size)
2057 return -1;
2058
2059 /* find the end of the word */
2060 *end = *begin + 1;
2061 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2062 (*end)++;
2063
2064 return 0;
2065 }
2066
2067 /*
2068 * This function splits the words in buffer->text, stores the list with
2069 * newline separator into out, and saves the offsets of the original words
2070 * in buffer->orig.
2071 */
2072 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2073 regex_t *word_regex)
2074 {
2075 int i, j;
2076 long alloc = 0;
2077
2078 out->size = 0;
2079 out->ptr = NULL;
2080
2081 /* fake an empty "0th" word */
2082 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2083 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2084 buffer->orig_nr = 1;
2085
2086 for (i = 0; i < buffer->text.size; i++) {
2087 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2088 return;
2089
2090 /* store original boundaries */
2091 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2092 buffer->orig_alloc);
2093 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2094 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2095 buffer->orig_nr++;
2096
2097 /* store one word */
2098 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2099 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2100 out->ptr[out->size + j - i] = '\n';
2101 out->size += j - i + 1;
2102
2103 i = j - 1;
2104 }
2105 }
2106
2107 /* this executes the word diff on the accumulated buffers */
2108 static void diff_words_show(struct diff_words_data *diff_words)
2109 {
2110 xpparam_t xpp;
2111 xdemitconf_t xecfg;
2112 mmfile_t minus, plus;
2113 struct diff_words_style *style = diff_words->style;
2114
2115 struct diff_options *opt = diff_words->opt;
2116 const char *line_prefix;
2117
2118 assert(opt);
2119 line_prefix = diff_line_prefix(opt);
2120
2121 /* special case: only removal */
2122 if (!diff_words->plus.text.size) {
2123 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2124 line_prefix, strlen(line_prefix), 0);
2125 fn_out_diff_words_write_helper(diff_words->opt,
2126 &style->old_word, style->newline,
2127 diff_words->minus.text.size,
2128 diff_words->minus.text.ptr);
2129 diff_words->minus.text.size = 0;
2130 return;
2131 }
2132
2133 diff_words->current_plus = diff_words->plus.text.ptr;
2134 diff_words->last_minus = 0;
2135
2136 memset(&xpp, 0, sizeof(xpp));
2137 memset(&xecfg, 0, sizeof(xecfg));
2138 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2139 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2140 xpp.flags = 0;
2141 /* as only the hunk header will be parsed, we need a 0-context */
2142 xecfg.ctxlen = 0;
2143 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2144 diff_words, &xpp, &xecfg))
2145 die("unable to generate word diff");
2146 free(minus.ptr);
2147 free(plus.ptr);
2148 if (diff_words->current_plus != diff_words->plus.text.ptr +
2149 diff_words->plus.text.size) {
2150 if (color_words_output_graph_prefix(diff_words))
2151 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2152 line_prefix, strlen(line_prefix), 0);
2153 fn_out_diff_words_write_helper(diff_words->opt,
2154 &style->ctx, style->newline,
2155 diff_words->plus.text.ptr + diff_words->plus.text.size
2156 - diff_words->current_plus, diff_words->current_plus);
2157 }
2158 diff_words->minus.text.size = diff_words->plus.text.size = 0;
2159 }
2160
2161 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2162 static void diff_words_flush(struct emit_callback *ecbdata)
2163 {
2164 struct diff_options *wo = ecbdata->diff_words->opt;
2165
2166 if (ecbdata->diff_words->minus.text.size ||
2167 ecbdata->diff_words->plus.text.size)
2168 diff_words_show(ecbdata->diff_words);
2169
2170 if (wo->emitted_symbols) {
2171 struct diff_options *o = ecbdata->opt;
2172 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2173 int i;
2174
2175 /*
2176 * NEEDSWORK:
2177 * Instead of appending each, concat all words to a line?
2178 */
2179 for (i = 0; i < wol->nr; i++)
2180 append_emitted_diff_symbol(o, &wol->buf[i]);
2181
2182 for (i = 0; i < wol->nr; i++)
2183 free((void *)wol->buf[i].line);
2184
2185 wol->nr = 0;
2186 }
2187 }
2188
2189 static void diff_filespec_load_driver(struct diff_filespec *one,
2190 struct index_state *istate)
2191 {
2192 /* Use already-loaded driver */
2193 if (one->driver)
2194 return;
2195
2196 if (S_ISREG(one->mode))
2197 one->driver = userdiff_find_by_path(istate, one->path);
2198
2199 /* Fallback to default settings */
2200 if (!one->driver)
2201 one->driver = userdiff_find_by_name("default");
2202 }
2203
2204 static const char *userdiff_word_regex(struct diff_filespec *one,
2205 struct index_state *istate)
2206 {
2207 diff_filespec_load_driver(one, istate);
2208 return one->driver->word_regex;
2209 }
2210
2211 static void init_diff_words_data(struct emit_callback *ecbdata,
2212 struct diff_options *orig_opts,
2213 struct diff_filespec *one,
2214 struct diff_filespec *two)
2215 {
2216 int i;
2217 struct diff_options *o = xmalloc(sizeof(struct diff_options));
2218 memcpy(o, orig_opts, sizeof(struct diff_options));
2219
2220 ecbdata->diff_words =
2221 xcalloc(1, sizeof(struct diff_words_data));
2222 ecbdata->diff_words->type = o->word_diff;
2223 ecbdata->diff_words->opt = o;
2224
2225 if (orig_opts->emitted_symbols)
2226 o->emitted_symbols =
2227 xcalloc(1, sizeof(struct emitted_diff_symbols));
2228
2229 if (!o->word_regex)
2230 o->word_regex = userdiff_word_regex(one, o->repo->index);
2231 if (!o->word_regex)
2232 o->word_regex = userdiff_word_regex(two, o->repo->index);
2233 if (!o->word_regex)
2234 o->word_regex = diff_word_regex_cfg;
2235 if (o->word_regex) {
2236 ecbdata->diff_words->word_regex = (regex_t *)
2237 xmalloc(sizeof(regex_t));
2238 if (regcomp(ecbdata->diff_words->word_regex,
2239 o->word_regex,
2240 REG_EXTENDED | REG_NEWLINE))
2241 die("invalid regular expression: %s",
2242 o->word_regex);
2243 }
2244 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2245 if (o->word_diff == diff_words_styles[i].type) {
2246 ecbdata->diff_words->style =
2247 &diff_words_styles[i];
2248 break;
2249 }
2250 }
2251 if (want_color(o->use_color)) {
2252 struct diff_words_style *st = ecbdata->diff_words->style;
2253 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2254 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2255 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2256 }
2257 }
2258
2259 static void free_diff_words_data(struct emit_callback *ecbdata)
2260 {
2261 if (ecbdata->diff_words) {
2262 diff_words_flush(ecbdata);
2263 free (ecbdata->diff_words->opt->emitted_symbols);
2264 free (ecbdata->diff_words->opt);
2265 free (ecbdata->diff_words->minus.text.ptr);
2266 free (ecbdata->diff_words->minus.orig);
2267 free (ecbdata->diff_words->plus.text.ptr);
2268 free (ecbdata->diff_words->plus.orig);
2269 if (ecbdata->diff_words->word_regex) {
2270 regfree(ecbdata->diff_words->word_regex);
2271 free(ecbdata->diff_words->word_regex);
2272 }
2273 FREE_AND_NULL(ecbdata->diff_words);
2274 }
2275 }
2276
2277 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2278 {
2279 if (want_color(diff_use_color))
2280 return diff_colors[ix];
2281 return "";
2282 }
2283
2284 const char *diff_line_prefix(struct diff_options *opt)
2285 {
2286 struct strbuf *msgbuf;
2287 if (!opt->output_prefix)
2288 return "";
2289
2290 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2291 return msgbuf->buf;
2292 }
2293
2294 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2295 {
2296 const char *cp;
2297 unsigned long allot;
2298 size_t l = len;
2299
2300 cp = line;
2301 allot = l;
2302 while (0 < l) {
2303 (void) utf8_width(&cp, &l);
2304 if (!cp)
2305 break; /* truncated in the middle? */
2306 }
2307 return allot - l;
2308 }
2309
2310 static void find_lno(const char *line, struct emit_callback *ecbdata)
2311 {
2312 const char *p;
2313 ecbdata->lno_in_preimage = 0;
2314 ecbdata->lno_in_postimage = 0;
2315 p = strchr(line, '-');
2316 if (!p)
2317 return; /* cannot happen */
2318 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2319 p = strchr(p, '+');
2320 if (!p)
2321 return; /* cannot happen */
2322 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2323 }
2324
2325 static void fn_out_consume(void *priv, char *line, unsigned long len)
2326 {
2327 struct emit_callback *ecbdata = priv;
2328 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2329 struct diff_options *o = ecbdata->opt;
2330
2331 o->found_changes = 1;
2332
2333 if (ecbdata->header) {
2334 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2335 ecbdata->header->buf, ecbdata->header->len, 0);
2336 strbuf_reset(ecbdata->header);
2337 ecbdata->header = NULL;
2338 }
2339
2340 if (ecbdata->label_path[0]) {
2341 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2342 ecbdata->label_path[0],
2343 strlen(ecbdata->label_path[0]), 0);
2344 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2345 ecbdata->label_path[1],
2346 strlen(ecbdata->label_path[1]), 0);
2347 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2348 }
2349
2350 if (diff_suppress_blank_empty
2351 && len == 2 && line[0] == ' ' && line[1] == '\n') {
2352 line[0] = '\n';
2353 len = 1;
2354 }
2355
2356 if (line[0] == '@') {
2357 if (ecbdata->diff_words)
2358 diff_words_flush(ecbdata);
2359 len = sane_truncate_line(ecbdata, line, len);
2360 find_lno(line, ecbdata);
2361 emit_hunk_header(ecbdata, line, len);
2362 return;
2363 }
2364
2365 if (ecbdata->diff_words) {
2366 enum diff_symbol s =
2367 ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2368 DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2369 if (line[0] == '-') {
2370 diff_words_append(line, len,
2371 &ecbdata->diff_words->minus);
2372 return;
2373 } else if (line[0] == '+') {
2374 diff_words_append(line, len,
2375 &ecbdata->diff_words->plus);
2376 return;
2377 } else if (starts_with(line, "\\ ")) {
2378 /*
2379 * Eat the "no newline at eof" marker as if we
2380 * saw a "+" or "-" line with nothing on it,
2381 * and return without diff_words_flush() to
2382 * defer processing. If this is the end of
2383 * preimage, more "+" lines may come after it.
2384 */
2385 return;
2386 }
2387 diff_words_flush(ecbdata);
2388 emit_diff_symbol(o, s, line, len, 0);
2389 return;
2390 }
2391
2392 switch (line[0]) {
2393 case '+':
2394 ecbdata->lno_in_postimage++;
2395 emit_add_line(reset, ecbdata, line + 1, len - 1);
2396 break;
2397 case '-':
2398 ecbdata->lno_in_preimage++;
2399 emit_del_line(reset, ecbdata, line + 1, len - 1);
2400 break;
2401 case ' ':
2402 ecbdata->lno_in_postimage++;
2403 ecbdata->lno_in_preimage++;
2404 emit_context_line(reset, ecbdata, line + 1, len - 1);
2405 break;
2406 default:
2407 /* incomplete line at the end */
2408 ecbdata->lno_in_preimage++;
2409 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2410 line, len, 0);
2411 break;
2412 }
2413 }
2414
2415 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2416 {
2417 const char *old_name = a;
2418 const char *new_name = b;
2419 int pfx_length, sfx_length;
2420 int pfx_adjust_for_slash;
2421 int len_a = strlen(a);
2422 int len_b = strlen(b);
2423 int a_midlen, b_midlen;
2424 int qlen_a = quote_c_style(a, NULL, NULL, 0);
2425 int qlen_b = quote_c_style(b, NULL, NULL, 0);
2426
2427 if (qlen_a || qlen_b) {
2428 quote_c_style(a, name, NULL, 0);
2429 strbuf_addstr(name, " => ");
2430 quote_c_style(b, name, NULL, 0);
2431 return;
2432 }
2433
2434 /* Find common prefix */
2435 pfx_length = 0;
2436 while (*old_name && *new_name && *old_name == *new_name) {
2437 if (*old_name == '/')
2438 pfx_length = old_name - a + 1;
2439 old_name++;
2440 new_name++;
2441 }
2442
2443 /* Find common suffix */
2444 old_name = a + len_a;
2445 new_name = b + len_b;
2446 sfx_length = 0;
2447 /*
2448 * If there is a common prefix, it must end in a slash. In
2449 * that case we let this loop run 1 into the prefix to see the
2450 * same slash.
2451 *
2452 * If there is no common prefix, we cannot do this as it would
2453 * underrun the input strings.
2454 */
2455 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2456 while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2457 b + pfx_length - pfx_adjust_for_slash <= new_name &&
2458 *old_name == *new_name) {
2459 if (*old_name == '/')
2460 sfx_length = len_a - (old_name - a);
2461 old_name--;
2462 new_name--;
2463 }
2464
2465 /*
2466 * pfx{mid-a => mid-b}sfx
2467 * {pfx-a => pfx-b}sfx
2468 * pfx{sfx-a => sfx-b}
2469 * name-a => name-b
2470 */
2471 a_midlen = len_a - pfx_length - sfx_length;
2472 b_midlen = len_b - pfx_length - sfx_length;
2473 if (a_midlen < 0)
2474 a_midlen = 0;
2475 if (b_midlen < 0)
2476 b_midlen = 0;
2477
2478 strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2479 if (pfx_length + sfx_length) {
2480 strbuf_add(name, a, pfx_length);
2481 strbuf_addch(name, '{');
2482 }
2483 strbuf_add(name, a + pfx_length, a_midlen);
2484 strbuf_addstr(name, " => ");
2485 strbuf_add(name, b + pfx_length, b_midlen);
2486 if (pfx_length + sfx_length) {
2487 strbuf_addch(name, '}');
2488 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2489 }
2490 }
2491
2492 struct diffstat_t {
2493 int nr;
2494 int alloc;
2495 struct diffstat_file {
2496 char *from_name;
2497 char *name;
2498 char *print_name;
2499 const char *comments;
2500 unsigned is_unmerged:1;
2501 unsigned is_binary:1;
2502 unsigned is_renamed:1;
2503 unsigned is_interesting:1;
2504 uintmax_t added, deleted;
2505 } **files;
2506 };
2507
2508 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2509 const char *name_a,
2510 const char *name_b)
2511 {
2512 struct diffstat_file *x;
2513 x = xcalloc(1, sizeof(*x));
2514 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2515 diffstat->files[diffstat->nr++] = x;
2516 if (name_b) {
2517 x->from_name = xstrdup(name_a);
2518 x->name = xstrdup(name_b);
2519 x->is_renamed = 1;
2520 }
2521 else {
2522 x->from_name = NULL;
2523 x->name = xstrdup(name_a);
2524 }
2525 return x;
2526 }
2527
2528 static void diffstat_consume(void *priv, char *line, unsigned long len)
2529 {
2530 struct diffstat_t *diffstat = priv;
2531 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2532
2533 if (line[0] == '+')
2534 x->added++;
2535 else if (line[0] == '-')
2536 x->deleted++;
2537 }
2538
2539 const char mime_boundary_leader[] = "------------";
2540
2541 static int scale_linear(int it, int width, int max_change)
2542 {
2543 if (!it)
2544 return 0;
2545 /*
2546 * make sure that at least one '-' or '+' is printed if
2547 * there is any change to this path. The easiest way is to
2548 * scale linearly as if the alloted width is one column shorter
2549 * than it is, and then add 1 to the result.
2550 */
2551 return 1 + (it * (width - 1) / max_change);
2552 }
2553
2554 static void show_graph(struct strbuf *out, char ch, int cnt,
2555 const char *set, const char *reset)
2556 {
2557 if (cnt <= 0)
2558 return;
2559 strbuf_addstr(out, set);
2560 strbuf_addchars(out, ch, cnt);
2561 strbuf_addstr(out, reset);
2562 }
2563
2564 static void fill_print_name(struct diffstat_file *file)
2565 {
2566 struct strbuf pname = STRBUF_INIT;
2567
2568 if (file->print_name)
2569 return;
2570
2571 if (file->is_renamed)
2572 pprint_rename(&pname, file->from_name, file->name);
2573 else
2574 quote_c_style(file->name, &pname, NULL, 0);
2575
2576 if (file->comments)
2577 strbuf_addf(&pname, " (%s)", file->comments);
2578
2579 file->print_name = strbuf_detach(&pname, NULL);
2580 }
2581
2582 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2583 int files, int insertions, int deletions)
2584 {
2585 struct strbuf sb = STRBUF_INIT;
2586
2587 if (!files) {
2588 assert(insertions == 0 && deletions == 0);
2589 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2590 NULL, 0, 0);
2591 return;
2592 }
2593
2594 strbuf_addf(&sb,
2595 (files == 1) ? " %d file changed" : " %d files changed",
2596 files);
2597
2598 /*
2599 * For binary diff, the caller may want to print "x files
2600 * changed" with insertions == 0 && deletions == 0.
2601 *
2602 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2603 * is probably less confusing (i.e skip over "2 files changed
2604 * but nothing about added/removed lines? Is this a bug in Git?").
2605 */
2606 if (insertions || deletions == 0) {
2607 strbuf_addf(&sb,
2608 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2609 insertions);
2610 }
2611
2612 if (deletions || insertions == 0) {
2613 strbuf_addf(&sb,
2614 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2615 deletions);
2616 }
2617 strbuf_addch(&sb, '\n');
2618 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2619 sb.buf, sb.len, 0);
2620 strbuf_release(&sb);
2621 }
2622
2623 void print_stat_summary(FILE *fp, int files,
2624 int insertions, int deletions)
2625 {
2626 struct diff_options o;
2627 memset(&o, 0, sizeof(o));
2628 o.file = fp;
2629
2630 print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2631 }
2632
2633 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2634 {
2635 int i, len, add, del, adds = 0, dels = 0;
2636 uintmax_t max_change = 0, max_len = 0;
2637 int total_files = data->nr, count;
2638 int width, name_width, graph_width, number_width = 0, bin_width = 0;
2639 const char *reset, *add_c, *del_c;
2640 int extra_shown = 0;
2641 const char *line_prefix = diff_line_prefix(options);
2642 struct strbuf out = STRBUF_INIT;
2643
2644 if (data->nr == 0)
2645 return;
2646
2647 count = options->stat_count ? options->stat_count : data->nr;
2648
2649 reset = diff_get_color_opt(options, DIFF_RESET);
2650 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2651 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2652
2653 /*
2654 * Find the longest filename and max number of changes
2655 */
2656 for (i = 0; (i < count) && (i < data->nr); i++) {
2657 struct diffstat_file *file = data->files[i];
2658 uintmax_t change = file->added + file->deleted;
2659
2660 if (!file->is_interesting && (change == 0)) {
2661 count++; /* not shown == room for one more */
2662 continue;
2663 }
2664 fill_print_name(file);
2665 len = strlen(file->print_name);
2666 if (max_len < len)
2667 max_len = len;
2668
2669 if (file->is_unmerged) {
2670 /* "Unmerged" is 8 characters */
2671 bin_width = bin_width < 8 ? 8 : bin_width;
2672 continue;
2673 }
2674 if (file->is_binary) {
2675 /* "Bin XXX -> YYY bytes" */
2676 int w = 14 + decimal_width(file->added)
2677 + decimal_width(file->deleted);
2678 bin_width = bin_width < w ? w : bin_width;
2679 /* Display change counts aligned with "Bin" */
2680 number_width = 3;
2681 continue;
2682 }
2683
2684 if (max_change < change)
2685 max_change = change;
2686 }
2687 count = i; /* where we can stop scanning in data->files[] */
2688
2689 /*
2690 * We have width = stat_width or term_columns() columns total.
2691 * We want a maximum of min(max_len, stat_name_width) for the name part.
2692 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2693 * We also need 1 for " " and 4 + decimal_width(max_change)
2694 * for " | NNNN " and one the empty column at the end, altogether
2695 * 6 + decimal_width(max_change).
2696 *
2697 * If there's not enough space, we will use the smaller of
2698 * stat_name_width (if set) and 5/8*width for the filename,
2699 * and the rest for constant elements + graph part, but no more
2700 * than stat_graph_width for the graph part.
2701 * (5/8 gives 50 for filename and 30 for the constant parts + graph
2702 * for the standard terminal size).
2703 *
2704 * In other words: stat_width limits the maximum width, and
2705 * stat_name_width fixes the maximum width of the filename,
2706 * and is also used to divide available columns if there
2707 * aren't enough.
2708 *
2709 * Binary files are displayed with "Bin XXX -> YYY bytes"
2710 * instead of the change count and graph. This part is treated
2711 * similarly to the graph part, except that it is not
2712 * "scaled". If total width is too small to accommodate the
2713 * guaranteed minimum width of the filename part and the
2714 * separators and this message, this message will "overflow"
2715 * making the line longer than the maximum width.
2716 */
2717
2718 if (options->stat_width == -1)
2719 width = term_columns() - strlen(line_prefix);
2720 else
2721 width = options->stat_width ? options->stat_width : 80;
2722 number_width = decimal_width(max_change) > number_width ?
2723 decimal_width(max_change) : number_width;
2724
2725 if (options->stat_graph_width == -1)
2726 options->stat_graph_width = diff_stat_graph_width;
2727
2728 /*
2729 * Guarantee 3/8*16==6 for the graph part
2730 * and 5/8*16==10 for the filename part
2731 */
2732 if (width < 16 + 6 + number_width)
2733 width = 16 + 6 + number_width;
2734
2735 /*
2736 * First assign sizes that are wanted, ignoring available width.
2737 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2738 * starting from "XXX" should fit in graph_width.
2739 */
2740 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2741 if (options->stat_graph_width &&
2742 options->stat_graph_width < graph_width)
2743 graph_width = options->stat_graph_width;
2744
2745 name_width = (options->stat_name_width > 0 &&
2746 options->stat_name_width < max_len) ?
2747 options->stat_name_width : max_len;
2748
2749 /*
2750 * Adjust adjustable widths not to exceed maximum width
2751 */
2752 if (name_width + number_width + 6 + graph_width > width) {
2753 if (graph_width > width * 3/8 - number_width - 6) {
2754 graph_width = width * 3/8 - number_width - 6;
2755 if (graph_width < 6)
2756 graph_width = 6;
2757 }
2758
2759 if (options->stat_graph_width &&
2760 graph_width > options->stat_graph_width)
2761 graph_width = options->stat_graph_width;
2762 if (name_width > width - number_width - 6 - graph_width)
2763 name_width = width - number_width - 6 - graph_width;
2764 else
2765 graph_width = width - number_width - 6 - name_width;
2766 }
2767
2768 /*
2769 * From here name_width is the width of the name area,
2770 * and graph_width is the width of the graph area.
2771 * max_change is used to scale graph properly.
2772 */
2773 for (i = 0; i < count; i++) {
2774 const char *prefix = "";
2775 struct diffstat_file *file = data->files[i];
2776 char *name = file->print_name;
2777 uintmax_t added = file->added;
2778 uintmax_t deleted = file->deleted;
2779 int name_len;
2780
2781 if (!file->is_interesting && (added + deleted == 0))
2782 continue;
2783
2784 /*
2785 * "scale" the filename
2786 */
2787 len = name_width;
2788 name_len = strlen(name);
2789 if (name_width < name_len) {
2790 char *slash;
2791 prefix = "...";
2792 len -= 3;
2793 name += name_len - len;
2794 slash = strchr(name, '/');
2795 if (slash)
2796 name = slash;
2797 }
2798
2799 if (file->is_binary) {
2800 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2801 strbuf_addf(&out, " %*s", number_width, "Bin");
2802 if (!added && !deleted) {
2803 strbuf_addch(&out, '\n');
2804 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2805 out.buf, out.len, 0);
2806 strbuf_reset(&out);
2807 continue;
2808 }
2809 strbuf_addf(&out, " %s%"PRIuMAX"%s",
2810 del_c, deleted, reset);
2811 strbuf_addstr(&out, " -> ");
2812 strbuf_addf(&out, "%s%"PRIuMAX"%s",
2813 add_c, added, reset);
2814 strbuf_addstr(&out, " bytes\n");
2815 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2816 out.buf, out.len, 0);
2817 strbuf_reset(&out);
2818 continue;
2819 }
2820 else if (file->is_unmerged) {
2821 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2822 strbuf_addstr(&out, " Unmerged\n");
2823 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2824 out.buf, out.len, 0);
2825 strbuf_reset(&out);
2826 continue;
2827 }
2828
2829 /*
2830 * scale the add/delete
2831 */
2832 add = added;
2833 del = deleted;
2834
2835 if (graph_width <= max_change) {
2836 int total = scale_linear(add + del, graph_width, max_change);
2837 if (total < 2 && add && del)
2838 /* width >= 2 due to the sanity check */
2839 total = 2;
2840 if (add < del) {
2841 add = scale_linear(add, graph_width, max_change);
2842 del = total - add;
2843 } else {
2844 del = scale_linear(del, graph_width, max_change);
2845 add = total - del;
2846 }
2847 }
2848 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2849 strbuf_addf(&out, " %*"PRIuMAX"%s",
2850 number_width, added + deleted,
2851 added + deleted ? " " : "");
2852 show_graph(&out, '+', add, add_c, reset);
2853 show_graph(&out, '-', del, del_c, reset);
2854 strbuf_addch(&out, '\n');
2855 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2856 out.buf, out.len, 0);
2857 strbuf_reset(&out);
2858 }
2859
2860 for (i = 0; i < data->nr; i++) {
2861 struct diffstat_file *file = data->files[i];
2862 uintmax_t added = file->added;
2863 uintmax_t deleted = file->deleted;
2864
2865 if (file->is_unmerged ||
2866 (!file->is_interesting && (added + deleted == 0))) {
2867 total_files--;
2868 continue;
2869 }
2870
2871 if (!file->is_binary) {
2872 adds += added;
2873 dels += deleted;
2874 }
2875 if (i < count)
2876 continue;
2877 if (!extra_shown)
2878 emit_diff_symbol(options,
2879 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2880 NULL, 0, 0);
2881 extra_shown = 1;
2882 }
2883
2884 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2885 strbuf_release(&out);
2886 }
2887
2888 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2889 {
2890 int i, adds = 0, dels = 0, total_files = data->nr;
2891
2892 if (data->nr == 0)
2893 return;
2894
2895 for (i = 0; i < data->nr; i++) {
2896 int added = data->files[i]->added;
2897 int deleted = data->files[i]->deleted;
2898
2899 if (data->files[i]->is_unmerged ||
2900 (!data->files[i]->is_interesting && (added + deleted == 0))) {
2901 total_files--;
2902 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2903 adds += added;
2904 dels += deleted;
2905 }
2906 }
2907 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2908 }
2909
2910 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2911 {
2912 int i;
2913
2914 if (data->nr == 0)
2915 return;
2916
2917 for (i = 0; i < data->nr; i++) {
2918 struct diffstat_file *file = data->files[i];
2919
2920 fprintf(options->file, "%s", diff_line_prefix(options));
2921
2922 if (file->is_binary)
2923 fprintf(options->file, "-\t-\t");
2924 else
2925 fprintf(options->file,
2926 "%"PRIuMAX"\t%"PRIuMAX"\t",
2927 file->added, file->deleted);
2928 if (options->line_termination) {
2929 fill_print_name(file);
2930 if (!file->is_renamed)
2931 write_name_quoted(file->name, options->file,
2932 options->line_termination);
2933 else {
2934 fputs(file->print_name, options->file);
2935 putc(options->line_termination, options->file);
2936 }
2937 } else {
2938 if (file->is_renamed) {
2939 putc('\0', options->file);
2940 write_name_quoted(file->from_name, options->file, '\0');
2941 }
2942 write_name_quoted(file->name, options->file, '\0');
2943 }
2944 }
2945 }
2946
2947 struct dirstat_file {
2948 const char *name;
2949 unsigned long changed;
2950 };
2951
2952 struct dirstat_dir {
2953 struct dirstat_file *files;
2954 int alloc, nr, permille, cumulative;
2955 };
2956
2957 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2958 unsigned long changed, const char *base, int baselen)
2959 {
2960 unsigned long sum_changes = 0;
2961 unsigned int sources = 0;
2962 const char *line_prefix = diff_line_prefix(opt);
2963
2964 while (dir->nr) {
2965 struct dirstat_file *f = dir->files;
2966 int namelen = strlen(f->name);
2967 unsigned long changes;
2968 char *slash;
2969
2970 if (namelen < baselen)
2971 break;
2972 if (memcmp(f->name, base, baselen))
2973 break;
2974 slash = strchr(f->name + baselen, '/');
2975 if (slash) {
2976 int newbaselen = slash + 1 - f->name;
2977 changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2978 sources++;
2979 } else {
2980 changes = f->changed;
2981 dir->files++;
2982 dir->nr--;
2983 sources += 2;
2984 }
2985 sum_changes += changes;
2986 }
2987
2988 /*
2989 * We don't report dirstat's for
2990 * - the top level
2991 * - or cases where everything came from a single directory
2992 * under this directory (sources == 1).
2993 */
2994 if (baselen && sources != 1) {
2995 if (sum_changes) {
2996 int permille = sum_changes * 1000 / changed;
2997 if (permille >= dir->permille) {
2998 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2999 permille / 10, permille % 10, baselen, base);
3000 if (!dir->cumulative)
3001 return 0;
3002 }
3003 }
3004 }
3005 return sum_changes;
3006 }
3007
3008 static int dirstat_compare(const void *_a, const void *_b)
3009 {
3010 const struct dirstat_file *a = _a;
3011 const struct dirstat_file *b = _b;
3012 return strcmp(a->name, b->name);
3013 }
3014
3015 static void show_dirstat(struct diff_options *options)
3016 {
3017 int i;
3018 unsigned long changed;
3019 struct dirstat_dir dir;
3020 struct diff_queue_struct *q = &diff_queued_diff;
3021
3022 dir.files = NULL;
3023 dir.alloc = 0;
3024 dir.nr = 0;
3025 dir.permille = options->dirstat_permille;
3026 dir.cumulative = options->flags.dirstat_cumulative;
3027
3028 changed = 0;
3029 for (i = 0; i < q->nr; i++) {
3030 struct diff_filepair *p = q->queue[i];
3031 const char *name;
3032 unsigned long copied, added, damage;
3033
3034 name = p->two->path ? p->two->path : p->one->path;
3035
3036 if (p->one->oid_valid && p->two->oid_valid &&
3037 oideq(&p->one->oid, &p->two->oid)) {
3038 /*
3039 * The SHA1 has not changed, so pre-/post-content is
3040 * identical. We can therefore skip looking at the
3041 * file contents altogether.
3042 */
3043 damage = 0;
3044 goto found_damage;
3045 }
3046
3047 if (options->flags.dirstat_by_file) {
3048 /*
3049 * In --dirstat-by-file mode, we don't really need to
3050 * look at the actual file contents at all.
3051 * The fact that the SHA1 changed is enough for us to
3052 * add this file to the list of results
3053 * (with each file contributing equal damage).
3054 */
3055 damage = 1;
3056 goto found_damage;
3057 }
3058
3059 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3060 diff_populate_filespec(options->repo, p->one, 0);
3061 diff_populate_filespec(options->repo, p->two, 0);
3062 diffcore_count_changes(options->repo,
3063 p->one, p->two, NULL, NULL,
3064 &copied, &added);
3065 diff_free_filespec_data(p->one);
3066 diff_free_filespec_data(p->two);
3067 } else if (DIFF_FILE_VALID(p->one)) {
3068 diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3069 copied = added = 0;
3070 diff_free_filespec_data(p->one);
3071 } else if (DIFF_FILE_VALID(p->two)) {
3072 diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3073 copied = 0;
3074 added = p->two->size;
3075 diff_free_filespec_data(p->two);
3076 } else
3077 continue;
3078
3079 /*
3080 * Original minus copied is the removed material,
3081 * added is the new material. They are both damages
3082 * made to the preimage.
3083 * If the resulting damage is zero, we know that
3084 * diffcore_count_changes() considers the two entries to
3085 * be identical, but since the oid changed, we
3086 * know that there must have been _some_ kind of change,
3087 * so we force all entries to have damage > 0.
3088 */
3089 damage = (p->one->size - copied) + added;
3090 if (!damage)
3091 damage = 1;
3092
3093 found_damage:
3094 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3095 dir.files[dir.nr].name = name;
3096 dir.files[dir.nr].changed = damage;
3097 changed += damage;
3098 dir.nr++;
3099 }
3100
3101 /* This can happen even with many files, if everything was renames */
3102 if (!changed)
3103 return;
3104
3105 /* Show all directories with more than x% of the changes */
3106 QSORT(dir.files, dir.nr, dirstat_compare);
3107 gather_dirstat(options, &dir, changed, "", 0);
3108 }
3109
3110 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3111 {
3112 int i;
3113 unsigned long changed;
3114 struct dirstat_dir dir;
3115
3116 if (data->nr == 0)
3117 return;
3118
3119 dir.files = NULL;
3120 dir.alloc = 0;
3121 dir.nr = 0;
3122 dir.permille = options->dirstat_permille;
3123 dir.cumulative = options->flags.dirstat_cumulative;
3124
3125 changed = 0;
3126 for (i = 0; i < data->nr; i++) {
3127 struct diffstat_file *file = data->files[i];
3128 unsigned long damage = file->added + file->deleted;
3129 if (file->is_binary)
3130 /*
3131 * binary files counts bytes, not lines. Must find some
3132 * way to normalize binary bytes vs. textual lines.
3133 * The following heuristic assumes that there are 64
3134 * bytes per "line".
3135 * This is stupid and ugly, but very cheap...
3136 */
3137 damage = DIV_ROUND_UP(damage, 64);
3138 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3139 dir.files[dir.nr].name = file->name;
3140 dir.files[dir.nr].changed = damage;
3141 changed += damage;
3142 dir.nr++;
3143 }
3144
3145 /* This can happen even with many files, if everything was renames */
3146 if (!changed)
3147 return;
3148
3149 /* Show all directories with more than x% of the changes */
3150 QSORT(dir.files, dir.nr, dirstat_compare);
3151 gather_dirstat(options, &dir, changed, "", 0);
3152 }
3153
3154 static void free_diffstat_info(struct diffstat_t *diffstat)
3155 {
3156 int i;
3157 for (i = 0; i < diffstat->nr; i++) {
3158 struct diffstat_file *f = diffstat->files[i];
3159 free(f->print_name);
3160 free(f->name);
3161 free(f->from_name);
3162 free(f);
3163 }
3164 free(diffstat->files);
3165 }
3166
3167 struct checkdiff_t {
3168 const char *filename;
3169 int lineno;
3170 int conflict_marker_size;
3171 struct diff_options *o;
3172 unsigned ws_rule;
3173 unsigned status;
3174 };
3175
3176 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3177 {
3178 char firstchar;
3179 int cnt;
3180
3181 if (len < marker_size + 1)
3182 return 0;
3183 firstchar = line[0];
3184 switch (firstchar) {
3185 case '=': case '>': case '<': case '|':
3186 break;
3187 default:
3188 return 0;
3189 }
3190 for (cnt = 1; cnt < marker_size; cnt++)
3191 if (line[cnt] != firstchar)
3192 return 0;
3193 /* line[1] thru line[marker_size-1] are same as firstchar */
3194 if (len < marker_size + 1 || !isspace(line[marker_size]))
3195 return 0;
3196 return 1;
3197 }
3198
3199 static void checkdiff_consume_hunk(void *priv,
3200 long ob, long on, long nb, long nn,
3201 const char *func, long funclen)
3202
3203 {
3204 struct checkdiff_t *data = priv;
3205 data->lineno = nb - 1;
3206 }
3207
3208 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3209 {
3210 struct checkdiff_t *data = priv;
3211 int marker_size = data->conflict_marker_size;
3212 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3213 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3214 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3215 char *err;
3216 const char *line_prefix;
3217
3218 assert(data->o);
3219 line_prefix = diff_line_prefix(data->o);
3220
3221 if (line[0] == '+') {
3222 unsigned bad;
3223 data->lineno++;
3224 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3225 data->status |= 1;
3226 fprintf(data->o->file,
3227 "%s%s:%d: leftover conflict marker\n",
3228 line_prefix, data->filename, data->lineno);
3229 }
3230 bad = ws_check(line + 1, len - 1, data->ws_rule);
3231 if (!bad)
3232 return;
3233 data->status |= bad;
3234 err = whitespace_error_string(bad);
3235 fprintf(data->o->file, "%s%s:%d: %s.\n",
3236 line_prefix, data->filename, data->lineno, err);
3237 free(err);
3238 emit_line(data->o, set, reset, line, 1);
3239 ws_check_emit(line + 1, len - 1, data->ws_rule,
3240 data->o->file, set, reset, ws);
3241 } else if (line[0] == ' ') {
3242 data->lineno++;
3243 }
3244 }
3245
3246 static unsigned char *deflate_it(char *data,
3247 unsigned long size,
3248 unsigned long *result_size)
3249 {
3250 int bound;
3251 unsigned char *deflated;
3252 git_zstream stream;
3253
3254 git_deflate_init(&stream, zlib_compression_level);
3255 bound = git_deflate_bound(&stream, size);
3256 deflated = xmalloc(bound);
3257 stream.next_out = deflated;
3258 stream.avail_out = bound;
3259
3260 stream.next_in = (unsigned char *)data;
3261 stream.avail_in = size;
3262 while (git_deflate(&stream, Z_FINISH) == Z_OK)
3263 ; /* nothing */
3264 git_deflate_end(&stream);
3265 *result_size = stream.total_out;
3266 return deflated;
3267 }
3268
3269 static void emit_binary_diff_body(struct diff_options *o,
3270 mmfile_t *one, mmfile_t *two)
3271 {
3272 void *cp;
3273 void *delta;
3274 void *deflated;
3275 void *data;
3276 unsigned long orig_size;
3277 unsigned long delta_size;
3278 unsigned long deflate_size;
3279 unsigned long data_size;
3280
3281 /* We could do deflated delta, or we could do just deflated two,
3282 * whichever is smaller.
3283 */
3284 delta = NULL;
3285 deflated = deflate_it(two->ptr, two->size, &deflate_size);
3286 if (one->size && two->size) {
3287 delta = diff_delta(one->ptr, one->size,
3288 two->ptr, two->size,
3289 &delta_size, deflate_size);
3290 if (delta) {
3291 void *to_free = delta;
3292 orig_size = delta_size;
3293 delta = deflate_it(delta, delta_size, &delta_size);
3294 free(to_free);
3295 }
3296 }
3297
3298 if (delta && delta_size < deflate_size) {
3299 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3300 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3301 s, strlen(s), 0);
3302 free(s);
3303 free(deflated);
3304 data = delta;
3305 data_size = delta_size;
3306 } else {
3307 char *s = xstrfmt("%lu", two->size);
3308 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3309 s, strlen(s), 0);
3310 free(s);
3311 free(delta);
3312 data = deflated;
3313 data_size = deflate_size;
3314 }
3315
3316 /* emit data encoded in base85 */
3317 cp = data;
3318 while (data_size) {
3319 int len;
3320 int bytes = (52 < data_size) ? 52 : data_size;
3321 char line[71];
3322 data_size -= bytes;
3323 if (bytes <= 26)
3324 line[0] = bytes + 'A' - 1;
3325 else
3326 line[0] = bytes - 26 + 'a' - 1;
3327 encode_85(line + 1, cp, bytes);
3328 cp = (char *) cp + bytes;
3329
3330 len = strlen(line);
3331 line[len++] = '\n';
3332 line[len] = '\0';
3333
3334 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3335 line, len, 0);
3336 }
3337 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3338 free(data);
3339 }
3340
3341 static void emit_binary_diff(struct diff_options *o,
3342 mmfile_t *one, mmfile_t *two)
3343 {
3344 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3345 emit_binary_diff_body(o, one, two);
3346 emit_binary_diff_body(o, two, one);
3347 }
3348
3349 int diff_filespec_is_binary(struct repository *r,
3350 struct diff_filespec *one)
3351 {
3352 if (one->is_binary == -1) {
3353 diff_filespec_load_driver(one, r->index);
3354 if (one->driver->binary != -1)
3355 one->is_binary = one->driver->binary;
3356 else {
3357 if (!one->data && DIFF_FILE_VALID(one))
3358 diff_populate_filespec(r, one, CHECK_BINARY);
3359 if (one->is_binary == -1 && one->data)
3360 one->is_binary = buffer_is_binary(one->data,
3361 one->size);
3362 if (one->is_binary == -1)
3363 one->is_binary = 0;
3364 }
3365 }
3366 return one->is_binary;
3367 }
3368
3369 static const struct userdiff_funcname *
3370 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3371 {
3372 diff_filespec_load_driver(one, o->repo->index);
3373 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3374 }
3375
3376 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3377 {
3378 if (!options->a_prefix)
3379 options->a_prefix = a;
3380 if (!options->b_prefix)
3381 options->b_prefix = b;
3382 }
3383
3384 struct userdiff_driver *get_textconv(struct repository *r,
3385 struct diff_filespec *one)
3386 {
3387 if (!DIFF_FILE_VALID(one))
3388 return NULL;
3389
3390 diff_filespec_load_driver(one, r->index);
3391 return userdiff_get_textconv(r, one->driver);
3392 }
3393
3394 static void builtin_diff(const char *name_a,
3395 const char *name_b,
3396 struct diff_filespec *one,
3397 struct diff_filespec *two,
3398 const char *xfrm_msg,
3399 int must_show_header,
3400 struct diff_options *o,
3401 int complete_rewrite)
3402 {
3403 mmfile_t mf1, mf2;
3404 const char *lbl[2];
3405 char *a_one, *b_two;
3406 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3407 const char *reset = diff_get_color_opt(o, DIFF_RESET);
3408 const char *a_prefix, *b_prefix;
3409 struct userdiff_driver *textconv_one = NULL;
3410 struct userdiff_driver *textconv_two = NULL;
3411 struct strbuf header = STRBUF_INIT;
3412 const char *line_prefix = diff_line_prefix(o);
3413
3414 diff_set_mnemonic_prefix(o, "a/", "b/");
3415 if (o->flags.reverse_diff) {
3416 a_prefix = o->b_prefix;
3417 b_prefix = o->a_prefix;
3418 } else {
3419 a_prefix = o->a_prefix;
3420 b_prefix = o->b_prefix;
3421 }
3422
3423 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3424 (!one->mode || S_ISGITLINK(one->mode)) &&
3425 (!two->mode || S_ISGITLINK(two->mode))) {
3426 show_submodule_summary(o, one->path ? one->path : two->path,
3427 &one->oid, &two->oid,
3428 two->dirty_submodule);
3429 return;
3430 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3431 (!one->mode || S_ISGITLINK(one->mode)) &&
3432 (!two->mode || S_ISGITLINK(two->mode))) {
3433 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3434 &one->oid, &two->oid,
3435 two->dirty_submodule);
3436 return;
3437 }
3438
3439 if (o->flags.allow_textconv) {
3440 textconv_one = get_textconv(o->repo, one);
3441 textconv_two = get_textconv(o->repo, two);
3442 }
3443
3444 /* Never use a non-valid filename anywhere if at all possible */
3445 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3446 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3447
3448 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3449 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3450 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3451 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3452 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3453 if (lbl[0][0] == '/') {
3454 /* /dev/null */
3455 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3456 if (xfrm_msg)
3457 strbuf_addstr(&header, xfrm_msg);
3458 must_show_header = 1;
3459 }
3460 else if (lbl[1][0] == '/') {
3461 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3462 if (xfrm_msg)
3463 strbuf_addstr(&header, xfrm_msg);
3464 must_show_header = 1;
3465 }
3466 else {
3467 if (one->mode != two->mode) {
3468 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3469 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3470 must_show_header = 1;
3471 }
3472 if (xfrm_msg)
3473 strbuf_addstr(&header, xfrm_msg);
3474
3475 /*
3476 * we do not run diff between different kind
3477 * of objects.
3478 */
3479 if ((one->mode ^ two->mode) & S_IFMT)
3480 goto free_ab_and_return;
3481 if (complete_rewrite &&
3482 (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3483 (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3484 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3485 header.buf, header.len, 0);
3486 strbuf_reset(&header);
3487 emit_rewrite_diff(name_a, name_b, one, two,
3488 textconv_one, textconv_two, o);
3489 o->found_changes = 1;
3490 goto free_ab_and_return;
3491 }
3492 }
3493
3494 if (o->irreversible_delete && lbl[1][0] == '/') {
3495 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3496 header.len, 0);
3497 strbuf_reset(&header);
3498 goto free_ab_and_return;
3499 } else if (!o->flags.text &&
3500 ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3501 (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3502 struct strbuf sb = STRBUF_INIT;
3503 if (!one->data && !two->data &&
3504 S_ISREG(one->mode) && S_ISREG(two->mode) &&
3505 !o->flags.binary) {
3506 if (oideq(&one->oid, &two->oid)) {
3507 if (must_show_header)
3508 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3509 header.buf, header.len,
3510 0);
3511 goto free_ab_and_return;
3512 }
3513 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3514 header.buf, header.len, 0);
3515 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3516 diff_line_prefix(o), lbl[0], lbl[1]);
3517 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3518 sb.buf, sb.len, 0);
3519 strbuf_release(&sb);
3520 goto free_ab_and_return;
3521 }
3522 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3523 fill_mmfile(o->repo, &mf2, two) < 0)
3524 die("unable to read files to diff");
3525 /* Quite common confusing case */
3526 if (mf1.size == mf2.size &&
3527 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3528 if (must_show_header)
3529 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3530 header.buf, header.len, 0);
3531 goto free_ab_and_return;
3532 }
3533 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3534 strbuf_reset(&header);
3535 if (o->flags.binary)
3536 emit_binary_diff(o, &mf1, &mf2);
3537 else {
3538 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3539 diff_line_prefix(o), lbl[0], lbl[1]);
3540 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3541 sb.buf, sb.len, 0);
3542 strbuf_release(&sb);
3543 }
3544 o->found_changes = 1;
3545 } else {
3546 /* Crazy xdl interfaces.. */
3547 const char *diffopts;
3548 const char *v;
3549 xpparam_t xpp;
3550 xdemitconf_t xecfg;
3551 struct emit_callback ecbdata;
3552 const struct userdiff_funcname *pe;
3553
3554 if (must_show_header) {
3555 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3556 header.buf, header.len, 0);
3557 strbuf_reset(&header);
3558 }
3559
3560 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3561 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3562
3563 pe = diff_funcname_pattern(o, one);
3564 if (!pe)
3565 pe = diff_funcname_pattern(o, two);
3566
3567 memset(&xpp, 0, sizeof(xpp));
3568 memset(&xecfg, 0, sizeof(xecfg));
3569 memset(&ecbdata, 0, sizeof(ecbdata));
3570 if (o->flags.suppress_diff_headers)
3571 lbl[0] = NULL;
3572 ecbdata.label_path = lbl;
3573 ecbdata.color_diff = want_color(o->use_color);
3574 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3575 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3576 check_blank_at_eof(&mf1, &mf2, &ecbdata);
3577 ecbdata.opt = o;
3578 if (header.len && !o->flags.suppress_diff_headers)
3579 ecbdata.header = &header;
3580 xpp.flags = o->xdl_opts;
3581 xpp.anchors = o->anchors;
3582 xpp.anchors_nr = o->anchors_nr;
3583 xecfg.ctxlen = o->context;
3584 xecfg.interhunkctxlen = o->interhunkcontext;
3585 xecfg.flags = XDL_EMIT_FUNCNAMES;
3586 if (o->flags.funccontext)
3587 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3588 if (pe)
3589 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3590
3591 diffopts = getenv("GIT_DIFF_OPTS");
3592 if (!diffopts)
3593 ;
3594 else if (skip_prefix(diffopts, "--unified=", &v))
3595 xecfg.ctxlen = strtoul(v, NULL, 10);
3596 else if (skip_prefix(diffopts, "-u", &v))
3597 xecfg.ctxlen = strtoul(v, NULL, 10);
3598
3599 if (o->word_diff)
3600 init_diff_words_data(&ecbdata, o, one, two);
3601 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3602 &ecbdata, &xpp, &xecfg))
3603 die("unable to generate diff for %s", one->path);
3604 if (o->word_diff)
3605 free_diff_words_data(&ecbdata);
3606 if (textconv_one)
3607 free(mf1.ptr);
3608 if (textconv_two)
3609 free(mf2.ptr);
3610 xdiff_clear_find_func(&xecfg);
3611 }
3612
3613 free_ab_and_return:
3614 strbuf_release(&header);
3615 diff_free_filespec_data(one);
3616 diff_free_filespec_data(two);
3617 free(a_one);
3618 free(b_two);
3619 return;
3620 }
3621
3622 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3623 {
3624 if (!is_renamed) {
3625 if (p->status == DIFF_STATUS_ADDED) {
3626 if (S_ISLNK(p->two->mode))
3627 return "new +l";
3628 else if ((p->two->mode & 0777) == 0755)
3629 return "new +x";
3630 else
3631 return "new";
3632 } else if (p->status == DIFF_STATUS_DELETED)
3633 return "gone";
3634 }
3635 if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3636 return "mode -l";
3637 else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3638 return "mode +l";
3639 else if ((p->one->mode & 0777) == 0644 &&
3640 (p->two->mode & 0777) == 0755)
3641 return "mode +x";
3642 else if ((p->one->mode & 0777) == 0755 &&
3643 (p->two->mode & 0777) == 0644)
3644 return "mode -x";
3645 return NULL;
3646 }
3647
3648 static void builtin_diffstat(const char *name_a, const char *name_b,
3649 struct diff_filespec *one,
3650 struct diff_filespec *two,
3651 struct diffstat_t *diffstat,
3652 struct diff_options *o,
3653 struct diff_filepair *p)
3654 {
3655 mmfile_t mf1, mf2;
3656 struct diffstat_file *data;
3657 int same_contents;
3658 int complete_rewrite = 0;
3659
3660 if (!DIFF_PAIR_UNMERGED(p)) {
3661 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3662 complete_rewrite = 1;
3663 }
3664
3665 data = diffstat_add(diffstat, name_a, name_b);
3666 data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3667 if (o->flags.stat_with_summary)
3668 data->comments = get_compact_summary(p, data->is_renamed);
3669
3670 if (!one || !two) {
3671 data->is_unmerged = 1;
3672 return;
3673 }
3674
3675 same_contents = oideq(&one->oid, &two->oid);
3676
3677 if (diff_filespec_is_binary(o->repo, one) ||
3678 diff_filespec_is_binary(o->repo, two)) {
3679 data->is_binary = 1;
3680 if (same_contents) {
3681 data->added = 0;
3682 data->deleted = 0;
3683 } else {
3684 data->added = diff_filespec_size(o->repo, two);
3685 data->deleted = diff_filespec_size(o->repo, one);
3686 }
3687 }
3688
3689 else if (complete_rewrite) {
3690 diff_populate_filespec(o->repo, one, 0);
3691 diff_populate_filespec(o->repo, two, 0);
3692 data->deleted = count_lines(one->data, one->size);
3693 data->added = count_lines(two->data, two->size);
3694 }
3695
3696 else if (!same_contents) {
3697 /* Crazy xdl interfaces.. */
3698 xpparam_t xpp;
3699 xdemitconf_t xecfg;
3700
3701 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3702 fill_mmfile(o->repo, &mf2, two) < 0)
3703 die("unable to read files to diff");
3704
3705 memset(&xpp, 0, sizeof(xpp));
3706 memset(&xecfg, 0, sizeof(xecfg));
3707 xpp.flags = o->xdl_opts;
3708 xpp.anchors = o->anchors;
3709 xpp.anchors_nr = o->anchors_nr;
3710 xecfg.ctxlen = o->context;
3711 xecfg.interhunkctxlen = o->interhunkcontext;
3712 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3713 diffstat_consume, diffstat, &xpp, &xecfg))
3714 die("unable to generate diffstat for %s", one->path);
3715 }
3716
3717 diff_free_filespec_data(one);
3718 diff_free_filespec_data(two);