diff --color-moved-ws: fix another memory leak
[git/git.git] / diff.c
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include "cache.h"
5 #include "config.h"
6 #include "tempfile.h"
7 #include "quote.h"
8 #include "diff.h"
9 #include "diffcore.h"
10 #include "delta.h"
11 #include "xdiff-interface.h"
12 #include "color.h"
13 #include "attr.h"
14 #include "run-command.h"
15 #include "utf8.h"
16 #include "object-store.h"
17 #include "userdiff.h"
18 #include "submodule-config.h"
19 #include "submodule.h"
20 #include "hashmap.h"
21 #include "ll-merge.h"
22 #include "string-list.h"
23 #include "argv-array.h"
24 #include "graph.h"
25 #include "packfile.h"
26 #include "help.h"
27
28 #ifdef NO_FAST_WORKING_DIRECTORY
29 #define FAST_WORKING_DIRECTORY 0
30 #else
31 #define FAST_WORKING_DIRECTORY 1
32 #endif
33
34 static int diff_detect_rename_default;
35 static int diff_indent_heuristic = 1;
36 static int diff_rename_limit_default = 400;
37 static int diff_suppress_blank_empty;
38 static int diff_use_color_default = -1;
39 static int diff_color_moved_default;
40 static int diff_color_moved_ws_default;
41 static int diff_context_default = 3;
42 static int diff_interhunk_context_default;
43 static const char *diff_word_regex_cfg;
44 static const char *external_diff_cmd_cfg;
45 static const char *diff_order_file_cfg;
46 int diff_auto_refresh_index = 1;
47 static int diff_mnemonic_prefix;
48 static int diff_no_prefix;
49 static int diff_stat_graph_width;
50 static int diff_dirstat_permille_default = 30;
51 static struct diff_options default_diff_options;
52 static long diff_algorithm;
53 static unsigned ws_error_highlight_default = WSEH_NEW;
54
55 static char diff_colors[][COLOR_MAXLEN] = {
56 GIT_COLOR_RESET,
57 GIT_COLOR_NORMAL, /* CONTEXT */
58 GIT_COLOR_BOLD, /* METAINFO */
59 GIT_COLOR_CYAN, /* FRAGINFO */
60 GIT_COLOR_RED, /* OLD */
61 GIT_COLOR_GREEN, /* NEW */
62 GIT_COLOR_YELLOW, /* COMMIT */
63 GIT_COLOR_BG_RED, /* WHITESPACE */
64 GIT_COLOR_NORMAL, /* FUNCINFO */
65 GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
66 GIT_COLOR_BOLD_BLUE, /* OLD_MOVED ALTERNATIVE */
67 GIT_COLOR_FAINT, /* OLD_MOVED_DIM */
68 GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
69 GIT_COLOR_BOLD_CYAN, /* NEW_MOVED */
70 GIT_COLOR_BOLD_YELLOW, /* NEW_MOVED ALTERNATIVE */
71 GIT_COLOR_FAINT, /* NEW_MOVED_DIM */
72 GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
73 GIT_COLOR_FAINT, /* CONTEXT_DIM */
74 GIT_COLOR_FAINT_RED, /* OLD_DIM */
75 GIT_COLOR_FAINT_GREEN, /* NEW_DIM */
76 GIT_COLOR_BOLD, /* CONTEXT_BOLD */
77 GIT_COLOR_BOLD_RED, /* OLD_BOLD */
78 GIT_COLOR_BOLD_GREEN, /* NEW_BOLD */
79 };
80
81 static const char *color_diff_slots[] = {
82 [DIFF_CONTEXT] = "context",
83 [DIFF_METAINFO] = "meta",
84 [DIFF_FRAGINFO] = "frag",
85 [DIFF_FILE_OLD] = "old",
86 [DIFF_FILE_NEW] = "new",
87 [DIFF_COMMIT] = "commit",
88 [DIFF_WHITESPACE] = "whitespace",
89 [DIFF_FUNCINFO] = "func",
90 [DIFF_FILE_OLD_MOVED] = "oldMoved",
91 [DIFF_FILE_OLD_MOVED_ALT] = "oldMovedAlternative",
92 [DIFF_FILE_OLD_MOVED_DIM] = "oldMovedDimmed",
93 [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
94 [DIFF_FILE_NEW_MOVED] = "newMoved",
95 [DIFF_FILE_NEW_MOVED_ALT] = "newMovedAlternative",
96 [DIFF_FILE_NEW_MOVED_DIM] = "newMovedDimmed",
97 [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
98 [DIFF_CONTEXT_DIM] = "contextDimmed",
99 [DIFF_FILE_OLD_DIM] = "oldDimmed",
100 [DIFF_FILE_NEW_DIM] = "newDimmed",
101 [DIFF_CONTEXT_BOLD] = "contextBold",
102 [DIFF_FILE_OLD_BOLD] = "oldBold",
103 [DIFF_FILE_NEW_BOLD] = "newBold",
104 };
105
106 static NORETURN void die_want_option(const char *option_name)
107 {
108 die(_("option '%s' requires a value"), option_name);
109 }
110
111 define_list_config_array_extra(color_diff_slots, {"plain"});
112
113 static int parse_diff_color_slot(const char *var)
114 {
115 if (!strcasecmp(var, "plain"))
116 return DIFF_CONTEXT;
117 return LOOKUP_CONFIG(color_diff_slots, var);
118 }
119
120 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
121 struct strbuf *errmsg)
122 {
123 char *params_copy = xstrdup(params_string);
124 struct string_list params = STRING_LIST_INIT_NODUP;
125 int ret = 0;
126 int i;
127
128 if (*params_copy)
129 string_list_split_in_place(&params, params_copy, ',', -1);
130 for (i = 0; i < params.nr; i++) {
131 const char *p = params.items[i].string;
132 if (!strcmp(p, "changes")) {
133 options->flags.dirstat_by_line = 0;
134 options->flags.dirstat_by_file = 0;
135 } else if (!strcmp(p, "lines")) {
136 options->flags.dirstat_by_line = 1;
137 options->flags.dirstat_by_file = 0;
138 } else if (!strcmp(p, "files")) {
139 options->flags.dirstat_by_line = 0;
140 options->flags.dirstat_by_file = 1;
141 } else if (!strcmp(p, "noncumulative")) {
142 options->flags.dirstat_cumulative = 0;
143 } else if (!strcmp(p, "cumulative")) {
144 options->flags.dirstat_cumulative = 1;
145 } else if (isdigit(*p)) {
146 char *end;
147 int permille = strtoul(p, &end, 10) * 10;
148 if (*end == '.' && isdigit(*++end)) {
149 /* only use first digit */
150 permille += *end - '0';
151 /* .. and ignore any further digits */
152 while (isdigit(*++end))
153 ; /* nothing */
154 }
155 if (!*end)
156 options->dirstat_permille = permille;
157 else {
158 strbuf_addf(errmsg, _(" Failed to parse dirstat cut-off percentage '%s'\n"),
159 p);
160 ret++;
161 }
162 } else {
163 strbuf_addf(errmsg, _(" Unknown dirstat parameter '%s'\n"), p);
164 ret++;
165 }
166
167 }
168 string_list_clear(&params, 0);
169 free(params_copy);
170 return ret;
171 }
172
173 static int parse_submodule_params(struct diff_options *options, const char *value)
174 {
175 if (!strcmp(value, "log"))
176 options->submodule_format = DIFF_SUBMODULE_LOG;
177 else if (!strcmp(value, "short"))
178 options->submodule_format = DIFF_SUBMODULE_SHORT;
179 else if (!strcmp(value, "diff"))
180 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
181 else
182 return -1;
183 return 0;
184 }
185
186 int git_config_rename(const char *var, const char *value)
187 {
188 if (!value)
189 return DIFF_DETECT_RENAME;
190 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
191 return DIFF_DETECT_COPY;
192 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
193 }
194
195 long parse_algorithm_value(const char *value)
196 {
197 if (!value)
198 return -1;
199 else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
200 return 0;
201 else if (!strcasecmp(value, "minimal"))
202 return XDF_NEED_MINIMAL;
203 else if (!strcasecmp(value, "patience"))
204 return XDF_PATIENCE_DIFF;
205 else if (!strcasecmp(value, "histogram"))
206 return XDF_HISTOGRAM_DIFF;
207 return -1;
208 }
209
210 static int parse_one_token(const char **arg, const char *token)
211 {
212 const char *rest;
213 if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
214 *arg = rest;
215 return 1;
216 }
217 return 0;
218 }
219
220 static int parse_ws_error_highlight(const char *arg)
221 {
222 const char *orig_arg = arg;
223 unsigned val = 0;
224
225 while (*arg) {
226 if (parse_one_token(&arg, "none"))
227 val = 0;
228 else if (parse_one_token(&arg, "default"))
229 val = WSEH_NEW;
230 else if (parse_one_token(&arg, "all"))
231 val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
232 else if (parse_one_token(&arg, "new"))
233 val |= WSEH_NEW;
234 else if (parse_one_token(&arg, "old"))
235 val |= WSEH_OLD;
236 else if (parse_one_token(&arg, "context"))
237 val |= WSEH_CONTEXT;
238 else {
239 return -1 - (int)(arg - orig_arg);
240 }
241 if (*arg)
242 arg++;
243 }
244 return val;
245 }
246
247 /*
248 * These are to give UI layer defaults.
249 * The core-level commands such as git-diff-files should
250 * never be affected by the setting of diff.renames
251 * the user happens to have in the configuration file.
252 */
253 void init_diff_ui_defaults(void)
254 {
255 diff_detect_rename_default = DIFF_DETECT_RENAME;
256 }
257
258 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
259 {
260 if (!strcmp(var, "diff.indentheuristic"))
261 diff_indent_heuristic = git_config_bool(var, value);
262 return 0;
263 }
264
265 static int parse_color_moved(const char *arg)
266 {
267 switch (git_parse_maybe_bool(arg)) {
268 case 0:
269 return COLOR_MOVED_NO;
270 case 1:
271 return COLOR_MOVED_DEFAULT;
272 default:
273 break;
274 }
275
276 if (!strcmp(arg, "no"))
277 return COLOR_MOVED_NO;
278 else if (!strcmp(arg, "plain"))
279 return COLOR_MOVED_PLAIN;
280 else if (!strcmp(arg, "blocks"))
281 return COLOR_MOVED_BLOCKS;
282 else if (!strcmp(arg, "zebra"))
283 return COLOR_MOVED_ZEBRA;
284 else if (!strcmp(arg, "default"))
285 return COLOR_MOVED_DEFAULT;
286 else if (!strcmp(arg, "dimmed-zebra"))
287 return COLOR_MOVED_ZEBRA_DIM;
288 else if (!strcmp(arg, "dimmed_zebra"))
289 return COLOR_MOVED_ZEBRA_DIM;
290 else
291 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
292 }
293
294 static int parse_color_moved_ws(const char *arg)
295 {
296 int ret = 0;
297 struct string_list l = STRING_LIST_INIT_DUP;
298 struct string_list_item *i;
299
300 string_list_split(&l, arg, ',', -1);
301
302 for_each_string_list_item(i, &l) {
303 struct strbuf sb = STRBUF_INIT;
304 strbuf_addstr(&sb, i->string);
305 strbuf_trim(&sb);
306
307 if (!strcmp(sb.buf, "ignore-space-change"))
308 ret |= XDF_IGNORE_WHITESPACE_CHANGE;
309 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
310 ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
311 else if (!strcmp(sb.buf, "ignore-all-space"))
312 ret |= XDF_IGNORE_WHITESPACE;
313 else if (!strcmp(sb.buf, "allow-indentation-change"))
314 ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
315 else
316 error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
317
318 strbuf_release(&sb);
319 }
320
321 if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
322 (ret & XDF_WHITESPACE_FLAGS))
323 die(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
324
325 string_list_clear(&l, 0);
326
327 return ret;
328 }
329
330 int git_diff_ui_config(const char *var, const char *value, void *cb)
331 {
332 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
333 diff_use_color_default = git_config_colorbool(var, value);
334 return 0;
335 }
336 if (!strcmp(var, "diff.colormoved")) {
337 int cm = parse_color_moved(value);
338 if (cm < 0)
339 return -1;
340 diff_color_moved_default = cm;
341 return 0;
342 }
343 if (!strcmp(var, "diff.colormovedws")) {
344 int cm = parse_color_moved_ws(value);
345 if (cm < 0)
346 return -1;
347 diff_color_moved_ws_default = cm;
348 return 0;
349 }
350 if (!strcmp(var, "diff.context")) {
351 diff_context_default = git_config_int(var, value);
352 if (diff_context_default < 0)
353 return -1;
354 return 0;
355 }
356 if (!strcmp(var, "diff.interhunkcontext")) {
357 diff_interhunk_context_default = git_config_int(var, value);
358 if (diff_interhunk_context_default < 0)
359 return -1;
360 return 0;
361 }
362 if (!strcmp(var, "diff.renames")) {
363 diff_detect_rename_default = git_config_rename(var, value);
364 return 0;
365 }
366 if (!strcmp(var, "diff.autorefreshindex")) {
367 diff_auto_refresh_index = git_config_bool(var, value);
368 return 0;
369 }
370 if (!strcmp(var, "diff.mnemonicprefix")) {
371 diff_mnemonic_prefix = git_config_bool(var, value);
372 return 0;
373 }
374 if (!strcmp(var, "diff.noprefix")) {
375 diff_no_prefix = git_config_bool(var, value);
376 return 0;
377 }
378 if (!strcmp(var, "diff.statgraphwidth")) {
379 diff_stat_graph_width = git_config_int(var, value);
380 return 0;
381 }
382 if (!strcmp(var, "diff.external"))
383 return git_config_string(&external_diff_cmd_cfg, var, value);
384 if (!strcmp(var, "diff.wordregex"))
385 return git_config_string(&diff_word_regex_cfg, var, value);
386 if (!strcmp(var, "diff.orderfile"))
387 return git_config_pathname(&diff_order_file_cfg, var, value);
388
389 if (!strcmp(var, "diff.ignoresubmodules"))
390 handle_ignore_submodules_arg(&default_diff_options, value);
391
392 if (!strcmp(var, "diff.submodule")) {
393 if (parse_submodule_params(&default_diff_options, value))
394 warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
395 value);
396 return 0;
397 }
398
399 if (!strcmp(var, "diff.algorithm")) {
400 diff_algorithm = parse_algorithm_value(value);
401 if (diff_algorithm < 0)
402 return -1;
403 return 0;
404 }
405
406 if (!strcmp(var, "diff.wserrorhighlight")) {
407 int val = parse_ws_error_highlight(value);
408 if (val < 0)
409 return -1;
410 ws_error_highlight_default = val;
411 return 0;
412 }
413
414 if (git_color_config(var, value, cb) < 0)
415 return -1;
416
417 return git_diff_basic_config(var, value, cb);
418 }
419
420 int git_diff_basic_config(const char *var, const char *value, void *cb)
421 {
422 const char *name;
423
424 if (!strcmp(var, "diff.renamelimit")) {
425 diff_rename_limit_default = git_config_int(var, value);
426 return 0;
427 }
428
429 if (userdiff_config(var, value) < 0)
430 return -1;
431
432 if (skip_prefix(var, "diff.color.", &name) ||
433 skip_prefix(var, "color.diff.", &name)) {
434 int slot = parse_diff_color_slot(name);
435 if (slot < 0)
436 return 0;
437 if (!value)
438 return config_error_nonbool(var);
439 return color_parse(value, diff_colors[slot]);
440 }
441
442 /* like GNU diff's --suppress-blank-empty option */
443 if (!strcmp(var, "diff.suppressblankempty") ||
444 /* for backwards compatibility */
445 !strcmp(var, "diff.suppress-blank-empty")) {
446 diff_suppress_blank_empty = git_config_bool(var, value);
447 return 0;
448 }
449
450 if (!strcmp(var, "diff.dirstat")) {
451 struct strbuf errmsg = STRBUF_INIT;
452 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
453 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
454 warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
455 errmsg.buf);
456 strbuf_release(&errmsg);
457 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
458 return 0;
459 }
460
461 if (git_diff_heuristic_config(var, value, cb) < 0)
462 return -1;
463
464 return git_default_config(var, value, cb);
465 }
466
467 static char *quote_two(const char *one, const char *two)
468 {
469 int need_one = quote_c_style(one, NULL, NULL, 1);
470 int need_two = quote_c_style(two, NULL, NULL, 1);
471 struct strbuf res = STRBUF_INIT;
472
473 if (need_one + need_two) {
474 strbuf_addch(&res, '"');
475 quote_c_style(one, &res, NULL, 1);
476 quote_c_style(two, &res, NULL, 1);
477 strbuf_addch(&res, '"');
478 } else {
479 strbuf_addstr(&res, one);
480 strbuf_addstr(&res, two);
481 }
482 return strbuf_detach(&res, NULL);
483 }
484
485 static const char *external_diff(void)
486 {
487 static const char *external_diff_cmd = NULL;
488 static int done_preparing = 0;
489
490 if (done_preparing)
491 return external_diff_cmd;
492 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
493 if (!external_diff_cmd)
494 external_diff_cmd = external_diff_cmd_cfg;
495 done_preparing = 1;
496 return external_diff_cmd;
497 }
498
499 /*
500 * Keep track of files used for diffing. Sometimes such an entry
501 * refers to a temporary file, sometimes to an existing file, and
502 * sometimes to "/dev/null".
503 */
504 static struct diff_tempfile {
505 /*
506 * filename external diff should read from, or NULL if this
507 * entry is currently not in use:
508 */
509 const char *name;
510
511 char hex[GIT_MAX_HEXSZ + 1];
512 char mode[10];
513
514 /*
515 * If this diff_tempfile instance refers to a temporary file,
516 * this tempfile object is used to manage its lifetime.
517 */
518 struct tempfile *tempfile;
519 } diff_temp[2];
520
521 struct emit_callback {
522 int color_diff;
523 unsigned ws_rule;
524 int blank_at_eof_in_preimage;
525 int blank_at_eof_in_postimage;
526 int lno_in_preimage;
527 int lno_in_postimage;
528 const char **label_path;
529 struct diff_words_data *diff_words;
530 struct diff_options *opt;
531 struct strbuf *header;
532 };
533
534 static int count_lines(const char *data, int size)
535 {
536 int count, ch, completely_empty = 1, nl_just_seen = 0;
537 count = 0;
538 while (0 < size--) {
539 ch = *data++;
540 if (ch == '\n') {
541 count++;
542 nl_just_seen = 1;
543 completely_empty = 0;
544 }
545 else {
546 nl_just_seen = 0;
547 completely_empty = 0;
548 }
549 }
550 if (completely_empty)
551 return 0;
552 if (!nl_just_seen)
553 count++; /* no trailing newline */
554 return count;
555 }
556
557 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
558 {
559 if (!DIFF_FILE_VALID(one)) {
560 mf->ptr = (char *)""; /* does not matter */
561 mf->size = 0;
562 return 0;
563 }
564 else if (diff_populate_filespec(one, 0))
565 return -1;
566
567 mf->ptr = one->data;
568 mf->size = one->size;
569 return 0;
570 }
571
572 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
573 static unsigned long diff_filespec_size(struct diff_filespec *one)
574 {
575 if (!DIFF_FILE_VALID(one))
576 return 0;
577 diff_populate_filespec(one, CHECK_SIZE_ONLY);
578 return one->size;
579 }
580
581 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
582 {
583 char *ptr = mf->ptr;
584 long size = mf->size;
585 int cnt = 0;
586
587 if (!size)
588 return cnt;
589 ptr += size - 1; /* pointing at the very end */
590 if (*ptr != '\n')
591 ; /* incomplete line */
592 else
593 ptr--; /* skip the last LF */
594 while (mf->ptr < ptr) {
595 char *prev_eol;
596 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
597 if (*prev_eol == '\n')
598 break;
599 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
600 break;
601 cnt++;
602 ptr = prev_eol - 1;
603 }
604 return cnt;
605 }
606
607 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
608 struct emit_callback *ecbdata)
609 {
610 int l1, l2, at;
611 unsigned ws_rule = ecbdata->ws_rule;
612 l1 = count_trailing_blank(mf1, ws_rule);
613 l2 = count_trailing_blank(mf2, ws_rule);
614 if (l2 <= l1) {
615 ecbdata->blank_at_eof_in_preimage = 0;
616 ecbdata->blank_at_eof_in_postimage = 0;
617 return;
618 }
619 at = count_lines(mf1->ptr, mf1->size);
620 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
621
622 at = count_lines(mf2->ptr, mf2->size);
623 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
624 }
625
626 static void emit_line_0(struct diff_options *o,
627 const char *set_sign, const char *set, unsigned reverse, const char *reset,
628 int first, const char *line, int len)
629 {
630 int has_trailing_newline, has_trailing_carriage_return;
631 int needs_reset = 0; /* at the end of the line */
632 FILE *file = o->file;
633
634 fputs(diff_line_prefix(o), file);
635
636 has_trailing_newline = (len > 0 && line[len-1] == '\n');
637 if (has_trailing_newline)
638 len--;
639
640 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
641 if (has_trailing_carriage_return)
642 len--;
643
644 if (!len && !first)
645 goto end_of_line;
646
647 if (reverse && want_color(o->use_color)) {
648 fputs(GIT_COLOR_REVERSE, file);
649 needs_reset = 1;
650 }
651
652 if (set_sign) {
653 fputs(set_sign, file);
654 needs_reset = 1;
655 }
656
657 if (first)
658 fputc(first, file);
659
660 if (!len)
661 goto end_of_line;
662
663 if (set) {
664 if (set_sign && set != set_sign)
665 fputs(reset, file);
666 fputs(set, file);
667 needs_reset = 1;
668 }
669 fwrite(line, len, 1, file);
670 needs_reset = 1; /* 'line' may contain color codes. */
671
672 end_of_line:
673 if (needs_reset)
674 fputs(reset, file);
675 if (has_trailing_carriage_return)
676 fputc('\r', file);
677 if (has_trailing_newline)
678 fputc('\n', file);
679 }
680
681 static void emit_line(struct diff_options *o, const char *set, const char *reset,
682 const char *line, int len)
683 {
684 emit_line_0(o, set, NULL, 0, reset, 0, line, len);
685 }
686
687 enum diff_symbol {
688 DIFF_SYMBOL_BINARY_DIFF_HEADER,
689 DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
690 DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
691 DIFF_SYMBOL_BINARY_DIFF_BODY,
692 DIFF_SYMBOL_BINARY_DIFF_FOOTER,
693 DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
694 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
695 DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
696 DIFF_SYMBOL_STATS_LINE,
697 DIFF_SYMBOL_WORD_DIFF,
698 DIFF_SYMBOL_STAT_SEP,
699 DIFF_SYMBOL_SUMMARY,
700 DIFF_SYMBOL_SUBMODULE_ADD,
701 DIFF_SYMBOL_SUBMODULE_DEL,
702 DIFF_SYMBOL_SUBMODULE_UNTRACKED,
703 DIFF_SYMBOL_SUBMODULE_MODIFIED,
704 DIFF_SYMBOL_SUBMODULE_HEADER,
705 DIFF_SYMBOL_SUBMODULE_ERROR,
706 DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
707 DIFF_SYMBOL_REWRITE_DIFF,
708 DIFF_SYMBOL_BINARY_FILES,
709 DIFF_SYMBOL_HEADER,
710 DIFF_SYMBOL_FILEPAIR_PLUS,
711 DIFF_SYMBOL_FILEPAIR_MINUS,
712 DIFF_SYMBOL_WORDS_PORCELAIN,
713 DIFF_SYMBOL_WORDS,
714 DIFF_SYMBOL_CONTEXT,
715 DIFF_SYMBOL_CONTEXT_INCOMPLETE,
716 DIFF_SYMBOL_PLUS,
717 DIFF_SYMBOL_MINUS,
718 DIFF_SYMBOL_NO_LF_EOF,
719 DIFF_SYMBOL_CONTEXT_FRAGINFO,
720 DIFF_SYMBOL_CONTEXT_MARKER,
721 DIFF_SYMBOL_SEPARATOR
722 };
723 /*
724 * Flags for content lines:
725 * 0..12 are whitespace rules
726 * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
727 * 16 is marking if the line is blank at EOF
728 */
729 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF (1<<16)
730 #define DIFF_SYMBOL_MOVED_LINE (1<<17)
731 #define DIFF_SYMBOL_MOVED_LINE_ALT (1<<18)
732 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING (1<<19)
733 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
734
735 /*
736 * This struct is used when we need to buffer the output of the diff output.
737 *
738 * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
739 * into the pre/post image file. This pointer could be a union with the
740 * line pointer. By storing an offset into the file instead of the literal line,
741 * we can decrease the memory footprint for the buffered output. At first we
742 * may want to only have indirection for the content lines, but we could also
743 * enhance the state for emitting prefabricated lines, e.g. the similarity
744 * score line or hunk/file headers would only need to store a number or path
745 * and then the output can be constructed later on depending on state.
746 */
747 struct emitted_diff_symbol {
748 const char *line;
749 int len;
750 int flags;
751 enum diff_symbol s;
752 };
753 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
754
755 struct emitted_diff_symbols {
756 struct emitted_diff_symbol *buf;
757 int nr, alloc;
758 };
759 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
760
761 static void append_emitted_diff_symbol(struct diff_options *o,
762 struct emitted_diff_symbol *e)
763 {
764 struct emitted_diff_symbol *f;
765
766 ALLOC_GROW(o->emitted_symbols->buf,
767 o->emitted_symbols->nr + 1,
768 o->emitted_symbols->alloc);
769 f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
770
771 memcpy(f, e, sizeof(struct emitted_diff_symbol));
772 f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
773 }
774
775 struct moved_entry {
776 struct hashmap_entry ent;
777 const struct emitted_diff_symbol *es;
778 struct moved_entry *next_line;
779 };
780
781 /**
782 * The struct ws_delta holds white space differences between moved lines, i.e.
783 * between '+' and '-' lines that have been detected to be a move.
784 * The string contains the difference in leading white spaces, before the
785 * rest of the line is compared using the white space config for move
786 * coloring. The current_longer indicates if the first string in the
787 * comparision is longer than the second.
788 */
789 struct ws_delta {
790 char *string;
791 unsigned int current_longer : 1;
792 };
793 #define WS_DELTA_INIT { NULL, 0 }
794
795 struct moved_block {
796 struct moved_entry *match;
797 struct ws_delta wsd;
798 };
799
800 static void moved_block_clear(struct moved_block *b)
801 {
802 FREE_AND_NULL(b->wsd.string);
803 b->match = NULL;
804 }
805
806 static int compute_ws_delta(const struct emitted_diff_symbol *a,
807 const struct emitted_diff_symbol *b,
808 struct ws_delta *out)
809 {
810 const struct emitted_diff_symbol *longer = a->len > b->len ? a : b;
811 const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
812 int d = longer->len - shorter->len;
813
814 if (strncmp(longer->line + d, shorter->line, shorter->len))
815 return 0;
816
817 out->string = xmemdupz(longer->line, d);
818 out->current_longer = (a == longer);
819
820 return 1;
821 }
822
823 static int cmp_in_block_with_wsd(const struct diff_options *o,
824 const struct moved_entry *cur,
825 const struct moved_entry *match,
826 struct moved_block *pmb,
827 int n)
828 {
829 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
830 int al = cur->es->len, cl = l->len;
831 const char *a = cur->es->line,
832 *b = match->es->line,
833 *c = l->line;
834
835 int wslen;
836
837 /*
838 * We need to check if 'cur' is equal to 'match'.
839 * As those are from the same (+/-) side, we do not need to adjust for
840 * indent changes. However these were found using fuzzy matching
841 * so we do have to check if they are equal.
842 */
843 if (strcmp(a, b))
844 return 1;
845
846 if (!pmb->wsd.string)
847 /*
848 * The white space delta is not active? This can happen
849 * when we exit early in this function.
850 */
851 return 1;
852
853 /*
854 * The indent changes of the block are known and stored in
855 * pmb->wsd; however we need to check if the indent changes of the
856 * current line are still the same as before.
857 *
858 * To do so we need to compare 'l' to 'cur', adjusting the
859 * one of them for the white spaces, depending which was longer.
860 */
861
862 wslen = strlen(pmb->wsd.string);
863 if (pmb->wsd.current_longer) {
864 c += wslen;
865 cl -= wslen;
866 } else {
867 a += wslen;
868 al -= wslen;
869 }
870
871 if (al != cl || memcmp(a, c, al))
872 return 1;
873
874 return 0;
875 }
876
877 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
878 const void *entry,
879 const void *entry_or_key,
880 const void *keydata)
881 {
882 const struct diff_options *diffopt = hashmap_cmp_fn_data;
883 const struct moved_entry *a = entry;
884 const struct moved_entry *b = entry_or_key;
885 unsigned flags = diffopt->color_moved_ws_handling
886 & XDF_WHITESPACE_FLAGS;
887
888 if (diffopt->color_moved_ws_handling &
889 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
890 /*
891 * As there is not specific white space config given,
892 * we'd need to check for a new block, so ignore all
893 * white space. The setup of the white space
894 * configuration for the next block is done else where
895 */
896 flags |= XDF_IGNORE_WHITESPACE;
897
898 return !xdiff_compare_lines(a->es->line, a->es->len,
899 b->es->line, b->es->len,
900 flags);
901 }
902
903 static struct moved_entry *prepare_entry(struct diff_options *o,
904 int line_no)
905 {
906 struct moved_entry *ret = xmalloc(sizeof(*ret));
907 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
908 unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
909
910 ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
911 ret->es = l;
912 ret->next_line = NULL;
913
914 return ret;
915 }
916
917 static void add_lines_to_move_detection(struct diff_options *o,
918 struct hashmap *add_lines,
919 struct hashmap *del_lines)
920 {
921 struct moved_entry *prev_line = NULL;
922
923 int n;
924 for (n = 0; n < o->emitted_symbols->nr; n++) {
925 struct hashmap *hm;
926 struct moved_entry *key;
927
928 switch (o->emitted_symbols->buf[n].s) {
929 case DIFF_SYMBOL_PLUS:
930 hm = add_lines;
931 break;
932 case DIFF_SYMBOL_MINUS:
933 hm = del_lines;
934 break;
935 default:
936 prev_line = NULL;
937 continue;
938 }
939
940 key = prepare_entry(o, n);
941 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
942 prev_line->next_line = key;
943
944 hashmap_add(hm, key);
945 prev_line = key;
946 }
947 }
948
949 static void pmb_advance_or_null(struct diff_options *o,
950 struct moved_entry *match,
951 struct hashmap *hm,
952 struct moved_block *pmb,
953 int pmb_nr)
954 {
955 int i;
956 for (i = 0; i < pmb_nr; i++) {
957 struct moved_entry *prev = pmb[i].match;
958 struct moved_entry *cur = (prev && prev->next_line) ?
959 prev->next_line : NULL;
960 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
961 pmb[i].match = cur;
962 } else {
963 pmb[i].match = NULL;
964 }
965 }
966 }
967
968 static void pmb_advance_or_null_multi_match(struct diff_options *o,
969 struct moved_entry *match,
970 struct hashmap *hm,
971 struct moved_block *pmb,
972 int pmb_nr, int n)
973 {
974 int i;
975 char *got_match = xcalloc(1, pmb_nr);
976
977 for (; match; match = hashmap_get_next(hm, match)) {
978 for (i = 0; i < pmb_nr; i++) {
979 struct moved_entry *prev = pmb[i].match;
980 struct moved_entry *cur = (prev && prev->next_line) ?
981 prev->next_line : NULL;
982 if (!cur)
983 continue;
984 if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
985 got_match[i] |= 1;
986 }
987 }
988
989 for (i = 0; i < pmb_nr; i++) {
990 if (got_match[i]) {
991 /* Advance to the next line */
992 pmb[i].match = pmb[i].match->next_line;
993 } else {
994 moved_block_clear(&pmb[i]);
995 }
996 }
997
998 free(got_match);
999 }
1000
1001 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1002 int pmb_nr)
1003 {
1004 int lp, rp;
1005
1006 /* Shrink the set of potential block to the remaining running */
1007 for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1008 while (lp < pmb_nr && pmb[lp].match)
1009 lp++;
1010 /* lp points at the first NULL now */
1011
1012 while (rp > -1 && !pmb[rp].match)
1013 rp--;
1014 /* rp points at the last non-NULL */
1015
1016 if (lp < pmb_nr && rp > -1 && lp < rp) {
1017 pmb[lp] = pmb[rp];
1018 pmb[rp].match = NULL;
1019 pmb[rp].wsd.string = NULL;
1020 rp--;
1021 lp++;
1022 }
1023 }
1024
1025 /* Remember the number of running sets */
1026 return rp + 1;
1027 }
1028
1029 /*
1030 * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1031 *
1032 * Otherwise, if the last block has fewer alphanumeric characters than
1033 * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1034 * that block.
1035 *
1036 * The last block consists of the (n - block_length)'th line up to but not
1037 * including the nth line.
1038 *
1039 * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1040 * Think of a way to unify them.
1041 */
1042 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1043 {
1044 int i, alnum_count = 0;
1045 if (o->color_moved == COLOR_MOVED_PLAIN)
1046 return;
1047 for (i = 1; i < block_length + 1; i++) {
1048 const char *c = o->emitted_symbols->buf[n - i].line;
1049 for (; *c; c++) {
1050 if (!isalnum(*c))
1051 continue;
1052 alnum_count++;
1053 if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1054 return;
1055 }
1056 }
1057 for (i = 1; i < block_length + 1; i++)
1058 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1059 }
1060
1061 /* Find blocks of moved code, delegate actual coloring decision to helper */
1062 static void mark_color_as_moved(struct diff_options *o,
1063 struct hashmap *add_lines,
1064 struct hashmap *del_lines)
1065 {
1066 struct moved_block *pmb = NULL; /* potentially moved blocks */
1067 int pmb_nr = 0, pmb_alloc = 0;
1068 int n, flipped_block = 1, block_length = 0;
1069
1070
1071 for (n = 0; n < o->emitted_symbols->nr; n++) {
1072 struct hashmap *hm = NULL;
1073 struct moved_entry *key;
1074 struct moved_entry *match = NULL;
1075 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1076
1077 switch (l->s) {
1078 case DIFF_SYMBOL_PLUS:
1079 hm = del_lines;
1080 key = prepare_entry(o, n);
1081 match = hashmap_get(hm, key, NULL);
1082 free(key);
1083 break;
1084 case DIFF_SYMBOL_MINUS:
1085 hm = add_lines;
1086 key = prepare_entry(o, n);
1087 match = hashmap_get(hm, key, NULL);
1088 free(key);
1089 break;
1090 default:
1091 flipped_block = 1;
1092 }
1093
1094 if (!match) {
1095 int i;
1096
1097 adjust_last_block(o, n, block_length);
1098 for(i = 0; i < pmb_nr; i++)
1099 moved_block_clear(&pmb[i]);
1100 pmb_nr = 0;
1101 block_length = 0;
1102 continue;
1103 }
1104
1105 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1106
1107 if (o->color_moved == COLOR_MOVED_PLAIN)
1108 continue;
1109
1110 if (o->color_moved_ws_handling &
1111 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1112 pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1113 else
1114 pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1115
1116 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1117
1118 if (pmb_nr == 0) {
1119 /*
1120 * The current line is the start of a new block.
1121 * Setup the set of potential blocks.
1122 */
1123 for (; match; match = hashmap_get_next(hm, match)) {
1124 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1125 if (o->color_moved_ws_handling &
1126 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1127 if (compute_ws_delta(l, match->es,
1128 &pmb[pmb_nr].wsd))
1129 pmb[pmb_nr++].match = match;
1130 } else {
1131 pmb[pmb_nr].wsd.string = NULL;
1132 pmb[pmb_nr++].match = match;
1133 }
1134 }
1135
1136 flipped_block = (flipped_block + 1) % 2;
1137
1138 adjust_last_block(o, n, block_length);
1139 block_length = 0;
1140 }
1141
1142 block_length++;
1143
1144 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1145 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1146 }
1147 adjust_last_block(o, n, block_length);
1148
1149 for(n = 0; n < pmb_nr; n++)
1150 moved_block_clear(&pmb[n]);
1151 free(pmb);
1152 }
1153
1154 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1155 (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1156 static void dim_moved_lines(struct diff_options *o)
1157 {
1158 int n;
1159 for (n = 0; n < o->emitted_symbols->nr; n++) {
1160 struct emitted_diff_symbol *prev = (n != 0) ?
1161 &o->emitted_symbols->buf[n - 1] : NULL;
1162 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1163 struct emitted_diff_symbol *next =
1164 (n < o->emitted_symbols->nr - 1) ?
1165 &o->emitted_symbols->buf[n + 1] : NULL;
1166
1167 /* Not a plus or minus line? */
1168 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1169 continue;
1170
1171 /* Not a moved line? */
1172 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1173 continue;
1174
1175 /*
1176 * If prev or next are not a plus or minus line,
1177 * pretend they don't exist
1178 */
1179 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1180 prev->s != DIFF_SYMBOL_MINUS)
1181 prev = NULL;
1182 if (next && next->s != DIFF_SYMBOL_PLUS &&
1183 next->s != DIFF_SYMBOL_MINUS)
1184 next = NULL;
1185
1186 /* Inside a block? */
1187 if ((prev &&
1188 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1189 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1190 (next &&
1191 (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1192 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1193 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1194 continue;
1195 }
1196
1197 /* Check if we are at an interesting bound: */
1198 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1199 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1200 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1201 continue;
1202 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1203 (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1204 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1205 continue;
1206
1207 /*
1208 * The boundary to prev and next are not interesting,
1209 * so this line is not interesting as a whole
1210 */
1211 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1212 }
1213 }
1214
1215 static void emit_line_ws_markup(struct diff_options *o,
1216 const char *set_sign, const char *set,
1217 const char *reset,
1218 char sign, const char *line, int len,
1219 unsigned ws_rule, int blank_at_eof)
1220 {
1221 const char *ws = NULL;
1222
1223 if (o->ws_error_highlight & ws_rule) {
1224 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1225 if (!*ws)
1226 ws = NULL;
1227 }
1228
1229 if (!ws && !set_sign)
1230 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1231 else if (!ws) {
1232 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1233 } else if (blank_at_eof)
1234 /* Blank line at EOF - paint '+' as well */
1235 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1236 else {
1237 /* Emit just the prefix, then the rest. */
1238 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1239 sign, "", 0);
1240 ws_check_emit(line, len, ws_rule,
1241 o->file, set, reset, ws);
1242 }
1243 }
1244
1245 static void emit_diff_symbol_from_struct(struct diff_options *o,
1246 struct emitted_diff_symbol *eds)
1247 {
1248 static const char *nneof = " No newline at end of file\n";
1249 const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1250 struct strbuf sb = STRBUF_INIT;
1251
1252 enum diff_symbol s = eds->s;
1253 const char *line = eds->line;
1254 int len = eds->len;
1255 unsigned flags = eds->flags;
1256
1257 switch (s) {
1258 case DIFF_SYMBOL_NO_LF_EOF:
1259 context = diff_get_color_opt(o, DIFF_CONTEXT);
1260 reset = diff_get_color_opt(o, DIFF_RESET);
1261 putc('\n', o->file);
1262 emit_line_0(o, context, NULL, 0, reset, '\\',
1263 nneof, strlen(nneof));
1264 break;
1265 case DIFF_SYMBOL_SUBMODULE_HEADER:
1266 case DIFF_SYMBOL_SUBMODULE_ERROR:
1267 case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1268 case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1269 case DIFF_SYMBOL_SUMMARY:
1270 case DIFF_SYMBOL_STATS_LINE:
1271 case DIFF_SYMBOL_BINARY_DIFF_BODY:
1272 case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1273 emit_line(o, "", "", line, len);
1274 break;
1275 case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1276 case DIFF_SYMBOL_CONTEXT_MARKER:
1277 context = diff_get_color_opt(o, DIFF_CONTEXT);
1278 reset = diff_get_color_opt(o, DIFF_RESET);
1279 emit_line(o, context, reset, line, len);
1280 break;
1281 case DIFF_SYMBOL_SEPARATOR:
1282 fprintf(o->file, "%s%c",
1283 diff_line_prefix(o),
1284 o->line_termination);
1285 break;
1286 case DIFF_SYMBOL_CONTEXT:
1287 set = diff_get_color_opt(o, DIFF_CONTEXT);
1288 reset = diff_get_color_opt(o, DIFF_RESET);
1289 set_sign = NULL;
1290 if (o->flags.dual_color_diffed_diffs) {
1291 char c = !len ? 0 : line[0];
1292
1293 if (c == '+')
1294 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1295 else if (c == '@')
1296 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1297 else if (c == '-')
1298 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1299 }
1300 emit_line_ws_markup(o, set_sign, set, reset,
1301 o->output_indicators[OUTPUT_INDICATOR_CONTEXT],
1302 line, len,
1303 flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1304 break;
1305 case DIFF_SYMBOL_PLUS:
1306 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1307 DIFF_SYMBOL_MOVED_LINE_ALT |
1308 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1309 case DIFF_SYMBOL_MOVED_LINE |
1310 DIFF_SYMBOL_MOVED_LINE_ALT |
1311 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1312 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1313 break;
1314 case DIFF_SYMBOL_MOVED_LINE |
1315 DIFF_SYMBOL_MOVED_LINE_ALT:
1316 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1317 break;
1318 case DIFF_SYMBOL_MOVED_LINE |
1319 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1320 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1321 break;
1322 case DIFF_SYMBOL_MOVED_LINE:
1323 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1324 break;
1325 default:
1326 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1327 }
1328 reset = diff_get_color_opt(o, DIFF_RESET);
1329 if (!o->flags.dual_color_diffed_diffs)
1330 set_sign = NULL;
1331 else {
1332 char c = !len ? 0 : line[0];
1333
1334 set_sign = set;
1335 if (c == '-')
1336 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1337 else if (c == '@')
1338 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1339 else if (c == '+')
1340 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1341 else
1342 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1343 flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1344 }
1345 emit_line_ws_markup(o, set_sign, set, reset,
1346 o->output_indicators[OUTPUT_INDICATOR_NEW],
1347 line, len,
1348 flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1349 flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1350 break;
1351 case DIFF_SYMBOL_MINUS:
1352 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1353 DIFF_SYMBOL_MOVED_LINE_ALT |
1354 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1355 case DIFF_SYMBOL_MOVED_LINE |
1356 DIFF_SYMBOL_MOVED_LINE_ALT |
1357 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1358 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1359 break;
1360 case DIFF_SYMBOL_MOVED_LINE |
1361 DIFF_SYMBOL_MOVED_LINE_ALT:
1362 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1363 break;
1364 case DIFF_SYMBOL_MOVED_LINE |
1365 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1366 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1367 break;
1368 case DIFF_SYMBOL_MOVED_LINE:
1369 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1370 break;
1371 default:
1372 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1373 }
1374 reset = diff_get_color_opt(o, DIFF_RESET);
1375 if (!o->flags.dual_color_diffed_diffs)
1376 set_sign = NULL;
1377 else {
1378 char c = !len ? 0 : line[0];
1379
1380 set_sign = set;
1381 if (c == '+')
1382 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1383 else if (c == '@')
1384 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1385 else if (c == '-')
1386 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1387 else
1388 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1389 }
1390 emit_line_ws_markup(o, set_sign, set, reset,
1391 o->output_indicators[OUTPUT_INDICATOR_OLD],
1392 line, len,
1393 flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1394 break;
1395 case DIFF_SYMBOL_WORDS_PORCELAIN:
1396 context = diff_get_color_opt(o, DIFF_CONTEXT);
1397 reset = diff_get_color_opt(o, DIFF_RESET);
1398 emit_line(o, context, reset, line, len);
1399 fputs("~\n", o->file);
1400 break;
1401 case DIFF_SYMBOL_WORDS:
1402 context = diff_get_color_opt(o, DIFF_CONTEXT);
1403 reset = diff_get_color_opt(o, DIFF_RESET);
1404 /*
1405 * Skip the prefix character, if any. With
1406 * diff_suppress_blank_empty, there may be
1407 * none.
1408 */
1409 if (line[0] != '\n') {
1410 line++;
1411 len--;
1412 }
1413 emit_line(o, context, reset, line, len);
1414 break;
1415 case DIFF_SYMBOL_FILEPAIR_PLUS:
1416 meta = diff_get_color_opt(o, DIFF_METAINFO);
1417 reset = diff_get_color_opt(o, DIFF_RESET);
1418 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1419 line, reset,
1420 strchr(line, ' ') ? "\t" : "");
1421 break;
1422 case DIFF_SYMBOL_FILEPAIR_MINUS:
1423 meta = diff_get_color_opt(o, DIFF_METAINFO);
1424 reset = diff_get_color_opt(o, DIFF_RESET);
1425 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1426 line, reset,
1427 strchr(line, ' ') ? "\t" : "");
1428 break;
1429 case DIFF_SYMBOL_BINARY_FILES:
1430 case DIFF_SYMBOL_HEADER:
1431 fprintf(o->file, "%s", line);
1432 break;
1433 case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1434 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1435 break;
1436 case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1437 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1438 break;
1439 case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1440 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1441 break;
1442 case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1443 fputs(diff_line_prefix(o), o->file);
1444 fputc('\n', o->file);
1445 break;
1446 case DIFF_SYMBOL_REWRITE_DIFF:
1447 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1448 reset = diff_get_color_opt(o, DIFF_RESET);
1449 emit_line(o, fraginfo, reset, line, len);
1450 break;
1451 case DIFF_SYMBOL_SUBMODULE_ADD:
1452 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1453 reset = diff_get_color_opt(o, DIFF_RESET);
1454 emit_line(o, set, reset, line, len);
1455 break;
1456 case DIFF_SYMBOL_SUBMODULE_DEL:
1457 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1458 reset = diff_get_color_opt(o, DIFF_RESET);
1459 emit_line(o, set, reset, line, len);
1460 break;
1461 case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1462 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1463 diff_line_prefix(o), line);
1464 break;
1465 case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1466 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1467 diff_line_prefix(o), line);
1468 break;
1469 case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1470 emit_line(o, "", "", " 0 files changed\n",
1471 strlen(" 0 files changed\n"));
1472 break;
1473 case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1474 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1475 break;
1476 case DIFF_SYMBOL_WORD_DIFF:
1477 fprintf(o->file, "%.*s", len, line);
1478 break;
1479 case DIFF_SYMBOL_STAT_SEP:
1480 fputs(o->stat_sep, o->file);
1481 break;
1482 default:
1483 BUG("unknown diff symbol");
1484 }
1485 strbuf_release(&sb);
1486 }
1487
1488 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1489 const char *line, int len, unsigned flags)
1490 {
1491 struct emitted_diff_symbol e = {line, len, flags, s};
1492
1493 if (o->emitted_symbols)
1494 append_emitted_diff_symbol(o, &e);
1495 else
1496 emit_diff_symbol_from_struct(o, &e);
1497 }
1498
1499 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1500 {
1501 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1502 }
1503
1504 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1505 {
1506 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1507 }
1508
1509 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1510 {
1511 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1512 path, strlen(path), 0);
1513 }
1514
1515 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1516 {
1517 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1518 path, strlen(path), 0);
1519 }
1520
1521 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1522 {
1523 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1524 header, strlen(header), 0);
1525 }
1526
1527 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1528 {
1529 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1530 }
1531
1532 void diff_emit_submodule_pipethrough(struct diff_options *o,
1533 const char *line, int len)
1534 {
1535 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1536 }
1537
1538 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1539 {
1540 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1541 ecbdata->blank_at_eof_in_preimage &&
1542 ecbdata->blank_at_eof_in_postimage &&
1543 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1544 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1545 return 0;
1546 return ws_blank_line(line, len, ecbdata->ws_rule);
1547 }
1548
1549 static void emit_add_line(const char *reset,
1550 struct emit_callback *ecbdata,
1551 const char *line, int len)
1552 {
1553 unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1554 if (new_blank_line_at_eof(ecbdata, line, len))
1555 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1556
1557 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1558 }
1559
1560 static void emit_del_line(const char *reset,
1561 struct emit_callback *ecbdata,
1562 const char *line, int len)
1563 {
1564 unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1565 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1566 }
1567
1568 static void emit_context_line(const char *reset,
1569 struct emit_callback *ecbdata,
1570 const char *line, int len)
1571 {
1572 unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1573 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1574 }
1575
1576 static void emit_hunk_header(struct emit_callback *ecbdata,
1577 const char *line, int len)
1578 {
1579 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1580 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1581 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1582 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1583 const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1584 static const char atat[2] = { '@', '@' };
1585 const char *cp, *ep;
1586 struct strbuf msgbuf = STRBUF_INIT;
1587 int org_len = len;
1588 int i = 1;
1589
1590 /*
1591 * As a hunk header must begin with "@@ -<old>, +<new> @@",
1592 * it always is at least 10 bytes long.
1593 */
1594 if (len < 10 ||
1595 memcmp(line, atat, 2) ||
1596 !(ep = memmem(line + 2, len - 2, atat, 2))) {
1597 emit_diff_symbol(ecbdata->opt,
1598 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1599 return;
1600 }
1601 ep += 2; /* skip over @@ */
1602
1603 /* The hunk header in fraginfo color */
1604 if (ecbdata->opt->flags.dual_color_diffed_diffs)
1605 strbuf_addstr(&msgbuf, reverse);
1606 strbuf_addstr(&msgbuf, frag);
1607 strbuf_add(&msgbuf, line, ep - line);
1608 strbuf_addstr(&msgbuf, reset);
1609
1610 /*
1611 * trailing "\r\n"
1612 */
1613 for ( ; i < 3; i++)
1614 if (line[len - i] == '\r' || line[len - i] == '\n')
1615 len--;
1616
1617 /* blank before the func header */
1618 for (cp = ep; ep - line < len; ep++)
1619 if (*ep != ' ' && *ep != '\t')
1620 break;
1621 if (ep != cp) {
1622 strbuf_addstr(&msgbuf, context);
1623 strbuf_add(&msgbuf, cp, ep - cp);
1624 strbuf_addstr(&msgbuf, reset);
1625 }
1626
1627 if (ep < line + len) {
1628 strbuf_addstr(&msgbuf, func);
1629 strbuf_add(&msgbuf, ep, line + len - ep);
1630 strbuf_addstr(&msgbuf, reset);
1631 }
1632
1633 strbuf_add(&msgbuf, line + len, org_len - len);
1634 strbuf_complete_line(&msgbuf);
1635 emit_diff_symbol(ecbdata->opt,
1636 DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1637 strbuf_release(&msgbuf);
1638 }
1639
1640 static struct diff_tempfile *claim_diff_tempfile(void) {
1641 int i;
1642 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1643 if (!diff_temp[i].name)
1644 return diff_temp + i;
1645 BUG("diff is failing to clean up its tempfiles");
1646 }
1647
1648 static void remove_tempfile(void)
1649 {
1650 int i;
1651 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1652 if (is_tempfile_active(diff_temp[i].tempfile))
1653 delete_tempfile(&diff_temp[i].tempfile);
1654 diff_temp[i].name = NULL;
1655 }
1656 }
1657
1658 static void add_line_count(struct strbuf *out, int count)
1659 {
1660 switch (count) {
1661 case 0:
1662 strbuf_addstr(out, "0,0");
1663 break;
1664 case 1:
1665 strbuf_addstr(out, "1");
1666 break;
1667 default:
1668 strbuf_addf(out, "1,%d", count);
1669 break;
1670 }
1671 }
1672
1673 static void emit_rewrite_lines(struct emit_callback *ecb,
1674 int prefix, const char *data, int size)
1675 {
1676 const char *endp = NULL;
1677 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1678
1679 while (0 < size) {
1680 int len;
1681
1682 endp = memchr(data, '\n', size);
1683 len = endp ? (endp - data + 1) : size;
1684 if (prefix != '+') {
1685 ecb->lno_in_preimage++;
1686 emit_del_line(reset, ecb, data, len);
1687 } else {
1688 ecb->lno_in_postimage++;
1689 emit_add_line(reset, ecb, data, len);
1690 }
1691 size -= len;
1692 data += len;
1693 }
1694 if (!endp)
1695 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1696 }
1697
1698 static void emit_rewrite_diff(const char *name_a,
1699 const char *name_b,
1700 struct diff_filespec *one,
1701 struct diff_filespec *two,
1702 struct userdiff_driver *textconv_one,
1703 struct userdiff_driver *textconv_two,
1704 struct diff_options *o)
1705 {
1706 int lc_a, lc_b;
1707 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1708 const char *a_prefix, *b_prefix;
1709 char *data_one, *data_two;
1710 size_t size_one, size_two;
1711 struct emit_callback ecbdata;
1712 struct strbuf out = STRBUF_INIT;
1713
1714 if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1715 a_prefix = o->b_prefix;
1716 b_prefix = o->a_prefix;
1717 } else {
1718 a_prefix = o->a_prefix;
1719 b_prefix = o->b_prefix;
1720 }
1721
1722 name_a += (*name_a == '/');
1723 name_b += (*name_b == '/');
1724
1725 strbuf_reset(&a_name);
1726 strbuf_reset(&b_name);
1727 quote_two_c_style(&a_name, a_prefix, name_a, 0);
1728 quote_two_c_style(&b_name, b_prefix, name_b, 0);
1729
1730 size_one = fill_textconv(textconv_one, one, &data_one);
1731 size_two = fill_textconv(textconv_two, two, &data_two);
1732
1733 memset(&ecbdata, 0, sizeof(ecbdata));
1734 ecbdata.color_diff = want_color(o->use_color);
1735 ecbdata.ws_rule = whitespace_rule(name_b);
1736 ecbdata.opt = o;
1737 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1738 mmfile_t mf1, mf2;
1739 mf1.ptr = (char *)data_one;
1740 mf2.ptr = (char *)data_two;
1741 mf1.size = size_one;
1742 mf2.size = size_two;
1743 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1744 }
1745 ecbdata.lno_in_preimage = 1;
1746 ecbdata.lno_in_postimage = 1;
1747
1748 lc_a = count_lines(data_one, size_one);
1749 lc_b = count_lines(data_two, size_two);
1750
1751 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1752 a_name.buf, a_name.len, 0);
1753 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1754 b_name.buf, b_name.len, 0);
1755
1756 strbuf_addstr(&out, "@@ -");
1757 if (!o->irreversible_delete)
1758 add_line_count(&out, lc_a);
1759 else
1760 strbuf_addstr(&out, "?,?");
1761 strbuf_addstr(&out, " +");
1762 add_line_count(&out, lc_b);
1763 strbuf_addstr(&out, " @@\n");
1764 emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1765 strbuf_release(&out);
1766
1767 if (lc_a && !o->irreversible_delete)
1768 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1769 if (lc_b)
1770 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1771 if (textconv_one)
1772 free((char *)data_one);
1773 if (textconv_two)
1774 free((char *)data_two);
1775 }
1776
1777 struct diff_words_buffer {
1778 mmfile_t text;
1779 unsigned long alloc;
1780 struct diff_words_orig {
1781 const char *begin, *end;
1782 } *orig;
1783 int orig_nr, orig_alloc;
1784 };
1785
1786 static void diff_words_append(char *line, unsigned long len,
1787 struct diff_words_buffer *buffer)
1788 {
1789 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1790 line++;
1791 len--;
1792 memcpy(buffer->text.ptr + buffer->text.size, line, len);
1793 buffer->text.size += len;
1794 buffer->text.ptr[buffer->text.size] = '\0';
1795 }
1796
1797 struct diff_words_style_elem {
1798 const char *prefix;
1799 const char *suffix;
1800 const char *color; /* NULL; filled in by the setup code if
1801 * color is enabled */
1802 };
1803
1804 struct diff_words_style {
1805 enum diff_words_type type;
1806 struct diff_words_style_elem new_word, old_word, ctx;
1807 const char *newline;
1808 };
1809
1810 static struct diff_words_style diff_words_styles[] = {
1811 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1812 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1813 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1814 };
1815
1816 struct diff_words_data {
1817 struct diff_words_buffer minus, plus;
1818 const char *current_plus;
1819 int last_minus;
1820 struct diff_options *opt;
1821 regex_t *word_regex;
1822 enum diff_words_type type;
1823 struct diff_words_style *style;
1824 };
1825
1826 static int fn_out_diff_words_write_helper(struct diff_options *o,
1827 struct diff_words_style_elem *st_el,
1828 const char *newline,
1829 size_t count, const char *buf)
1830 {
1831 int print = 0;
1832 struct strbuf sb = STRBUF_INIT;
1833
1834 while (count) {
1835 char *p = memchr(buf, '\n', count);
1836 if (print)
1837 strbuf_addstr(&sb, diff_line_prefix(o));
1838
1839 if (p != buf) {
1840 const char *reset = st_el->color && *st_el->color ?
1841 GIT_COLOR_RESET : NULL;
1842 if (st_el->color && *st_el->color)
1843 strbuf_addstr(&sb, st_el->color);
1844 strbuf_addstr(&sb, st_el->prefix);
1845 strbuf_add(&sb, buf, p ? p - buf : count);
1846 strbuf_addstr(&sb, st_el->suffix);
1847 if (reset)
1848 strbuf_addstr(&sb, reset);
1849 }
1850 if (!p)
1851 goto out;
1852
1853 strbuf_addstr(&sb, newline);
1854 count -= p + 1 - buf;
1855 buf = p + 1;
1856 print = 1;
1857 if (count) {
1858 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1859 sb.buf, sb.len, 0);
1860 strbuf_reset(&sb);
1861 }
1862 }
1863
1864 out:
1865 if (sb.len)
1866 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1867 sb.buf, sb.len, 0);
1868 strbuf_release(&sb);
1869 return 0;
1870 }
1871
1872 /*
1873 * '--color-words' algorithm can be described as:
1874 *
1875 * 1. collect the minus/plus lines of a diff hunk, divided into
1876 * minus-lines and plus-lines;
1877 *
1878 * 2. break both minus-lines and plus-lines into words and
1879 * place them into two mmfile_t with one word for each line;
1880 *
1881 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1882 *
1883 * And for the common parts of the both file, we output the plus side text.
1884 * diff_words->current_plus is used to trace the current position of the plus file
1885 * which printed. diff_words->last_minus is used to trace the last minus word
1886 * printed.
1887 *
1888 * For '--graph' to work with '--color-words', we need to output the graph prefix
1889 * on each line of color words output. Generally, there are two conditions on
1890 * which we should output the prefix.
1891 *
1892 * 1. diff_words->last_minus == 0 &&
1893 * diff_words->current_plus == diff_words->plus.text.ptr
1894 *
1895 * that is: the plus text must start as a new line, and if there is no minus
1896 * word printed, a graph prefix must be printed.
1897 *
1898 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
1899 * *(diff_words->current_plus - 1) == '\n'
1900 *
1901 * that is: a graph prefix must be printed following a '\n'
1902 */
1903 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1904 {
1905 if ((diff_words->last_minus == 0 &&
1906 diff_words->current_plus == diff_words->plus.text.ptr) ||
1907 (diff_words->current_plus > diff_words->plus.text.ptr &&
1908 *(diff_words->current_plus - 1) == '\n')) {
1909 return 1;
1910 } else {
1911 return 0;
1912 }
1913 }
1914
1915 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1916 {
1917 struct diff_words_data *diff_words = priv;
1918 struct diff_words_style *style = diff_words->style;
1919 int minus_first, minus_len, plus_first, plus_len;
1920 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1921 struct diff_options *opt = diff_words->opt;
1922 const char *line_prefix;
1923
1924 if (line[0] != '@' || parse_hunk_header(line, len,
1925 &minus_first, &minus_len, &plus_first, &plus_len))
1926 return;
1927
1928 assert(opt);
1929 line_prefix = diff_line_prefix(opt);
1930
1931 /* POSIX requires that first be decremented by one if len == 0... */
1932 if (minus_len) {
1933 minus_begin = diff_words->minus.orig[minus_first].begin;
1934 minus_end =
1935 diff_words->minus.orig[minus_first + minus_len - 1].end;
1936 } else
1937 minus_begin = minus_end =
1938 diff_words->minus.orig[minus_first].end;
1939
1940 if (plus_len) {
1941 plus_begin = diff_words->plus.orig[plus_first].begin;
1942 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1943 } else
1944 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1945
1946 if (color_words_output_graph_prefix(diff_words)) {
1947 fputs(line_prefix, diff_words->opt->file);
1948 }
1949 if (diff_words->current_plus != plus_begin) {
1950 fn_out_diff_words_write_helper(diff_words->opt,
1951 &style->ctx, style->newline,
1952 plus_begin - diff_words->current_plus,
1953 diff_words->current_plus);
1954 }
1955 if (minus_begin != minus_end) {
1956 fn_out_diff_words_write_helper(diff_words->opt,
1957 &style->old_word, style->newline,
1958 minus_end - minus_begin, minus_begin);
1959 }
1960 if (plus_begin != plus_end) {
1961 fn_out_diff_words_write_helper(diff_words->opt,
1962 &style->new_word, style->newline,
1963 plus_end - plus_begin, plus_begin);
1964 }
1965
1966 diff_words->current_plus = plus_end;
1967 diff_words->last_minus = minus_first;
1968 }
1969
1970 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1971 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1972 int *begin, int *end)
1973 {
1974 if (word_regex && *begin < buffer->size) {
1975 regmatch_t match[1];
1976 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1977 buffer->size - *begin, 1, match, 0)) {
1978 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1979 '\n', match[0].rm_eo - match[0].rm_so);
1980 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1981 *begin += match[0].rm_so;
1982 return *begin >= *end;
1983 }
1984 return -1;
1985 }
1986
1987 /* find the next word */
1988 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1989 (*begin)++;
1990 if (*begin >= buffer->size)
1991 return -1;
1992
1993 /* find the end of the word */
1994 *end = *begin + 1;
1995 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1996 (*end)++;
1997
1998 return 0;
1999 }
2000
2001 /*
2002 * This function splits the words in buffer->text, stores the list with
2003 * newline separator into out, and saves the offsets of the original words
2004 * in buffer->orig.
2005 */
2006 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2007 regex_t *word_regex)
2008 {
2009 int i, j;
2010 long alloc = 0;
2011
2012 out->size = 0;
2013 out->ptr = NULL;
2014
2015 /* fake an empty "0th" word */
2016 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2017 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2018 buffer->orig_nr = 1;
2019
2020 for (i = 0; i < buffer->text.size; i++) {
2021 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2022 return;
2023
2024 /* store original boundaries */
2025 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2026 buffer->orig_alloc);
2027 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2028 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2029 buffer->orig_nr++;
2030
2031 /* store one word */
2032 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2033 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2034 out->ptr[out->size + j - i] = '\n';
2035 out->size += j - i + 1;
2036
2037 i = j - 1;
2038 }
2039 }
2040
2041 /* this executes the word diff on the accumulated buffers */
2042 static void diff_words_show(struct diff_words_data *diff_words)
2043 {
2044 xpparam_t xpp;
2045 xdemitconf_t xecfg;
2046 mmfile_t minus, plus;
2047 struct diff_words_style *style = diff_words->style;
2048
2049 struct diff_options *opt = diff_words->opt;
2050 const char *line_prefix;
2051
2052 assert(opt);
2053 line_prefix = diff_line_prefix(opt);
2054
2055 /* special case: only removal */
2056 if (!diff_words->plus.text.size) {
2057 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2058 line_prefix, strlen(line_prefix), 0);
2059 fn_out_diff_words_write_helper(diff_words->opt,
2060 &style->old_word, style->newline,
2061 diff_words->minus.text.size,
2062 diff_words->minus.text.ptr);
2063 diff_words->minus.text.size = 0;
2064 return;
2065 }
2066
2067 diff_words->current_plus = diff_words->plus.text.ptr;
2068 diff_words->last_minus = 0;
2069
2070 memset(&xpp, 0, sizeof(xpp));
2071 memset(&xecfg, 0, sizeof(xecfg));
2072 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2073 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2074 xpp.flags = 0;
2075 /* as only the hunk header will be parsed, we need a 0-context */
2076 xecfg.ctxlen = 0;
2077 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
2078 &xpp, &xecfg))
2079 die("unable to generate word diff");
2080 free(minus.ptr);
2081 free(plus.ptr);
2082 if (diff_words->current_plus != diff_words->plus.text.ptr +
2083 diff_words->plus.text.size) {
2084 if (color_words_output_graph_prefix(diff_words))
2085 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2086 line_prefix, strlen(line_prefix), 0);
2087 fn_out_diff_words_write_helper(diff_words->opt,
2088 &style->ctx, style->newline,
2089 diff_words->plus.text.ptr + diff_words->plus.text.size
2090 - diff_words->current_plus, diff_words->current_plus);
2091 }
2092 diff_words->minus.text.size = diff_words->plus.text.size = 0;
2093 }
2094
2095 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2096 static void diff_words_flush(struct emit_callback *ecbdata)
2097 {
2098 struct diff_options *wo = ecbdata->diff_words->opt;
2099
2100 if (ecbdata->diff_words->minus.text.size ||
2101 ecbdata->diff_words->plus.text.size)
2102 diff_words_show(ecbdata->diff_words);
2103
2104 if (wo->emitted_symbols) {
2105 struct diff_options *o = ecbdata->opt;
2106 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2107 int i;
2108
2109 /*
2110 * NEEDSWORK:
2111 * Instead of appending each, concat all words to a line?
2112 */
2113 for (i = 0; i < wol->nr; i++)
2114 append_emitted_diff_symbol(o, &wol->buf[i]);
2115
2116 for (i = 0; i < wol->nr; i++)
2117 free((void *)wol->buf[i].line);
2118
2119 wol->nr = 0;
2120 }
2121 }
2122
2123 static void diff_filespec_load_driver(struct diff_filespec *one)
2124 {
2125 /* Use already-loaded driver */
2126 if (one->driver)
2127 return;
2128
2129 if (S_ISREG(one->mode))
2130 one->driver = userdiff_find_by_path(one->path);
2131
2132 /* Fallback to default settings */
2133 if (!one->driver)
2134 one->driver = userdiff_find_by_name("default");
2135 }
2136
2137 static const char *userdiff_word_regex(struct diff_filespec *one)
2138 {
2139 diff_filespec_load_driver(one);
2140 return one->driver->word_regex;
2141 }
2142
2143 static void init_diff_words_data(struct emit_callback *ecbdata,
2144 struct diff_options *orig_opts,
2145 struct diff_filespec *one,
2146 struct diff_filespec *two)
2147 {
2148 int i;
2149 struct diff_options *o = xmalloc(sizeof(struct diff_options));
2150 memcpy(o, orig_opts, sizeof(struct diff_options));
2151
2152 ecbdata->diff_words =
2153 xcalloc(1, sizeof(struct diff_words_data));
2154 ecbdata->diff_words->type = o->word_diff;
2155 ecbdata->diff_words->opt = o;
2156
2157 if (orig_opts->emitted_symbols)
2158 o->emitted_symbols =
2159 xcalloc(1, sizeof(struct emitted_diff_symbols));
2160
2161 if (!o->word_regex)
2162 o->word_regex = userdiff_word_regex(one);
2163 if (!o->word_regex)
2164 o->word_regex = userdiff_word_regex(two);
2165 if (!o->word_regex)
2166 o->word_regex = diff_word_regex_cfg;
2167 if (o->word_regex) {
2168 ecbdata->diff_words->word_regex = (regex_t *)
2169 xmalloc(sizeof(regex_t));
2170 if (regcomp(ecbdata->diff_words->word_regex,
2171 o->word_regex,
2172 REG_EXTENDED | REG_NEWLINE))
2173 die("invalid regular expression: %s",
2174 o->word_regex);
2175 }
2176 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2177 if (o->word_diff == diff_words_styles[i].type) {
2178 ecbdata->diff_words->style =
2179 &diff_words_styles[i];
2180 break;
2181 }
2182 }
2183 if (want_color(o->use_color)) {
2184 struct diff_words_style *st = ecbdata->diff_words->style;
2185 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2186 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2187 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2188 }
2189 }
2190
2191 static void free_diff_words_data(struct emit_callback *ecbdata)
2192 {
2193 if (ecbdata->diff_words) {
2194 diff_words_flush(ecbdata);
2195 free (ecbdata->diff_words->opt->emitted_symbols);
2196 free (ecbdata->diff_words->opt);
2197 free (ecbdata->diff_words->minus.text.ptr);
2198 free (ecbdata->diff_words->minus.orig);
2199 free (ecbdata->diff_words->plus.text.ptr);
2200 free (ecbdata->diff_words->plus.orig);
2201 if (ecbdata->diff_words->word_regex) {
2202 regfree(ecbdata->diff_words->word_regex);
2203 free(ecbdata->diff_words->word_regex);
2204 }
2205 FREE_AND_NULL(ecbdata->diff_words);
2206 }
2207 }
2208
2209 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2210 {
2211 if (want_color(diff_use_color))
2212 return diff_colors[ix];
2213 return "";
2214 }
2215
2216 const char *diff_line_prefix(struct diff_options *opt)
2217 {
2218 struct strbuf *msgbuf;
2219 if (!opt->output_prefix)
2220 return "";
2221
2222 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2223 return msgbuf->buf;
2224 }
2225
2226 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2227 {
2228 const char *cp;
2229 unsigned long allot;
2230 size_t l = len;
2231
2232 cp = line;
2233 allot = l;
2234 while (0 < l) {
2235 (void) utf8_width(&cp, &l);
2236 if (!cp)
2237 break; /* truncated in the middle? */
2238 }
2239 return allot - l;
2240 }
2241
2242 static void find_lno(const char *line, struct emit_callback *ecbdata)
2243 {
2244 const char *p;
2245 ecbdata->lno_in_preimage = 0;
2246 ecbdata->lno_in_postimage = 0;
2247 p = strchr(line, '-');
2248 if (!p)
2249 return; /* cannot happen */
2250 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2251 p = strchr(p, '+');
2252 if (!p)
2253 return; /* cannot happen */
2254 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2255 }
2256
2257 static void fn_out_consume(void *priv, char *line, unsigned long len)
2258 {
2259 struct emit_callback *ecbdata = priv;
2260 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2261 struct diff_options *o = ecbdata->opt;
2262
2263 o->found_changes = 1;
2264
2265 if (ecbdata->header) {
2266 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2267 ecbdata->header->buf, ecbdata->header->len, 0);
2268 strbuf_reset(ecbdata->header);
2269 ecbdata->header = NULL;
2270 }
2271
2272 if (ecbdata->label_path[0]) {
2273 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2274 ecbdata->label_path[0],
2275 strlen(ecbdata->label_path[0]), 0);
2276 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2277 ecbdata->label_path[1],
2278 strlen(ecbdata->label_path[1]), 0);
2279 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2280 }
2281
2282 if (diff_suppress_blank_empty
2283 && len == 2 && line[0] == ' ' && line[1] == '\n') {
2284 line[0] = '\n';
2285 len = 1;
2286 }
2287
2288 if (line[0] == '@') {
2289 if (ecbdata->diff_words)
2290 diff_words_flush(ecbdata);
2291 len = sane_truncate_line(ecbdata, line, len);
2292 find_lno(line, ecbdata);
2293 emit_hunk_header(ecbdata, line, len);
2294 return;
2295 }
2296
2297 if (ecbdata->diff_words) {
2298 enum diff_symbol s =
2299 ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2300 DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2301 if (line[0] == '-') {
2302 diff_words_append(line, len,
2303 &ecbdata->diff_words->minus);
2304 return;
2305 } else if (line[0] == '+') {
2306 diff_words_append(line, len,
2307 &ecbdata->diff_words->plus);
2308 return;
2309 } else if (starts_with(line, "\\ ")) {
2310 /*
2311 * Eat the "no newline at eof" marker as if we
2312 * saw a "+" or "-" line with nothing on it,
2313 * and return without diff_words_flush() to
2314 * defer processing. If this is the end of
2315 * preimage, more "+" lines may come after it.
2316 */
2317 return;
2318 }
2319 diff_words_flush(ecbdata);
2320 emit_diff_symbol(o, s, line, len, 0);
2321 return;
2322 }
2323
2324 switch (line[0]) {
2325 case '+':
2326 ecbdata->lno_in_postimage++;
2327 emit_add_line(reset, ecbdata, line + 1, len - 1);
2328 break;
2329 case '-':
2330 ecbdata->lno_in_preimage++;
2331 emit_del_line(reset, ecbdata, line + 1, len - 1);
2332 break;
2333 case ' ':
2334 ecbdata->lno_in_postimage++;
2335 ecbdata->lno_in_preimage++;
2336 emit_context_line(reset, ecbdata, line + 1, len - 1);
2337 break;
2338 default:
2339 /* incomplete line at the end */
2340 ecbdata->lno_in_preimage++;
2341 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2342 line, len, 0);
2343 break;
2344 }
2345 }
2346
2347 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2348 {
2349 const char *old_name = a;
2350 const char *new_name = b;
2351 int pfx_length, sfx_length;
2352 int pfx_adjust_for_slash;
2353 int len_a = strlen(a);
2354 int len_b = strlen(b);
2355 int a_midlen, b_midlen;
2356 int qlen_a = quote_c_style(a, NULL, NULL, 0);
2357 int qlen_b = quote_c_style(b, NULL, NULL, 0);
2358
2359 if (qlen_a || qlen_b) {
2360 quote_c_style(a, name, NULL, 0);
2361 strbuf_addstr(name, " => ");
2362 quote_c_style(b, name, NULL, 0);
2363 return;
2364 }
2365
2366 /* Find common prefix */
2367 pfx_length = 0;
2368 while (*old_name && *new_name && *old_name == *new_name) {
2369 if (*old_name == '/')
2370 pfx_length = old_name - a + 1;
2371 old_name++;
2372 new_name++;
2373 }
2374
2375 /* Find common suffix */
2376 old_name = a + len_a;
2377 new_name = b + len_b;
2378 sfx_length = 0;
2379 /*
2380 * If there is a common prefix, it must end in a slash. In
2381 * that case we let this loop run 1 into the prefix to see the
2382 * same slash.
2383 *
2384 * If there is no common prefix, we cannot do this as it would
2385 * underrun the input strings.
2386 */
2387 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2388 while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2389 b + pfx_length - pfx_adjust_for_slash <= new_name &&
2390 *old_name == *new_name) {
2391 if (*old_name == '/')
2392 sfx_length = len_a - (old_name - a);
2393 old_name--;
2394 new_name--;
2395 }
2396
2397 /*
2398 * pfx{mid-a => mid-b}sfx
2399 * {pfx-a => pfx-b}sfx
2400 * pfx{sfx-a => sfx-b}
2401 * name-a => name-b
2402 */
2403 a_midlen = len_a - pfx_length - sfx_length;
2404 b_midlen = len_b - pfx_length - sfx_length;
2405 if (a_midlen < 0)
2406 a_midlen = 0;
2407 if (b_midlen < 0)
2408 b_midlen = 0;
2409
2410 strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2411 if (pfx_length + sfx_length) {
2412 strbuf_add(name, a, pfx_length);
2413 strbuf_addch(name, '{');
2414 }
2415 strbuf_add(name, a + pfx_length, a_midlen);
2416 strbuf_addstr(name, " => ");
2417 strbuf_add(name, b + pfx_length, b_midlen);
2418 if (pfx_length + sfx_length) {
2419 strbuf_addch(name, '}');
2420 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2421 }
2422 }
2423
2424 struct diffstat_t {
2425 int nr;
2426 int alloc;
2427 struct diffstat_file {
2428 char *from_name;
2429 char *name;
2430 char *print_name;
2431 const char *comments;
2432 unsigned is_unmerged:1;
2433 unsigned is_binary:1;
2434 unsigned is_renamed:1;
2435 unsigned is_interesting:1;
2436 uintmax_t added, deleted;
2437 } **files;
2438 };
2439
2440 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2441 const char *name_a,
2442 const char *name_b)
2443 {
2444 struct diffstat_file *x;
2445 x = xcalloc(1, sizeof(*x));
2446 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2447 diffstat->files[diffstat->nr++] = x;
2448 if (name_b) {
2449 x->from_name = xstrdup(name_a);
2450 x->name = xstrdup(name_b);
2451 x->is_renamed = 1;
2452 }
2453 else {
2454 x->from_name = NULL;
2455 x->name = xstrdup(name_a);
2456 }
2457 return x;
2458 }
2459
2460 static void diffstat_consume(void *priv, char *line, unsigned long len)
2461 {
2462 struct diffstat_t *diffstat = priv;
2463 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2464
2465 if (line[0] == '+')
2466 x->added++;
2467 else if (line[0] == '-')
2468 x->deleted++;
2469 }
2470
2471 const char mime_boundary_leader[] = "------------";
2472
2473 static int scale_linear(int it, int width, int max_change)
2474 {
2475 if (!it)
2476 return 0;
2477 /*
2478 * make sure that at least one '-' or '+' is printed if
2479 * there is any change to this path. The easiest way is to
2480 * scale linearly as if the alloted width is one column shorter
2481 * than it is, and then add 1 to the result.
2482 */
2483 return 1 + (it * (width - 1) / max_change);
2484 }
2485
2486 static void show_graph(struct strbuf *out, char ch, int cnt,
2487 const char *set, const char *reset)
2488 {
2489 if (cnt <= 0)
2490 return;
2491 strbuf_addstr(out, set);
2492 strbuf_addchars(out, ch, cnt);
2493 strbuf_addstr(out, reset);
2494 }
2495
2496 static void fill_print_name(struct diffstat_file *file)
2497 {
2498 struct strbuf pname = STRBUF_INIT;
2499
2500 if (file->print_name)
2501 return;
2502
2503 if (file->is_renamed)
2504 pprint_rename(&pname, file->from_name, file->name);
2505 else
2506 quote_c_style(file->name, &pname, NULL, 0);
2507
2508 if (file->comments)
2509 strbuf_addf(&pname, " (%s)", file->comments);
2510
2511 file->print_name = strbuf_detach(&pname, NULL);
2512 }
2513
2514 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2515 int files, int insertions, int deletions)
2516 {
2517 struct strbuf sb = STRBUF_INIT;
2518
2519 if (!files) {
2520 assert(insertions == 0 && deletions == 0);
2521 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2522 NULL, 0, 0);
2523 return;
2524 }
2525
2526 strbuf_addf(&sb,
2527 (files == 1) ? " %d file changed" : " %d files changed",
2528 files);
2529
2530 /*
2531 * For binary diff, the caller may want to print "x files
2532 * changed" with insertions == 0 && deletions == 0.
2533 *
2534 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2535 * is probably less confusing (i.e skip over "2 files changed
2536 * but nothing about added/removed lines? Is this a bug in Git?").
2537 */
2538 if (insertions || deletions == 0) {
2539 strbuf_addf(&sb,
2540 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2541 insertions);
2542 }
2543
2544 if (deletions || insertions == 0) {
2545 strbuf_addf(&sb,
2546 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2547 deletions);
2548 }
2549 strbuf_addch(&sb, '\n');
2550 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2551 sb.buf, sb.len, 0);
2552 strbuf_release(&sb);
2553 }
2554
2555 void print_stat_summary(FILE *fp, int files,
2556 int insertions, int deletions)
2557 {
2558 struct diff_options o;
2559 memset(&o, 0, sizeof(o));
2560 o.file = fp;
2561
2562 print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2563 }
2564
2565 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2566 {
2567 int i, len, add, del, adds = 0, dels = 0;
2568 uintmax_t max_change = 0, max_len = 0;
2569 int total_files = data->nr, count;
2570 int width, name_width, graph_width, number_width = 0, bin_width = 0;
2571 const char *reset, *add_c, *del_c;
2572 int extra_shown = 0;
2573 const char *line_prefix = diff_line_prefix(options);
2574 struct strbuf out = STRBUF_INIT;
2575
2576 if (data->nr == 0)
2577 return;
2578
2579 count = options->stat_count ? options->stat_count : data->nr;
2580
2581 reset = diff_get_color_opt(options, DIFF_RESET);
2582 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2583 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2584
2585 /*
2586 * Find the longest filename and max number of changes
2587 */
2588 for (i = 0; (i < count) && (i < data->nr); i++) {
2589 struct diffstat_file *file = data->files[i];
2590 uintmax_t change = file->added + file->deleted;
2591
2592 if (!file->is_interesting && (change == 0)) {
2593 count++; /* not shown == room for one more */
2594 continue;
2595 }
2596 fill_print_name(file);
2597 len = strlen(file->print_name);
2598 if (max_len < len)
2599 max_len = len;
2600
2601 if (file->is_unmerged) {
2602 /* "Unmerged" is 8 characters */
2603 bin_width = bin_width < 8 ? 8 : bin_width;
2604 continue;
2605 }
2606 if (file->is_binary) {
2607 /* "Bin XXX -> YYY bytes" */
2608 int w = 14 + decimal_width(file->added)
2609 + decimal_width(file->deleted);
2610 bin_width = bin_width < w ? w : bin_width;
2611 /* Display change counts aligned with "Bin" */
2612 number_width = 3;
2613 continue;
2614 }
2615
2616 if (max_change < change)
2617 max_change = change;
2618 }
2619 count = i; /* where we can stop scanning in data->files[] */
2620
2621 /*
2622 * We have width = stat_width or term_columns() columns total.
2623 * We want a maximum of min(max_len, stat_name_width) for the name part.
2624 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2625 * We also need 1 for " " and 4 + decimal_width(max_change)
2626 * for " | NNNN " and one the empty column at the end, altogether
2627 * 6 + decimal_width(max_change).
2628 *
2629 * If there's not enough space, we will use the smaller of
2630 * stat_name_width (if set) and 5/8*width for the filename,
2631 * and the rest for constant elements + graph part, but no more
2632 * than stat_graph_width for the graph part.
2633 * (5/8 gives 50 for filename and 30 for the constant parts + graph
2634 * for the standard terminal size).
2635 *
2636 * In other words: stat_width limits the maximum width, and
2637 * stat_name_width fixes the maximum width of the filename,
2638 * and is also used to divide available columns if there
2639 * aren't enough.
2640 *
2641 * Binary files are displayed with "Bin XXX -> YYY bytes"
2642 * instead of the change count and graph. This part is treated
2643 * similarly to the graph part, except that it is not
2644 * "scaled". If total width is too small to accommodate the
2645 * guaranteed minimum width of the filename part and the
2646 * separators and this message, this message will "overflow"
2647 * making the line longer than the maximum width.
2648 */
2649
2650 if (options->stat_width == -1)
2651 width = term_columns() - strlen(line_prefix);
2652 else
2653 width = options->stat_width ? options->stat_width : 80;
2654 number_width = decimal_width(max_change) > number_width ?
2655 decimal_width(max_change) : number_width;
2656
2657 if (options->stat_graph_width == -1)
2658 options->stat_graph_width = diff_stat_graph_width;
2659
2660 /*
2661 * Guarantee 3/8*16==6 for the graph part
2662 * and 5/8*16==10 for the filename part
2663 */
2664 if (width < 16 + 6 + number_width)
2665 width = 16 + 6 + number_width;
2666
2667 /*
2668 * First assign sizes that are wanted, ignoring available width.
2669 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2670 * starting from "XXX" should fit in graph_width.
2671 */
2672 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2673 if (options->stat_graph_width &&
2674 options->stat_graph_width < graph_width)
2675 graph_width = options->stat_graph_width;
2676
2677 name_width = (options->stat_name_width > 0 &&
2678 options->stat_name_width < max_len) ?
2679 options->stat_name_width : max_len;
2680
2681 /*
2682 * Adjust adjustable widths not to exceed maximum width
2683 */
2684 if (name_width + number_width + 6 + graph_width > width) {
2685 if (graph_width > width * 3/8 - number_width - 6) {
2686 graph_width = width * 3/8 - number_width - 6;
2687 if (graph_width < 6)
2688 graph_width = 6;
2689 }
2690
2691 if (options->stat_graph_width &&
2692 graph_width > options->stat_graph_width)
2693 graph_width = options->stat_graph_width;
2694 if (name_width > width - number_width - 6 - graph_width)
2695 name_width = width - number_width - 6 - graph_width;
2696 else
2697 graph_width = width - number_width - 6 - name_width;
2698 }
2699
2700 /*
2701 * From here name_width is the width of the name area,
2702 * and graph_width is the width of the graph area.
2703 * max_change is used to scale graph properly.
2704 */
2705 for (i = 0; i < count; i++) {
2706 const char *prefix = "";
2707 struct diffstat_file *file = data->files[i];
2708 char *name = file->print_name;
2709 uintmax_t added = file->added;
2710 uintmax_t deleted = file->deleted;
2711 int name_len;
2712
2713 if (!file->is_interesting && (added + deleted == 0))
2714 continue;
2715
2716 /*
2717 * "scale" the filename
2718 */
2719 len = name_width;
2720 name_len = strlen(name);
2721 if (name_width < name_len) {
2722 char *slash;
2723 prefix = "...";
2724 len -= 3;
2725 name += name_len - len;
2726 slash = strchr(name, '/');
2727 if (slash)
2728 name = slash;
2729 }
2730
2731 if (file->is_binary) {
2732 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2733 strbuf_addf(&out, " %*s", number_width, "Bin");
2734 if (!added && !deleted) {
2735 strbuf_addch(&out, '\n');
2736 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2737 out.buf, out.len, 0);
2738 strbuf_reset(&out);
2739 continue;
2740 }
2741 strbuf_addf(&out, " %s%"PRIuMAX"%s",
2742 del_c, deleted, reset);
2743 strbuf_addstr(&out, " -> ");
2744 strbuf_addf(&out, "%s%"PRIuMAX"%s",
2745 add_c, added, reset);
2746 strbuf_addstr(&out, " bytes\n");
2747 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2748 out.buf, out.len, 0);
2749 strbuf_reset(&out);
2750 continue;
2751 }
2752 else if (file->is_unmerged) {
2753 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2754 strbuf_addstr(&out, " Unmerged\n");
2755 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2756 out.buf, out.len, 0);
2757 strbuf_reset(&out);
2758 continue;
2759 }
2760
2761 /*
2762 * scale the add/delete
2763 */
2764 add = added;
2765 del = deleted;
2766
2767 if (graph_width <= max_change) {
2768 int total = scale_linear(add + del, graph_width, max_change);
2769 if (total < 2 && add && del)
2770 /* width >= 2 due to the sanity check */
2771 total = 2;
2772 if (add < del) {
2773 add = scale_linear(add, graph_width, max_change);
2774 del = total - add;
2775 } else {
2776 del = scale_linear(del, graph_width, max_change);
2777 add = total - del;
2778 }
2779 }
2780 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2781 strbuf_addf(&out, " %*"PRIuMAX"%s",
2782 number_width, added + deleted,
2783 added + deleted ? " " : "");
2784 show_graph(&out, '+', add, add_c, reset);
2785 show_graph(&out, '-', del, del_c, reset);
2786 strbuf_addch(&out, '\n');
2787 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2788 out.buf, out.len, 0);
2789 strbuf_reset(&out);
2790 }
2791
2792 for (i = 0; i < data->nr; i++) {
2793 struct diffstat_file *file = data->files[i];
2794 uintmax_t added = file->added;
2795 uintmax_t deleted = file->deleted;
2796
2797 if (file->is_unmerged ||
2798 (!file->is_interesting && (added + deleted == 0))) {
2799 total_files--;
2800 continue;
2801 }
2802
2803 if (!file->is_binary) {
2804 adds += added;
2805 dels += deleted;
2806 }
2807 if (i < count)
2808 continue;
2809 if (!extra_shown)
2810 emit_diff_symbol(options,
2811 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2812 NULL, 0, 0);
2813 extra_shown = 1;
2814 }
2815
2816 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2817 strbuf_release(&out);
2818 }
2819
2820 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2821 {
2822 int i, adds = 0, dels = 0, total_files = data->nr;
2823
2824 if (data->nr == 0)
2825 return;
2826
2827 for (i = 0; i < data->nr; i++) {
2828 int added = data->files[i]->added;
2829 int deleted = data->files[i]->deleted;
2830
2831 if (data->files[i]->is_unmerged ||
2832 (!data->files[i]->is_interesting && (added + deleted == 0))) {
2833 total_files--;
2834 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2835 adds += added;
2836 dels += deleted;
2837 }
2838 }
2839 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2840 }
2841
2842 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2843 {
2844 int i;
2845
2846 if (data->nr == 0)
2847 return;
2848
2849 for (i = 0; i < data->nr; i++) {
2850 struct diffstat_file *file = data->files[i];
2851
2852 fprintf(options->file, "%s", diff_line_prefix(options));
2853
2854 if (file->is_binary)
2855 fprintf(options->file, "-\t-\t");
2856 else
2857 fprintf(options->file,
2858 "%"PRIuMAX"\t%"PRIuMAX"\t",
2859 file->added, file->deleted);
2860 if (options->line_termination) {
2861 fill_print_name(file);
2862 if (!file->is_renamed)
2863 write_name_quoted(file->name, options->file,
2864 options->line_termination);
2865 else {
2866 fputs(file->print_name, options->file);
2867 putc(options->line_termination, options->file);
2868 }
2869 } else {
2870 if (file->is_renamed) {
2871 putc('\0', options->file);
2872 write_name_quoted(file->from_name, options->file, '\0');
2873 }
2874 write_name_quoted(file->name, options->file, '\0');
2875 }
2876 }
2877 }
2878
2879 struct dirstat_file {
2880 const char *name;
2881 unsigned long changed;
2882 };
2883
2884 struct dirstat_dir {
2885 struct dirstat_file *files;
2886 int alloc, nr, permille, cumulative;
2887 };
2888
2889 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2890 unsigned long changed, const char *base, int baselen)
2891 {
2892 unsigned long sum_changes = 0;
2893 unsigned int sources = 0;
2894 const char *line_prefix = diff_line_prefix(opt);
2895
2896 while (dir->nr) {
2897 struct dirstat_file *f = dir->files;
2898 int namelen = strlen(f->name);
2899 unsigned long changes;
2900 char *slash;
2901
2902 if (namelen < baselen)
2903 break;
2904 if (memcmp(f->name, base, baselen))
2905 break;
2906 slash = strchr(f->name + baselen, '/');
2907 if (slash) {
2908 int newbaselen = slash + 1 - f->name;
2909 changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2910 sources++;
2911 } else {
2912 changes = f->changed;
2913 dir->files++;
2914 dir->nr--;
2915 sources += 2;
2916 }
2917 sum_changes += changes;
2918 }
2919
2920 /*
2921 * We don't report dirstat's for
2922 * - the top level
2923 * - or cases where everything came from a single directory
2924 * under this directory (sources == 1).
2925 */
2926 if (baselen && sources != 1) {
2927 if (sum_changes) {
2928 int permille = sum_changes * 1000 / changed;
2929 if (permille >= dir->permille) {
2930 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2931 permille / 10, permille % 10, baselen, base);
2932 if (!dir->cumulative)
2933 return 0;
2934 }
2935 }
2936 }
2937 return sum_changes;
2938 }
2939
2940 static int dirstat_compare(const void *_a, const void *_b)
2941 {
2942 const struct dirstat_file *a = _a;
2943 const struct dirstat_file *b = _b;
2944 return strcmp(a->name, b->name);
2945 }
2946
2947 static void show_dirstat(struct diff_options *options)
2948 {
2949 int i;
2950 unsigned long changed;
2951 struct dirstat_dir dir;
2952 struct diff_queue_struct *q = &diff_queued_diff;
2953
2954 dir.files = NULL;
2955 dir.alloc = 0;
2956 dir.nr = 0;
2957 dir.permille = options->dirstat_permille;
2958 dir.cumulative = options->flags.dirstat_cumulative;
2959
2960 changed = 0;
2961 for (i = 0; i < q->nr; i++) {
2962 struct diff_filepair *p = q->queue[i];
2963 const char *name;
2964 unsigned long copied, added, damage;
2965
2966 name = p->two->path ? p->two->path : p->one->path;
2967
2968 if (p->one->oid_valid && p->two->oid_valid &&
2969 oideq(&p->one->oid, &p->two->oid)) {
2970 /*
2971 * The SHA1 has not changed, so pre-/post-content is
2972 * identical. We can therefore skip looking at the
2973 * file contents altogether.
2974 */
2975 damage = 0;
2976 goto found_damage;
2977 }
2978
2979 if (options->flags.dirstat_by_file) {
2980 /*
2981 * In --dirstat-by-file mode, we don't really need to
2982 * look at the actual file contents at all.
2983 * The fact that the SHA1 changed is enough for us to
2984 * add this file to the list of results
2985 * (with each file contributing equal damage).
2986 */
2987 damage = 1;
2988 goto found_damage;
2989 }
2990
2991 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2992 diff_populate_filespec(p->one, 0);
2993 diff_populate_filespec(p->two, 0);
2994 diffcore_count_changes(p->one, p->two, NULL, NULL,
2995 &copied, &added);
2996 diff_free_filespec_data(p->one);
2997 diff_free_filespec_data(p->two);
2998 } else if (DIFF_FILE_VALID(p->one)) {
2999 diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
3000 copied = added = 0;
3001 diff_free_filespec_data(p->one);
3002 } else if (DIFF_FILE_VALID(p->two)) {
3003 diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
3004 copied = 0;
3005 added = p->two->size;
3006 diff_free_filespec_data(p->two);
3007 } else
3008 continue;
3009
3010 /*
3011 * Original minus copied is the removed material,
3012 * added is the new material. They are both damages
3013 * made to the preimage.
3014 * If the resulting damage is zero, we know that
3015 * diffcore_count_changes() considers the two entries to
3016 * be identical, but since the oid changed, we
3017 * know that there must have been _some_ kind of change,
3018 * so we force all entries to have damage > 0.
3019 */
3020 damage = (p->one->size - copied) + added;
3021 if (!damage)
3022 damage = 1;
3023
3024 found_damage:
3025 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3026 dir.files[dir.nr].name = name;
3027 dir.files[dir.nr].changed = damage;
3028 changed += damage;
3029 dir.nr++;
3030 }
3031
3032 /* This can happen even with many files, if everything was renames */
3033 if (!changed)
3034 return;
3035
3036 /* Show all directories with more than x% of the changes */
3037 QSORT(dir.files, dir.nr, dirstat_compare);
3038 gather_dirstat(options, &dir, changed, "", 0);
3039 }
3040
3041 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3042 {
3043 int i;
3044 unsigned long changed;
3045 struct dirstat_dir dir;
3046
3047 if (data->nr == 0)
3048 return;
3049
3050 dir.files = NULL;
3051 dir.alloc = 0;
3052 dir.nr = 0;
3053 dir.permille = options->dirstat_permille;
3054 dir.cumulative = options->flags.dirstat_cumulative;
3055
3056 changed = 0;
3057 for (i = 0; i < data->nr; i++) {
3058 struct diffstat_file *file = data->files[i];
3059 unsigned long damage = file->added + file->deleted;
3060 if (file->is_binary)
3061 /*
3062 * binary files counts bytes, not lines. Must find some
3063 * way to normalize binary bytes vs. textual lines.
3064 * The following heuristic assumes that there are 64
3065 * bytes per "line".
3066 * This is stupid and ugly, but very cheap...
3067 */
3068 damage = DIV_ROUND_UP(damage, 64);
3069 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3070 dir.files[dir.nr].name = file->name;
3071 dir.files[dir.nr].changed = damage;
3072 changed += damage;
3073 dir.nr++;
3074 }
3075
3076 /* This can happen even with many files, if everything was renames */
3077 if (!changed)
3078 return;
3079
3080 /* Show all directories with more than x% of the changes */
3081 QSORT(dir.files, dir.nr, dirstat_compare);
3082 gather_dirstat(options, &dir, changed, "", 0);
3083 }
3084
3085 static void free_diffstat_info(struct diffstat_t *diffstat)
3086 {
3087 int i;
3088 for (i = 0; i < diffstat->nr; i++) {
3089 struct diffstat_file *f = diffstat->files[i];
3090 free(f->print_name);
3091 free(f->name);
3092 free(f->from_name);
3093 free(f);
3094 }
3095 free(diffstat->files);
3096 }
3097
3098 struct checkdiff_t {
3099 const char *filename;
3100 int lineno;
3101 int conflict_marker_size;
3102 struct diff_options *o;
3103 unsigned ws_rule;
3104 unsigned status;
3105 };
3106
3107 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3108 {
3109 char firstchar;
3110 int cnt;
3111
3112 if (len < marker_size + 1)
3113 return 0;
3114 firstchar = line[0];
3115 switch (firstchar) {
3116 case '=': case '>': case '<': case '|':
3117 break;
3118 default:
3119 return 0;
3120 }
3121 for (cnt = 1; cnt < marker_size; cnt++)
3122 if (line[cnt] != firstchar)
3123 return 0;
3124 /* line[1] thru line[marker_size-1] are same as firstchar */
3125 if (len < marker_size + 1 || !isspace(line[marker_size]))
3126 return 0;
3127 return 1;
3128 }
3129
3130 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3131 {
3132 struct checkdiff_t *data = priv;
3133 int marker_size = data->conflict_marker_size;
3134 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3135 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3136 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3137 char *err;
3138 const char *line_prefix;
3139
3140 assert(data->o);
3141 line_prefix = diff_line_prefix(data->o);
3142
3143 if (line[0] == '+') {
3144 unsigned bad;
3145 data->lineno++;
3146 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3147 data->status |= 1;
3148 fprintf(data->o->file,
3149 "%s%s:%d: leftover conflict marker\n",
3150 line_prefix, data->filename, data->lineno);
3151 }
3152 bad = ws_check(line + 1, len - 1, data->ws_rule);
3153 if (!bad)
3154 return;
3155 data->status |= bad;
3156 err = whitespace_error_string(bad);
3157 fprintf(data->o->file, "%s%s:%d: %s.\n",
3158 line_prefix, data->filename, data->lineno, err);
3159 free(err);
3160 emit_line(data->o, set, reset, line, 1);
3161 ws_check_emit(line + 1, len - 1, data->ws_rule,
3162 data->o->file, set, reset, ws);
3163 } else if (line[0] == ' ') {
3164 data->lineno++;
3165 } else if (line[0] == '@') {
3166 char *plus = strchr(line, '+');
3167 if (plus)
3168 data->lineno = strtol(plus, NULL, 10) - 1;
3169 else
3170 die("invalid diff");
3171 }
3172 }
3173
3174 static unsigned char *deflate_it(char *data,
3175 unsigned long size,
3176 unsigned long *result_size)
3177 {
3178 int bound;
3179 unsigned char *deflated;
3180 git_zstream stream;
3181
3182 git_deflate_init(&stream, zlib_compression_level);
3183 bound = git_deflate_bound(&stream, size);
3184 deflated = xmalloc(bound);
3185 stream.next_out = deflated;
3186 stream.avail_out = bound;
3187
3188 stream.next_in = (unsigned char *)data;
3189 stream.avail_in = size;
3190 while (git_deflate(&stream, Z_FINISH) == Z_OK)
3191 ; /* nothing */
3192 git_deflate_end(&stream);
3193 *result_size = stream.total_out;
3194 return deflated;
3195 }
3196
3197 static void emit_binary_diff_body(struct diff_options *o,
3198 mmfile_t *one, mmfile_t *two)
3199 {
3200 void *cp;
3201 void *delta;
3202 void *deflated;
3203 void *data;
3204 unsigned long orig_size;
3205 unsigned long delta_size;
3206 unsigned long deflate_size;
3207 unsigned long data_size;
3208
3209 /* We could do deflated delta, or we could do just deflated two,
3210 * whichever is smaller.
3211 */
3212 delta = NULL;
3213 deflated = deflate_it(two->ptr, two->size, &deflate_size);
3214 if (one->size && two->size) {
3215 delta = diff_delta(one->ptr, one->size,
3216 two->ptr, two->size,
3217 &delta_size, deflate_size);
3218 if (delta) {
3219 void *to_free = delta;
3220 orig_size = delta_size;
3221 delta = deflate_it(delta, delta_size, &delta_size);
3222 free(to_free);
3223 }
3224 }
3225
3226 if (delta && delta_size < deflate_size) {
3227 char *s = xstrfmt("%lu", orig_size);
3228 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3229 s, strlen(s), 0);
3230 free(s);
3231 free(deflated);
3232 data = delta;
3233 data_size = delta_size;
3234 } else {
3235 char *s = xstrfmt("%lu", two->size);
3236 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3237 s, strlen(s), 0);
3238 free(s);
3239 free(delta);
3240 data = deflated;
3241 data_size = deflate_size;
3242 }
3243
3244 /* emit data encoded in base85 */
3245 cp = data;
3246 while (data_size) {
3247 int len;
3248 int bytes = (52 < data_size) ? 52 : data_size;
3249 char line[71];
3250 data_size -= bytes;
3251 if (bytes <= 26)
3252 line[0] = bytes + 'A' - 1;
3253 else
3254 line[0] = bytes - 26 + 'a' - 1;
3255 encode_85(line + 1, cp, bytes);
3256 cp = (char *) cp + bytes;
3257
3258 len = strlen(line);
3259 line[len++] = '\n';
3260 line[len] = '\0';
3261
3262 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3263 line, len, 0);
3264 }
3265 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3266 free(data);
3267 }
3268
3269 static void emit_binary_diff(struct diff_options *o,
3270 mmfile_t *one, mmfile_t *two)
3271 {
3272 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3273 emit_binary_diff_body(o, one, two);
3274 emit_binary_diff_body(o, two, one);
3275 }
3276
3277 int diff_filespec_is_binary(struct diff_filespec *one)
3278 {
3279 if (one->is_binary == -1) {
3280 diff_filespec_load_driver(one);
3281 if (one->driver->binary != -1)
3282 one->is_binary = one->driver->binary;
3283 else {
3284 if (!one->data && DIFF_FILE_VALID(one))
3285 diff_populate_filespec(one, CHECK_BINARY);
3286 if (one->is_binary == -1 && one->data)
3287 one->is_binary = buffer_is_binary(one->data,
3288 one->size);
3289 if (one->is_binary == -1)
3290 one->is_binary = 0;
3291 }
3292 }
3293 return one->is_binary;
3294 }
3295
3296 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3297 {
3298 diff_filespec_load_driver(one);
3299 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3300 }
3301
3302 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3303 {
3304 if (!options->a_prefix)
3305 options->a_prefix = a;
3306 if (!options->b_prefix)
3307 options->b_prefix = b;
3308 }
3309
3310 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3311 {
3312 if (!DIFF_FILE_VALID(one))
3313 return NULL;
3314
3315 diff_filespec_load_driver(one);
3316 return userdiff_get_textconv(one->driver);
3317 }
3318
3319 static void builtin_diff(const char *name_a,
3320 const char *name_b,
3321 struct diff_filespec *one,
3322 struct diff_filespec *two,
3323 const char *xfrm_msg,
3324 int must_show_header,
3325 struct diff_options *o,
3326 int complete_rewrite)
3327 {
3328 mmfile_t mf1, mf2;
3329 const char *lbl[2];
3330 char *a_one, *b_two;
3331 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3332 const char *reset = diff_get_color_opt(o, DIFF_RESET);
3333 const char *a_prefix, *b_prefix;
3334 struct userdiff_driver *textconv_one = NULL;
3335 struct userdiff_driver *textconv_two = NULL;
3336 struct strbuf header = STRBUF_INIT;
3337 const char *line_prefix = diff_line_prefix(o);
3338
3339 diff_set_mnemonic_prefix(o, "a/", "b/");
3340 if (o->flags.reverse_diff) {
3341 a_prefix = o->b_prefix;
3342 b_prefix = o->a_prefix;
3343 } else {
3344 a_prefix = o->a_prefix;
3345 b_prefix = o->b_prefix;
3346 }
3347
3348 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3349 (!one->mode || S_ISGITLINK(one->mode)) &&
3350 (!two->mode || S_ISGITLINK(two->mode))) {
3351 show_submodule_summary(o, one->path ? one->path : two->path,
3352 &one->oid, &two->oid,
3353 two->dirty_submodule);
3354 return;
3355 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3356 (!one->mode || S_ISGITLINK(one->mode)) &&
3357 (!two->mode || S_ISGITLINK(two->mode))) {
3358 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3359 &one->oid, &two->oid,
3360 two->dirty_submodule);
3361 return;
3362 }
3363
3364 if (o->flags.allow_textconv) {
3365 textconv_one = get_textconv(one);
3366 textconv_two = get_textconv(two);
3367 }
3368
3369 /* Never use a non-valid filename anywhere if at all possible */
3370 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3371 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3372
3373 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3374 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3375 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3376 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3377 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3378 if (lbl[0][0] == '/') {
3379 /* /dev/null */
3380 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3381 if (xfrm_msg)
3382 strbuf_addstr(&header, xfrm_msg);
3383 must_show_header = 1;
3384 }
3385 else if (lbl[1][0] == '/') {
3386 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3387 if (xfrm_msg)
3388 strbuf_addstr(&header, xfrm_msg);
3389 must_show_header = 1;
3390 }
3391 else {
3392 if (one->mode != two->mode) {
3393 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3394 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3395 must_show_header = 1;
3396 }
3397 if (xfrm_msg)
3398 strbuf_addstr(&header, xfrm_msg);
3399
3400 /*
3401 * we do not run diff between different kind
3402 * of objects.
3403 */
3404 if ((one->mode ^ two->mode) & S_IFMT)
3405 goto free_ab_and_return;
3406 if (complete_rewrite &&
3407 (textconv_one || !diff_filespec_is_binary(one)) &&
3408 (textconv_two || !diff_filespec_is_binary(two))) {
3409 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3410 header.buf, header.len, 0);
3411 strbuf_reset(&header);
3412 emit_rewrite_diff(name_a, name_b, one, two,
3413 textconv_one, textconv_two, o);
3414 o->found_changes = 1;
3415 goto free_ab_and_return;
3416 }
3417 }
3418
3419 if (o->irreversible_delete && lbl[1][0] == '/') {
3420 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3421 header.len, 0);
3422 strbuf_reset(&header);
3423 goto free_ab_and_return;
3424 } else if (!o->flags.text &&
3425 ( (!textconv_one && diff_filespec_is_binary(one)) ||
3426 (!textconv_two && diff_filespec_is_binary(two)) )) {
3427 struct strbuf sb = STRBUF_INIT;
3428 if (!one->data && !two->data &&
3429 S_ISREG(one->mode) && S_ISREG(two->mode) &&
3430 !o->flags.binary) {
3431 if (oideq(&one->oid, &two->oid)) {
3432 if (must_show_header)
3433 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3434 header.buf, header.len,
3435 0);
3436 goto free_ab_and_return;
3437 }
3438 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3439 header.buf, header.len, 0);
3440 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3441 diff_line_prefix(o), lbl[0], lbl[1]);
3442 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3443 sb.buf, sb.len, 0);
3444 strbuf_release(&sb);
3445 goto free_ab_and_return;
3446 }
3447 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3448 die("unable to read files to diff");
3449 /* Quite common confusing case */
3450 if (mf1.size == mf2.size &&
3451 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3452 if (must_show_header)
3453 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3454 header.buf, header.len, 0);
3455 goto free_ab_and_return;
3456 }
3457 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3458 strbuf_reset(&header);
3459 if (o->flags.binary)
3460 emit_binary_diff(o, &mf1, &mf2);
3461 else {
3462 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3463 diff_line_prefix(o), lbl[0], lbl[1]);
3464 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3465 sb.buf, sb.len, 0);
3466 strbuf_release(&sb);
3467 }
3468 o->found_changes = 1;
3469 } else {
3470 /* Crazy xdl interfaces.. */
3471 const char *diffopts = getenv("GIT_DIFF_OPTS");
3472 const char *v;
3473 xpparam_t xpp;
3474 xdemitconf_t xecfg;
3475 struct emit_callback ecbdata;
3476 const struct userdiff_funcname *pe;
3477
3478 if (must_show_header) {
3479 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3480 header.buf, header.len, 0);
3481 strbuf_reset(&header);
3482 }
3483
3484 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3485 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3486
3487 pe = diff_funcname_pattern(one);
3488 if (!pe)
3489 pe = diff_funcname_pattern(two);
3490
3491 memset(&xpp, 0, sizeof(xpp));
3492 memset(&xecfg, 0, sizeof(xecfg));
3493 memset(&ecbdata, 0, sizeof(ecbdata));
3494 if (o->flags.suppress_diff_headers)
3495 lbl[0] = NULL;
3496 ecbdata.label_path = lbl;
3497 ecbdata.color_diff = want_color(o->use_color);
3498 ecbdata.ws_rule = whitespace_rule(name_b);
3499 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3500 check_blank_at_eof(&mf1, &mf2, &ecbdata);
3501 ecbdata.opt = o;
3502 if (header.len && !o->flags.suppress_diff_headers)
3503 ecbdata.header = &header;
3504 xpp.flags = o->xdl_opts;
3505 xpp.anchors = o->anchors;
3506 xpp.anchors_nr = o->anchors_nr;
3507 xecfg.ctxlen = o->context;
3508 xecfg.interhunkctxlen = o->interhunkcontext;
3509 xecfg.flags = XDL_EMIT_FUNCNAMES;
3510 if (o->flags.funccontext)
3511 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3512 if (pe)
3513 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3514 if (!diffopts)
3515 ;
3516 else if (skip_prefix(diffopts, "--unified=", &v))
3517 xecfg.ctxlen = strtoul(v, NULL, 10);
3518 else if (skip_prefix(diffopts, "-u", &v))
3519 xecfg.ctxlen = strtoul(v, NULL, 10);
3520 if (o->word_diff)
3521 init_diff_words_data(&ecbdata, o, one, two);
3522 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3523 &xpp, &xecfg))
3524 die("unable to generate diff for %s", one->path);
3525 if (o->word_diff)
3526 free_diff_words_data(&ecbdata);
3527 if (textconv_one)
3528 free(mf1.ptr);
3529 if (textconv_two)
3530 free(mf2.ptr);
3531 xdiff_clear_find_func(&xecfg);
3532 }
3533
3534 free_ab_and_return:
3535 strbuf_release(&header);
3536 diff_free_filespec_data(one);
3537 diff_free_filespec_data(two);
3538 free(a_one);
3539 free(b_two);
3540 return;
3541 }
3542
3543 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3544 {
3545 if (!is_renamed) {
3546 if (p->status == DIFF_STATUS_ADDED) {
3547 if (S_ISLNK(p->two->mode))
3548 return "new +l";
3549 else if ((p->two->mode & 0777) == 0755)
3550 return "new +x";
3551 else
3552 return "new";
3553 } else if (p->status == DIFF_STATUS_DELETED)
3554 return "gone";
3555 }
3556 if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3557 return "mode -l";
3558 else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3559 return "mode +l";
3560 else if ((p->one->mode & 0777) == 0644 &&
3561 (p->two->mode & 0777) == 0755)
3562 return "mode +x";
3563 else if ((p->one->mode & 0777) == 0755 &&
3564 (p->two->mode & 0777) == 0644)
3565 return "mode -x";
3566 return NULL;
3567 }
3568
3569 static void builtin_diffstat(const char *name_a, const char *name_b,
3570 struct diff_filespec *one,
3571 struct diff_filespec *two,
3572 struct diffstat_t *diffstat,
3573 struct diff_options *o,
3574 struct diff_filepair *p)
3575 {
3576 mmfile_t mf1, mf2;
3577 struct diffstat_file *data;
3578 int same_contents;
3579 int complete_rewrite = 0;
3580
3581 if (!DIFF_PAIR_UNMERGED(p)) {
3582 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3583 complete_rewrite = 1;
3584 }
3585
3586 data = diffstat_add(diffstat, name_a, name_b);
3587 data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3588 if (o->flags.stat_with_summary)
3589 data->comments = get_compact_summary(p, data->is_renamed);
3590
3591 if (!one || !two) {
3592 data->is_unmerged = 1;
3593 return;
3594 }
3595
3596 same_contents = oideq(&one->oid, &two->oid);
3597
3598 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3599 data->is_binary = 1;
3600 if (same_contents) {
3601 data->added = 0;
3602 data->deleted = 0;
3603 } else {
3604 data->added = diff_filespec_size(two);
3605 data->deleted = diff_filespec_size(one);
3606 }
3607 }
3608
3609 else if (complete_rewrite) {
3610 diff_populate_filespec(one, 0);
3611 diff_populate_filespec(two, 0);
3612 data->deleted = count_lines(one->data, one->size);
3613 data->added = count_lines(two->data, two->size);
3614 }
3615
3616 else if (!same_contents) {
3617 /* Crazy xdl interfaces.. */
3618 xpparam_t xpp;
3619 xdemitconf_t xecfg;
3620
3621 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3622 die("unable to read files to diff");
3623
3624 memset(&xpp, 0, sizeof(xpp));
3625 memset(&xecfg, 0, sizeof(xecfg));
3626 xpp.flags = o->xdl_opts;
3627 xpp.anchors = o->anchors;
3628 xpp.anchors_nr = o->anchors_nr;
3629 xecfg.ctxlen = o->context;
3630 xecfg.interhunkctxlen = o->interhunkcontext;
3631 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3632 &xpp, &xecfg))
3633 die("unable to generate diffstat for %s", one->path);
3634 }
3635
3636 diff_free_filespec_data(one);
3637 diff_free_filespec_data(two);
3638 }
3639
3640 static void builtin_checkdiff(const char *name_a, const char *name_b,
3641 const char *attr_path,
3642 struct diff_filespec *one,
3643 struct diff_filespec *two,
3644 struct diff_options *o)
3645 {
3646 mmfile_t mf1, mf2;
3647 struct checkdiff_t data;
3648
3649 if (!two)
3650 return;
3651
3652 memset(&data, 0, sizeof(data));
3653 data.filename = name_b ? name_b : name_a;
3654 data.lineno = 0;
3655 data.o = o;
3656 data.ws_rule = whitespace_rule(attr_path);
3657 data.conflict_marker_size = ll_merge_marker_size(attr_path);
3658
3659 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3660 die("unable to read files to diff");
3661
3662 /*
3663 * All the other codepaths check both sides, but not checking
3664 * the "old" side here is deliberate. We are checking the newly
3665 * introduced changes, and as long as the "new" side is text, we
3666 * can and should check what it introduces.
3667 */
3668 if (diff_filespec_is_binary(two))
3669 goto free_and_return;
3670 else {
3671 /* Crazy xdl interfaces.. */
3672 xpparam_t xpp;
3673 xdemitconf_t xecfg;
3674
3675 memset(&xpp, 0, sizeof(xpp));
3676 memset(&xecfg, 0, sizeof(xecfg));
3677 xecfg.ctxlen = 1; /* at least one context line */
3678 xpp.flags = 0;
3679 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3680 &xpp, &xecfg))
3681 die("unable to generate checkdiff for %s", one->path);
3682
3683 if (data.ws_rule & WS_BLANK_AT_EOF) {
3684 struct emit_callback ecbdata;
3685 int blank_at_eof;
3686
3687 ecbdata.ws_rule = data.ws_rule;
3688 check_blank_at_eof(&mf1, &mf2, &ecbdata);
3689 blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3690
3691 if (blank_at_eof) {
3692 static char *err;
3693 if (!err)
3694 err = whitespace_error_string(WS_BLANK_AT_EOF);
3695 fprintf(o->file, "%s:%d: %s.\n",
3696 data.filename, blank_at_eof, err);
3697 data.status = 1; /* report errors */
3698 }
3699 }
3700 }
3701 free_and_return:
3702 diff_free_filespec_data(one);
3703 diff_free_filespec_data(two);
3704 if (data.status)
3705 o->flags.check_failed = 1;
3706 }
3707
3708 struct diff_filespec *alloc_filespec(const char *path)
3709 {
3710 struct diff_filespec *spec;
3711
3712 FLEXPTR_ALLOC_STR(spec, path, path);
3713 spec->count = 1;
3714 spec->is_binary = -1;
3715 return spec;
3716 }
3717
3718 void free_filespec(struct diff_filespec *spec)
3719 {
3720 if (!--spec->count) {
3721 diff_free_filespec_data(spec);
3722 free(spec);
3723 }
3724 }
3725
3726 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3727 int oid_valid, unsigned short mode)
3728 {
3729 if (mode) {
3730 spec->mode = canon_mode(mode);
3731 oidcpy(&spec