diff-parseopt: correct variable types that are used by parseopt
[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 "parse-options.h"
27 #include "help.h"
28
29 #ifdef NO_FAST_WORKING_DIRECTORY
30 #define FAST_WORKING_DIRECTORY 0
31 #else
32 #define FAST_WORKING_DIRECTORY 1
33 #endif
34
35 static int diff_detect_rename_default;
36 static int diff_indent_heuristic = 1;
37 static int diff_rename_limit_default = 400;
38 static int diff_suppress_blank_empty;
39 static int diff_use_color_default = -1;
40 static int diff_color_moved_default;
41 static int diff_color_moved_ws_default;
42 static int diff_context_default = 3;
43 static int diff_interhunk_context_default;
44 static const char *diff_word_regex_cfg;
45 static const char *external_diff_cmd_cfg;
46 static const char *diff_order_file_cfg;
47 int diff_auto_refresh_index = 1;
48 static int diff_mnemonic_prefix;
49 static int diff_no_prefix;
50 static int diff_stat_graph_width;
51 static int diff_dirstat_permille_default = 30;
52 static struct diff_options default_diff_options;
53 static long diff_algorithm;
54 static unsigned ws_error_highlight_default = WSEH_NEW;
55
56 static char diff_colors[][COLOR_MAXLEN] = {
57 GIT_COLOR_RESET,
58 GIT_COLOR_NORMAL, /* CONTEXT */
59 GIT_COLOR_BOLD, /* METAINFO */
60 GIT_COLOR_CYAN, /* FRAGINFO */
61 GIT_COLOR_RED, /* OLD */
62 GIT_COLOR_GREEN, /* NEW */
63 GIT_COLOR_YELLOW, /* COMMIT */
64 GIT_COLOR_BG_RED, /* WHITESPACE */
65 GIT_COLOR_NORMAL, /* FUNCINFO */
66 GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
67 GIT_COLOR_BOLD_BLUE, /* OLD_MOVED ALTERNATIVE */
68 GIT_COLOR_FAINT, /* OLD_MOVED_DIM */
69 GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
70 GIT_COLOR_BOLD_CYAN, /* NEW_MOVED */
71 GIT_COLOR_BOLD_YELLOW, /* NEW_MOVED ALTERNATIVE */
72 GIT_COLOR_FAINT, /* NEW_MOVED_DIM */
73 GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
74 GIT_COLOR_FAINT, /* CONTEXT_DIM */
75 GIT_COLOR_FAINT_RED, /* OLD_DIM */
76 GIT_COLOR_FAINT_GREEN, /* NEW_DIM */
77 GIT_COLOR_BOLD, /* CONTEXT_BOLD */
78 GIT_COLOR_BOLD_RED, /* OLD_BOLD */
79 GIT_COLOR_BOLD_GREEN, /* NEW_BOLD */
80 };
81
82 static const char *color_diff_slots[] = {
83 [DIFF_CONTEXT] = "context",
84 [DIFF_METAINFO] = "meta",
85 [DIFF_FRAGINFO] = "frag",
86 [DIFF_FILE_OLD] = "old",
87 [DIFF_FILE_NEW] = "new",
88 [DIFF_COMMIT] = "commit",
89 [DIFF_WHITESPACE] = "whitespace",
90 [DIFF_FUNCINFO] = "func",
91 [DIFF_FILE_OLD_MOVED] = "oldMoved",
92 [DIFF_FILE_OLD_MOVED_ALT] = "oldMovedAlternative",
93 [DIFF_FILE_OLD_MOVED_DIM] = "oldMovedDimmed",
94 [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
95 [DIFF_FILE_NEW_MOVED] = "newMoved",
96 [DIFF_FILE_NEW_MOVED_ALT] = "newMovedAlternative",
97 [DIFF_FILE_NEW_MOVED_DIM] = "newMovedDimmed",
98 [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
99 [DIFF_CONTEXT_DIM] = "contextDimmed",
100 [DIFF_FILE_OLD_DIM] = "oldDimmed",
101 [DIFF_FILE_NEW_DIM] = "newDimmed",
102 [DIFF_CONTEXT_BOLD] = "contextBold",
103 [DIFF_FILE_OLD_BOLD] = "oldBold",
104 [DIFF_FILE_NEW_BOLD] = "newBold",
105 };
106
107 static NORETURN void die_want_option(const char *option_name)
108 {
109 die(_("option '%s' requires a value"), option_name);
110 }
111
112 define_list_config_array_extra(color_diff_slots, {"plain"});
113
114 static int parse_diff_color_slot(const char *var)
115 {
116 if (!strcasecmp(var, "plain"))
117 return DIFF_CONTEXT;
118 return LOOKUP_CONFIG(color_diff_slots, var);
119 }
120
121 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
122 struct strbuf *errmsg)
123 {
124 char *params_copy = xstrdup(params_string);
125 struct string_list params = STRING_LIST_INIT_NODUP;
126 int ret = 0;
127 int i;
128
129 if (*params_copy)
130 string_list_split_in_place(&params, params_copy, ',', -1);
131 for (i = 0; i < params.nr; i++) {
132 const char *p = params.items[i].string;
133 if (!strcmp(p, "changes")) {
134 options->flags.dirstat_by_line = 0;
135 options->flags.dirstat_by_file = 0;
136 } else if (!strcmp(p, "lines")) {
137 options->flags.dirstat_by_line = 1;
138 options->flags.dirstat_by_file = 0;
139 } else if (!strcmp(p, "files")) {
140 options->flags.dirstat_by_line = 0;
141 options->flags.dirstat_by_file = 1;
142 } else if (!strcmp(p, "noncumulative")) {
143 options->flags.dirstat_cumulative = 0;
144 } else if (!strcmp(p, "cumulative")) {
145 options->flags.dirstat_cumulative = 1;
146 } else if (isdigit(*p)) {
147 char *end;
148 int permille = strtoul(p, &end, 10) * 10;
149 if (*end == '.' && isdigit(*++end)) {
150 /* only use first digit */
151 permille += *end - '0';
152 /* .. and ignore any further digits */
153 while (isdigit(*++end))
154 ; /* nothing */
155 }
156 if (!*end)
157 options->dirstat_permille = permille;
158 else {
159 strbuf_addf(errmsg, _(" Failed to parse dirstat cut-off percentage '%s'\n"),
160 p);
161 ret++;
162 }
163 } else {
164 strbuf_addf(errmsg, _(" Unknown dirstat parameter '%s'\n"), p);
165 ret++;
166 }
167
168 }
169 string_list_clear(&params, 0);
170 free(params_copy);
171 return ret;
172 }
173
174 static int parse_submodule_params(struct diff_options *options, const char *value)
175 {
176 if (!strcmp(value, "log"))
177 options->submodule_format = DIFF_SUBMODULE_LOG;
178 else if (!strcmp(value, "short"))
179 options->submodule_format = DIFF_SUBMODULE_SHORT;
180 else if (!strcmp(value, "diff"))
181 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
182 else
183 return -1;
184 return 0;
185 }
186
187 int git_config_rename(const char *var, const char *value)
188 {
189 if (!value)
190 return DIFF_DETECT_RENAME;
191 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
192 return DIFF_DETECT_COPY;
193 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
194 }
195
196 long parse_algorithm_value(const char *value)
197 {
198 if (!value)
199 return -1;
200 else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
201 return 0;
202 else if (!strcasecmp(value, "minimal"))
203 return XDF_NEED_MINIMAL;
204 else if (!strcasecmp(value, "patience"))
205 return XDF_PATIENCE_DIFF;
206 else if (!strcasecmp(value, "histogram"))
207 return XDF_HISTOGRAM_DIFF;
208 return -1;
209 }
210
211 static int parse_one_token(const char **arg, const char *token)
212 {
213 const char *rest;
214 if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
215 *arg = rest;
216 return 1;
217 }
218 return 0;
219 }
220
221 static int parse_ws_error_highlight(const char *arg)
222 {
223 const char *orig_arg = arg;
224 unsigned val = 0;
225
226 while (*arg) {
227 if (parse_one_token(&arg, "none"))
228 val = 0;
229 else if (parse_one_token(&arg, "default"))
230 val = WSEH_NEW;
231 else if (parse_one_token(&arg, "all"))
232 val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
233 else if (parse_one_token(&arg, "new"))
234 val |= WSEH_NEW;
235 else if (parse_one_token(&arg, "old"))
236 val |= WSEH_OLD;
237 else if (parse_one_token(&arg, "context"))
238 val |= WSEH_CONTEXT;
239 else {
240 return -1 - (int)(arg - orig_arg);
241 }
242 if (*arg)
243 arg++;
244 }
245 return val;
246 }
247
248 /*
249 * These are to give UI layer defaults.
250 * The core-level commands such as git-diff-files should
251 * never be affected by the setting of diff.renames
252 * the user happens to have in the configuration file.
253 */
254 void init_diff_ui_defaults(void)
255 {
256 diff_detect_rename_default = DIFF_DETECT_RENAME;
257 }
258
259 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
260 {
261 if (!strcmp(var, "diff.indentheuristic"))
262 diff_indent_heuristic = git_config_bool(var, value);
263 return 0;
264 }
265
266 static int parse_color_moved(const char *arg)
267 {
268 switch (git_parse_maybe_bool(arg)) {
269 case 0:
270 return COLOR_MOVED_NO;
271 case 1:
272 return COLOR_MOVED_DEFAULT;
273 default:
274 break;
275 }
276
277 if (!strcmp(arg, "no"))
278 return COLOR_MOVED_NO;
279 else if (!strcmp(arg, "plain"))
280 return COLOR_MOVED_PLAIN;
281 else if (!strcmp(arg, "blocks"))
282 return COLOR_MOVED_BLOCKS;
283 else if (!strcmp(arg, "zebra"))
284 return COLOR_MOVED_ZEBRA;
285 else if (!strcmp(arg, "default"))
286 return COLOR_MOVED_DEFAULT;
287 else if (!strcmp(arg, "dimmed-zebra"))
288 return COLOR_MOVED_ZEBRA_DIM;
289 else if (!strcmp(arg, "dimmed_zebra"))
290 return COLOR_MOVED_ZEBRA_DIM;
291 else
292 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
293 }
294
295 static unsigned parse_color_moved_ws(const char *arg)
296 {
297 int ret = 0;
298 struct string_list l = STRING_LIST_INIT_DUP;
299 struct string_list_item *i;
300
301 string_list_split(&l, arg, ',', -1);
302
303 for_each_string_list_item(i, &l) {
304 struct strbuf sb = STRBUF_INIT;
305 strbuf_addstr(&sb, i->string);
306 strbuf_trim(&sb);
307
308 if (!strcmp(sb.buf, "ignore-space-change"))
309 ret |= XDF_IGNORE_WHITESPACE_CHANGE;
310 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
311 ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
312 else if (!strcmp(sb.buf, "ignore-all-space"))
313 ret |= XDF_IGNORE_WHITESPACE;
314 else if (!strcmp(sb.buf, "allow-indentation-change"))
315 ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
316 else {
317 ret |= COLOR_MOVED_WS_ERROR;
318 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);
319 }
320
321 strbuf_release(&sb);
322 }
323
324 if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
325 (ret & XDF_WHITESPACE_FLAGS)) {
326 error(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
327 ret |= COLOR_MOVED_WS_ERROR;
328 }
329
330 string_list_clear(&l, 0);
331
332 return ret;
333 }
334
335 int git_diff_ui_config(const char *var, const char *value, void *cb)
336 {
337 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
338 diff_use_color_default = git_config_colorbool(var, value);
339 return 0;
340 }
341 if (!strcmp(var, "diff.colormoved")) {
342 int cm = parse_color_moved(value);
343 if (cm < 0)
344 return -1;
345 diff_color_moved_default = cm;
346 return 0;
347 }
348 if (!strcmp(var, "diff.colormovedws")) {
349 unsigned cm = parse_color_moved_ws(value);
350 if (cm & COLOR_MOVED_WS_ERROR)
351 return -1;
352 diff_color_moved_ws_default = cm;
353 return 0;
354 }
355 if (!strcmp(var, "diff.context")) {
356 diff_context_default = git_config_int(var, value);
357 if (diff_context_default < 0)
358 return -1;
359 return 0;
360 }
361 if (!strcmp(var, "diff.interhunkcontext")) {
362 diff_interhunk_context_default = git_config_int(var, value);
363 if (diff_interhunk_context_default < 0)
364 return -1;
365 return 0;
366 }
367 if (!strcmp(var, "diff.renames")) {
368 diff_detect_rename_default = git_config_rename(var, value);
369 return 0;
370 }
371 if (!strcmp(var, "diff.autorefreshindex")) {
372 diff_auto_refresh_index = git_config_bool(var, value);
373 return 0;
374 }
375 if (!strcmp(var, "diff.mnemonicprefix")) {
376 diff_mnemonic_prefix = git_config_bool(var, value);
377 return 0;
378 }
379 if (!strcmp(var, "diff.noprefix")) {
380 diff_no_prefix = git_config_bool(var, value);
381 return 0;
382 }
383 if (!strcmp(var, "diff.statgraphwidth")) {
384 diff_stat_graph_width = git_config_int(var, value);
385 return 0;
386 }
387 if (!strcmp(var, "diff.external"))
388 return git_config_string(&external_diff_cmd_cfg, var, value);
389 if (!strcmp(var, "diff.wordregex"))
390 return git_config_string(&diff_word_regex_cfg, var, value);
391 if (!strcmp(var, "diff.orderfile"))
392 return git_config_pathname(&diff_order_file_cfg, var, value);
393
394 if (!strcmp(var, "diff.ignoresubmodules"))
395 handle_ignore_submodules_arg(&default_diff_options, value);
396
397 if (!strcmp(var, "diff.submodule")) {
398 if (parse_submodule_params(&default_diff_options, value))
399 warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
400 value);
401 return 0;
402 }
403
404 if (!strcmp(var, "diff.algorithm")) {
405 diff_algorithm = parse_algorithm_value(value);
406 if (diff_algorithm < 0)
407 return -1;
408 return 0;
409 }
410
411 if (!strcmp(var, "diff.wserrorhighlight")) {
412 int val = parse_ws_error_highlight(value);
413 if (val < 0)
414 return -1;
415 ws_error_highlight_default = val;
416 return 0;
417 }
418
419 if (git_color_config(var, value, cb) < 0)
420 return -1;
421
422 return git_diff_basic_config(var, value, cb);
423 }
424
425 int git_diff_basic_config(const char *var, const char *value, void *cb)
426 {
427 const char *name;
428
429 if (!strcmp(var, "diff.renamelimit")) {
430 diff_rename_limit_default = git_config_int(var, value);
431 return 0;
432 }
433
434 if (userdiff_config(var, value) < 0)
435 return -1;
436
437 if (skip_prefix(var, "diff.color.", &name) ||
438 skip_prefix(var, "color.diff.", &name)) {
439 int slot = parse_diff_color_slot(name);
440 if (slot < 0)
441 return 0;
442 if (!value)
443 return config_error_nonbool(var);
444 return color_parse(value, diff_colors[slot]);
445 }
446
447 /* like GNU diff's --suppress-blank-empty option */
448 if (!strcmp(var, "diff.suppressblankempty") ||
449 /* for backwards compatibility */
450 !strcmp(var, "diff.suppress-blank-empty")) {
451 diff_suppress_blank_empty = git_config_bool(var, value);
452 return 0;
453 }
454
455 if (!strcmp(var, "diff.dirstat")) {
456 struct strbuf errmsg = STRBUF_INIT;
457 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
458 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
459 warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
460 errmsg.buf);
461 strbuf_release(&errmsg);
462 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
463 return 0;
464 }
465
466 if (git_diff_heuristic_config(var, value, cb) < 0)
467 return -1;
468
469 return git_default_config(var, value, cb);
470 }
471
472 static char *quote_two(const char *one, const char *two)
473 {
474 int need_one = quote_c_style(one, NULL, NULL, 1);
475 int need_two = quote_c_style(two, NULL, NULL, 1);
476 struct strbuf res = STRBUF_INIT;
477
478 if (need_one + need_two) {
479 strbuf_addch(&res, '"');
480 quote_c_style(one, &res, NULL, 1);
481 quote_c_style(two, &res, NULL, 1);
482 strbuf_addch(&res, '"');
483 } else {
484 strbuf_addstr(&res, one);
485 strbuf_addstr(&res, two);
486 }
487 return strbuf_detach(&res, NULL);
488 }
489
490 static const char *external_diff(void)
491 {
492 static const char *external_diff_cmd = NULL;
493 static int done_preparing = 0;
494
495 if (done_preparing)
496 return external_diff_cmd;
497 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
498 if (!external_diff_cmd)
499 external_diff_cmd = external_diff_cmd_cfg;
500 done_preparing = 1;
501 return external_diff_cmd;
502 }
503
504 /*
505 * Keep track of files used for diffing. Sometimes such an entry
506 * refers to a temporary file, sometimes to an existing file, and
507 * sometimes to "/dev/null".
508 */
509 static struct diff_tempfile {
510 /*
511 * filename external diff should read from, or NULL if this
512 * entry is currently not in use:
513 */
514 const char *name;
515
516 char hex[GIT_MAX_HEXSZ + 1];
517 char mode[10];
518
519 /*
520 * If this diff_tempfile instance refers to a temporary file,
521 * this tempfile object is used to manage its lifetime.
522 */
523 struct tempfile *tempfile;
524 } diff_temp[2];
525
526 struct emit_callback {
527 int color_diff;
528 unsigned ws_rule;
529 int blank_at_eof_in_preimage;
530 int blank_at_eof_in_postimage;
531 int lno_in_preimage;
532 int lno_in_postimage;
533 const char **label_path;
534 struct diff_words_data *diff_words;
535 struct diff_options *opt;
536 struct strbuf *header;
537 };
538
539 static int count_lines(const char *data, int size)
540 {
541 int count, ch, completely_empty = 1, nl_just_seen = 0;
542 count = 0;
543 while (0 < size--) {
544 ch = *data++;
545 if (ch == '\n') {
546 count++;
547 nl_just_seen = 1;
548 completely_empty = 0;
549 }
550 else {
551 nl_just_seen = 0;
552 completely_empty = 0;
553 }
554 }
555 if (completely_empty)
556 return 0;
557 if (!nl_just_seen)
558 count++; /* no trailing newline */
559 return count;
560 }
561
562 static int fill_mmfile(struct repository *r, mmfile_t *mf,
563 struct diff_filespec *one)
564 {
565 if (!DIFF_FILE_VALID(one)) {
566 mf->ptr = (char *)""; /* does not matter */
567 mf->size = 0;
568 return 0;
569 }
570 else if (diff_populate_filespec(r, one, 0))
571 return -1;
572
573 mf->ptr = one->data;
574 mf->size = one->size;
575 return 0;
576 }
577
578 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
579 static unsigned long diff_filespec_size(struct repository *r,
580 struct diff_filespec *one)
581 {
582 if (!DIFF_FILE_VALID(one))
583 return 0;
584 diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
585 return one->size;
586 }
587
588 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
589 {
590 char *ptr = mf->ptr;
591 long size = mf->size;
592 int cnt = 0;
593
594 if (!size)
595 return cnt;
596 ptr += size - 1; /* pointing at the very end */
597 if (*ptr != '\n')
598 ; /* incomplete line */
599 else
600 ptr--; /* skip the last LF */
601 while (mf->ptr < ptr) {
602 char *prev_eol;
603 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
604 if (*prev_eol == '\n')
605 break;
606 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
607 break;
608 cnt++;
609 ptr = prev_eol - 1;
610 }
611 return cnt;
612 }
613
614 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
615 struct emit_callback *ecbdata)
616 {
617 int l1, l2, at;
618 unsigned ws_rule = ecbdata->ws_rule;
619 l1 = count_trailing_blank(mf1, ws_rule);
620 l2 = count_trailing_blank(mf2, ws_rule);
621 if (l2 <= l1) {
622 ecbdata->blank_at_eof_in_preimage = 0;
623 ecbdata->blank_at_eof_in_postimage = 0;
624 return;
625 }
626 at = count_lines(mf1->ptr, mf1->size);
627 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
628
629 at = count_lines(mf2->ptr, mf2->size);
630 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
631 }
632
633 static void emit_line_0(struct diff_options *o,
634 const char *set_sign, const char *set, unsigned reverse, const char *reset,
635 int first, const char *line, int len)
636 {
637 int has_trailing_newline, has_trailing_carriage_return;
638 int needs_reset = 0; /* at the end of the line */
639 FILE *file = o->file;
640
641 fputs(diff_line_prefix(o), file);
642
643 has_trailing_newline = (len > 0 && line[len-1] == '\n');
644 if (has_trailing_newline)
645 len--;
646
647 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
648 if (has_trailing_carriage_return)
649 len--;
650
651 if (!len && !first)
652 goto end_of_line;
653
654 if (reverse && want_color(o->use_color)) {
655 fputs(GIT_COLOR_REVERSE, file);
656 needs_reset = 1;
657 }
658
659 if (set_sign) {
660 fputs(set_sign, file);
661 needs_reset = 1;
662 }
663
664 if (first)
665 fputc(first, file);
666
667 if (!len)
668 goto end_of_line;
669
670 if (set) {
671 if (set_sign && set != set_sign)
672 fputs(reset, file);
673 fputs(set, file);
674 needs_reset = 1;
675 }
676 fwrite(line, len, 1, file);
677 needs_reset = 1; /* 'line' may contain color codes. */
678
679 end_of_line:
680 if (needs_reset)
681 fputs(reset, file);
682 if (has_trailing_carriage_return)
683 fputc('\r', file);
684 if (has_trailing_newline)
685 fputc('\n', file);
686 }
687
688 static void emit_line(struct diff_options *o, const char *set, const char *reset,
689 const char *line, int len)
690 {
691 emit_line_0(o, set, NULL, 0, reset, 0, line, len);
692 }
693
694 enum diff_symbol {
695 DIFF_SYMBOL_BINARY_DIFF_HEADER,
696 DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
697 DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
698 DIFF_SYMBOL_BINARY_DIFF_BODY,
699 DIFF_SYMBOL_BINARY_DIFF_FOOTER,
700 DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
701 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
702 DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
703 DIFF_SYMBOL_STATS_LINE,
704 DIFF_SYMBOL_WORD_DIFF,
705 DIFF_SYMBOL_STAT_SEP,
706 DIFF_SYMBOL_SUMMARY,
707 DIFF_SYMBOL_SUBMODULE_ADD,
708 DIFF_SYMBOL_SUBMODULE_DEL,
709 DIFF_SYMBOL_SUBMODULE_UNTRACKED,
710 DIFF_SYMBOL_SUBMODULE_MODIFIED,
711 DIFF_SYMBOL_SUBMODULE_HEADER,
712 DIFF_SYMBOL_SUBMODULE_ERROR,
713 DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
714 DIFF_SYMBOL_REWRITE_DIFF,
715 DIFF_SYMBOL_BINARY_FILES,
716 DIFF_SYMBOL_HEADER,
717 DIFF_SYMBOL_FILEPAIR_PLUS,
718 DIFF_SYMBOL_FILEPAIR_MINUS,
719 DIFF_SYMBOL_WORDS_PORCELAIN,
720 DIFF_SYMBOL_WORDS,
721 DIFF_SYMBOL_CONTEXT,
722 DIFF_SYMBOL_CONTEXT_INCOMPLETE,
723 DIFF_SYMBOL_PLUS,
724 DIFF_SYMBOL_MINUS,
725 DIFF_SYMBOL_NO_LF_EOF,
726 DIFF_SYMBOL_CONTEXT_FRAGINFO,
727 DIFF_SYMBOL_CONTEXT_MARKER,
728 DIFF_SYMBOL_SEPARATOR
729 };
730 /*
731 * Flags for content lines:
732 * 0..12 are whitespace rules
733 * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
734 * 16 is marking if the line is blank at EOF
735 */
736 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF (1<<16)
737 #define DIFF_SYMBOL_MOVED_LINE (1<<17)
738 #define DIFF_SYMBOL_MOVED_LINE_ALT (1<<18)
739 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING (1<<19)
740 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
741
742 /*
743 * This struct is used when we need to buffer the output of the diff output.
744 *
745 * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
746 * into the pre/post image file. This pointer could be a union with the
747 * line pointer. By storing an offset into the file instead of the literal line,
748 * we can decrease the memory footprint for the buffered output. At first we
749 * may want to only have indirection for the content lines, but we could also
750 * enhance the state for emitting prefabricated lines, e.g. the similarity
751 * score line or hunk/file headers would only need to store a number or path
752 * and then the output can be constructed later on depending on state.
753 */
754 struct emitted_diff_symbol {
755 const char *line;
756 int len;
757 int flags;
758 enum diff_symbol s;
759 };
760 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
761
762 struct emitted_diff_symbols {
763 struct emitted_diff_symbol *buf;
764 int nr, alloc;
765 };
766 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
767
768 static void append_emitted_diff_symbol(struct diff_options *o,
769 struct emitted_diff_symbol *e)
770 {
771 struct emitted_diff_symbol *f;
772
773 ALLOC_GROW(o->emitted_symbols->buf,
774 o->emitted_symbols->nr + 1,
775 o->emitted_symbols->alloc);
776 f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
777
778 memcpy(f, e, sizeof(struct emitted_diff_symbol));
779 f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
780 }
781
782 struct moved_entry {
783 struct hashmap_entry ent;
784 const struct emitted_diff_symbol *es;
785 struct moved_entry *next_line;
786 };
787
788 /**
789 * The struct ws_delta holds white space differences between moved lines, i.e.
790 * between '+' and '-' lines that have been detected to be a move.
791 * The string contains the difference in leading white spaces, before the
792 * rest of the line is compared using the white space config for move
793 * coloring. The current_longer indicates if the first string in the
794 * comparision is longer than the second.
795 */
796 struct ws_delta {
797 char *string;
798 unsigned int current_longer : 1;
799 };
800 #define WS_DELTA_INIT { NULL, 0 }
801
802 struct moved_block {
803 struct moved_entry *match;
804 struct ws_delta wsd;
805 };
806
807 static void moved_block_clear(struct moved_block *b)
808 {
809 FREE_AND_NULL(b->wsd.string);
810 b->match = NULL;
811 }
812
813 static int compute_ws_delta(const struct emitted_diff_symbol *a,
814 const struct emitted_diff_symbol *b,
815 struct ws_delta *out)
816 {
817 const struct emitted_diff_symbol *longer = a->len > b->len ? a : b;
818 const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
819 int d = longer->len - shorter->len;
820
821 if (strncmp(longer->line + d, shorter->line, shorter->len))
822 return 0;
823
824 out->string = xmemdupz(longer->line, d);
825 out->current_longer = (a == longer);
826
827 return 1;
828 }
829
830 static int cmp_in_block_with_wsd(const struct diff_options *o,
831 const struct moved_entry *cur,
832 const struct moved_entry *match,
833 struct moved_block *pmb,
834 int n)
835 {
836 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
837 int al = cur->es->len, cl = l->len;
838 const char *a = cur->es->line,
839 *b = match->es->line,
840 *c = l->line;
841
842 int wslen;
843
844 /*
845 * We need to check if 'cur' is equal to 'match'.
846 * As those are from the same (+/-) side, we do not need to adjust for
847 * indent changes. However these were found using fuzzy matching
848 * so we do have to check if they are equal.
849 */
850 if (strcmp(a, b))
851 return 1;
852
853 if (!pmb->wsd.string)
854 /*
855 * The white space delta is not active? This can happen
856 * when we exit early in this function.
857 */
858 return 1;
859
860 /*
861 * The indent changes of the block are known and stored in
862 * pmb->wsd; however we need to check if the indent changes of the
863 * current line are still the same as before.
864 *
865 * To do so we need to compare 'l' to 'cur', adjusting the
866 * one of them for the white spaces, depending which was longer.
867 */
868
869 wslen = strlen(pmb->wsd.string);
870 if (pmb->wsd.current_longer) {
871 c += wslen;
872 cl -= wslen;
873 } else {
874 a += wslen;
875 al -= wslen;
876 }
877
878 if (al != cl || memcmp(a, c, al))
879 return 1;
880
881 return 0;
882 }
883
884 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
885 const void *entry,
886 const void *entry_or_key,
887 const void *keydata)
888 {
889 const struct diff_options *diffopt = hashmap_cmp_fn_data;
890 const struct moved_entry *a = entry;
891 const struct moved_entry *b = entry_or_key;
892 unsigned flags = diffopt->color_moved_ws_handling
893 & XDF_WHITESPACE_FLAGS;
894
895 if (diffopt->color_moved_ws_handling &
896 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
897 /*
898 * As there is not specific white space config given,
899 * we'd need to check for a new block, so ignore all
900 * white space. The setup of the white space
901 * configuration for the next block is done else where
902 */
903 flags |= XDF_IGNORE_WHITESPACE;
904
905 return !xdiff_compare_lines(a->es->line, a->es->len,
906 b->es->line, b->es->len,
907 flags);
908 }
909
910 static struct moved_entry *prepare_entry(struct diff_options *o,
911 int line_no)
912 {
913 struct moved_entry *ret = xmalloc(sizeof(*ret));
914 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
915 unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
916
917 ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
918 ret->es = l;
919 ret->next_line = NULL;
920
921 return ret;
922 }
923
924 static void add_lines_to_move_detection(struct diff_options *o,
925 struct hashmap *add_lines,
926 struct hashmap *del_lines)
927 {
928 struct moved_entry *prev_line = NULL;
929
930 int n;
931 for (n = 0; n < o->emitted_symbols->nr; n++) {
932 struct hashmap *hm;
933 struct moved_entry *key;
934
935 switch (o->emitted_symbols->buf[n].s) {
936 case DIFF_SYMBOL_PLUS:
937 hm = add_lines;
938 break;
939 case DIFF_SYMBOL_MINUS:
940 hm = del_lines;
941 break;
942 default:
943 prev_line = NULL;
944 continue;
945 }
946
947 key = prepare_entry(o, n);
948 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
949 prev_line->next_line = key;
950
951 hashmap_add(hm, key);
952 prev_line = key;
953 }
954 }
955
956 static void pmb_advance_or_null(struct diff_options *o,
957 struct moved_entry *match,
958 struct hashmap *hm,
959 struct moved_block *pmb,
960 int pmb_nr)
961 {
962 int i;
963 for (i = 0; i < pmb_nr; i++) {
964 struct moved_entry *prev = pmb[i].match;
965 struct moved_entry *cur = (prev && prev->next_line) ?
966 prev->next_line : NULL;
967 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
968 pmb[i].match = cur;
969 } else {
970 pmb[i].match = NULL;
971 }
972 }
973 }
974
975 static void pmb_advance_or_null_multi_match(struct diff_options *o,
976 struct moved_entry *match,
977 struct hashmap *hm,
978 struct moved_block *pmb,
979 int pmb_nr, int n)
980 {
981 int i;
982 char *got_match = xcalloc(1, pmb_nr);
983
984 for (; match; match = hashmap_get_next(hm, match)) {
985 for (i = 0; i < pmb_nr; i++) {
986 struct moved_entry *prev = pmb[i].match;
987 struct moved_entry *cur = (prev && prev->next_line) ?
988 prev->next_line : NULL;
989 if (!cur)
990 continue;
991 if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
992 got_match[i] |= 1;
993 }
994 }
995
996 for (i = 0; i < pmb_nr; i++) {
997 if (got_match[i]) {
998 /* Advance to the next line */
999 pmb[i].match = pmb[i].match->next_line;
1000 } else {
1001 moved_block_clear(&pmb[i]);
1002 }
1003 }
1004
1005 free(got_match);
1006 }
1007
1008 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1009 int pmb_nr)
1010 {
1011 int lp, rp;
1012
1013 /* Shrink the set of potential block to the remaining running */
1014 for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1015 while (lp < pmb_nr && pmb[lp].match)
1016 lp++;
1017 /* lp points at the first NULL now */
1018
1019 while (rp > -1 && !pmb[rp].match)
1020 rp--;
1021 /* rp points at the last non-NULL */
1022
1023 if (lp < pmb_nr && rp > -1 && lp < rp) {
1024 pmb[lp] = pmb[rp];
1025 pmb[rp].match = NULL;
1026 pmb[rp].wsd.string = NULL;
1027 rp--;
1028 lp++;
1029 }
1030 }
1031
1032 /* Remember the number of running sets */
1033 return rp + 1;
1034 }
1035
1036 /*
1037 * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1038 *
1039 * Otherwise, if the last block has fewer alphanumeric characters than
1040 * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1041 * that block.
1042 *
1043 * The last block consists of the (n - block_length)'th line up to but not
1044 * including the nth line.
1045 *
1046 * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1047 * Think of a way to unify them.
1048 */
1049 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1050 {
1051 int i, alnum_count = 0;
1052 if (o->color_moved == COLOR_MOVED_PLAIN)
1053 return;
1054 for (i = 1; i < block_length + 1; i++) {
1055 const char *c = o->emitted_symbols->buf[n - i].line;
1056 for (; *c; c++) {
1057 if (!isalnum(*c))
1058 continue;
1059 alnum_count++;
1060 if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1061 return;
1062 }
1063 }
1064 for (i = 1; i < block_length + 1; i++)
1065 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1066 }
1067
1068 /* Find blocks of moved code, delegate actual coloring decision to helper */
1069 static void mark_color_as_moved(struct diff_options *o,
1070 struct hashmap *add_lines,
1071 struct hashmap *del_lines)
1072 {
1073 struct moved_block *pmb = NULL; /* potentially moved blocks */
1074 int pmb_nr = 0, pmb_alloc = 0;
1075 int n, flipped_block = 1, block_length = 0;
1076
1077
1078 for (n = 0; n < o->emitted_symbols->nr; n++) {
1079 struct hashmap *hm = NULL;
1080 struct moved_entry *key;
1081 struct moved_entry *match = NULL;
1082 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1083
1084 switch (l->s) {
1085 case DIFF_SYMBOL_PLUS:
1086 hm = del_lines;
1087 key = prepare_entry(o, n);
1088 match = hashmap_get(hm, key, NULL);
1089 free(key);
1090 break;
1091 case DIFF_SYMBOL_MINUS:
1092 hm = add_lines;
1093 key = prepare_entry(o, n);
1094 match = hashmap_get(hm, key, NULL);
1095 free(key);
1096 break;
1097 default:
1098 flipped_block = 1;
1099 }
1100
1101 if (!match) {
1102 int i;
1103
1104 adjust_last_block(o, n, block_length);
1105 for(i = 0; i < pmb_nr; i++)
1106 moved_block_clear(&pmb[i]);
1107 pmb_nr = 0;
1108 block_length = 0;
1109 continue;
1110 }
1111
1112 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1113
1114 if (o->color_moved == COLOR_MOVED_PLAIN)
1115 continue;
1116
1117 if (o->color_moved_ws_handling &
1118 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1119 pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1120 else
1121 pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1122
1123 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1124
1125 if (pmb_nr == 0) {
1126 /*
1127 * The current line is the start of a new block.
1128 * Setup the set of potential blocks.
1129 */
1130 for (; match; match = hashmap_get_next(hm, match)) {
1131 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1132 if (o->color_moved_ws_handling &
1133 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1134 if (compute_ws_delta(l, match->es,
1135 &pmb[pmb_nr].wsd))
1136 pmb[pmb_nr++].match = match;
1137 } else {
1138 pmb[pmb_nr].wsd.string = NULL;
1139 pmb[pmb_nr++].match = match;
1140 }
1141 }
1142
1143 flipped_block = (flipped_block + 1) % 2;
1144
1145 adjust_last_block(o, n, block_length);
1146 block_length = 0;
1147 }
1148
1149 block_length++;
1150
1151 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1152 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1153 }
1154 adjust_last_block(o, n, block_length);
1155
1156 for(n = 0; n < pmb_nr; n++)
1157 moved_block_clear(&pmb[n]);
1158 free(pmb);
1159 }
1160
1161 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1162 (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1163 static void dim_moved_lines(struct diff_options *o)
1164 {
1165 int n;
1166 for (n = 0; n < o->emitted_symbols->nr; n++) {
1167 struct emitted_diff_symbol *prev = (n != 0) ?
1168 &o->emitted_symbols->buf[n - 1] : NULL;
1169 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1170 struct emitted_diff_symbol *next =
1171 (n < o->emitted_symbols->nr - 1) ?
1172 &o->emitted_symbols->buf[n + 1] : NULL;
1173
1174 /* Not a plus or minus line? */
1175 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1176 continue;
1177
1178 /* Not a moved line? */
1179 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1180 continue;
1181
1182 /*
1183 * If prev or next are not a plus or minus line,
1184 * pretend they don't exist
1185 */
1186 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1187 prev->s != DIFF_SYMBOL_MINUS)
1188 prev = NULL;
1189 if (next && next->s != DIFF_SYMBOL_PLUS &&
1190 next->s != DIFF_SYMBOL_MINUS)
1191 next = NULL;
1192
1193 /* Inside a block? */
1194 if ((prev &&
1195 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1196 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1197 (next &&
1198 (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1199 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1200 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1201 continue;
1202 }
1203
1204 /* Check if we are at an interesting bound: */
1205 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1206 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1207 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1208 continue;
1209 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1210 (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1211 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1212 continue;
1213
1214 /*
1215 * The boundary to prev and next are not interesting,
1216 * so this line is not interesting as a whole
1217 */
1218 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1219 }
1220 }
1221
1222 static void emit_line_ws_markup(struct diff_options *o,
1223 const char *set_sign, const char *set,
1224 const char *reset,
1225 int sign_index, const char *line, int len,
1226 unsigned ws_rule, int blank_at_eof)
1227 {
1228 const char *ws = NULL;
1229 int sign = o->output_indicators[sign_index];
1230
1231 if (o->ws_error_highlight & ws_rule) {
1232 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1233 if (!*ws)
1234 ws = NULL;
1235 }
1236
1237 if (!ws && !set_sign)
1238 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1239 else if (!ws) {
1240 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1241 } else if (blank_at_eof)
1242 /* Blank line at EOF - paint '+' as well */
1243 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1244 else {
1245 /* Emit just the prefix, then the rest. */
1246 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1247 sign, "", 0);
1248 ws_check_emit(line, len, ws_rule,
1249 o->file, set, reset, ws);
1250 }
1251 }
1252
1253 static void emit_diff_symbol_from_struct(struct diff_options *o,
1254 struct emitted_diff_symbol *eds)
1255 {
1256 static const char *nneof = " No newline at end of file\n";
1257 const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1258 struct strbuf sb = STRBUF_INIT;
1259
1260 enum diff_symbol s = eds->s;
1261 const char *line = eds->line;
1262 int len = eds->len;
1263 unsigned flags = eds->flags;
1264
1265 switch (s) {
1266 case DIFF_SYMBOL_NO_LF_EOF:
1267 context = diff_get_color_opt(o, DIFF_CONTEXT);
1268 reset = diff_get_color_opt(o, DIFF_RESET);
1269 putc('\n', o->file);
1270 emit_line_0(o, context, NULL, 0, reset, '\\',
1271 nneof, strlen(nneof));
1272 break;
1273 case DIFF_SYMBOL_SUBMODULE_HEADER:
1274 case DIFF_SYMBOL_SUBMODULE_ERROR:
1275 case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1276 case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1277 case DIFF_SYMBOL_SUMMARY:
1278 case DIFF_SYMBOL_STATS_LINE:
1279 case DIFF_SYMBOL_BINARY_DIFF_BODY:
1280 case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1281 emit_line(o, "", "", line, len);
1282 break;
1283 case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1284 case DIFF_SYMBOL_CONTEXT_MARKER:
1285 context = diff_get_color_opt(o, DIFF_CONTEXT);
1286 reset = diff_get_color_opt(o, DIFF_RESET);
1287 emit_line(o, context, reset, line, len);
1288 break;
1289 case DIFF_SYMBOL_SEPARATOR:
1290 fprintf(o->file, "%s%c",
1291 diff_line_prefix(o),
1292 o->line_termination);
1293 break;
1294 case DIFF_SYMBOL_CONTEXT:
1295 set = diff_get_color_opt(o, DIFF_CONTEXT);
1296 reset = diff_get_color_opt(o, DIFF_RESET);
1297 set_sign = NULL;
1298 if (o->flags.dual_color_diffed_diffs) {
1299 char c = !len ? 0 : line[0];
1300
1301 if (c == '+')
1302 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1303 else if (c == '@')
1304 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1305 else if (c == '-')
1306 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1307 }
1308 emit_line_ws_markup(o, set_sign, set, reset,
1309 OUTPUT_INDICATOR_CONTEXT, line, len,
1310 flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1311 break;
1312 case DIFF_SYMBOL_PLUS:
1313 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1314 DIFF_SYMBOL_MOVED_LINE_ALT |
1315 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1316 case DIFF_SYMBOL_MOVED_LINE |
1317 DIFF_SYMBOL_MOVED_LINE_ALT |
1318 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1319 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1320 break;
1321 case DIFF_SYMBOL_MOVED_LINE |
1322 DIFF_SYMBOL_MOVED_LINE_ALT:
1323 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1324 break;
1325 case DIFF_SYMBOL_MOVED_LINE |
1326 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1327 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1328 break;
1329 case DIFF_SYMBOL_MOVED_LINE:
1330 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1331 break;
1332 default:
1333 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1334 }
1335 reset = diff_get_color_opt(o, DIFF_RESET);
1336 if (!o->flags.dual_color_diffed_diffs)
1337 set_sign = NULL;
1338 else {
1339 char c = !len ? 0 : line[0];
1340
1341 set_sign = set;
1342 if (c == '-')
1343 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1344 else if (c == '@')
1345 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1346 else if (c == '+')
1347 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1348 else
1349 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1350 flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1351 }
1352 emit_line_ws_markup(o, set_sign, set, reset,
1353 OUTPUT_INDICATOR_NEW, line, len,
1354 flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1355 flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1356 break;
1357 case DIFF_SYMBOL_MINUS:
1358 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1359 DIFF_SYMBOL_MOVED_LINE_ALT |
1360 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1361 case DIFF_SYMBOL_MOVED_LINE |
1362 DIFF_SYMBOL_MOVED_LINE_ALT |
1363 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1364 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1365 break;
1366 case DIFF_SYMBOL_MOVED_LINE |
1367 DIFF_SYMBOL_MOVED_LINE_ALT:
1368 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1369 break;
1370 case DIFF_SYMBOL_MOVED_LINE |
1371 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1372 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1373 break;
1374 case DIFF_SYMBOL_MOVED_LINE:
1375 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1376 break;
1377 default:
1378 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1379 }
1380 reset = diff_get_color_opt(o, DIFF_RESET);
1381 if (!o->flags.dual_color_diffed_diffs)
1382 set_sign = NULL;
1383 else {
1384 char c = !len ? 0 : line[0];
1385
1386 set_sign = set;
1387 if (c == '+')
1388 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1389 else if (c == '@')
1390 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1391 else if (c == '-')
1392 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1393 else
1394 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1395 }
1396 emit_line_ws_markup(o, set_sign, set, reset,
1397 OUTPUT_INDICATOR_OLD, line, len,
1398 flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1399 break;
1400 case DIFF_SYMBOL_WORDS_PORCELAIN:
1401 context = diff_get_color_opt(o, DIFF_CONTEXT);
1402 reset = diff_get_color_opt(o, DIFF_RESET);
1403 emit_line(o, context, reset, line, len);
1404 fputs("~\n", o->file);
1405 break;
1406 case DIFF_SYMBOL_WORDS:
1407 context = diff_get_color_opt(o, DIFF_CONTEXT);
1408 reset = diff_get_color_opt(o, DIFF_RESET);
1409 /*
1410 * Skip the prefix character, if any. With
1411 * diff_suppress_blank_empty, there may be
1412 * none.
1413 */
1414 if (line[0] != '\n') {
1415 line++;
1416 len--;
1417 }
1418 emit_line(o, context, reset, line, len);
1419 break;
1420 case DIFF_SYMBOL_FILEPAIR_PLUS:
1421 meta = diff_get_color_opt(o, DIFF_METAINFO);
1422 reset = diff_get_color_opt(o, DIFF_RESET);
1423 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1424 line, reset,
1425 strchr(line, ' ') ? "\t" : "");
1426 break;
1427 case DIFF_SYMBOL_FILEPAIR_MINUS:
1428 meta = diff_get_color_opt(o, DIFF_METAINFO);
1429 reset = diff_get_color_opt(o, DIFF_RESET);
1430 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1431 line, reset,
1432 strchr(line, ' ') ? "\t" : "");
1433 break;
1434 case DIFF_SYMBOL_BINARY_FILES:
1435 case DIFF_SYMBOL_HEADER:
1436 fprintf(o->file, "%s", line);
1437 break;
1438 case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1439 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1440 break;
1441 case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1442 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1443 break;
1444 case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1445 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1446 break;
1447 case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1448 fputs(diff_line_prefix(o), o->file);
1449 fputc('\n', o->file);
1450 break;
1451 case DIFF_SYMBOL_REWRITE_DIFF:
1452 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1453 reset = diff_get_color_opt(o, DIFF_RESET);
1454 emit_line(o, fraginfo, reset, line, len);
1455 break;
1456 case DIFF_SYMBOL_SUBMODULE_ADD:
1457 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1458 reset = diff_get_color_opt(o, DIFF_RESET);
1459 emit_line(o, set, reset, line, len);
1460 break;
1461 case DIFF_SYMBOL_SUBMODULE_DEL:
1462 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1463 reset = diff_get_color_opt(o, DIFF_RESET);
1464 emit_line(o, set, reset, line, len);
1465 break;
1466 case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1467 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1468 diff_line_prefix(o), line);
1469 break;
1470 case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1471 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1472 diff_line_prefix(o), line);
1473 break;
1474 case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1475 emit_line(o, "", "", " 0 files changed\n",
1476 strlen(" 0 files changed\n"));
1477 break;
1478 case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1479 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1480 break;
1481 case DIFF_SYMBOL_WORD_DIFF:
1482 fprintf(o->file, "%.*s", len, line);
1483 break;
1484 case DIFF_SYMBOL_STAT_SEP:
1485 fputs(o->stat_sep, o->file);
1486 break;
1487 default:
1488 BUG("unknown diff symbol");
1489 }
1490 strbuf_release(&sb);
1491 }
1492
1493 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1494 const char *line, int len, unsigned flags)
1495 {
1496 struct emitted_diff_symbol e = {line, len, flags, s};
1497
1498 if (o->emitted_symbols)
1499 append_emitted_diff_symbol(o, &e);
1500 else
1501 emit_diff_symbol_from_struct(o, &e);
1502 }
1503
1504 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1505 {
1506 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1507 }
1508
1509 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1510 {
1511 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1512 }
1513
1514 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1515 {
1516 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1517 path, strlen(path), 0);
1518 }
1519
1520 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1521 {
1522 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1523 path, strlen(path), 0);
1524 }
1525
1526 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1527 {
1528 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1529 header, strlen(header), 0);
1530 }
1531
1532 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1533 {
1534 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1535 }
1536
1537 void diff_emit_submodule_pipethrough(struct diff_options *o,
1538 const char *line, int len)
1539 {
1540 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1541 }
1542
1543 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1544 {
1545 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1546 ecbdata->blank_at_eof_in_preimage &&
1547 ecbdata->blank_at_eof_in_postimage &&
1548 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1549 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1550 return 0;
1551 return ws_blank_line(line, len, ecbdata->ws_rule);
1552 }
1553
1554 static void emit_add_line(const char *reset,
1555 struct emit_callback *ecbdata,
1556 const char *line, int len)
1557 {
1558 unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1559 if (new_blank_line_at_eof(ecbdata, line, len))
1560 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1561
1562 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1563 }
1564
1565 static void emit_del_line(const char *reset,
1566 struct emit_callback *ecbdata,
1567 const char *line, int len)
1568 {
1569 unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1570 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1571 }
1572
1573 static void emit_context_line(const char *reset,
1574 struct emit_callback *ecbdata,
1575 const char *line, int len)
1576 {
1577 unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1578 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1579 }
1580
1581 static void emit_hunk_header(struct emit_callback *ecbdata,
1582 const char *line, int len)
1583 {
1584 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1585 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1586 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1587 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1588 const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1589 static const char atat[2] = { '@', '@' };
1590 const char *cp, *ep;
1591 struct strbuf msgbuf = STRBUF_INIT;
1592 int org_len = len;
1593 int i = 1;
1594
1595 /*
1596 * As a hunk header must begin with "@@ -<old>, +<new> @@",
1597 * it always is at least 10 bytes long.
1598 */
1599 if (len < 10 ||
1600 memcmp(line, atat, 2) ||
1601 !(ep = memmem(line + 2, len - 2, atat, 2))) {
1602 emit_diff_symbol(ecbdata->opt,
1603 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1604 return;
1605 }
1606 ep += 2; /* skip over @@ */
1607
1608 /* The hunk header in fraginfo color */
1609 if (ecbdata->opt->flags.dual_color_diffed_diffs)
1610 strbuf_addstr(&msgbuf, reverse);
1611 strbuf_addstr(&msgbuf, frag);
1612 strbuf_add(&msgbuf, line, ep - line);
1613 strbuf_addstr(&msgbuf, reset);
1614
1615 /*
1616 * trailing "\r\n"
1617 */
1618 for ( ; i < 3; i++)
1619 if (line[len - i] == '\r' || line[len - i] == '\n')
1620 len--;
1621
1622 /* blank before the func header */
1623 for (cp = ep; ep - line < len; ep++)
1624 if (*ep != ' ' && *ep != '\t')
1625 break;
1626 if (ep != cp) {
1627 strbuf_addstr(&msgbuf, context);
1628 strbuf_add(&msgbuf, cp, ep - cp);
1629 strbuf_addstr(&msgbuf, reset);
1630 }
1631
1632 if (ep < line + len) {
1633 strbuf_addstr(&msgbuf, func);
1634 strbuf_add(&msgbuf, ep, line + len - ep);
1635 strbuf_addstr(&msgbuf, reset);
1636 }
1637
1638 strbuf_add(&msgbuf, line + len, org_len - len);
1639 strbuf_complete_line(&msgbuf);
1640 emit_diff_symbol(ecbdata->opt,
1641 DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1642 strbuf_release(&msgbuf);
1643 }
1644
1645 static struct diff_tempfile *claim_diff_tempfile(void)
1646 {
1647 int i;
1648 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1649 if (!diff_temp[i].name)
1650 return diff_temp + i;
1651 BUG("diff is failing to clean up its tempfiles");
1652 }
1653
1654 static void remove_tempfile(void)
1655 {
1656 int i;
1657 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1658 if (is_tempfile_active(diff_temp[i].tempfile))
1659 delete_tempfile(&diff_temp[i].tempfile);
1660 diff_temp[i].name = NULL;
1661 }
1662 }
1663
1664 static void add_line_count(struct strbuf *out, int count)
1665 {
1666 switch (count) {
1667 case 0:
1668 strbuf_addstr(out, "0,0");
1669 break;
1670 case 1:
1671 strbuf_addstr(out, "1");
1672 break;
1673 default:
1674 strbuf_addf(out, "1,%d", count);
1675 break;
1676 }
1677 }
1678
1679 static void emit_rewrite_lines(struct emit_callback *ecb,
1680 int prefix, const char *data, int size)
1681 {
1682 const char *endp = NULL;
1683 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1684
1685 while (0 < size) {
1686 int len;
1687
1688 endp = memchr(data, '\n', size);
1689 len = endp ? (endp - data + 1) : size;
1690 if (prefix != '+') {
1691 ecb->lno_in_preimage++;
1692 emit_del_line(reset, ecb, data, len);
1693 } else {
1694 ecb->lno_in_postimage++;
1695 emit_add_line(reset, ecb, data, len);
1696 }
1697 size -= len;
1698 data += len;
1699 }
1700 if (!endp)
1701 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1702 }
1703
1704 static void emit_rewrite_diff(const char *name_a,
1705 const char *name_b,
1706 struct diff_filespec *one,
1707 struct diff_filespec *two,
1708 struct userdiff_driver *textconv_one,
1709 struct userdiff_driver *textconv_two,
1710 struct diff_options *o)
1711 {
1712 int lc_a, lc_b;
1713 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1714 const char *a_prefix, *b_prefix;
1715 char *data_one, *data_two;
1716 size_t size_one, size_two;
1717 struct emit_callback ecbdata;
1718 struct strbuf out = STRBUF_INIT;
1719
1720 if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1721 a_prefix = o->b_prefix;
1722 b_prefix = o->a_prefix;
1723 } else {
1724 a_prefix = o->a_prefix;
1725 b_prefix = o->b_prefix;
1726 }
1727
1728 name_a += (*name_a == '/');
1729 name_b += (*name_b == '/');
1730
1731 strbuf_reset(&a_name);
1732 strbuf_reset(&b_name);
1733 quote_two_c_style(&a_name, a_prefix, name_a, 0);
1734 quote_two_c_style(&b_name, b_prefix, name_b, 0);
1735
1736 size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1737 size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1738
1739 memset(&ecbdata, 0, sizeof(ecbdata));
1740 ecbdata.color_diff = want_color(o->use_color);
1741 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1742 ecbdata.opt = o;
1743 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1744 mmfile_t mf1, mf2;
1745 mf1.ptr = (char *)data_one;
1746 mf2.ptr = (char *)data_two;
1747 mf1.size = size_one;
1748 mf2.size = size_two;
1749 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1750 }
1751 ecbdata.lno_in_preimage = 1;
1752 ecbdata.lno_in_postimage = 1;
1753
1754 lc_a = count_lines(data_one, size_one);
1755 lc_b = count_lines(data_two, size_two);
1756
1757 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1758 a_name.buf, a_name.len, 0);
1759 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1760 b_name.buf, b_name.len, 0);
1761
1762 strbuf_addstr(&out, "@@ -");
1763 if (!o->irreversible_delete)
1764 add_line_count(&out, lc_a);
1765 else
1766 strbuf_addstr(&out, "?,?");
1767 strbuf_addstr(&out, " +");
1768 add_line_count(&out, lc_b);
1769 strbuf_addstr(&out, " @@\n");
1770 emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1771 strbuf_release(&out);
1772
1773 if (lc_a && !o->irreversible_delete)
1774 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1775 if (lc_b)
1776 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1777 if (textconv_one)
1778 free((char *)data_one);
1779 if (textconv_two)
1780 free((char *)data_two);
1781 }
1782
1783 struct diff_words_buffer {
1784 mmfile_t text;
1785 unsigned long alloc;
1786 struct diff_words_orig {
1787 const char *begin, *end;
1788 } *orig;
1789 int orig_nr, orig_alloc;
1790 };
1791
1792 static void diff_words_append(char *line, unsigned long len,
1793 struct diff_words_buffer *buffer)
1794 {
1795 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1796 line++;
1797 len--;
1798 memcpy(buffer->text.ptr + buffer->text.size, line, len);
1799 buffer->text.size += len;
1800 buffer->text.ptr[buffer->text.size] = '\0';
1801 }
1802
1803 struct diff_words_style_elem {
1804 const char *prefix;
1805 const char *suffix;
1806 const char *color; /* NULL; filled in by the setup code if
1807 * color is enabled */
1808 };
1809
1810 struct diff_words_style {
1811 enum diff_words_type type;
1812 struct diff_words_style_elem new_word, old_word, ctx;
1813 const char *newline;
1814 };
1815
1816 static struct diff_words_style diff_words_styles[] = {
1817 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1818 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1819 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1820 };
1821
1822 struct diff_words_data {
1823 struct diff_words_buffer minus, plus;
1824 const char *current_plus;
1825 int last_minus;
1826 struct diff_options *opt;
1827 regex_t *word_regex;
1828 enum diff_words_type type;
1829 struct diff_words_style *style;
1830 };
1831
1832 static int fn_out_diff_words_write_helper(struct diff_options *o,
1833 struct diff_words_style_elem *st_el,
1834 const char *newline,
1835 size_t count, const char *buf)
1836 {
1837 int print = 0;
1838 struct strbuf sb = STRBUF_INIT;
1839
1840 while (count) {
1841 char *p = memchr(buf, '\n', count);
1842 if (print)
1843 strbuf_addstr(&sb, diff_line_prefix(o));
1844
1845 if (p != buf) {
1846 const char *reset = st_el->color && *st_el->color ?
1847 GIT_COLOR_RESET : NULL;
1848 if (st_el->color && *st_el->color)
1849 strbuf_addstr(&sb, st_el->color);
1850 strbuf_addstr(&sb, st_el->prefix);
1851 strbuf_add(&sb, buf, p ? p - buf : count);
1852 strbuf_addstr(&sb, st_el->suffix);
1853 if (reset)
1854 strbuf_addstr(&sb, reset);
1855 }
1856 if (!p)
1857 goto out;
1858
1859 strbuf_addstr(&sb, newline);
1860 count -= p + 1 - buf;
1861 buf = p + 1;
1862 print = 1;
1863 if (count) {
1864 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1865 sb.buf, sb.len, 0);
1866 strbuf_reset(&sb);
1867 }
1868 }
1869
1870 out:
1871 if (sb.len)
1872 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1873 sb.buf, sb.len, 0);
1874 strbuf_release(&sb);
1875 return 0;
1876 }
1877
1878 /*
1879 * '--color-words' algorithm can be described as:
1880 *
1881 * 1. collect the minus/plus lines of a diff hunk, divided into
1882 * minus-lines and plus-lines;
1883 *
1884 * 2. break both minus-lines and plus-lines into words and
1885 * place them into two mmfile_t with one word for each line;
1886 *
1887 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1888 *
1889 * And for the common parts of the both file, we output the plus side text.
1890 * diff_words->current_plus is used to trace the current position of the plus file
1891 * which printed. diff_words->last_minus is used to trace the last minus word
1892 * printed.
1893 *
1894 * For '--graph' to work with '--color-words', we need to output the graph prefix
1895 * on each line of color words output. Generally, there are two conditions on
1896 * which we should output the prefix.
1897 *
1898 * 1. diff_words->last_minus == 0 &&
1899 * diff_words->current_plus == diff_words->plus.text.ptr
1900 *
1901 * that is: the plus text must start as a new line, and if there is no minus
1902 * word printed, a graph prefix must be printed.
1903 *
1904 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
1905 * *(diff_words->current_plus - 1) == '\n'
1906 *
1907 * that is: a graph prefix must be printed following a '\n'
1908 */
1909 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1910 {
1911 if ((diff_words->last_minus == 0 &&
1912 diff_words->current_plus == diff_words->plus.text.ptr) ||
1913 (diff_words->current_plus > diff_words->plus.text.ptr &&
1914 *(diff_words->current_plus - 1) == '\n')) {
1915 return 1;
1916 } else {
1917 return 0;
1918 }
1919 }
1920
1921 static void fn_out_diff_words_aux(void *priv,
1922 long minus_first, long minus_len,
1923 long plus_first, long plus_len,
1924 const char *func, long funclen)
1925 {
1926 struct diff_words_data *diff_words = priv;
1927 struct diff_words_style *style = diff_words->style;
1928 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1929 struct diff_options *opt = diff_words->opt;
1930 const char *line_prefix;
1931
1932 assert(opt);
1933 line_prefix = diff_line_prefix(opt);
1934
1935 /* POSIX requires that first be decremented by one if len == 0... */
1936 if (minus_len) {
1937 minus_begin = diff_words->minus.orig[minus_first].begin;
1938 minus_end =
1939 diff_words->minus.orig[minus_first + minus_len - 1].end;
1940 } else
1941 minus_begin = minus_end =
1942 diff_words->minus.orig[minus_first].end;
1943
1944 if (plus_len) {
1945 plus_begin = diff_words->plus.orig[plus_first].begin;
1946 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1947 } else
1948 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1949
1950 if (color_words_output_graph_prefix(diff_words)) {
1951 fputs(line_prefix, diff_words->opt->file);
1952 }
1953 if (diff_words->current_plus != plus_begin) {
1954 fn_out_diff_words_write_helper(diff_words->opt,
1955 &style->ctx, style->newline,
1956 plus_begin - diff_words->current_plus,
1957 diff_words->current_plus);
1958 }
1959 if (minus_begin != minus_end) {
1960 fn_out_diff_words_write_helper(diff_words->opt,
1961 &style->old_word, style->newline,
1962 minus_end - minus_begin, minus_begin);
1963 }
1964 if (plus_begin != plus_end) {
1965 fn_out_diff_words_write_helper(diff_words->opt,
1966 &style->new_word, style->newline,
1967 plus_end - plus_begin, plus_begin);
1968 }
1969
1970 diff_words->current_plus = plus_end;
1971 diff_words->last_minus = minus_first;
1972 }
1973
1974 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1975 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1976 int *begin, int *end)
1977 {
1978 if (word_regex && *begin < buffer->size) {
1979 regmatch_t match[1];
1980 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1981 buffer->size - *begin, 1, match, 0)) {
1982 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1983 '\n', match[0].rm_eo - match[0].rm_so);
1984 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1985 *begin += match[0].rm_so;
1986 return *begin >= *end;
1987 }
1988 return -1;
1989 }
1990
1991 /* find the next word */
1992 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1993 (*begin)++;
1994 if (*begin >= buffer->size)
1995 return -1;
1996
1997 /* find the end of the word */
1998 *end = *begin + 1;
1999 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2000 (*end)++;
2001
2002 return 0;
2003 }
2004
2005 /*
2006 * This function splits the words in buffer->text, stores the list with
2007 * newline separator into out, and saves the offsets of the original words
2008 * in buffer->orig.
2009 */
2010 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2011 regex_t *word_regex)
2012 {
2013 int i, j;
2014 long alloc = 0;
2015
2016 out->size = 0;
2017 out->ptr = NULL;
2018
2019 /* fake an empty "0th" word */
2020 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2021 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2022 buffer->orig_nr = 1;
2023
2024 for (i = 0; i < buffer->text.size; i++) {
2025 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2026 return;
2027
2028 /* store original boundaries */
2029 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2030 buffer->orig_alloc);
2031 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2032 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2033 buffer->orig_nr++;
2034
2035 /* store one word */
2036 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2037 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2038 out->ptr[out->size + j - i] = '\n';
2039 out->size += j - i + 1;
2040
2041 i = j - 1;
2042 }
2043 }
2044
2045 /* this executes the word diff on the accumulated buffers */
2046 static void diff_words_show(struct diff_words_data *diff_words)
2047 {
2048 xpparam_t xpp;
2049 xdemitconf_t xecfg;
2050 mmfile_t minus, plus;
2051 struct diff_words_style *style = diff_words->style;
2052
2053 struct diff_options *opt = diff_words->opt;
2054 const char *line_prefix;
2055
2056 assert(opt);
2057 line_prefix = diff_line_prefix(opt);
2058
2059 /* special case: only removal */
2060 if (!diff_words->plus.text.size) {
2061 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2062 line_prefix, strlen(line_prefix), 0);
2063 fn_out_diff_words_write_helper(diff_words->opt,
2064 &style->old_word, style->newline,
2065 diff_words->minus.text.size,
2066 diff_words->minus.text.ptr);
2067 diff_words->minus.text.size = 0;
2068 return;
2069 }
2070
2071 diff_words->current_plus = diff_words->plus.text.ptr;
2072 diff_words->last_minus = 0;
2073
2074 memset(&xpp, 0, sizeof(xpp));
2075 memset(&xecfg, 0, sizeof(xecfg));
2076 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2077 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2078 xpp.flags = 0;
2079 /* as only the hunk header will be parsed, we need a 0-context */
2080 xecfg.ctxlen = 0;
2081 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2082 diff_words, &xpp, &xecfg))
2083 die("unable to generate word diff");
2084 free(minus.ptr);
2085 free(plus.ptr);
2086 if (diff_words->current_plus != diff_words->plus.text.ptr +
2087 diff_words->plus.text.size) {
2088 if (color_words_output_graph_prefix(diff_words))
2089 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2090 line_prefix, strlen(line_prefix), 0);
2091 fn_out_diff_words_write_helper(diff_words->opt,
2092 &style->ctx, style->newline,
2093 diff_words->plus.text.ptr + diff_words->plus.text.size
2094 - diff_words->current_plus, diff_words->current_plus);
2095 }
2096 diff_words->minus.text.size = diff_words->plus.text.size = 0;
2097 }
2098
2099 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2100 static void diff_words_flush(struct emit_callback *ecbdata)
2101 {
2102 struct diff_options *wo = ecbdata->diff_words->opt;
2103
2104 if (ecbdata->diff_words->minus.text.size ||
2105 ecbdata->diff_words->plus.text.size)
2106 diff_words_show(ecbdata->diff_words);
2107
2108 if (wo->emitted_symbols) {
2109 struct diff_options *o = ecbdata->opt;
2110 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2111 int i;
2112
2113 /*
2114 * NEEDSWORK:
2115 * Instead of appending each, concat all words to a line?
2116 */
2117 for (i = 0; i < wol->nr; i++)
2118 append_emitted_diff_symbol(o, &wol->buf[i]);
2119
2120 for (i = 0; i < wol->nr; i++)
2121 free((void *)wol->buf[i].line);
2122
2123 wol->nr = 0;
2124 }
2125 }
2126
2127 static void diff_filespec_load_driver(struct diff_filespec *one,
2128 struct index_state *istate)
2129 {
2130 /* Use already-loaded driver */
2131 if (one->driver)
2132 return;
2133
2134 if (S_ISREG(one->mode))
2135 one->driver = userdiff_find_by_path(istate, one->path);
2136
2137 /* Fallback to default settings */
2138 if (!one->driver)
2139 one->driver = userdiff_find_by_name("default");
2140 }
2141
2142 static const char *userdiff_word_regex(struct diff_filespec *one,
2143 struct index_state *istate)
2144 {
2145 diff_filespec_load_driver(one, istate);
2146 return one->driver->word_regex;
2147 }
2148
2149 static void init_diff_words_data(struct emit_callback *ecbdata,
2150 struct diff_options *orig_opts,
2151 struct diff_filespec *one,
2152 struct diff_filespec *two)
2153 {
2154 int i;
2155 struct diff_options *o = xmalloc(sizeof(struct diff_options));
2156 memcpy(o, orig_opts, sizeof(struct diff_options));
2157
2158 ecbdata->diff_words =
2159 xcalloc(1, sizeof(struct diff_words_data));
2160 ecbdata->diff_words->type = o->word_diff;
2161 ecbdata->diff_words->opt = o;
2162
2163 if (orig_opts->emitted_symbols)
2164 o->emitted_symbols =
2165 xcalloc(1, sizeof(struct emitted_diff_symbols));
2166
2167 if (!o->word_regex)
2168 o->word_regex = userdiff_word_regex(one, o->repo->index);
2169 if (!o->word_regex)
2170 o->word_regex = userdiff_word_regex(two, o->repo->index);
2171 if (!o->word_regex)
2172 o->word_regex = diff_word_regex_cfg;
2173 if (o->word_regex) {
2174 ecbdata->diff_words->word_regex = (regex_t *)
2175 xmalloc(sizeof(regex_t));
2176 if (regcomp(ecbdata->diff_words->word_regex,
2177 o->word_regex,
2178 REG_EXTENDED | REG_NEWLINE))
2179 die("invalid regular expression: %s",
2180 o->word_regex);
2181 }
2182 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2183 if (o->word_diff == diff_words_styles[i].type) {
2184 ecbdata->diff_words->style =
2185 &diff_words_styles[i];
2186 break;
2187 }
2188 }
2189 if (want_color(o->use_color)) {
2190 struct diff_words_style *st = ecbdata->diff_words->style;
2191 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2192 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2193 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2194 }
2195 }
2196
2197 static void free_diff_words_data(struct emit_callback *ecbdata)
2198 {
2199 if (ecbdata->diff_words) {
2200 diff_words_flush(ecbdata);
2201 free (ecbdata->diff_words->opt->emitted_symbols);
2202 free (ecbdata->diff_words->opt);
2203 free (ecbdata->diff_words->minus.text.ptr);
2204 free (ecbdata->diff_words->minus.orig);
2205 free (ecbdata->diff_words->plus.text.ptr);
2206 free (ecbdata->diff_words->plus.orig);
2207 if (ecbdata->diff_words->word_regex) {
2208 regfree(ecbdata->diff_words->word_regex);
2209 free(ecbdata->diff_words->word_regex);
2210 }
2211 FREE_AND_NULL(ecbdata->diff_words);
2212 }
2213 }
2214
2215 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2216 {
2217 if (want_color(diff_use_color))
2218 return diff_colors[ix];
2219 return "";
2220 }
2221
2222 const char *diff_line_prefix(struct diff_options *opt)
2223 {
2224 struct strbuf *msgbuf;
2225 if (!opt->output_prefix)
2226 return "";
2227
2228 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2229 return msgbuf->buf;
2230 }
2231
2232 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2233 {
2234 const char *cp;
2235 unsigned long allot;
2236 size_t l = len;
2237
2238 cp = line;
2239 allot = l;
2240 while (0 < l) {
2241 (void) utf8_width(&cp, &l);
2242 if (!cp)
2243 break; /* truncated in the middle? */
2244 }
2245 return allot - l;
2246 }
2247
2248 static void find_lno(const char *line, struct emit_callback *ecbdata)
2249 {
2250 const char *p;
2251 ecbdata->lno_in_preimage = 0;
2252 ecbdata->lno_in_postimage = 0;
2253 p = strchr(line, '-');
2254 if (!p)
2255 return; /* cannot happen */
2256 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2257 p = strchr(p, '+');
2258 if (!p)
2259 return; /* cannot happen */
2260 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2261 }
2262
2263 static void fn_out_consume(void *priv, char *line, unsigned long len)
2264 {
2265 struct emit_callback *ecbdata = priv;
2266 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2267 struct diff_options *o = ecbdata->opt;
2268
2269 o->found_changes = 1;
2270
2271 if (ecbdata->header) {
2272 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2273 ecbdata->header->buf, ecbdata->header->len, 0);
2274 strbuf_reset(ecbdata->header);
2275 ecbdata->header = NULL;
2276 }
2277
2278 if (ecbdata->label_path[0]) {
2279 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2280 ecbdata->label_path[0],
2281 strlen(ecbdata->label_path[0]), 0);
2282 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2283 ecbdata->label_path[1],
2284 strlen(ecbdata->label_path[1]), 0);
2285 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2286 }
2287
2288 if (diff_suppress_blank_empty
2289 && len == 2 && line[0] == ' ' && line[1] == '\n') {
2290 line[0] = '\n';
2291 len = 1;
2292 }
2293
2294 if (line[0] == '@') {
2295 if (ecbdata->diff_words)
2296 diff_words_flush(ecbdata);
2297 len = sane_truncate_line(ecbdata, line, len);
2298 find_lno(line, ecbdata);
2299 emit_hunk_header(ecbdata, line, len);
2300 return;
2301 }
2302
2303 if (ecbdata->diff_words) {
2304 enum diff_symbol s =
2305 ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2306 DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2307 if (line[0] == '-') {
2308 diff_words_append(line, len,
2309 &ecbdata->diff_words->minus);
2310 return;
2311 } else if (line[0] == '+') {
2312 diff_words_append(line, len,
2313 &ecbdata->diff_words->plus);
2314 return;
2315 } else if (starts_with(line, "\\ ")) {
2316 /*
2317 * Eat the "no newline at eof" marker as if we
2318 * saw a "+" or "-" line with nothing on it,
2319 * and return without diff_words_flush() to
2320 * defer processing. If this is the end of
2321 * preimage, more "+" lines may come after it.
2322 */
2323 return;
2324 }
2325 diff_words_flush(ecbdata);
2326 emit_diff_symbol(o, s, line, len, 0);
2327 return;
2328 }
2329
2330 switch (line[0]) {
2331 case '+':
2332 ecbdata->lno_in_postimage++;
2333 emit_add_line(reset, ecbdata, line + 1, len - 1);
2334 break;
2335 case '-':
2336 ecbdata->lno_in_preimage++;
2337 emit_del_line(reset, ecbdata, line + 1, len - 1);
2338 break;
2339 case ' ':
2340 ecbdata->lno_in_postimage++;
2341 ecbdata->lno_in_preimage++;
2342 emit_context_line(reset, ecbdata, line + 1, len - 1);
2343 break;
2344 default:
2345 /* incomplete line at the end */
2346 ecbdata->lno_in_preimage++;
2347 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2348 line, len, 0);
2349 break;
2350 }
2351 }
2352
2353 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2354 {
2355 const char *old_name = a;
2356 const char *new_name = b;
2357 int pfx_length, sfx_length;
2358 int pfx_adjust_for_slash;
2359 int len_a = strlen(a);
2360 int len_b = strlen(b);
2361 int a_midlen, b_midlen;
2362 int qlen_a = quote_c_style(a, NULL, NULL, 0);
2363 int qlen_b = quote_c_style(b, NULL, NULL, 0);
2364
2365 if (qlen_a || qlen_b) {
2366 quote_c_style(a, name, NULL, 0);
2367 strbuf_addstr(name, " => ");
2368 quote_c_style(b, name, NULL, 0);
2369 return;
2370 }
2371
2372 /* Find common prefix */
2373 pfx_length = 0;
2374 while (*old_name && *new_name && *old_name == *new_name) {
2375 if (*old_name == '/')
2376 pfx_length = old_name - a + 1;
2377 old_name++;
2378 new_name++;
2379 }
2380
2381 /* Find common suffix */
2382 old_name = a + len_a;
2383 new_name = b + len_b;
2384 sfx_length = 0;
2385 /*
2386 * If there is a common prefix, it must end in a slash. In
2387 * that case we let this loop run 1 into the prefix to see the
2388 * same slash.
2389 *
2390 * If there is no common prefix, we cannot do this as it would
2391 * underrun the input strings.
2392 */
2393 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2394 while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2395 b + pfx_length - pfx_adjust_for_slash <= new_name &&
2396 *old_name == *new_name) {
2397 if (*old_name == '/')
2398 sfx_length = len_a - (old_name - a);
2399 old_name--;
2400 new_name--;
2401 }
2402
2403 /*
2404 * pfx{mid-a => mid-b}sfx
2405 * {pfx-a => pfx-b}sfx
2406 * pfx{sfx-a => sfx-b}
2407 * name-a => name-b
2408 */
2409 a_midlen = len_a - pfx_length - sfx_length;
2410 b_midlen = len_b - pfx_length - sfx_length;
2411 if (a_midlen < 0)
2412 a_midlen = 0;
2413 if (b_midlen < 0)
2414 b_midlen = 0;
2415
2416 strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2417 if (pfx_length + sfx_length) {
2418 strbuf_add(name, a, pfx_length);
2419 strbuf_addch(name, '{');
2420 }
2421 strbuf_add(name, a + pfx_length, a_midlen);
2422 strbuf_addstr(name, " => ");
2423 strbuf_add(name, b + pfx_length, b_midlen);
2424 if (pfx_length + sfx_length) {
2425 strbuf_addch(name, '}');
2426 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2427 }
2428 }
2429
2430 struct diffstat_t {
2431 int nr;
2432 int alloc;
2433 struct diffstat_file {
2434 char *from_name;
2435 char *name;
2436 char *print_name;
2437 const char *comments;
2438 unsigned is_unmerged:1;
2439 unsigned is_binary:1;
2440 unsigned is_renamed:1;
2441 unsigned is_interesting:1;
2442 uintmax_t added, deleted;
2443 } **files;
2444 };
2445
2446 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2447 const char *name_a,
2448 const char *name_b)
2449 {
2450 struct diffstat_file *x;
2451 x = xcalloc(1, sizeof(*x));
2452 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2453 diffstat->files[diffstat->nr++] = x;
2454 if (name_b) {
2455 x->from_name = xstrdup(name_a);
2456 x->name = xstrdup(name_b);
2457 x->is_renamed = 1;
2458 }
2459 else {
2460 x->from_name = NULL;
2461 x->name = xstrdup(name_a);
2462 }
2463 return x;
2464 }
2465
2466 static void diffstat_consume(void *priv, char *line, unsigned long len)
2467 {
2468 struct diffstat_t *diffstat = priv;
2469 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2470
2471 if (line[0] == '+')
2472 x->added++;
2473 else if (line[0] == '-')
2474 x->deleted++;
2475 }
2476
2477 const char mime_boundary_leader[] = "------------";
2478
2479 static int scale_linear(int it, int width, int max_change)
2480 {
2481 if (!it)
2482 return 0;
2483 /*
2484 * make sure that at least one '-' or '+' is printed if
2485 * there is any change to this path. The easiest way is to
2486 * scale linearly as if the alloted width is one column shorter
2487 * than it is, and then add 1 to the result.
2488 */
2489 return 1 + (it * (width - 1) / max_change);
2490 }
2491
2492 static void show_graph(struct strbuf *out, char ch, int cnt,
2493 const char *set, const char *reset)
2494 {
2495 if (cnt <= 0)
2496 return;
2497 strbuf_addstr(out, set);
2498 strbuf_addchars(out, ch, cnt);
2499 strbuf_addstr(out, reset);
2500 }
2501
2502 static void fill_print_name(struct diffstat_file *file)
2503 {
2504 struct strbuf pname = STRBUF_INIT;
2505
2506 if (file->print_name)
2507 return;
2508
2509 if (file->is_renamed)
2510 pprint_rename(&pname, file->from_name, file->name);
2511 else
2512 quote_c_style(file->name, &pname, NULL, 0);
2513
2514 if (file->comments)
2515 strbuf_addf(&pname, " (%s)", file->comments);
2516
2517 file->print_name = strbuf_detach(&pname, NULL);
2518 }
2519
2520 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2521 int files, int insertions, int deletions)
2522 {
2523 struct strbuf sb = STRBUF_INIT;
2524
2525 if (!files) {
2526 assert(insertions == 0 && deletions == 0);
2527 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2528 NULL, 0, 0);
2529 return;
2530 }
2531
2532 strbuf_addf(&sb,
2533 (files == 1) ? " %d file changed" : " %d files changed",
2534 files);
2535
2536 /*
2537 * For binary diff, the caller may want to print "x files
2538 * changed" with insertions == 0 && deletions == 0.
2539 *
2540 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2541 * is probably less confusing (i.e skip over "2 files changed
2542 * but nothing about added/removed lines? Is this a bug in Git?").
2543 */
2544 if (insertions || deletions == 0) {
2545 strbuf_addf(&sb,
2546 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2547 insertions);
2548 }
2549
2550 if (deletions || insertions == 0) {
2551 strbuf_addf(&sb,
2552 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2553 deletions);
2554 }
2555 strbuf_addch(&sb, '\n');
2556 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2557 sb.buf, sb.len, 0);
2558 strbuf_release(&sb);
2559 }
2560
2561 void print_stat_summary(FILE *fp, int files,
2562 int insertions, int deletions)
2563 {
2564 struct diff_options o;
2565 memset(&o, 0, sizeof(o));
2566 o.file = fp;
2567
2568 print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2569 }
2570
2571 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2572 {
2573 int i, len, add, del, adds = 0, dels = 0;
2574 uintmax_t max_change = 0, max_len = 0;
2575 int total_files = data->nr, count;
2576 int width, name_width, graph_width, number_width = 0, bin_width = 0;
2577 const char *reset, *add_c, *del_c;
2578 int extra_shown = 0;
2579 const char *line_prefix = diff_line_prefix(options);
2580 struct strbuf out = STRBUF_INIT;
2581
2582 if (data->nr == 0)
2583 return;
2584
2585 count = options->stat_count ? options->stat_count : data->nr;
2586
2587 reset = diff_get_color_opt(options, DIFF_RESET);
2588 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2589 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2590
2591 /*
2592 * Find the longest filename and max number of changes
2593 */
2594 for (i = 0; (i < count) && (i < data->nr); i++) {
2595 struct diffstat_file *file = data->files[i];
2596 uintmax_t change = file->added + file->deleted;
2597
2598 if (!file->is_interesting && (change == 0)) {
2599 count++; /* not shown == room for one more */
2600 continue;
2601 }
2602 fill_print_name(file);
2603 len = strlen(file->print_name);
2604 if (max_len < len)
2605 max_len = len;
2606
2607 if (file->is_unmerged) {
2608 /* "Unmerged" is 8 characters */
2609 bin_width = bin_width < 8 ? 8 : bin_width;
2610 continue;
2611 }
2612 if (file->is_binary) {
2613 /* "Bin XXX -> YYY bytes" */
2614 int w = 14 + decimal_width(file->added)
2615 + decimal_width(file->deleted);
2616 bin_width = bin_width < w ? w : bin_width;
2617 /* Display change counts aligned with "Bin" */
2618 number_width = 3;
2619 continue;
2620 }
2621
2622 if (max_change < change)
2623 max_change = change;
2624 }
2625 count = i; /* where we can stop scanning in data->files[] */
2626
2627 /*
2628 * We have width = stat_width or term_columns() columns total.
2629 * We want a maximum of min(max_len, stat_name_width) for the name part.
2630 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2631 * We also need 1 for " " and 4 + decimal_width(max_change)
2632 * for " | NNNN " and one the empty column at the end, altogether
2633 * 6 + decimal_width(max_change).
2634 *
2635 * If there's not enough space, we will use the smaller of
2636 * stat_name_width (if set) and 5/8*width for the filename,
2637 * and the rest for constant elements + graph part, but no more
2638 * than stat_graph_width for the graph part.
2639 * (5/8 gives 50 for filename and 30 for the constant parts + graph
2640 * for the standard terminal size).
2641 *
2642 * In other words: stat_width limits the maximum width, and
2643 * stat_name_width fixes the maximum width of the filename,
2644 * and is also used to divide available columns if there
2645 * aren't enough.
2646 *
2647 * Binary files are displayed with "Bin XXX -> YYY bytes"
2648 * instead of the change count and graph. This part is treated
2649 * similarly to the graph part, except that it is not
2650 * "scaled". If total width is too small to accommodate the
2651 * guaranteed minimum width of the filename part and the
2652 * separators and this message, this message will "overflow"
2653 * making the line longer than the maximum width.
2654 */
2655
2656 if (options->stat_width == -1)
2657 width = term_columns() - strlen(line_prefix);
2658 else
2659 width = options->stat_width ? options->stat_width : 80;
2660 number_width = decimal_width(max_change) > number_width ?
2661 decimal_width(max_change) : number_width;
2662
2663 if (options->stat_graph_width == -1)
2664 options->stat_graph_width = diff_stat_graph_width;
2665
2666 /*
2667 * Guarantee 3/8*16==6 for the graph part
2668 * and 5/8*16==10 for the filename part
2669 */
2670 if (width < 16 + 6 + number_width)
2671 width = 16 + 6 + number_width;
2672
2673 /*
2674 * First assign sizes that are wanted, ignoring available width.
2675 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2676 * starting from "XXX" should fit in graph_width.
2677 */
2678 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2679 if (options->stat_graph_width &&
2680 options->stat_graph_width < graph_width)
2681 graph_width = options->stat_graph_width;
2682
2683 name_width = (options->stat_name_width > 0 &&
2684 options->stat_name_width < max_len) ?
2685 options->stat_name_width : max_len;
2686
2687 /*
2688 * Adjust adjustable widths not to exceed maximum width
2689 */
2690 if (name_width + number_width + 6 + graph_width > width) {
2691 if (graph_width > width * 3/8 - number_width - 6) {
2692 graph_width = width * 3/8 - number_width - 6;
2693 if (graph_width < 6)
2694 graph_width = 6;
2695 }
2696
2697 if (options->stat_graph_width &&
2698 graph_width > options->stat_graph_width)
2699 graph_width = options->stat_graph_width;
2700 if (name_width > width - number_width - 6 - graph_width)
2701 name_width = width - number_width - 6 - graph_width;
2702 else
2703 graph_width = width - number_width - 6 - name_width;
2704 }
2705
2706 /*
2707 * From here name_width is the width of the name area,
2708 * and graph_width is the width of the graph area.
2709 * max_change is used to scale graph properly.
2710 */
2711 for (i = 0; i < count; i++) {
2712 const char *prefix = "";
2713 struct diffstat_file *file = data->files[i];
2714 char *name = file->print_name;
2715 uintmax_t added = file->added;
2716 uintmax_t deleted = file->deleted;
2717 int name_len;
2718
2719 if (!file->is_interesting && (added + deleted == 0))
2720 continue;
2721
2722 /*
2723 * "scale" the filename
2724 */
2725 len = name_width;
2726 name_len = strlen(name);
2727 if (name_width < name_len) {
2728 char *slash;
2729 prefix = "...";
2730 len -= 3;
2731 name += name_len - len;
2732 slash = strchr(name, '/');
2733 if (slash)
2734 name = slash;
2735 }
2736
2737 if (file->is_binary) {
2738 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2739 strbuf_addf(&out, " %*s", number_width, "Bin");
2740 if (!added && !deleted) {
2741 strbuf_addch(&out, '\n');
2742 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2743 out.buf, out.len, 0);
2744 strbuf_reset(&out);
2745 continue;
2746 }
2747 strbuf_addf(&out, " %s%"PRIuMAX"%s",
2748 del_c, deleted, reset);
2749 strbuf_addstr(&out, " -> ");
2750 strbuf_addf(&out, "%s%"PRIuMAX"%s",
2751 add_c, added, reset);
2752 strbuf_addstr(&out, " bytes\n");
2753 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2754 out.buf, out.len, 0);
2755 strbuf_reset(&out);
2756 continue;
2757 }
2758 else if (file->is_unmerged) {
2759 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2760 strbuf_addstr(&out, " Unmerged\n");
2761 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2762 out.buf, out.len, 0);
2763 strbuf_reset(&out);
2764 continue;
2765 }
2766
2767 /*
2768 * scale the add/delete
2769 */
2770 add = added;
2771 del = deleted;
2772
2773 if (graph_width <= max_change) {
2774 int total = scale_linear(add + del, graph_width, max_change);
2775 if (total < 2 && add && del)
2776 /* width >= 2 due to the sanity check */
2777 total = 2;
2778 if (add < del) {
2779 add = scale_linear(add, graph_width, max_change);
2780 del = total - add;
2781 } else {
2782 del = scale_linear(del, graph_width, max_change);
2783 add = total - del;
2784 }
2785 }
2786 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2787 strbuf_addf(&out, " %*"PRIuMAX"%s",
2788 number_width, added + deleted,
2789 added + deleted ? " " : "");
2790 show_graph(&out, '+', add, add_c, reset);
2791 show_graph(&out, '-', del, del_c, reset);
2792 strbuf_addch(&out, '\n');
2793 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2794 out.buf, out.len, 0);
2795 strbuf_reset(&out);
2796 }
2797
2798 for (i = 0; i < data->nr; i++) {
2799 struct diffstat_file *file = data->files[i];
2800 uintmax_t added = file->added;
2801 uintmax_t deleted = file->deleted;
2802
2803 if (file->is_unmerged ||
2804 (!file->is_interesting && (added + deleted == 0))) {
2805 total_files--;
2806 continue;
2807 }
2808
2809 if (!file->is_binary) {
2810 adds += added;
2811 dels += deleted;
2812 }
2813 if (i < count)
2814 continue;
2815 if (!extra_shown)
2816 emit_diff_symbol(options,
2817 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2818 NULL, 0, 0);
2819 extra_shown = 1;
2820 }
2821
2822 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2823 strbuf_release(&out);
2824 }
2825
2826 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2827 {
2828 int i, adds = 0, dels = 0, total_files = data->nr;
2829
2830 if (data->nr == 0)
2831 return;
2832
2833 for (i = 0; i < data->nr; i++) {
2834 int added = data->files[i]->added;
2835 int deleted = data->files[i]->deleted;
2836
2837 if (data->files[i]->is_unmerged ||
2838 (!data->files[i]->is_interesting && (added + deleted == 0))) {
2839 total_files--;
2840 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2841 adds += added;
2842 dels += deleted;
2843 }
2844 }
2845 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2846 }
2847
2848 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2849 {
2850 int i;
2851
2852 if (data->nr == 0)
2853 return;
2854
2855 for (i = 0; i < data->nr; i++) {
2856 struct diffstat_file *file = data->files[i];
2857
2858 fprintf(options->file, "%s", diff_line_prefix(options));
2859
2860 if (file->is_binary)
2861 fprintf(options->file, "-\t-\t");
2862 else
2863 fprintf(options->file,
2864 "%"PRIuMAX"\t%"PRIuMAX"\t",
2865 file->added, file->deleted);
2866 if (options->line_termination) {
2867 fill_print_name(file);
2868 if (!file->is_renamed)
2869 write_name_quoted(file->name, options->file,
2870 options->line_termination);
2871 else {
2872 fputs(file->print_name, options->file);
2873 putc(options->line_termination, options->file);
2874 }
2875 } else {
2876 if (file->is_renamed) {
2877 putc('\0', options->file);
2878 write_name_quoted(file->from_name, options->file, '\0');
2879 }
2880 write_name_quoted(file->name, options->file, '\0');
2881 }
2882 }
2883 }
2884
2885 struct dirstat_file {
2886 const char *name;
2887 unsigned long changed;
2888 };
2889
2890 struct dirstat_dir {
2891 struct dirstat_file *files;
2892 int alloc, nr, permille, cumulative;
2893 };
2894
2895 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2896 unsigned long changed, const char *base, int baselen)
2897 {
2898 unsigned long sum_changes = 0;
2899 unsigned int sources = 0;
2900 const char *line_prefix = diff_line_prefix(opt);
2901
2902 while (dir->nr) {
2903 struct dirstat_file *f = dir->files;
2904 int namelen = strlen(f->name);
2905 unsigned long changes;
2906 char *slash;
2907
2908 if (namelen < baselen)
2909 break;
2910 if (memcmp(f->name, base, baselen))
2911 break;
2912 slash = strchr(f->name + baselen, '/');
2913 if (slash) {
2914 int newbaselen = slash + 1 - f->name;
2915 changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2916 sources++;
2917 } else {
2918 changes = f->changed;
2919 dir->files++;
2920 dir->nr--;
2921 sources += 2;
2922 }
2923 sum_changes += changes;
2924 }
2925
2926 /*
2927 * We don't report dirstat's for
2928 * - the top level
2929 * - or cases where everything came from a single directory
2930 * under this directory (sources == 1).
2931 */
2932 if (baselen && sources != 1) {
2933 if (sum_changes) {
2934 int permille = sum_changes * 1000 / changed;
2935 if (permille >= dir->permille) {
2936 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2937 permille / 10, permille % 10, baselen, base);
2938 if (!dir->cumulative)
2939 return 0;
2940 }
2941 }
2942 }
2943 return sum_changes;
2944 }
2945
2946 static int dirstat_compare(const void *_a, const void *_b)
2947 {
2948 const struct dirstat_file *a = _a;
2949 const struct dirstat_file *b = _b;
2950 return strcmp(a->name, b->name);
2951 }
2952
2953 static void show_dirstat(struct diff_options *options)
2954 {
2955 int i;
2956 unsigned long changed;
2957 struct dirstat_dir dir;
2958 struct diff_queue_struct *q = &diff_queued_diff;
2959
2960 dir.files = NULL;
2961 dir.alloc = 0;
2962 dir.nr = 0;
2963 dir.permille = options->dirstat_permille;
2964 dir.cumulative = options->flags.dirstat_cumulative;
2965
2966 changed = 0;
2967 for (i = 0; i < q->nr; i++) {
2968 struct diff_filepair *p = q->queue[i];
2969 const char *name;
2970 unsigned long copied, added, damage;
2971
2972 name = p->two->path ? p->two->path : p->one->path;
2973
2974 if (p->one->oid_valid && p->two->oid_valid &&
2975 oideq(&p->one->oid, &p->two->oid)) {
2976 /*
2977 * The SHA1 has not changed, so pre-/post-content is
2978 * identical. We can therefore skip looking at the
2979 * file contents altogether.
2980 */
2981 damage = 0;
2982 goto found_damage;
2983 }
2984
2985 if (options->flags.dirstat_by_file) {
2986 /*
2987 * In --dirstat-by-file mode, we don't really need to
2988 * look at the actual file contents at all.
2989 * The fact that the SHA1 changed is enough for us to
2990 * add this file to the list of results
2991 * (with each file contributing equal damage).
2992 */
2993 damage = 1;
2994 goto found_damage;
2995 }
2996
2997 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2998 diff_populate_filespec(options->repo, p->one, 0);
2999 diff_populate_filespec(options->repo, p->two, 0);
3000 diffcore_count_changes(options->repo,
3001 p->one, p->two, NULL, NULL,
3002 &copied, &added);
3003 diff_free_filespec_data(p->one);
3004 diff_free_filespec_data(p->two);
3005 } else if (DIFF_FILE_VALID(p->one)) {
3006 diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3007 copied = added = 0;
3008 diff_free_filespec_data(p->one);
3009 } else if (DIFF_FILE_VALID(p->two)) {
3010 diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3011 copied = 0;
3012 added = p->two->size;
3013 diff_free_filespec_data(p->two);
3014 } else
3015 continue;
3016
3017 /*
3018 * Original minus copied is the removed material,
3019 * added is the new material. They are both damages
3020 * made to the preimage.
3021 * If the resulting damage is zero, we know that
3022 * diffcore_count_changes() considers the two entries to
3023 * be identical, but since the oid changed, we
3024 * know that there must have been _some_ kind of change,
3025 * so we force all entries to have damage > 0.
3026 */
3027 damage = (p->one->size - copied) + added;
3028 if (!damage)
3029 damage = 1;
3030
3031 found_damage:
3032 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3033 dir.files[dir.nr].name = name;
3034 dir.files[dir.nr].changed = damage;
3035 changed += damage;
3036 dir.nr++;
3037 }
3038
3039 /* This can happen even with many files, if everything was renames */
3040 if (!changed)
3041 return;
3042
3043 /* Show all directories with more than x% of the changes */
3044 QSORT(dir.files, dir.nr, dirstat_compare);
3045 gather_dirstat(options, &dir, changed, "", 0);
3046 }
3047
3048 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3049 {
3050 int i;
3051 unsigned long changed;
3052 struct dirstat_dir dir;
3053
3054 if (data->nr == 0)
3055 return;
3056
3057 dir.files = NULL;
3058 dir.alloc = 0;
3059 dir.nr = 0;
3060 dir.permille = options->dirstat_permille;
3061 dir.cumulative = options->flags.dirstat_cumulative;
3062
3063 changed = 0;
3064 for (i = 0; i < data->nr; i++) {
3065 struct diffstat_file *file = data->files[i];
3066 unsigned long damage = file->added + file->deleted;
3067 if (file->is_binary)
3068 /*
3069 * binary files counts bytes, not lines. Must find some
3070 * way to normalize binary bytes vs. textual lines.
3071 * The following heuristic assumes that there are 64
3072 * bytes per "line".
3073 * This is stupid and ugly, but very cheap...
3074 */
3075 damage = DIV_ROUND_UP(damage, 64);
3076 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3077 dir.files[dir.nr].name = file->name;
3078 dir.files[dir.nr].changed = damage;
3079 changed += damage;
3080 dir.nr++;
3081 }
3082
3083 /* This can happen even with many files, if everything was renames */
3084 if (!changed)
3085 return;
3086
3087 /* Show all directories with more than x% of the changes */
3088 QSORT(dir.files, dir.nr, dirstat_compare);
3089 gather_dirstat(options, &dir, changed, "", 0);
3090 }
3091
3092 static void free_diffstat_info(struct diffstat_t *diffstat)
3093 {
3094 int i;
3095 for (i = 0; i < diffstat->nr; i++) {
3096 struct diffstat_file *f = diffstat->files[i];
3097 free(f->print_name);
3098 free(f->name);
3099 free(f->from_name);
3100 free(f);
3101 }
3102 free(diffstat->files);
3103 }
3104
3105 struct checkdiff_t {
3106 const char *filename;
3107 int lineno;
3108 int conflict_marker_size;
3109 struct diff_options *o;
3110 unsigned ws_rule;
3111 unsigned status;
3112 };
3113
3114 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3115 {
3116 char firstchar;
3117 int cnt;
3118
3119 if (len < marker_size + 1)
3120 return 0;
3121 firstchar = line[0];
3122 switch (firstchar) {
3123 case '=': case '>': case '<': case '|':
3124 break;
3125 default:
3126 return 0;
3127 }
3128 for (cnt = 1; cnt < marker_size; cnt++)
3129 if (line[cnt] != firstchar)
3130 return 0;
3131 /* line[1] thru line[marker_size-1] are same as firstchar */
3132 if (len < marker_size + 1 || !isspace(line[marker_size]))
3133 return 0;
3134 return 1;
3135 }
3136
3137 static void checkdiff_consume_hunk(void *priv,
3138 long ob, long on, long nb, long nn,
3139 const char *func, long funclen)
3140
3141 {
3142 struct checkdiff_t *data = priv;
3143 data->lineno = nb - 1;
3144 }
3145
3146 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3147 {
3148 struct checkdiff_t *data = priv;
3149 int marker_size = data->conflict_marker_size;
3150 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3151 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3152 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3153 char *err;
3154 const char *line_prefix;
3155
3156 assert(data->o);
3157 line_prefix = diff_line_prefix(data->o);
3158
3159 if (line[0] == '+') {
3160 unsigned bad;
3161 data->lineno++;
3162 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3163 data->status |= 1;
3164 fprintf(data->o->file,
3165 "%s%s:%d: leftover conflict marker\n",
3166 line_prefix, data->filename, data->lineno);
3167 }
3168 bad = ws_check(line + 1, len - 1, data->ws_rule);
3169 if (!bad)
3170 return;
3171 data->status |= bad;
3172 err = whitespace_error_string(bad);
3173 fprintf(data->o->file, "%s%s:%d: %s.\n",
3174 line_prefix, data->filename, data->lineno, err);
3175 free(err);
3176 emit_line(data->o, set, reset, line, 1);
3177 ws_check_emit(line + 1, len - 1, data->ws_rule,
3178 data->o->file, set, reset, ws);
3179 } else if (line[0] == ' ') {
3180 data->lineno++;
3181 }
3182 }
3183
3184 static unsigned char *deflate_it(char *data,
3185 unsigned long size,
3186 unsigned long *result_size)
3187 {
3188 int bound;
3189 unsigned char *deflated;
3190 git_zstream stream;
3191
3192 git_deflate_init(&stream, zlib_compression_level);
3193 bound = git_deflate_bound(&stream, size);
3194 deflated = xmalloc(bound);
3195 stream.next_out = deflated;
3196 stream.avail_out = bound;
3197
3198 stream.next_in = (unsigned char *)data;
3199 stream.avail_in = size;
3200 while (git_deflate(&stream, Z_FINISH) == Z_OK)
3201 ; /* nothing */
3202 git_deflate_end(&stream);
3203 *result_size = stream.total_out;
3204 return deflated;
3205 }
3206
3207 static void emit_binary_diff_body(struct diff_options *o,
3208 mmfile_t *one, mmfile_t *two)
3209 {
3210 void *cp;
3211 void *delta;
3212 void *deflated;
3213 void *data;
3214 unsigned long orig_size;
3215 unsigned long delta_size;
3216 unsigned long deflate_size;
3217 unsigned long data_size;
3218
3219 /* We could do deflated delta, or we could do just deflated two,
3220 * whichever is smaller.
3221 */
3222 delta = NULL;
3223 deflated = deflate_it(two->ptr, two->size, &deflate_size);
3224 if (one->size && two->size) {
3225 delta = diff_delta(one->ptr, one->size,
3226 two->ptr, two->size,
3227 &delta_size, deflate_size);
3228 if (delta) {
3229 void *to_free = delta;
3230 orig_size = delta_size;
3231 delta = deflate_it(delta, delta_size, &delta_size);
3232 free(to_free);
3233 }
3234 }
3235
3236 if (delta && delta_size < deflate_size) {
3237 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3238 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3239 s, strlen(s), 0);
3240 free(s);
3241 free(deflated);
3242 data = delta;
3243 data_size = delta_size;
3244 } else {
3245 char *s = xstrfmt("%lu", two->size);
3246 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3247 s, strlen(s), 0);
3248 free(s);
3249 free(delta);
3250 data = deflated;
3251 data_size = deflate_size;
3252 }
3253
3254 /* emit data encoded in base85 */
3255 cp = data;
3256 while (data_size) {
3257 int len;
3258 int bytes = (52 < data_size) ? 52 : data_size;
3259 char line[71];
3260 data_size -= bytes;
3261 if (bytes <= 26)
3262 line[0] = bytes + 'A' - 1;
3263 else
3264 line[0] = bytes - 26 + 'a' - 1;
3265 encode_85(line + 1, cp, bytes);
3266 cp = (char *) cp + bytes;
3267
3268 len = strlen(line);
3269 line[len++] = '\n';
3270 line[len] = '\0';
3271
3272 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3273 line, len, 0);
3274 }
3275 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3276 free(data);
3277 }
3278
3279 static void emit_binary_diff(struct diff_options *o,
3280 mmfile_t *one, mmfile_t *two)
3281 {
3282 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3283 emit_binary_diff_body(o, one, two);
3284 emit_binary_diff_body(o, two, one);
3285 }
3286
3287 int diff_filespec_is_binary(struct repository *r,
3288 struct diff_filespec *one)
3289 {
3290 if (one->is_binary == -1) {
3291 diff_filespec_load_driver(one, r->index);
3292 if (one->driver->binary != -1)
3293 one->is_binary = one->driver->binary;
3294 else {
3295 if (!one->data && DIFF_FILE_VALID(one))
3296 diff_populate_filespec(r, one, CHECK_BINARY);
3297 if (one->is_binary == -1 && one->data)
3298 one->is_binary = buffer_is_binary(one->data,
3299 one->size);
3300 if (one->is_binary == -1)
3301 one->is_binary = 0;
3302 }
3303 }
3304 return one->is_binary;
3305 }
3306
3307 static const struct userdiff_funcname *
3308 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3309 {
3310 diff_filespec_load_driver(one, o->repo->index);
3311 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3312 }
3313
3314 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3315 {
3316 if (!options->a_prefix)
3317 options->a_prefix = a;
3318 if (!options->b_prefix)
3319 options->b_prefix = b;
3320 }
3321
3322 struct userdiff_driver *get_textconv(struct repository *r,
3323 struct diff_filespec *one)
3324 {
3325 if (!DIFF_FILE_VALID(one))
3326 return NULL;
3327
3328 diff_filespec_load_driver(one, r->index);
3329 return userdiff_get_textconv(r, one->driver);
3330 }
3331
3332 static void builtin_diff(const char *name_a,
3333 const char *name_b,
3334 struct diff_filespec *one,
3335 struct diff_filespec *two,
3336 const char *xfrm_msg,
3337 int must_show_header,
3338 struct diff_options *o,
3339 int complete_rewrite)
3340 {
3341 mmfile_t mf1, mf2;
3342 const char *lbl[2];
3343 char *a_one, *b_two;
3344 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3345 const char *reset = diff_get_color_opt(o, DIFF_RESET);
3346 const char *a_prefix, *b_prefix;
3347 struct userdiff_driver *textconv_one = NULL;
3348 struct userdiff_driver *textconv_two = NULL;
3349 struct strbuf header = STRBUF_INIT;
3350 const char *line_prefix = diff_line_prefix(o);
3351
3352 diff_set_mnemonic_prefix(o, "a/", "b/");
3353 if (o->flags.reverse_diff) {
3354 a_prefix = o->b_prefix;
3355 b_prefix = o->a_prefix;
3356 } else {
3357 a_prefix = o->a_prefix;
3358 b_prefix = o->b_prefix;
3359 }
3360
3361 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3362 (!one->mode || S_ISGITLINK(one->mode)) &&
3363 (!two->mode || S_ISGITLINK(two->mode))) {
3364 show_submodule_summary(o, one->path ? one->path : two->path,
3365 &one->oid, &two->oid,
3366 two->dirty_submodule);
3367 return;
3368 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3369 (!one->mode || S_ISGITLINK(one->mode)) &&
3370 (!two->mode || S_ISGITLINK(two->mode))) {
3371 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3372 &one->oid, &two->oid,
3373 two->dirty_submodule);
3374 return;
3375 }
3376
3377 if (o->flags.allow_textconv) {
3378 textconv_one = get_textconv(o->repo, one);
3379 textconv_two = get_textconv(o->repo, two);
3380 }
3381
3382 /* Never use a non-valid filename anywhere if at all possible */
3383 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3384 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3385
3386 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3387 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3388 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3389 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3390 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3391 if (lbl[0][0] == '/') {
3392 /* /dev/null */
3393 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3394 if (xfrm_msg)
3395 strbuf_addstr(&header, xfrm_msg);
3396 must_show_header = 1;
3397 }
3398 else if (lbl[1][0] == '/') {
3399 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3400 if (xfrm_msg)
3401 strbuf_addstr(&header, xfrm_msg);
3402 must_show_header = 1;
3403 }
3404 else {
3405 if (one->mode != two->mode) {
3406 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3407 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3408 must_show_header = 1;
3409 }
3410 if (xfrm_msg)
3411 strbuf_addstr(&header, xfrm_msg);
3412
3413 /*
3414 * we do not run diff between different kind
3415 * of objects.
3416 */
3417 if ((one->mode ^ two->mode) & S_IFMT)
3418 goto free_ab_and_return;
3419 if (complete_rewrite &&
3420 (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3421 (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3422 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3423 header.buf, header.len, 0);
3424 strbuf_reset(&header);
3425 emit_rewrite_diff(name_a, name_b, one, two,
3426 textconv_one, textconv_two, o);
3427 o->found_changes = 1;
3428 goto free_ab_and_return;
3429 }
3430 }
3431
3432 if (o->irreversible_delete && lbl[1][0] == '/') {
3433 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3434 header.len, 0);
3435 strbuf_reset(&header);
3436 goto free_ab_and_return;
3437 } else if (!o->flags.text &&
3438 ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3439 (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3440 struct strbuf sb = STRBUF_INIT;
3441 if (!one->data && !two->data &&
3442 S_ISREG(one->mode) && S_ISREG(two->mode) &&
3443 !o->flags.binary) {
3444 if (oideq(&one->oid, &two->oid)) {
3445 if (must_show_header)
3446 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3447 header.buf, header.len,
3448 0);
3449 goto free_ab_and_return;
3450 }
3451 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3452 header.buf, header.len, 0);
3453 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3454 diff_line_prefix(o), lbl[0], lbl[1]);
3455 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3456 sb.buf, sb.len, 0);
3457 strbuf_release(&sb);
3458 goto free_ab_and_return;
3459 }
3460 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3461 fill_mmfile(o->repo, &mf2, two) < 0)
3462 die("unable to read files to diff");
3463 /* Quite common confusing case */
3464 if (mf1.size == mf2.size &&
3465 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3466 if (must_show_header)
3467 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3468 header.buf, header.len, 0);
3469 goto free_ab_and_return;
3470 }
3471 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3472 strbuf_reset(&header);
3473 if (o->flags.binary)
3474 emit_binary_diff(o, &mf1, &mf2);
3475 else {
3476 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3477 diff_line_prefix(o), lbl[0], lbl[1]);
3478 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3479 sb.buf, sb.len, 0);
3480 strbuf_release(&sb);
3481 }
3482 o->found_changes = 1;
3483 } else {
3484 /* Crazy xdl interfaces.. */
3485 const char *diffopts = getenv("GIT_DIFF_OPTS");
3486 const char *v;
3487 xpparam_t xpp;
3488 xdemitconf_t xecfg;
3489 struct emit_callback ecbdata;
3490 const struct userdiff_funcname *pe;
3491
3492 if (must_show_header) {
3493 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3494 header.buf, header.len, 0);
3495 strbuf_reset(&header);
3496 }
3497
3498 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3499 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3500
3501 pe = diff_funcname_pattern(o, one);
3502 if (!pe)
3503 pe = diff_funcname_pattern(o, two);
3504
3505 memset(&xpp, 0, sizeof(xpp));
3506 memset(&xecfg, 0, sizeof(xecfg));
3507 memset(&ecbdata, 0, sizeof(ecbdata));
3508 if (o->flags.suppress_diff_headers)
3509 lbl[0] = NULL;
3510 ecbdata.label_path = lbl;
3511 ecbdata.color_diff = want_color(o->use_color);
3512 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3513 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3514 check_blank_at_eof(&mf1, &mf2, &ecbdata);
3515 ecbdata.opt = o;
3516 if (header.len && !o->flags.suppress_diff_headers)
3517 ecbdata.header = &header;
3518 xpp.flags = o->xdl_opts;
3519 xpp.anchors = o->anchors;
3520 xpp.anchors_nr = o->anchors_nr;
3521 xecfg.ctxlen = o->context;
3522 xecfg.interhunkctxlen = o->interhunkcontext;
3523 xecfg.flags = XDL_EMIT_FUNCNAMES;
3524 if (o->flags.funccontext)
3525 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3526 if (pe)
3527 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3528 if (!diffopts)
3529 ;
3530 else if (skip_prefix(diffopts, "--unified=", &v))
3531 xecfg.ctxlen = strtoul(v, NULL, 10);
3532 else if (skip_prefix(diffopts, "-u", &v))
3533 xecfg.ctxlen = strtoul(v, NULL, 10);
3534 if (o->word_diff)
3535 init_diff_words_data(&ecbdata, o, one, two);
3536 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3537 &ecbdata, &xpp, &xecfg))
3538 die("unable to generate diff for %s", one->path);
3539 if (o->word_diff)
3540 free_diff_words_data(&ecbdata);
3541 if (textconv_one)
3542 free(mf1.ptr);
3543 if (textconv_two)
3544 free(mf2.ptr);
3545 xdiff_clear_find_func(&xecfg);
3546 }
3547
3548 free_ab_and_return:
3549 strbuf_release(&header);
3550 diff_free_filespec_data(one);
3551 diff_free_filespec_data(two);
3552 free(a_one);
3553 free(b_two);
3554 return;
3555 }
3556
3557 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3558 {
3559 if (!is_renamed) {
3560 if (p->status == DIFF_STATUS_ADDED) {
3561 if (S_ISLNK(p->two->mode))
3562 return "new +l";
3563 else if ((p->two->mode & 0777) == 0755)
3564 return "new +x";
3565 else
3566 return "new";
3567 } else if (p->status == DIFF_STATUS_DELETED)
3568 return "gone";
3569 }
3570 if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3571 return "mode -l";
3572 else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3573 return "mode +l";
3574 else if ((p->one->mode & 0777) == 0644 &&
3575 (p->two->mode & 0777) == 0755)
3576 return "mode +x";
3577 else if ((p->one->mode & 0777) == 0755 &&
3578 (p->two->mode & 0777) == 0644)
3579 return "mode -x";
3580 return NULL;
3581 }
3582
3583 static void builtin_diffstat(const char *name_a, const char *name_b,
3584 struct diff_filespec *one,
3585 struct diff_filespec *two,
3586 struct diffstat_t *diffstat,
3587 struct diff_options *o,
3588 struct diff_filepair *p)
3589 {
3590 mmfile_t mf1, mf2;
3591 struct diffstat_file *data;
3592 int same_contents;
3593 int complete_rewrite = 0;
3594
3595 if (!DIFF_PAIR_UNMERGED(p)) {
3596 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3597 complete_rewrite = 1;
3598 }
3599
3600 data = diffstat_add(diffstat, name_a, name_b);
3601 data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3602 if (o->flags.stat_with_summary)
3603 data->comments = get_compact_summary(p, data->is_renamed);
3604
3605 if (!one || !two) {
3606 data->is_unmerged = 1;
3607 return;
3608 }
3609
3610 same_contents = oideq(&one->oid, &two->oid);
3611
3612 if (diff_filespec_is_binary(o->repo, one) ||
3613 diff_filespec_is_binary(o->repo, two)) {
3614 data->is_binary = 1;
3615 if (same_contents) {
3616 data->added = 0;
3617 data->deleted = 0;
3618 } else {
3619 data->added = diff_filespec_size(o->repo, two);
3620 data->deleted = diff_filespec_size(o->repo, one);
3621 }
3622 }
3623
3624 else if (complete_rewrite) {
3625 diff_populate_filespec(o->repo, one, 0);
3626 diff_populate_filespec(o->repo, two, 0);
3627 data->deleted = count_lines(one->data, one->size);
3628 data->added = count_lines(two->data, two->size);
3629 }
3630
3631 else if (!same_contents) {
3632 /* Crazy xdl interfaces.. */
3633 xpparam_t xpp;
3634 xdemitconf_t xecfg;
3635
3636 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3637 fill_mmfile(o->repo, &mf2, two) < 0)
3638 die("unable to read files to diff");
3639
3640 memset(&xpp, 0, sizeof(xpp));
3641 memset(&xecfg, 0, sizeof(xecfg));
3642 xpp.flags = o->xdl_opts;
3643 xpp.anchors = o->anchors;
3644 xpp.anchors_nr = o->anchors_nr;
3645 xecfg.ctxlen = o->context;
3646 xecfg.interhunkctxlen = o->interhunkcontext;
3647 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3648 diffstat_consume, diffstat, &xpp, &xecfg))
3649 die("unable to generate diffstat for %s", one->path);
3650 }
3651
3652 diff_free_filespec_data(one);
3653 diff_free_filespec_data(two);
3654 }
3655
3656 static void builtin_checkdiff(const char *name_a, const char *name_b,
3657 const char *attr_path,
3658 struct diff_filespec *one,
3659 struct diff_filespec *two,
3660 struct diff_options *o)
3661 {
3662 mmfile_t mf1, mf2;
3663 struct checkdiff_t data;
3664
3665 if (!two)
3666 return;
3667
3668 memset(&data, 0, sizeof(data));
3669 data.filename = name_b ? name_b : name_a;
3670 data.lineno = 0;
3671 data.o = o;
3672 data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3673 data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3674
3675 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3676 fill_mmfile(o->repo, &mf2, two) < 0)
3677 die("unable to read files to diff");
3678
3679 /*
3680 * All the other codepaths check both sides, but not checking
3681 * the "old" side here is deliberate. We are checking the newly
3682 * introduced changes, and as long as the "new" side is text, we
3683 * can and should check what it introduces.
3684 */
3685 if (diff_filespec_is_binary(o->repo, two))
3686 goto free_and_return;
3687 else {
3688 /* Crazy xdl interfaces.. */
3689 xpparam_t xpp;
3690 xdemitconf_t xecfg;
3691
3692 memset(&xpp, 0, sizeof(xpp));
3693 memset(&xecfg, 0, sizeof(xecfg));
3694 xecfg.ctxlen = 1; /* at least one context line */
3695 xpp.flags = 0;
3696 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3697 checkdiff_consume, &data,
3698 &xpp, &xecfg))
3699 die("unable to generate checkdiff for %s", one->path);
3700
3701 if (data.ws_rule & WS_BLANK_AT_EOF) {
3702 struct emit_callback ecbdata;
3703 int blank_at_eof;
3704
3705 ecbdata.ws_rule = data.ws_rule;
3706 check_blank_at_eof(&mf1, &mf2, &ecbdata);
3707 blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3708
3709 if (blank_at_eof) {
3710 static char *err;
3711 if (!err)
3712 err = whitespace_error_string(WS_BLANK_AT_EOF);
3713 fprintf(o->file, "%s:%d: %s.\n",
3714 data.filename, blank_at_eof, err);
3715 data.status = 1; /* report errors */
3716 }
3717 }
3718 }
3719 free_and_return:
3720 diff_free_filespec_data(one);
3721 diff_free_filespec_data(two);
3722 if