diff.c: emit_diff_symbol learns DIFF_SYMBOL_REWRITE_DIFF
[git/git.git] / diff.c
CommitLineData
6973dcae
JH
1/*
2 * Copyright (C) 2005 Junio C Hamano
3 */
6973dcae 4#include "cache.h"
b2141fc1 5#include "config.h"
284098f1 6#include "tempfile.h"
6973dcae
JH
7#include "quote.h"
8#include "diff.h"
9#include "diffcore.h"
051308f6 10#include "delta.h"
6973dcae 11#include "xdiff-interface.h"
7c92fe0e 12#include "color.h"
8c701249 13#include "attr.h"
d5535ec7 14#include "run-command.h"
23707811 15#include "utf8.h"
be58e70d 16#include "userdiff.h"
851e18c3 17#include "submodule-config.h"
752c0c24 18#include "submodule.h"
a757c646 19#include "ll-merge.h"
02e8ca0e 20#include "string-list.h"
82fbf269 21#include "argv-array.h"
660e113c 22#include "graph.h"
6973dcae 23
1510fea7
SP
24#ifdef NO_FAST_WORKING_DIRECTORY
25#define FAST_WORKING_DIRECTORY 0
26#else
27#define FAST_WORKING_DIRECTORY 1
28#endif
29
96f1e58f 30static int diff_detect_rename_default;
33de7163 31static int diff_indent_heuristic = 1;
92c57e5c 32static int diff_rename_limit_default = 400;
a624eaa7 33static int diff_suppress_blank_empty;
d2aea137 34static int diff_use_color_default = -1;
6468a4e5 35static int diff_context_default = 3;
c4888677 36static int diff_interhunk_context_default;
98a4d87b 37static const char *diff_word_regex_cfg;
cbe02100 38static const char *external_diff_cmd_cfg;
6d8940b5 39static const char *diff_order_file_cfg;
aecbf914 40int diff_auto_refresh_index = 1;
a5a818ee 41static int diff_mnemonic_prefix;
f89504dd 42static int diff_no_prefix;
df44483a 43static int diff_stat_graph_width;
712d2c7d 44static int diff_dirstat_permille_default = 30;
be4f2b40 45static struct diff_options default_diff_options;
07ab4dec 46static long diff_algorithm;
a17505f2 47static unsigned ws_error_highlight_default = WSEH_NEW;
6973dcae 48
7c92fe0e 49static char diff_colors[][COLOR_MAXLEN] = {
dc6ebd4c 50 GIT_COLOR_RESET,
8dbf3eb6 51 GIT_COLOR_NORMAL, /* CONTEXT */
dc6ebd4c
AL
52 GIT_COLOR_BOLD, /* METAINFO */
53 GIT_COLOR_CYAN, /* FRAGINFO */
54 GIT_COLOR_RED, /* OLD */
55 GIT_COLOR_GREEN, /* NEW */
56 GIT_COLOR_YELLOW, /* COMMIT */
57 GIT_COLOR_BG_RED, /* WHITESPACE */
89cb73a1 58 GIT_COLOR_NORMAL, /* FUNCINFO */
cd112cef
JS
59};
60
a2f05c94
JNA
61static NORETURN void die_want_option(const char *option_name)
62{
63 die(_("option '%s' requires a value"), option_name);
64}
65
9e1a5ebe 66static int parse_diff_color_slot(const char *var)
801235c5 67{
74b15bfb 68 if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
8dbf3eb6 69 return DIFF_CONTEXT;
9e1a5ebe 70 if (!strcasecmp(var, "meta"))
801235c5 71 return DIFF_METAINFO;
9e1a5ebe 72 if (!strcasecmp(var, "frag"))
801235c5 73 return DIFF_FRAGINFO;
9e1a5ebe 74 if (!strcasecmp(var, "old"))
801235c5 75 return DIFF_FILE_OLD;
9e1a5ebe 76 if (!strcasecmp(var, "new"))
801235c5 77 return DIFF_FILE_NEW;
9e1a5ebe 78 if (!strcasecmp(var, "commit"))
ce436973 79 return DIFF_COMMIT;
9e1a5ebe 80 if (!strcasecmp(var, "whitespace"))
448c3ef1 81 return DIFF_WHITESPACE;
9e1a5ebe 82 if (!strcasecmp(var, "func"))
89cb73a1 83 return DIFF_FUNCINFO;
8b8e8624 84 return -1;
801235c5
JH
85}
86
02e8ca0e 87static int parse_dirstat_params(struct diff_options *options, const char *params_string,
51670fc8 88 struct strbuf *errmsg)
333f3fb0 89{
02e8ca0e
MH
90 char *params_copy = xstrdup(params_string);
91 struct string_list params = STRING_LIST_INIT_NODUP;
92 int ret = 0;
93 int i;
51670fc8 94
02e8ca0e
MH
95 if (*params_copy)
96 string_list_split_in_place(&params, params_copy, ',', -1);
97 for (i = 0; i < params.nr; i++) {
98 const char *p = params.items[i].string;
99 if (!strcmp(p, "changes")) {
1c57a627
JH
100 DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
101 DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
02e8ca0e 102 } else if (!strcmp(p, "lines")) {
1c57a627 103 DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
333f3fb0 104 DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
02e8ca0e 105 } else if (!strcmp(p, "files")) {
1c57a627 106 DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
333f3fb0 107 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
02e8ca0e 108 } else if (!strcmp(p, "noncumulative")) {
333f3fb0 109 DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
02e8ca0e 110 } else if (!strcmp(p, "cumulative")) {
333f3fb0
JH
111 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
112 } else if (isdigit(*p)) {
113 char *end;
51670fc8
JH
114 int permille = strtoul(p, &end, 10) * 10;
115 if (*end == '.' && isdigit(*++end)) {
712d2c7d 116 /* only use first digit */
51670fc8 117 permille += *end - '0';
712d2c7d 118 /* .. and ignore any further digits */
51670fc8 119 while (isdigit(*++end))
712d2c7d
JH
120 ; /* nothing */
121 }
02e8ca0e 122 if (!*end)
51670fc8
JH
123 options->dirstat_permille = permille;
124 else {
02e8ca0e
MH
125 strbuf_addf(errmsg, _(" Failed to parse dirstat cut-off percentage '%s'\n"),
126 p);
51670fc8
JH
127 ret++;
128 }
129 } else {
02e8ca0e 130 strbuf_addf(errmsg, _(" Unknown dirstat parameter '%s'\n"), p);
51670fc8 131 ret++;
333f3fb0 132 }
51670fc8 133
333f3fb0 134 }
02e8ca0e
MH
135 string_list_clear(&params, 0);
136 free(params_copy);
51670fc8 137 return ret;
333f3fb0
JH
138}
139
c47ef57c
RR
140static int parse_submodule_params(struct diff_options *options, const char *value)
141{
142 if (!strcmp(value, "log"))
61cfbc05 143 options->submodule_format = DIFF_SUBMODULE_LOG;
c47ef57c 144 else if (!strcmp(value, "short"))
61cfbc05 145 options->submodule_format = DIFF_SUBMODULE_SHORT;
fd47ae6a
JK
146 else if (!strcmp(value, "diff"))
147 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
c47ef57c
RR
148 else
149 return -1;
150 return 0;
151}
152
cced5fbc
LT
153static int git_config_rename(const char *var, const char *value)
154{
155 if (!value)
156 return DIFF_DETECT_RENAME;
157 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
158 return DIFF_DETECT_COPY;
159 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
160}
161
07924d4d 162long parse_algorithm_value(const char *value)
07ab4dec
MP
163{
164 if (!value)
165 return -1;
166 else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
167 return 0;
168 else if (!strcasecmp(value, "minimal"))
169 return XDF_NEED_MINIMAL;
170 else if (!strcasecmp(value, "patience"))
171 return XDF_PATIENCE_DIFF;
172 else if (!strcasecmp(value, "histogram"))
173 return XDF_HISTOGRAM_DIFF;
174 return -1;
175}
176
0b4b42e7
JH
177static int parse_one_token(const char **arg, const char *token)
178{
179 const char *rest;
180 if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
181 *arg = rest;
182 return 1;
183 }
184 return 0;
185}
186
187static int parse_ws_error_highlight(const char *arg)
188{
189 const char *orig_arg = arg;
190 unsigned val = 0;
191
192 while (*arg) {
193 if (parse_one_token(&arg, "none"))
194 val = 0;
195 else if (parse_one_token(&arg, "default"))
196 val = WSEH_NEW;
197 else if (parse_one_token(&arg, "all"))
198 val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
199 else if (parse_one_token(&arg, "new"))
200 val |= WSEH_NEW;
201 else if (parse_one_token(&arg, "old"))
202 val |= WSEH_OLD;
203 else if (parse_one_token(&arg, "context"))
204 val |= WSEH_CONTEXT;
205 else {
206 return -1 - (int)(arg - orig_arg);
207 }
208 if (*arg)
209 arg++;
210 }
211 return val;
212}
213
83ad63cf
JH
214/*
215 * These are to give UI layer defaults.
216 * The core-level commands such as git-diff-files should
217 * never be affected by the setting of diff.renames
218 * the user happens to have in the configuration file.
219 */
5404c116
MM
220void init_diff_ui_defaults(void)
221{
222 diff_detect_rename_default = 1;
223}
224
5b162879
MH
225int git_diff_heuristic_config(const char *var, const char *value, void *cb)
226{
3cde4e02 227 if (!strcmp(var, "diff.indentheuristic"))
5b162879 228 diff_indent_heuristic = git_config_bool(var, value);
5b162879
MH
229 return 0;
230}
231
ef90d6d4 232int git_diff_ui_config(const char *var, const char *value, void *cb)
801235c5 233{
a159ca0c 234 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
e269eb79 235 diff_use_color_default = git_config_colorbool(var, value);
801235c5
JH
236 return 0;
237 }
6468a4e5
JM
238 if (!strcmp(var, "diff.context")) {
239 diff_context_default = git_config_int(var, value);
240 if (diff_context_default < 0)
241 return -1;
242 return 0;
243 }
c4888677
VN
244 if (!strcmp(var, "diff.interhunkcontext")) {
245 diff_interhunk_context_default = git_config_int(var, value);
246 if (diff_interhunk_context_default < 0)
247 return -1;
248 return 0;
249 }
b68ea12e 250 if (!strcmp(var, "diff.renames")) {
cced5fbc 251 diff_detect_rename_default = git_config_rename(var, value);
b68ea12e
EW
252 return 0;
253 }
aecbf914
JH
254 if (!strcmp(var, "diff.autorefreshindex")) {
255 diff_auto_refresh_index = git_config_bool(var, value);
256 return 0;
257 }
a5a818ee
JH
258 if (!strcmp(var, "diff.mnemonicprefix")) {
259 diff_mnemonic_prefix = git_config_bool(var, value);
260 return 0;
261 }
f89504dd
EC
262 if (!strcmp(var, "diff.noprefix")) {
263 diff_no_prefix = git_config_bool(var, value);
264 return 0;
265 }
df44483a
ZJS
266 if (!strcmp(var, "diff.statgraphwidth")) {
267 diff_stat_graph_width = git_config_int(var, value);
268 return 0;
269 }
daec808c
BH
270 if (!strcmp(var, "diff.external"))
271 return git_config_string(&external_diff_cmd_cfg, var, value);
98a4d87b
BSSJ
272 if (!strcmp(var, "diff.wordregex"))
273 return git_config_string(&diff_word_regex_cfg, var, value);
6d8940b5
SB
274 if (!strcmp(var, "diff.orderfile"))
275 return git_config_pathname(&diff_order_file_cfg, var, value);
f1af60bd 276
be4f2b40
JS
277 if (!strcmp(var, "diff.ignoresubmodules"))
278 handle_ignore_submodules_arg(&default_diff_options, value);
279
c47ef57c
RR
280 if (!strcmp(var, "diff.submodule")) {
281 if (parse_submodule_params(&default_diff_options, value))
282 warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
283 value);
284 return 0;
285 }
286
07ab4dec
MP
287 if (!strcmp(var, "diff.algorithm")) {
288 diff_algorithm = parse_algorithm_value(value);
289 if (diff_algorithm < 0)
290 return -1;
291 return 0;
292 }
293
a17505f2
JH
294 if (!strcmp(var, "diff.wserrorhighlight")) {
295 int val = parse_ws_error_highlight(value);
296 if (val < 0)
297 return -1;
298 ws_error_highlight_default = val;
299 return 0;
300 }
301
3e1dd17a
JK
302 if (git_color_config(var, value, cb) < 0)
303 return -1;
304
ef90d6d4 305 return git_diff_basic_config(var, value, cb);
9a1805a8
JK
306}
307
ef90d6d4 308int git_diff_basic_config(const char *var, const char *value, void *cb)
9a1805a8 309{
ae021d87
JK
310 const char *name;
311
2b6ca6df
LT
312 if (!strcmp(var, "diff.renamelimit")) {
313 diff_rename_limit_default = git_config_int(var, value);
314 return 0;
315 }
316
6680a087
JK
317 if (userdiff_config(var, value) < 0)
318 return -1;
c7534ef4 319
ae021d87
JK
320 if (skip_prefix(var, "diff.color.", &name) ||
321 skip_prefix(var, "color.diff.", &name)) {
322 int slot = parse_diff_color_slot(name);
8b8e8624
JK
323 if (slot < 0)
324 return 0;
64f30e94
JH
325 if (!value)
326 return config_error_nonbool(var);
f6c5a296 327 return color_parse(value, diff_colors[slot]);
801235c5 328 }
f1af60bd 329
a624eaa7 330 /* like GNU diff's --suppress-blank-empty option */
950db879
JS
331 if (!strcmp(var, "diff.suppressblankempty") ||
332 /* for backwards compatibility */
333 !strcmp(var, "diff.suppress-blank-empty")) {
a624eaa7
JM
334 diff_suppress_blank_empty = git_config_bool(var, value);
335 return 0;
336 }
337
2d174951 338 if (!strcmp(var, "diff.dirstat")) {
51670fc8 339 struct strbuf errmsg = STRBUF_INIT;
712d2c7d 340 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
51670fc8 341 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
7478ac57 342 warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
51670fc8
JH
343 errmsg.buf);
344 strbuf_release(&errmsg);
712d2c7d 345 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
2d174951
JH
346 return 0;
347 }
348
59556548 349 if (starts_with(var, "submodule."))
aee9c7d6
JL
350 return parse_submodule_config_option(var, value);
351
cf5e7722
MB
352 if (git_diff_heuristic_config(var, value, cb) < 0)
353 return -1;
354
3e1dd17a 355 return git_default_config(var, value, cb);
801235c5
JH
356}
357
6973dcae
JH
358static char *quote_two(const char *one, const char *two)
359{
360 int need_one = quote_c_style(one, NULL, NULL, 1);
361 int need_two = quote_c_style(two, NULL, NULL, 1);
f285a2d7 362 struct strbuf res = STRBUF_INIT;
6973dcae
JH
363
364 if (need_one + need_two) {
663af342
PH
365 strbuf_addch(&res, '"');
366 quote_c_style(one, &res, NULL, 1);
367 quote_c_style(two, &res, NULL, 1);
368 strbuf_addch(&res, '"');
369 } else {
370 strbuf_addstr(&res, one);
371 strbuf_addstr(&res, two);
6973dcae 372 }
b315c5c0 373 return strbuf_detach(&res, NULL);
6973dcae
JH
374}
375
376static const char *external_diff(void)
377{
378 static const char *external_diff_cmd = NULL;
379 static int done_preparing = 0;
380
381 if (done_preparing)
382 return external_diff_cmd;
383 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
cbe02100
JS
384 if (!external_diff_cmd)
385 external_diff_cmd = external_diff_cmd_cfg;
6973dcae
JH
386 done_preparing = 1;
387 return external_diff_cmd;
388}
389
284098f1
MH
390/*
391 * Keep track of files used for diffing. Sometimes such an entry
392 * refers to a temporary file, sometimes to an existing file, and
393 * sometimes to "/dev/null".
394 */
6973dcae 395static struct diff_tempfile {
284098f1
MH
396 /*
397 * filename external diff should read from, or NULL if this
398 * entry is currently not in use:
399 */
400 const char *name;
401
dc01505f 402 char hex[GIT_MAX_HEXSZ + 1];
6973dcae 403 char mode[10];
284098f1
MH
404
405 /*
406 * If this diff_tempfile instance refers to a temporary file,
407 * this tempfile object is used to manage its lifetime.
408 */
409 struct tempfile tempfile;
6973dcae
JH
410} diff_temp[2];
411
6957eb9a
JH
412typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
413
414struct emit_callback {
6957eb9a
JH
415 int color_diff;
416 unsigned ws_rule;
417 int blank_at_eof_in_preimage;
418 int blank_at_eof_in_postimage;
419 int lno_in_preimage;
420 int lno_in_postimage;
421 sane_truncate_fn truncate;
422 const char **label_path;
423 struct diff_words_data *diff_words;
a3c158d4 424 struct diff_options *opt;
3e97c7c6 425 struct strbuf *header;
6957eb9a
JH
426};
427
6973dcae
JH
428static int count_lines(const char *data, int size)
429{
430 int count, ch, completely_empty = 1, nl_just_seen = 0;
431 count = 0;
432 while (0 < size--) {
433 ch = *data++;
434 if (ch == '\n') {
435 count++;
436 nl_just_seen = 1;
437 completely_empty = 0;
438 }
439 else {
440 nl_just_seen = 0;
441 completely_empty = 0;
442 }
443 }
444 if (completely_empty)
445 return 0;
446 if (!nl_just_seen)
447 count++; /* no trailing newline */
448 return count;
449}
450
6957eb9a
JH
451static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
452{
453 if (!DIFF_FILE_VALID(one)) {
454 mf->ptr = (char *)""; /* does not matter */
455 mf->size = 0;
456 return 0;
457 }
458 else if (diff_populate_filespec(one, 0))
459 return -1;
bb35fefb 460
6957eb9a
JH
461 mf->ptr = one->data;
462 mf->size = one->size;
463 return 0;
464}
465
abb371a1
JK
466/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
467static unsigned long diff_filespec_size(struct diff_filespec *one)
468{
469 if (!DIFF_FILE_VALID(one))
470 return 0;
8e5dd3d6 471 diff_populate_filespec(one, CHECK_SIZE_ONLY);
abb371a1
JK
472 return one->size;
473}
474
6957eb9a
JH
475static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
476{
477 char *ptr = mf->ptr;
478 long size = mf->size;
479 int cnt = 0;
480
481 if (!size)
482 return cnt;
483 ptr += size - 1; /* pointing at the very end */
484 if (*ptr != '\n')
485 ; /* incomplete line */
486 else
487 ptr--; /* skip the last LF */
488 while (mf->ptr < ptr) {
489 char *prev_eol;
490 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
491 if (*prev_eol == '\n')
492 break;
493 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
494 break;
495 cnt++;
496 ptr = prev_eol - 1;
497 }
498 return cnt;
499}
500
501static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
502 struct emit_callback *ecbdata)
503{
504 int l1, l2, at;
505 unsigned ws_rule = ecbdata->ws_rule;
506 l1 = count_trailing_blank(mf1, ws_rule);
507 l2 = count_trailing_blank(mf2, ws_rule);
508 if (l2 <= l1) {
509 ecbdata->blank_at_eof_in_preimage = 0;
510 ecbdata->blank_at_eof_in_postimage = 0;
511 return;
512 }
513 at = count_lines(mf1->ptr, mf1->size);
514 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
515
516 at = count_lines(mf2->ptr, mf2->size);
517 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
518}
519
a3c158d4 520static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
250f7993 521 int first, const char *line, int len)
6957eb9a
JH
522{
523 int has_trailing_newline, has_trailing_carriage_return;
250f7993 524 int nofirst;
a3c158d4
BY
525 FILE *file = o->file;
526
30997bb8 527 fputs(diff_line_prefix(o), file);
6957eb9a 528
250f7993
JH
529 if (len == 0) {
530 has_trailing_newline = (first == '\n');
531 has_trailing_carriage_return = (!has_trailing_newline &&
532 (first == '\r'));
533 nofirst = has_trailing_newline || has_trailing_carriage_return;
534 } else {
535 has_trailing_newline = (len > 0 && line[len-1] == '\n');
536 if (has_trailing_newline)
537 len--;
538 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
539 if (has_trailing_carriage_return)
540 len--;
541 nofirst = 0;
542 }
6957eb9a 543
06a47552
JH
544 if (len || !nofirst) {
545 fputs(set, file);
546 if (!nofirst)
547 fputc(first, file);
548 fwrite(line, len, 1, file);
549 fputs(reset, file);
550 }
6957eb9a
JH
551 if (has_trailing_carriage_return)
552 fputc('\r', file);
553 if (has_trailing_newline)
554 fputc('\n', file);
555}
556
a3c158d4 557static void emit_line(struct diff_options *o, const char *set, const char *reset,
250f7993
JH
558 const char *line, int len)
559{
a3c158d4 560 emit_line_0(o, set, reset, line[0], line+1, len-1);
250f7993
JH
561}
562
36a4cefd 563enum diff_symbol {
5af6ea95 564 DIFF_SYMBOL_REWRITE_DIFF,
4acaaa7a 565 DIFF_SYMBOL_BINARY_FILES,
a29b0a13 566 DIFF_SYMBOL_HEADER,
3ee8b7bf
SB
567 DIFF_SYMBOL_FILEPAIR_PLUS,
568 DIFF_SYMBOL_FILEPAIR_MINUS,
ff958679
SB
569 DIFF_SYMBOL_WORDS_PORCELAIN,
570 DIFF_SYMBOL_WORDS,
091f8e28 571 DIFF_SYMBOL_CONTEXT,
f2bb1218 572 DIFF_SYMBOL_CONTEXT_INCOMPLETE,
091f8e28
SB
573 DIFF_SYMBOL_PLUS,
574 DIFF_SYMBOL_MINUS,
b9cbfde6 575 DIFF_SYMBOL_NO_LF_EOF,
68abc6f1 576 DIFF_SYMBOL_CONTEXT_FRAGINFO,
c64b420b 577 DIFF_SYMBOL_CONTEXT_MARKER,
36a4cefd
SB
578 DIFF_SYMBOL_SEPARATOR
579};
091f8e28
SB
580/*
581 * Flags for content lines:
582 * 0..12 are whitespace rules
583 * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
584 * 16 is marking if the line is blank at EOF
585 */
586#define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF (1<<16)
587#define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
588
589static void emit_line_ws_markup(struct diff_options *o,
590 const char *set, const char *reset,
591 const char *line, int len, char sign,
592 unsigned ws_rule, int blank_at_eof)
593{
594 const char *ws = NULL;
595
596 if (o->ws_error_highlight & ws_rule) {
597 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
598 if (!*ws)
599 ws = NULL;
600 }
601
602 if (!ws)
603 emit_line_0(o, set, reset, sign, line, len);
604 else if (blank_at_eof)
605 /* Blank line at EOF - paint '+' as well */
606 emit_line_0(o, ws, reset, sign, line, len);
607 else {
608 /* Emit just the prefix, then the rest. */
609 emit_line_0(o, set, reset, sign, "", 0);
610 ws_check_emit(line, len, ws_rule,
611 o->file, set, reset, ws);
612 }
613}
36a4cefd
SB
614
615static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
091f8e28 616 const char *line, int len, unsigned flags)
36a4cefd 617{
b9cbfde6 618 static const char *nneof = " No newline at end of file\n";
5af6ea95 619 const char *context, *reset, *set, *meta, *fraginfo;
36a4cefd 620 switch (s) {
b9cbfde6
SB
621 case DIFF_SYMBOL_NO_LF_EOF:
622 context = diff_get_color_opt(o, DIFF_CONTEXT);
623 reset = diff_get_color_opt(o, DIFF_RESET);
624 putc('\n', o->file);
625 emit_line_0(o, context, reset, '\\',
626 nneof, strlen(nneof));
627 break;
68abc6f1
SB
628 case DIFF_SYMBOL_CONTEXT_FRAGINFO:
629 emit_line(o, "", "", line, len);
630 break;
f2bb1218 631 case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
c64b420b
SB
632 case DIFF_SYMBOL_CONTEXT_MARKER:
633 context = diff_get_color_opt(o, DIFF_CONTEXT);
634 reset = diff_get_color_opt(o, DIFF_RESET);
635 emit_line(o, context, reset, line, len);
636 break;
36a4cefd
SB
637 case DIFF_SYMBOL_SEPARATOR:
638 fprintf(o->file, "%s%c",
639 diff_line_prefix(o),
640 o->line_termination);
641 break;
091f8e28
SB
642 case DIFF_SYMBOL_CONTEXT:
643 set = diff_get_color_opt(o, DIFF_CONTEXT);
644 reset = diff_get_color_opt(o, DIFF_RESET);
645 emit_line_ws_markup(o, set, reset, line, len, ' ',
646 flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
647 break;
648 case DIFF_SYMBOL_PLUS:
649 set = diff_get_color_opt(o, DIFF_FILE_NEW);
650 reset = diff_get_color_opt(o, DIFF_RESET);
651 emit_line_ws_markup(o, set, reset, line, len, '+',
652 flags & DIFF_SYMBOL_CONTENT_WS_MASK,
653 flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
654 break;
655 case DIFF_SYMBOL_MINUS:
656 set = diff_get_color_opt(o, DIFF_FILE_OLD);
657 reset = diff_get_color_opt(o, DIFF_RESET);
658 emit_line_ws_markup(o, set, reset, line, len, '-',
659 flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
660 break;
ff958679
SB
661 case DIFF_SYMBOL_WORDS_PORCELAIN:
662 context = diff_get_color_opt(o, DIFF_CONTEXT);
663 reset = diff_get_color_opt(o, DIFF_RESET);
664 emit_line(o, context, reset, line, len);
665 fputs("~\n", o->file);
666 break;
667 case DIFF_SYMBOL_WORDS:
668 context = diff_get_color_opt(o, DIFF_CONTEXT);
669 reset = diff_get_color_opt(o, DIFF_RESET);
670 /*
671 * Skip the prefix character, if any. With
672 * diff_suppress_blank_empty, there may be
673 * none.
674 */
675 if (line[0] != '\n') {
676 line++;
677 len--;
678 }
679 emit_line(o, context, reset, line, len);
680 break;
3ee8b7bf
SB
681 case DIFF_SYMBOL_FILEPAIR_PLUS:
682 meta = diff_get_color_opt(o, DIFF_METAINFO);
683 reset = diff_get_color_opt(o, DIFF_RESET);
684 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
685 line, reset,
686 strchr(line, ' ') ? "\t" : "");
687 break;
688 case DIFF_SYMBOL_FILEPAIR_MINUS:
689 meta = diff_get_color_opt(o, DIFF_METAINFO);
690 reset = diff_get_color_opt(o, DIFF_RESET);
691 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
692 line, reset,
693 strchr(line, ' ') ? "\t" : "");
694 break;
4acaaa7a 695 case DIFF_SYMBOL_BINARY_FILES:
a29b0a13
SB
696 case DIFF_SYMBOL_HEADER:
697 fprintf(o->file, "%s", line);
698 break;
5af6ea95
SB
699 case DIFF_SYMBOL_REWRITE_DIFF:
700 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
701 reset = diff_get_color_opt(o, DIFF_RESET);
702 emit_line(o, fraginfo, reset, line, len);
703 break;
36a4cefd
SB
704 default:
705 die("BUG: unknown diff symbol");
706 }
707}
708
6957eb9a
JH
709static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
710{
711 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
712 ecbdata->blank_at_eof_in_preimage &&
713 ecbdata->blank_at_eof_in_postimage &&
714 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
715 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
716 return 0;
018cff70 717 return ws_blank_line(line, len, ecbdata->ws_rule);
6957eb9a
JH
718}
719
b8767f79 720static void emit_add_line(const char *reset,
0e383e18
JH
721 struct emit_callback *ecbdata,
722 const char *line, int len)
723{
091f8e28
SB
724 unsigned flags = WSEH_NEW | ecbdata->ws_rule;
725 if (new_blank_line_at_eof(ecbdata, line, len))
726 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
727
728 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
b8767f79 729}
0e383e18 730
b8767f79
JH
731static void emit_del_line(const char *reset,
732 struct emit_callback *ecbdata,
733 const char *line, int len)
734{
091f8e28
SB
735 unsigned flags = WSEH_OLD | ecbdata->ws_rule;
736 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
0e383e18
JH
737}
738
739static void emit_context_line(const char *reset,
740 struct emit_callback *ecbdata,
741 const char *line, int len)
742{
091f8e28
SB
743 unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
744 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
0e383e18
JH
745}
746
89cb73a1
BW
747static void emit_hunk_header(struct emit_callback *ecbdata,
748 const char *line, int len)
749{
8dbf3eb6 750 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
89cb73a1
BW
751 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
752 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
753 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
754 static const char atat[2] = { '@', '@' };
755 const char *cp, *ep;
2efcc977
BY
756 struct strbuf msgbuf = STRBUF_INIT;
757 int org_len = len;
758 int i = 1;
89cb73a1
BW
759
760 /*
761 * As a hunk header must begin with "@@ -<old>, +<new> @@",
762 * it always is at least 10 bytes long.
763 */
764 if (len < 10 ||
765 memcmp(line, atat, 2) ||
766 !(ep = memmem(line + 2, len - 2, atat, 2))) {
c64b420b 767 emit_diff_symbol(ecbdata->opt,
091f8e28 768 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
89cb73a1
BW
769 return;
770 }
771 ep += 2; /* skip over @@ */
772
773 /* The hunk header in fraginfo color */
cedc61a9 774 strbuf_addstr(&msgbuf, frag);
2efcc977 775 strbuf_add(&msgbuf, line, ep - line);
cedc61a9 776 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
777
778 /*
779 * trailing "\r\n"
780 */
781 for ( ; i < 3; i++)
782 if (line[len - i] == '\r' || line[len - i] == '\n')
783 len--;
89cb73a1
BW
784
785 /* blank before the func header */
786 for (cp = ep; ep - line < len; ep++)
787 if (*ep != ' ' && *ep != '\t')
788 break;
2efcc977 789 if (ep != cp) {
8dbf3eb6 790 strbuf_addstr(&msgbuf, context);
2efcc977 791 strbuf_add(&msgbuf, cp, ep - cp);
cedc61a9 792 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
793 }
794
795 if (ep < line + len) {
cedc61a9 796 strbuf_addstr(&msgbuf, func);
2efcc977 797 strbuf_add(&msgbuf, ep, line + len - ep);
cedc61a9 798 strbuf_addstr(&msgbuf, reset);
2efcc977 799 }
89cb73a1 800
2efcc977 801 strbuf_add(&msgbuf, line + len, org_len - len);
dfb7728f 802 strbuf_complete_line(&msgbuf);
68abc6f1 803 emit_diff_symbol(ecbdata->opt,
091f8e28 804 DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
2efcc977 805 strbuf_release(&msgbuf);
89cb73a1
BW
806}
807
479b0ae8
JK
808static struct diff_tempfile *claim_diff_tempfile(void) {
809 int i;
810 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
811 if (!diff_temp[i].name)
812 return diff_temp + i;
813 die("BUG: diff is failing to clean up its tempfiles");
814}
815
479b0ae8
JK
816static void remove_tempfile(void)
817{
818 int i;
a8344abe 819 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
284098f1
MH
820 if (is_tempfile_active(&diff_temp[i].tempfile))
821 delete_tempfile(&diff_temp[i].tempfile);
a8344abe
NR
822 diff_temp[i].name = NULL;
823 }
479b0ae8
JK
824}
825
5af6ea95 826static void add_line_count(struct strbuf *out, int count)
6973dcae
JH
827{
828 switch (count) {
829 case 0:
5af6ea95 830 strbuf_addstr(out, "0,0");
6973dcae
JH
831 break;
832 case 1:
5af6ea95 833 strbuf_addstr(out, "1");
6973dcae
JH
834 break;
835 default:
5af6ea95 836 strbuf_addf(out, "1,%d", count);
6973dcae
JH
837 break;
838 }
839}
840
7f7ee2ff
JH
841static void emit_rewrite_lines(struct emit_callback *ecb,
842 int prefix, const char *data, int size)
6973dcae 843{
7f7ee2ff 844 const char *endp = NULL;
7f7ee2ff
JH
845 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
846
847 while (0 < size) {
848 int len;
849
850 endp = memchr(data, '\n', size);
851 len = endp ? (endp - data + 1) : size;
852 if (prefix != '+') {
853 ecb->lno_in_preimage++;
0e383e18 854 emit_del_line(reset, ecb, data, len);
7f7ee2ff
JH
855 } else {
856 ecb->lno_in_postimage++;
857 emit_add_line(reset, ecb, data, len);
13e36ec5 858 }
7f7ee2ff
JH
859 size -= len;
860 data += len;
861 }
b9cbfde6 862 if (!endp)
091f8e28 863 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
6973dcae
JH
864}
865
866static void emit_rewrite_diff(const char *name_a,
867 const char *name_b,
868 struct diff_filespec *one,
13e36ec5 869 struct diff_filespec *two,
d9bae1a1
JK
870 struct userdiff_driver *textconv_one,
871 struct userdiff_driver *textconv_two,
eab9a40b 872 struct diff_options *o)
6973dcae
JH
873{
874 int lc_a, lc_b;
d5625091 875 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
a5a818ee 876 const char *a_prefix, *b_prefix;
840383b2 877 char *data_one, *data_two;
3aa1f7ca 878 size_t size_one, size_two;
7f7ee2ff 879 struct emit_callback ecbdata;
5af6ea95 880 struct strbuf out = STRBUF_INIT;
a5a818ee
JH
881
882 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
883 a_prefix = o->b_prefix;
884 b_prefix = o->a_prefix;
885 } else {
886 a_prefix = o->a_prefix;
887 b_prefix = o->b_prefix;
888 }
1a9eb3b9 889
8a13becc
JH
890 name_a += (*name_a == '/');
891 name_b += (*name_b == '/');
1a9eb3b9 892
d5625091
JH
893 strbuf_reset(&a_name);
894 strbuf_reset(&b_name);
a5a818ee
JH
895 quote_two_c_style(&a_name, a_prefix, name_a, 0);
896 quote_two_c_style(&b_name, b_prefix, name_b, 0);
d5625091 897
840383b2
JK
898 size_one = fill_textconv(textconv_one, one, &data_one);
899 size_two = fill_textconv(textconv_two, two, &data_two);
3aa1f7ca 900
d91ba8fa 901 memset(&ecbdata, 0, sizeof(ecbdata));
daa0c3d9 902 ecbdata.color_diff = want_color(o->use_color);
c189c4f2 903 ecbdata.ws_rule = whitespace_rule(name_b);
a3c158d4 904 ecbdata.opt = o;
d91ba8fa
JH
905 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
906 mmfile_t mf1, mf2;
907 mf1.ptr = (char *)data_one;
908 mf2.ptr = (char *)data_two;
909 mf1.size = size_one;
910 mf2.size = size_two;
911 check_blank_at_eof(&mf1, &mf2, &ecbdata);
912 }
913 ecbdata.lno_in_preimage = 1;
914 ecbdata.lno_in_postimage = 1;
915
3aa1f7ca
JK
916 lc_a = count_lines(data_one, size_one);
917 lc_b = count_lines(data_two, size_two);
3ee8b7bf
SB
918
919 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
920 a_name.buf, a_name.len, 0);
921 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
922 b_name.buf, b_name.len, 0);
923
5af6ea95 924 strbuf_addstr(&out, "@@ -");
467ddc14 925 if (!o->irreversible_delete)
5af6ea95 926 add_line_count(&out, lc_a);
467ddc14 927 else
5af6ea95
SB
928 strbuf_addstr(&out, "?,?");
929 strbuf_addstr(&out, " +");
930 add_line_count(&out, lc_b);
931 strbuf_addstr(&out, " @@\n");
932 emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
933 strbuf_release(&out);
934
467ddc14 935 if (lc_a && !o->irreversible_delete)
d91ba8fa 936 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
6973dcae 937 if (lc_b)
d91ba8fa 938 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
b76c056b 939 if (textconv_one)
aed6ca52 940 free((char *)data_one);
b76c056b 941 if (textconv_two)
aed6ca52 942 free((char *)data_two);
6973dcae
JH
943}
944
f59a59e2
JS
945struct diff_words_buffer {
946 mmfile_t text;
947 long alloc;
2e5d2003
JS
948 struct diff_words_orig {
949 const char *begin, *end;
950 } *orig;
951 int orig_nr, orig_alloc;
f59a59e2
JS
952};
953
954static void diff_words_append(char *line, unsigned long len,
955 struct diff_words_buffer *buffer)
956{
23c1575f 957 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
f59a59e2
JS
958 line++;
959 len--;
960 memcpy(buffer->text.ptr + buffer->text.size, line, len);
961 buffer->text.size += len;
2b6a5417 962 buffer->text.ptr[buffer->text.size] = '\0';
f59a59e2
JS
963}
964
9cba13ca 965struct diff_words_style_elem {
882749a0
TR
966 const char *prefix;
967 const char *suffix;
968 const char *color; /* NULL; filled in by the setup code if
969 * color is enabled */
970};
971
9cba13ca 972struct diff_words_style {
882749a0
TR
973 enum diff_words_type type;
974 struct diff_words_style_elem new, old, ctx;
975 const char *newline;
976};
977
c2e86add 978static struct diff_words_style diff_words_styles[] = {
882749a0
TR
979 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
980 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
981 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
982};
983
f59a59e2 984struct diff_words_data {
f59a59e2 985 struct diff_words_buffer minus, plus;
2e5d2003 986 const char *current_plus;
4297c0ae
BY
987 int last_minus;
988 struct diff_options *opt;
2b6a5417 989 regex_t *word_regex;
882749a0
TR
990 enum diff_words_type type;
991 struct diff_words_style *style;
f59a59e2
JS
992};
993
882749a0
TR
994static int fn_out_diff_words_write_helper(FILE *fp,
995 struct diff_words_style_elem *st_el,
996 const char *newline,
4297c0ae
BY
997 size_t count, const char *buf,
998 const char *line_prefix)
882749a0 999{
4297c0ae
BY
1000 int print = 0;
1001
882749a0
TR
1002 while (count) {
1003 char *p = memchr(buf, '\n', count);
4297c0ae
BY
1004 if (print)
1005 fputs(line_prefix, fp);
882749a0
TR
1006 if (p != buf) {
1007 if (st_el->color && fputs(st_el->color, fp) < 0)
1008 return -1;
1009 if (fputs(st_el->prefix, fp) < 0 ||
1010 fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
1011 fputs(st_el->suffix, fp) < 0)
1012 return -1;
1013 if (st_el->color && *st_el->color
1014 && fputs(GIT_COLOR_RESET, fp) < 0)
1015 return -1;
1016 }
1017 if (!p)
1018 return 0;
1019 if (fputs(newline, fp) < 0)
1020 return -1;
1021 count -= p + 1 - buf;
1022 buf = p + 1;
4297c0ae 1023 print = 1;
882749a0
TR
1024 }
1025 return 0;
1026}
1027
4297c0ae
BY
1028/*
1029 * '--color-words' algorithm can be described as:
1030 *
5621760f 1031 * 1. collect the minus/plus lines of a diff hunk, divided into
4297c0ae
BY
1032 * minus-lines and plus-lines;
1033 *
1034 * 2. break both minus-lines and plus-lines into words and
1035 * place them into two mmfile_t with one word for each line;
1036 *
1037 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1038 *
1039 * And for the common parts of the both file, we output the plus side text.
1040 * diff_words->current_plus is used to trace the current position of the plus file
1041 * which printed. diff_words->last_minus is used to trace the last minus word
1042 * printed.
1043 *
1044 * For '--graph' to work with '--color-words', we need to output the graph prefix
1045 * on each line of color words output. Generally, there are two conditions on
1046 * which we should output the prefix.
1047 *
1048 * 1. diff_words->last_minus == 0 &&
1049 * diff_words->current_plus == diff_words->plus.text.ptr
1050 *
1051 * that is: the plus text must start as a new line, and if there is no minus
1052 * word printed, a graph prefix must be printed.
1053 *
1054 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
1055 * *(diff_words->current_plus - 1) == '\n'
1056 *
1057 * that is: a graph prefix must be printed following a '\n'
1058 */
1059static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1060{
1061 if ((diff_words->last_minus == 0 &&
1062 diff_words->current_plus == diff_words->plus.text.ptr) ||
1063 (diff_words->current_plus > diff_words->plus.text.ptr &&
1064 *(diff_words->current_plus - 1) == '\n')) {
1065 return 1;
1066 } else {
1067 return 0;
1068 }
1069}
1070
f59a59e2 1071static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
f59a59e2 1072{
f59a59e2 1073 struct diff_words_data *diff_words = priv;
882749a0 1074 struct diff_words_style *style = diff_words->style;
2e5d2003
JS
1075 int minus_first, minus_len, plus_first, plus_len;
1076 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
4297c0ae 1077 struct diff_options *opt = diff_words->opt;
30997bb8 1078 const char *line_prefix;
f59a59e2 1079
2e5d2003
JS
1080 if (line[0] != '@' || parse_hunk_header(line, len,
1081 &minus_first, &minus_len, &plus_first, &plus_len))
f59a59e2
JS
1082 return;
1083
4297c0ae 1084 assert(opt);
30997bb8 1085 line_prefix = diff_line_prefix(opt);
4297c0ae 1086
2e5d2003
JS
1087 /* POSIX requires that first be decremented by one if len == 0... */
1088 if (minus_len) {
1089 minus_begin = diff_words->minus.orig[minus_first].begin;
1090 minus_end =
1091 diff_words->minus.orig[minus_first + minus_len - 1].end;
1092 } else
1093 minus_begin = minus_end =
1094 diff_words->minus.orig[minus_first].end;
1095
1096 if (plus_len) {
1097 plus_begin = diff_words->plus.orig[plus_first].begin;
1098 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1099 } else
1100 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1101
4297c0ae
BY
1102 if (color_words_output_graph_prefix(diff_words)) {
1103 fputs(line_prefix, diff_words->opt->file);
1104 }
1105 if (diff_words->current_plus != plus_begin) {
1106 fn_out_diff_words_write_helper(diff_words->opt->file,
882749a0
TR
1107 &style->ctx, style->newline,
1108 plus_begin - diff_words->current_plus,
4297c0ae
BY
1109 diff_words->current_plus, line_prefix);
1110 if (*(plus_begin - 1) == '\n')
1111 fputs(line_prefix, diff_words->opt->file);
1112 }
1113 if (minus_begin != minus_end) {
1114 fn_out_diff_words_write_helper(diff_words->opt->file,
882749a0 1115 &style->old, style->newline,
4297c0ae
BY
1116 minus_end - minus_begin, minus_begin,
1117 line_prefix);
1118 }
1119 if (plus_begin != plus_end) {
1120 fn_out_diff_words_write_helper(diff_words->opt->file,
882749a0 1121 &style->new, style->newline,
4297c0ae
BY
1122 plus_end - plus_begin, plus_begin,
1123 line_prefix);
1124 }
2e5d2003
JS
1125
1126 diff_words->current_plus = plus_end;
4297c0ae 1127 diff_words->last_minus = minus_first;
f59a59e2
JS
1128}
1129
2b6a5417
JS
1130/* This function starts looking at *begin, and returns 0 iff a word was found. */
1131static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1132 int *begin, int *end)
1133{
1134 if (word_regex && *begin < buffer->size) {
1135 regmatch_t match[1];
b7d36ffc
JS
1136 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1137 buffer->size - *begin, 1, match, 0)) {
2b6a5417
JS
1138 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1139 '\n', match[0].rm_eo - match[0].rm_so);
1140 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1141 *begin += match[0].rm_so;
1142 return *begin >= *end;
1143 }
1144 return -1;
f59a59e2
JS
1145 }
1146
2b6a5417
JS
1147 /* find the next word */
1148 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1149 (*begin)++;
1150 if (*begin >= buffer->size)
1151 return -1;
f59a59e2 1152
2b6a5417
JS
1153 /* find the end of the word */
1154 *end = *begin + 1;
1155 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1156 (*end)++;
1157
1158 return 0;
f59a59e2
JS
1159}
1160
23c1575f 1161/*
2e5d2003
JS
1162 * This function splits the words in buffer->text, stores the list with
1163 * newline separator into out, and saves the offsets of the original words
1164 * in buffer->orig.
23c1575f 1165 */
2b6a5417
JS
1166static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1167 regex_t *word_regex)
f59a59e2 1168{
2e5d2003 1169 int i, j;
2b6a5417 1170 long alloc = 0;
f59a59e2 1171
2e5d2003 1172 out->size = 0;
2b6a5417 1173 out->ptr = NULL;
f59a59e2 1174
2e5d2003
JS
1175 /* fake an empty "0th" word */
1176 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1177 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1178 buffer->orig_nr = 1;
1179
1180 for (i = 0; i < buffer->text.size; i++) {
2b6a5417
JS
1181 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1182 return;
2e5d2003
JS
1183
1184 /* store original boundaries */
1185 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1186 buffer->orig_alloc);
1187 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1188 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1189 buffer->orig_nr++;
1190
1191 /* store one word */
2b6a5417 1192 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2e5d2003
JS
1193 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1194 out->ptr[out->size + j - i] = '\n';
1195 out->size += j - i + 1;
1196
1197 i = j - 1;
f59a59e2
JS
1198 }
1199}
1200
1201/* this executes the word diff on the accumulated buffers */
1202static void diff_words_show(struct diff_words_data *diff_words)
1203{
1204 xpparam_t xpp;
1205 xdemitconf_t xecfg;
f59a59e2 1206 mmfile_t minus, plus;
882749a0 1207 struct diff_words_style *style = diff_words->style;
f59a59e2 1208
4297c0ae 1209 struct diff_options *opt = diff_words->opt;
30997bb8 1210 const char *line_prefix;
4297c0ae
BY
1211
1212 assert(opt);
30997bb8 1213 line_prefix = diff_line_prefix(opt);
4297c0ae 1214
2e5d2003
JS
1215 /* special case: only removal */
1216 if (!diff_words->plus.text.size) {
4297c0ae
BY
1217 fputs(line_prefix, diff_words->opt->file);
1218 fn_out_diff_words_write_helper(diff_words->opt->file,
882749a0 1219 &style->old, style->newline,
4297c0ae
BY
1220 diff_words->minus.text.size,
1221 diff_words->minus.text.ptr, line_prefix);
2e5d2003
JS
1222 diff_words->minus.text.size = 0;
1223 return;
1224 }
1225
1226 diff_words->current_plus = diff_words->plus.text.ptr;
4297c0ae 1227 diff_words->last_minus = 0;
f59a59e2 1228
9ccd0a88 1229 memset(&xpp, 0, sizeof(xpp));
30b25010 1230 memset(&xecfg, 0, sizeof(xecfg));
2b6a5417
JS
1231 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1232 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
582aa00b 1233 xpp.flags = 0;
2b6a5417 1234 /* as only the hunk header will be parsed, we need a 0-context */
2e5d2003 1235 xecfg.ctxlen = 0;
3efb9880
JK
1236 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1237 &xpp, &xecfg))
1238 die("unable to generate word diff");
f59a59e2
JS
1239 free(minus.ptr);
1240 free(plus.ptr);
2e5d2003 1241 if (diff_words->current_plus != diff_words->plus.text.ptr +
4297c0ae
BY
1242 diff_words->plus.text.size) {
1243 if (color_words_output_graph_prefix(diff_words))
1244 fputs(line_prefix, diff_words->opt->file);
1245 fn_out_diff_words_write_helper(diff_words->opt->file,
882749a0 1246 &style->ctx, style->newline,
2e5d2003 1247 diff_words->plus.text.ptr + diff_words->plus.text.size
4297c0ae
BY
1248 - diff_words->current_plus, diff_words->current_plus,
1249 line_prefix);
1250 }
f59a59e2 1251 diff_words->minus.text.size = diff_words->plus.text.size = 0;
f59a59e2
JS
1252}
1253
76fd2828
JH
1254/* In "color-words" mode, show word-diff of words accumulated in the buffer */
1255static void diff_words_flush(struct emit_callback *ecbdata)
1256{
1257 if (ecbdata->diff_words->minus.text.size ||
1258 ecbdata->diff_words->plus.text.size)
1259 diff_words_show(ecbdata->diff_words);
1260}
1261
77d1a520
TR
1262static void diff_filespec_load_driver(struct diff_filespec *one)
1263{
1264 /* Use already-loaded driver */
1265 if (one->driver)
1266 return;
1267
1268 if (S_ISREG(one->mode))
1269 one->driver = userdiff_find_by_path(one->path);
1270
1271 /* Fallback to default settings */
1272 if (!one->driver)
1273 one->driver = userdiff_find_by_name("default");
1274}
1275
1276static const char *userdiff_word_regex(struct diff_filespec *one)
1277{
1278 diff_filespec_load_driver(one);
1279 return one->driver->word_regex;
1280}
1281
1282static void init_diff_words_data(struct emit_callback *ecbdata,
6440d341 1283 struct diff_options *orig_opts,
77d1a520
TR
1284 struct diff_filespec *one,
1285 struct diff_filespec *two)
1286{
1287 int i;
6440d341
TR
1288 struct diff_options *o = xmalloc(sizeof(struct diff_options));
1289 memcpy(o, orig_opts, sizeof(struct diff_options));
77d1a520
TR
1290
1291 ecbdata->diff_words =
1292 xcalloc(1, sizeof(struct diff_words_data));
1293 ecbdata->diff_words->type = o->word_diff;
1294 ecbdata->diff_words->opt = o;
1295 if (!o->word_regex)
1296 o->word_regex = userdiff_word_regex(one);
1297 if (!o->word_regex)
1298 o->word_regex = userdiff_word_regex(two);
1299 if (!o->word_regex)
1300 o->word_regex = diff_word_regex_cfg;
1301 if (o->word_regex) {
1302 ecbdata->diff_words->word_regex = (regex_t *)
1303 xmalloc(sizeof(regex_t));
1304 if (regcomp(ecbdata->diff_words->word_regex,
1305 o->word_regex,
1306 REG_EXTENDED | REG_NEWLINE))
1307 die ("Invalid regular expression: %s",
1308 o->word_regex);
1309 }
1310 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1311 if (o->word_diff == diff_words_styles[i].type) {
1312 ecbdata->diff_words->style =
1313 &diff_words_styles[i];
1314 break;
1315 }
1316 }
1317 if (want_color(o->use_color)) {
1318 struct diff_words_style *st = ecbdata->diff_words->style;
1319 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1320 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
8dbf3eb6 1321 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
77d1a520
TR
1322 }
1323}
1324
f59a59e2
JS
1325static void free_diff_words_data(struct emit_callback *ecbdata)
1326{
1327 if (ecbdata->diff_words) {
76fd2828 1328 diff_words_flush(ecbdata);
6440d341 1329 free (ecbdata->diff_words->opt);
8e0f7003 1330 free (ecbdata->diff_words->minus.text.ptr);
2e5d2003 1331 free (ecbdata->diff_words->minus.orig);
8e0f7003 1332 free (ecbdata->diff_words->plus.text.ptr);
2e5d2003 1333 free (ecbdata->diff_words->plus.orig);
ef5644ea
BC
1334 if (ecbdata->diff_words->word_regex) {
1335 regfree(ecbdata->diff_words->word_regex);
1336 free(ecbdata->diff_words->word_regex);
1337 }
6a83d902 1338 FREE_AND_NULL(ecbdata->diff_words);
f59a59e2
JS
1339 }
1340}
1341
ce436973 1342const char *diff_get_color(int diff_use_color, enum color_diff ix)
cd112cef 1343{
daa0c3d9 1344 if (want_color(diff_use_color))
50f575fc
LT
1345 return diff_colors[ix];
1346 return "";
cd112cef
JS
1347}
1348
f1922234
JK
1349const char *diff_line_prefix(struct diff_options *opt)
1350{
1351 struct strbuf *msgbuf;
1352 if (!opt->output_prefix)
1353 return "";
1354
1355 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1356 return msgbuf->buf;
1357}
1358
23707811
JH
1359static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1360{
1361 const char *cp;
1362 unsigned long allot;
1363 size_t l = len;
1364
1365 if (ecb->truncate)
1366 return ecb->truncate(line, len);
1367 cp = line;
1368 allot = l;
1369 while (0 < l) {
1370 (void) utf8_width(&cp, &l);
1371 if (!cp)
1372 break; /* truncated in the middle? */
1373 }
1374 return allot - l;
1375}
1376
d68fe26f 1377static void find_lno(const char *line, struct emit_callback *ecbdata)
690ed843 1378{
d68fe26f
JH
1379 const char *p;
1380 ecbdata->lno_in_preimage = 0;
1381 ecbdata->lno_in_postimage = 0;
1382 p = strchr(line, '-');
690ed843 1383 if (!p)
d68fe26f
JH
1384 return; /* cannot happen */
1385 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1386 p = strchr(p, '+');
1387 if (!p)
1388 return; /* cannot happen */
1389 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
690ed843
JH
1390}
1391
cd112cef 1392static void fn_out_consume(void *priv, char *line, unsigned long len)
6973dcae 1393{
6973dcae 1394 struct emit_callback *ecbdata = priv;
ce436973 1395 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
7be57610 1396 struct diff_options *o = ecbdata->opt;
6973dcae 1397
ba16233c
SB
1398 o->found_changes = 1;
1399
3e97c7c6 1400 if (ecbdata->header) {
a29b0a13
SB
1401 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
1402 ecbdata->header->buf, ecbdata->header->len, 0);
3e97c7c6
GB
1403 strbuf_reset(ecbdata->header);
1404 ecbdata->header = NULL;
1405 }
34a5e1a2 1406
6973dcae 1407 if (ecbdata->label_path[0]) {
3ee8b7bf
SB
1408 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1409 ecbdata->label_path[0],
1410 strlen(ecbdata->label_path[0]), 0);
1411 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1412 ecbdata->label_path[1],
1413 strlen(ecbdata->label_path[1]), 0);
6973dcae
JH
1414 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1415 }
cd112cef 1416
a624eaa7
JM
1417 if (diff_suppress_blank_empty
1418 && len == 2 && line[0] == ' ' && line[1] == '\n') {
1419 line[0] = '\n';
1420 len = 1;
1421 }
1422
b8d9c1a6 1423 if (line[0] == '@') {
76fd2828
JH
1424 if (ecbdata->diff_words)
1425 diff_words_flush(ecbdata);
23707811 1426 len = sane_truncate_line(ecbdata, line, len);
d68fe26f 1427 find_lno(line, ecbdata);
89cb73a1 1428 emit_hunk_header(ecbdata, line, len);
448c3ef1 1429 return;
cd112cef 1430 }
448c3ef1 1431
448c3ef1 1432 if (ecbdata->diff_words) {
ff958679
SB
1433 enum diff_symbol s =
1434 ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
1435 DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
448c3ef1
JH
1436 if (line[0] == '-') {
1437 diff_words_append(line, len,
1438 &ecbdata->diff_words->minus);
1439 return;
1440 } else if (line[0] == '+') {
1441 diff_words_append(line, len,
1442 &ecbdata->diff_words->plus);
1443 return;
59556548 1444 } else if (starts_with(line, "\\ ")) {
c7c2bc0a
TR
1445 /*
1446 * Eat the "no newline at eof" marker as if we
1447 * saw a "+" or "-" line with nothing on it,
1448 * and return without diff_words_flush() to
1449 * defer processing. If this is the end of
1450 * preimage, more "+" lines may come after it.
1451 */
1452 return;
448c3ef1 1453 }
76fd2828 1454 diff_words_flush(ecbdata);
ff958679 1455 emit_diff_symbol(o, s, line, len, 0);
448c3ef1
JH
1456 return;
1457 }
448c3ef1 1458
0e383e18
JH
1459 switch (line[0]) {
1460 case '+':
d68fe26f 1461 ecbdata->lno_in_postimage++;
018cff70 1462 emit_add_line(reset, ecbdata, line + 1, len - 1);
0e383e18
JH
1463 break;
1464 case '-':
1465 ecbdata->lno_in_preimage++;
1466 emit_del_line(reset, ecbdata, line + 1, len - 1);
1467 break;
1468 case ' ':
1469 ecbdata->lno_in_postimage++;
1470 ecbdata->lno_in_preimage++;
1471 emit_context_line(reset, ecbdata, line + 1, len - 1);
1472 break;
1473 default:
1474 /* incomplete line at the end */
1475 ecbdata->lno_in_preimage++;
f2bb1218
SB
1476 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
1477 line, len, 0);
0e383e18 1478 break;
448c3ef1 1479 }
6973dcae
JH
1480}
1481
1482static char *pprint_rename(const char *a, const char *b)
1483{
1484 const char *old = a;
1485 const char *new = b;
f285a2d7 1486 struct strbuf name = STRBUF_INIT;
6973dcae 1487 int pfx_length, sfx_length;
dd281f09 1488 int pfx_adjust_for_slash;
6973dcae
JH
1489 int len_a = strlen(a);
1490 int len_b = strlen(b);
663af342 1491 int a_midlen, b_midlen;
e5bfbf9b
AJ
1492 int qlen_a = quote_c_style(a, NULL, NULL, 0);
1493 int qlen_b = quote_c_style(b, NULL, NULL, 0);
1494
1495 if (qlen_a || qlen_b) {
663af342
PH
1496 quote_c_style(a, &name, NULL, 0);
1497 strbuf_addstr(&name, " => ");
1498 quote_c_style(b, &name, NULL, 0);
b315c5c0 1499 return strbuf_detach(&name, NULL);
e5bfbf9b 1500 }
6973dcae
JH
1501
1502 /* Find common prefix */
1503 pfx_length = 0;
1504 while (*old && *new && *old == *new) {
1505 if (*old == '/')
1506 pfx_length = old - a + 1;
1507 old++;
1508 new++;
1509 }
1510
1511 /* Find common suffix */
1512 old = a + len_a;
1513 new = b + len_b;
1514 sfx_length = 0;
d020e27f 1515 /*
dd281f09
TR
1516 * If there is a common prefix, it must end in a slash. In
1517 * that case we let this loop run 1 into the prefix to see the
1518 * same slash.
1519 *
1520 * If there is no common prefix, we cannot do this as it would
1521 * underrun the input strings.
d020e27f 1522 */
dd281f09
TR
1523 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1524 while (a + pfx_length - pfx_adjust_for_slash <= old &&
1525 b + pfx_length - pfx_adjust_for_slash <= new &&
d020e27f 1526 *old == *new) {
6973dcae
JH
1527 if (*old == '/')
1528 sfx_length = len_a - (old - a);
1529 old--;
1530 new--;
1531 }
1532
1533 /*
1534 * pfx{mid-a => mid-b}sfx
1535 * {pfx-a => pfx-b}sfx
1536 * pfx{sfx-a => sfx-b}
1537 * name-a => name-b
1538 */
663af342
PH
1539 a_midlen = len_a - pfx_length - sfx_length;
1540 b_midlen = len_b - pfx_length - sfx_length;
1541 if (a_midlen < 0)
1542 a_midlen = 0;
1543 if (b_midlen < 0)
1544 b_midlen = 0;
1545
1546 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
6973dcae 1547 if (pfx_length + sfx_length) {
663af342
PH
1548 strbuf_add(&name, a, pfx_length);
1549 strbuf_addch(&name, '{');
6973dcae 1550 }
663af342
PH
1551 strbuf_add(&name, a + pfx_length, a_midlen);
1552 strbuf_addstr(&name, " => ");
1553 strbuf_add(&name, b + pfx_length, b_midlen);
1554 if (pfx_length + sfx_length) {
1555 strbuf_addch(&name, '}');
1556 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
6973dcae 1557 }
b315c5c0 1558 return strbuf_detach(&name, NULL);
6973dcae
JH
1559}
1560
1561struct diffstat_t {
6973dcae
JH
1562 int nr;
1563 int alloc;
1564 struct diffstat_file {
f604652e 1565 char *from_name;
6973dcae 1566 char *name;
f604652e 1567 char *print_name;
6973dcae
JH
1568 unsigned is_unmerged:1;
1569 unsigned is_binary:1;
1570 unsigned is_renamed:1;
74faaa16 1571 unsigned is_interesting:1;
0974c117 1572 uintmax_t added, deleted;
6973dcae
JH
1573 } **files;
1574};
1575
1576static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1577 const char *name_a,
1578 const char *name_b)
1579{
1580 struct diffstat_file *x;
1a4927c5 1581 x = xcalloc(1, sizeof(*x));
4c960a43 1582 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
6973dcae
JH
1583 diffstat->files[diffstat->nr++] = x;
1584 if (name_b) {
f604652e
JH
1585 x->from_name = xstrdup(name_a);
1586 x->name = xstrdup(name_b);
6973dcae
JH
1587 x->is_renamed = 1;
1588 }
f604652e
JH
1589 else {
1590 x->from_name = NULL;
9befac47 1591 x->name = xstrdup(name_a);
f604652e 1592 }
6973dcae
JH
1593 return x;
1594}
1595
1596static void diffstat_consume(void *priv, char *line, unsigned long len)
1597{
1598 struct diffstat_t *diffstat = priv;
1599 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1600
1601 if (line[0] == '+')
1602 x->added++;
1603 else if (line[0] == '-')
1604 x->deleted++;
1605}
1606
698ce6f8 1607const char mime_boundary_leader[] = "------------";
6973dcae 1608
a2540023
JH
1609static int scale_linear(int it, int width, int max_change)
1610{
2eeeef24
JH
1611 if (!it)
1612 return 0;
a2540023 1613 /*
2eeeef24
JH
1614 * make sure that at least one '-' or '+' is printed if
1615 * there is any change to this path. The easiest way is to
1616 * scale linearly as if the alloted width is one column shorter
1617 * than it is, and then add 1 to the result.
a2540023 1618 */
2eeeef24 1619 return 1 + (it * (width - 1) / max_change);
a2540023
JH
1620}
1621
c0c77734 1622static void show_name(FILE *file,
a408e0e6 1623 const char *prefix, const char *name, int len)
a2540023 1624{
a408e0e6 1625 fprintf(file, " %s%-*s |", prefix, len, name);
a2540023
JH
1626}
1627
c0c77734 1628static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
a2540023
JH
1629{
1630 if (cnt <= 0)
1631 return;
c0c77734 1632 fprintf(file, "%s", set);
a2540023 1633 while (cnt--)
c0c77734
DB
1634 putc(ch, file);
1635 fprintf(file, "%s", reset);
a2540023
JH
1636}
1637
f604652e
JH
1638static void fill_print_name(struct diffstat_file *file)
1639{
1640 char *pname;
1641
1642 if (file->print_name)
1643 return;
1644
1645 if (!file->is_renamed) {
f285a2d7 1646 struct strbuf buf = STRBUF_INIT;
f604652e
JH
1647 if (quote_c_style(file->name, &buf, NULL, 0)) {
1648 pname = strbuf_detach(&buf, NULL);
1649 } else {
1650 pname = file->name;
1651 strbuf_release(&buf);
1652 }
1653 } else {
1654 pname = pprint_rename(file->from_name, file->name);
1655 }
1656 file->print_name = pname;
1657}
1658
7f814632
NTND
1659int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1660{
1661 struct strbuf sb = STRBUF_INIT;
1662 int ret;
1663
1664 if (!files) {
1665 assert(insertions == 0 && deletions == 0);
218adaaa 1666 return fprintf(fp, "%s\n", " 0 files changed");
7f814632
NTND
1667 }
1668
1669 strbuf_addf(&sb,
218adaaa 1670 (files == 1) ? " %d file changed" : " %d files changed",
7f814632
NTND
1671 files);
1672
1673 /*
1674 * For binary diff, the caller may want to print "x files
1675 * changed" with insertions == 0 && deletions == 0.
1676 *
1677 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1678 * is probably less confusing (i.e skip over "2 files changed
1679 * but nothing about added/removed lines? Is this a bug in Git?").
1680 */
1681 if (insertions || deletions == 0) {
7f814632 1682 strbuf_addf(&sb,
218adaaa 1683 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
7f814632
NTND
1684 insertions);
1685 }
1686
1687 if (deletions || insertions == 0) {
7f814632 1688 strbuf_addf(&sb,
218adaaa 1689 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
7f814632
NTND
1690 deletions);
1691 }
1692 strbuf_addch(&sb, '\n');
1693 ret = fputs(sb.buf, fp);
1694 strbuf_release(&sb);
1695 return ret;
1696}
1697
4b25d091 1698static void show_stats(struct diffstat_t *data, struct diff_options *options)
6973dcae 1699{
eb3a9dd3 1700 int i, len, add, del, adds = 0, dels = 0;
0974c117 1701 uintmax_t max_change = 0, max_len = 0;
dc801e71
ZJS
1702 int total_files = data->nr, count;
1703 int width, name_width, graph_width, number_width = 0, bin_width = 0;
c0aa335c 1704 const char *reset, *add_c, *del_c;
7be57610 1705 const char *line_prefix = "";
e5f85df8 1706 int extra_shown = 0;
6973dcae
JH
1707
1708 if (data->nr == 0)
1709 return;
1710
30997bb8 1711 line_prefix = diff_line_prefix(options);
808e1db2 1712 count = options->stat_count ? options->stat_count : data->nr;
a2540023 1713
8f67f8ae 1714 reset = diff_get_color_opt(options, DIFF_RESET);
8f67f8ae
PH
1715 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1716 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
785f7432 1717
1b058bc3
ZJS
1718 /*
1719 * Find the longest filename and max number of changes
1720 */
808e1db2 1721 for (i = 0; (i < count) && (i < data->nr); i++) {
6973dcae 1722 struct diffstat_file *file = data->files[i];
0974c117 1723 uintmax_t change = file->added + file->deleted;
af0ed819
JH
1724
1725 if (!file->is_interesting && (change == 0)) {
808e1db2 1726 count++; /* not shown == room for one more */
358e460e
MG
1727 continue;
1728 }
f604652e
JH
1729 fill_print_name(file);
1730 len = strlen(file->print_name);
6973dcae
JH
1731 if (max_len < len)
1732 max_len = len;
1733
dc801e71
ZJS
1734 if (file->is_unmerged) {
1735 /* "Unmerged" is 8 characters */
1736 bin_width = bin_width < 8 ? 8 : bin_width;
6973dcae 1737 continue;
dc801e71
ZJS
1738 }
1739 if (file->is_binary) {
1740 /* "Bin XXX -> YYY bytes" */
1741 int w = 14 + decimal_width(file->added)
1742 + decimal_width(file->deleted);
1743 bin_width = bin_width < w ? w : bin_width;
1744 /* Display change counts aligned with "Bin" */
1745 number_width = 3;
1746 continue;
1747 }
1748
a2540023
JH
1749 if (max_change < change)
1750 max_change = change;
6973dcae 1751 }
a20d3c0d 1752 count = i; /* where we can stop scanning in data->files[] */
6973dcae 1753
1b058bc3
ZJS
1754 /*
1755 * We have width = stat_width or term_columns() columns total.
1756 * We want a maximum of min(max_len, stat_name_width) for the name part.
969fe57b 1757 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1b058bc3
ZJS
1758 * We also need 1 for " " and 4 + decimal_width(max_change)
1759 * for " | NNNN " and one the empty column at the end, altogether
1760 * 6 + decimal_width(max_change).
1761 *
1762 * If there's not enough space, we will use the smaller of
1763 * stat_name_width (if set) and 5/8*width for the filename,
969fe57b
ZJS
1764 * and the rest for constant elements + graph part, but no more
1765 * than stat_graph_width for the graph part.
1b058bc3
ZJS
1766 * (5/8 gives 50 for filename and 30 for the constant parts + graph
1767 * for the standard terminal size).
a2540023 1768 *
1b058bc3
ZJS
1769 * In other words: stat_width limits the maximum width, and
1770 * stat_name_width fixes the maximum width of the filename,
1771 * and is also used to divide available columns if there
1772 * aren't enough.
dc801e71
ZJS
1773 *
1774 * Binary files are displayed with "Bin XXX -> YYY bytes"
1775 * instead of the change count and graph. This part is treated
1776 * similarly to the graph part, except that it is not
41ccfdd9 1777 * "scaled". If total width is too small to accommodate the
dc801e71
ZJS
1778 * guaranteed minimum width of the filename part and the
1779 * separators and this message, this message will "overflow"
1780 * making the line longer than the maximum width.
a2540023 1781 */
1b058bc3
ZJS
1782
1783 if (options->stat_width == -1)
cd48dadb 1784 width = term_columns() - strlen(line_prefix);
a2540023 1785 else
1b058bc3 1786 width = options->stat_width ? options->stat_width : 80;
dc801e71
ZJS
1787 number_width = decimal_width(max_change) > number_width ?
1788 decimal_width(max_change) : number_width;
a2540023 1789
df44483a
ZJS
1790 if (options->stat_graph_width == -1)
1791 options->stat_graph_width = diff_stat_graph_width;
a2540023 1792
1b058bc3
ZJS
1793 /*
1794 * Guarantee 3/8*16==6 for the graph part
1795 * and 5/8*16==10 for the filename part
1796 */
1797 if (width < 16 + 6 + number_width)
1798 width = 16 + 6 + number_width;
1799
1800 /*
1801 * First assign sizes that are wanted, ignoring available width.
dc801e71
ZJS
1802 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1803 * starting from "XXX" should fit in graph_width.
1b058bc3 1804 */
dc801e71
ZJS
1805 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1806 if (options->stat_graph_width &&
1807 options->stat_graph_width < graph_width)
1808 graph_width = options->stat_graph_width;
1809
1b058bc3
ZJS
1810 name_width = (options->stat_name_width > 0 &&
1811 options->stat_name_width < max_len) ?
1812 options->stat_name_width : max_len;
1813
1814 /*
1815 * Adjust adjustable widths not to exceed maximum width
1816 */
1817 if (name_width + number_width + 6 + graph_width > width) {
678c5741 1818 if (graph_width > width * 3/8 - number_width - 6) {
1b058bc3 1819 graph_width = width * 3/8 - number_width - 6;
678c5741
LP
1820 if (graph_width < 6)
1821 graph_width = 6;
1822 }
1823
969fe57b
ZJS
1824 if (options->stat_graph_width &&
1825 graph_width > options->stat_graph_width)
1826 graph_width = options->stat_graph_width;
1b058bc3
ZJS
1827 if (name_width > width - number_width - 6 - graph_width)
1828 name_width = width - number_width - 6 - graph_width;
1829 else
1830 graph_width = width - number_width - 6 - name_width;
1831 }
1832
1833 /*
1834 * From here name_width is the width of the name area,
1835 * and graph_width is the width of the graph area.
1836 * max_change is used to scale graph properly.
1837 */
808e1db2 1838 for (i = 0; i < count; i++) {
d2543b8e 1839 const char *prefix = "";
af0ed819
JH
1840 struct diffstat_file *file = data->files[i];
1841 char *name = file->print_name;
1842 uintmax_t added = file->added;
1843 uintmax_t deleted = file->deleted;
a2540023 1844 int name_len;
6973dcae 1845
a20d3c0d 1846 if (!file->is_interesting && (added + deleted == 0))
358e460e 1847 continue;
a20d3c0d 1848
6973dcae
JH
1849 /*
1850 * "scale" the filename
1851 */
a2540023
JH
1852 len = name_width;
1853 name_len = strlen(name);
1854 if (name_width < name_len) {
6973dcae
JH
1855 char *slash;
1856 prefix = "...";
a2540023
JH
1857 len -= 3;
1858 name += name_len - len;
6973dcae
JH
1859 slash = strchr(name, '/');
1860 if (slash)
1861 name = slash;
1862 }
6973dcae 1863
af0ed819 1864 if (file->is_binary) {
7be57610 1865 fprintf(options->file, "%s", line_prefix);
a408e0e6 1866 show_name(options->file, prefix, name, len);
fc1320bf 1867 fprintf(options->file, " %*s", number_width, "Bin");
e18872b2
ZJS
1868 if (!added && !deleted) {
1869 putc('\n', options->file);
1870 continue;
1871 }
1872 fprintf(options->file, " %s%"PRIuMAX"%s",
0974c117 1873 del_c, deleted, reset);
c0c77734 1874 fprintf(options->file, " -> ");
0974c117
JK
1875 fprintf(options->file, "%s%"PRIuMAX"%s",
1876 add_c, added, reset);
c0c77734
DB
1877 fprintf(options->file, " bytes");
1878 fprintf(options->file, "\n");
f604652e 1879 continue;
6973dcae 1880 }
af0ed819 1881 else if (file->is_unmerged) {
7be57610 1882 fprintf(options->file, "%s", line_prefix);
a408e0e6 1883 show_name(options->file, prefix, name, len);
dc801e71 1884 fprintf(options->file, " Unmerged\n");
f604652e 1885 continue;
6973dcae 1886 }
6973dcae 1887
a2540023
JH
1888 /*
1889 * scale the add/delete
1890 */
6973dcae
JH
1891 add = added;
1892 del = deleted;
6973dcae 1893
1b058bc3 1894 if (graph_width <= max_change) {
d3c9cf32 1895 int total = scale_linear(add + del, graph_width, max_change);
2eeeef24
JH
1896 if (total < 2 && add && del)
1897 /* width >= 2 due to the sanity check */
1898 total = 2;
1899 if (add < del) {
1b058bc3 1900 add = scale_linear(add, graph_width, max_change);
2eeeef24
JH
1901 del = total - add;
1902 } else {
1b058bc3 1903 del = scale_linear(del, graph_width, max_change);
2eeeef24
JH
1904 add = total - del;
1905 }
6973dcae 1906 }
7be57610 1907 fprintf(options->file, "%s", line_prefix);
a408e0e6 1908 show_name(options->file, prefix, name, len);
dc801e71
ZJS
1909 fprintf(options->file, " %*"PRIuMAX"%s",
1910 number_width, added + deleted,
1911 added + deleted ? " " : "");
c0c77734
DB
1912 show_graph(options->file, '+', add, add_c, reset);
1913 show_graph(options->file, '-', del, del_c, reset);
1914 fprintf(options->file, "\n");
1915 }
a20d3c0d
JH
1916
1917 for (i = 0; i < data->nr; i++) {
af0ed819
JH
1918 struct diffstat_file *file = data->files[i];
1919 uintmax_t added = file->added;
1920 uintmax_t deleted = file->deleted;
82dfc2c4
JH
1921
1922 if (file->is_unmerged ||
1923 (!file->is_interesting && (added + deleted == 0))) {
808e1db2
MG
1924 total_files--;
1925 continue;
1926 }
a20d3c0d 1927
82dfc2c4 1928 if (!file->is_binary) {
a20d3c0d
JH
1929 adds += added;
1930 dels += deleted;
1931 }
1932 if (i < count)
1933 continue;
e5f85df8
JH
1934 if (!extra_shown)
1935 fprintf(options->file, "%s ...\n", line_prefix);
1936 extra_shown = 1;
808e1db2 1937 }
7be57610 1938 fprintf(options->file, "%s", line_prefix);
7f814632 1939 print_stat_summary(options->file, total_files, adds, dels);
6973dcae
JH
1940}
1941
2775d92c 1942static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
ebd124c6
NP
1943{
1944 int i, adds = 0, dels = 0, total_files = data->nr;
1945
1946 if (data->nr == 0)
1947 return;
1948
1949 for (i = 0; i < data->nr; i++) {
e18872b2
ZJS
1950 int added = data->files[i]->added;
1951 int deleted= data->files[i]->deleted;
1952
20c8cde4
JH
1953 if (data->files[i]->is_unmerged ||
1954 (!data->files[i]->is_interesting && (added + deleted == 0))) {
e18872b2 1955 total_files--;
de9658b5 1956 } else if (!data->files[i]->is_binary) { /* don't count bytes */
e18872b2
ZJS
1957 adds += added;
1958 dels += deleted;
ebd124c6 1959 }
ebd124c6 1960 }
30997bb8 1961 fprintf(options->file, "%s", diff_line_prefix(options));
7f814632 1962 print_stat_summary(options->file, total_files, adds, dels);
ebd124c6
NP
1963}
1964
4b25d091 1965static void show_numstat(struct diffstat_t *data, struct diff_options *options)
74e2abe5
JH
1966{
1967 int i;
1968
f604652e
JH
1969 if (data->nr == 0)
1970 return;
1971
74e2abe5
JH
1972 for (i = 0; i < data->nr; i++) {
1973 struct diffstat_file *file = data->files[i];
1974
30997bb8 1975 fprintf(options->file, "%s", diff_line_prefix(options));
7be57610 1976
bfddbc5e 1977 if (file->is_binary)
c0c77734 1978 fprintf(options->file, "-\t-\t");
bfddbc5e 1979 else
c0c77734 1980 fprintf(options->file,
0974c117
JK
1981 "%"PRIuMAX"\t%"PRIuMAX"\t",
1982 file->added, file->deleted);
f604652e
JH
1983 if (options->line_termination) {
1984 fill_print_name(file);
1985 if (!file->is_renamed)
c0c77734 1986 write_name_quoted(file->name, options->file,
f604652e
JH
1987 options->line_termination);
1988 else {
c0c77734
DB
1989 fputs(file->print_name, options->file);
1990 putc(options->line_termination, options->file);
f604652e 1991 }
663af342 1992 } else {
f604652e 1993 if (file->is_renamed) {
c0c77734
DB
1994 putc('\0', options->file);
1995 write_name_quoted(file->from_name, options->file, '\0');
f604652e 1996 }
c0c77734 1997 write_name_quoted(file->name, options->file, '\0');
663af342 1998 }
74e2abe5
JH
1999 }
2000}
2001
c04a7155
JH
2002struct dirstat_file {
2003 const char *name;
2004 unsigned long changed;
7df7c019
LT
2005};
2006
c04a7155
JH
2007struct dirstat_dir {
2008 struct dirstat_file *files;
712d2c7d 2009 int alloc, nr, permille, cumulative;
c04a7155
JH
2010};
2011
7be57610
BY
2012static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2013 unsigned long changed, const char *base, int baselen)
7df7c019
LT
2014{
2015 unsigned long this_dir = 0;
2016 unsigned int sources = 0;
30997bb8 2017 const char *line_prefix = diff_line_prefix(opt);
7df7c019
LT
2018
2019 while (dir->nr) {
c04a7155 2020 struct dirstat_file *f = dir->files;
7df7c019
LT
2021 int namelen = strlen(f->name);
2022 unsigned long this;
2023 char *slash;
2024
2025 if (namelen < baselen)
2026 break;
2027 if (memcmp(f->name, base, baselen))
2028 break;
2029 slash = strchr(f->name + baselen, '/');
2030 if (slash) {
2031 int newbaselen = slash + 1 - f->name;
7be57610 2032 this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
7df7c019
LT
2033 sources++;
2034 } else {
c04a7155 2035 this = f->changed;
7df7c019
LT
2036 dir->files++;
2037 dir->nr--;
2038 sources += 2;
2039 }
2040 this_dir += this;
2041 }
2042
2043 /*
2044 * We don't report dirstat's for
2045 * - the top level
2046 * - or cases where everything came from a single directory
2047 * under this directory (sources == 1).
2048 */
2049 if (baselen && sources != 1) {
58a8756a
JH
2050 if (this_dir) {
2051 int permille = this_dir * 1000 / changed;
712d2c7d 2052 if (permille >= dir->permille) {
7be57610 2053 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
712d2c7d 2054 permille / 10, permille % 10, baselen, base);
7df7c019
LT
2055 if (!dir->cumulative)
2056 return 0;
2057 }
2058 }
2059 }
2060 return this_dir;
2061}
2062
441bca0b
LT
2063static int dirstat_compare(const void *_a, const void *_b)
2064{
2065 const struct dirstat_file *a = _a;
2066 const struct dirstat_file *b = _b;
2067 return strcmp(a->name, b->name);
2068}
2069
c04a7155 2070static void show_dirstat(struct diff_options *options)
7df7c019
LT
2071{
2072 int i;
2073 unsigned long changed;
c04a7155
JH
2074 struct dirstat_dir dir;
2075 struct diff_queue_struct *q = &diff_queued_diff;
2076
2077 dir.files = NULL;
2078 dir.alloc = 0;
2079 dir.nr = 0;
712d2c7d 2080 dir.permille = options->dirstat_permille;
f88d225f 2081 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
7df7c019 2082
7df7c019 2083 changed = 0;
c04a7155
JH
2084 for (i = 0; i < q->nr; i++) {
2085 struct diff_filepair *p = q->queue[i];
2086 const char *name;
2087 unsigned long copied, added, damage;
0133dab7 2088 int content_changed;
c04a7155 2089
2ca86714 2090 name = p->two->path ? p->two->path : p->one->path;
c04a7155 2091
41c9560e 2092 if (p->one->oid_valid && p->two->oid_valid)
a0d12c44 2093 content_changed = oidcmp(&p->one->oid, &p->two->oid);
0133dab7
JH
2094 else
2095 content_changed = 1;
2096
2ff3a803
JH
2097 if (!content_changed) {
2098 /*
2099 * The SHA1 has not changed, so pre-/post-content is
2100 * identical. We can therefore skip looking at the
2101 * file contents altogether.
2102 */
2103 damage = 0;
2104 goto found_damage;
2105 }
2106
0133dab7
JH
2107 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2108 /*
2109 * In --dirstat-by-file mode, we don't really need to
2110 * look at the actual file contents at all.
2111 * The fact that the SHA1 changed is enough for us to
2112 * add this file to the list of results
2113 * (with each file contributing equal damage).
2114 */
2ff3a803 2115 damage = 1;
0133dab7
JH
2116 goto found_damage;
2117 }
c04a7155
JH
2118
2119 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2120 diff_populate_filespec(p->one, 0);
2121 diff_populate_filespec(p->two, 0);
974e0044 2122 diffcore_count_changes(p->one, p->two, NULL, NULL,
c04a7155
JH
2123 &copied, &added);
2124 diff_free_filespec_data(p->one);
2125 diff_free_filespec_data(p->two);
2126 } else if (DIFF_FILE_VALID(p->one)) {
8e5dd3d6 2127 diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
c04a7155
JH
2128 copied = added = 0;
2129 diff_free_filespec_data(p->one);
2130 } else if (DIFF_FILE_VALID(p->two)) {
8e5dd3d6 2131 diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
c04a7155
JH
2132 copied = 0;
2133 added = p->two->size;
2134 diff_free_filespec_data(p->two);
2135 } else
2b0b551d 2136 continue;
c04a7155
JH
2137
2138 /*
2139 * Original minus copied is the removed material,
2140 * added is the new material. They are both damages
0133dab7 2141 * made to the preimage.
2ff3a803
JH
2142 * If the resulting damage is zero, we know that
2143 * diffcore_count_changes() considers the two entries to
2144 * be identical, but since content_changed is true, we
2145 * know that there must have been _some_ kind of change,
2146 * so we force all entries to have damage > 0.
c04a7155
JH
2147 */
2148 damage = (p->one->size - copied) + added;
2ff3a803 2149 if (!damage)
fd33777b 2150 damage = 1;
c04a7155 2151
0133dab7 2152found_damage:
c04a7155
JH
2153 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2154 dir.files[dir.nr].name = name;
2155 dir.files[dir.nr].changed = damage;
2156 changed += damage;
2157 dir.nr++;
7df7c019
LT
2158 }
2159
2160 /* This can happen even with many files, if everything was renames */
2161 if (!changed)
2162 return;
2163
2164 /* Show all directories with more than x% of the changes */
9ed0d8d6 2165 QSORT(dir.files, dir.nr, dirstat_compare);
7be57610 2166 gather_dirstat(options, &dir, changed, "", 0);
7df7c019
LT
2167}
2168
1c57a627
JH
2169static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2170{
2171 int i;
2172 unsigned long changed;
2173 struct dirstat_dir dir;
2174
2175 if (data->nr == 0)
2176 return;
2177
2178 dir.files = NULL;
2179 dir.alloc = 0;
2180 dir.nr = 0;
2181 dir.permille = options->dirstat_permille;
2182 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2183
2184 changed = 0;
2185 for (i = 0; i < data->nr; i++) {
2186 struct diffstat_file *file = data->files[i];
2187 unsigned long damage = file->added + file->deleted;
2188 if (file->is_binary)
2189 /*
2190 * binary files counts bytes, not lines. Must find some
2191 * way to normalize binary bytes vs. textual lines.
2192 * The following heuristic assumes that there are 64
2193 * bytes per "line".
2194 * This is stupid and ugly, but very cheap...
2195 */
2196 damage = (damage + 63) / 64;
2197 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2198 dir.files[dir.nr].name = file->name;
2199 dir.files[dir.nr].changed = damage;
2200 changed += damage;
2201 dir.nr++;
2202 }
2203
2204 /* This can happen even with many files, if everything was renames */
2205 if (!changed)
2206 return;
2207
2208 /* Show all directories with more than x% of the changes */
9ed0d8d6 2209 QSORT(dir.files, dir.nr, dirstat_compare);
1c57a627
JH
2210 gather_dirstat(options, &dir, changed, "", 0);
2211}
2212
f604652e
JH
2213static void free_diffstat_info(struct diffstat_t *diffstat)
2214{
2215 int i;
2216 for (i = 0; i < diffstat->nr; i++) {
2217 struct diffstat_file *f = diffstat->files[i];
2218 if (f->name != f->print_name)
2219 free(f->print_name);
2220 free(f->name);
2221 free(f->from_name);
2222 free(f);
2223 }
2224 free(diffstat->files);
2225}
2226
88246898 2227struct checkdiff_t {
88246898 2228 const char *filename;
1ba111d1 2229 int lineno;
a757c646 2230 int conflict_marker_size;
1ba111d1 2231 struct diff_options *o;
cf1b7869 2232 unsigned ws_rule;
62c64895 2233 unsigned status;
88246898
JS
2234};
2235
a757c646 2236static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
04954043
JH
2237{
2238 char firstchar;
2239 int cnt;
2240
a757c646 2241 if (len < marker_size + 1)
04954043
JH
2242 return 0;
2243 firstchar = line[0];
2244 switch (firstchar) {
a757c646 2245 case '=': case '>': case '<': case '|':
04954043
JH
2246 break;
2247 default:
2248 return 0;
2249 }
a757c646 2250 for (cnt = 1; cnt < marker_size; cnt++)
04954043
JH
2251 if (line[cnt] != firstchar)
2252 return 0;
a757c646
JH
2253 /* line[1] thru line[marker_size-1] are same as firstchar */
2254 if (len < marker_size + 1 || !isspace(line[marker_size]))
04954043 2255 return 0;
04954043
JH
2256 return 1;
2257}
2258
88246898
JS
2259static void checkdiff_consume(void *priv, char *line, unsigned long len)
2260{
2261 struct checkdiff_t *data = priv;
a757c646 2262 int marker_size = data->conflict_marker_size;
f1c96261
JK
2263 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2264 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2265 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
c1795bb0 2266 char *err;
30997bb8 2267 const char *line_prefix;
7be57610
BY
2268
2269 assert(data->o);
30997bb8 2270 line_prefix = diff_line_prefix(data->o);
88246898
JS
2271
2272 if (line[0] == '+') {
18374e58 2273 unsigned bad;
0ef617f4 2274 data->lineno++;
a757c646 2275 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
04954043
JH
2276 data->status |= 1;
2277 fprintf(data->o->file,
7be57610
BY
2278 "%s%s:%d: leftover conflict marker\n",
2279 line_prefix, data->filename, data->lineno);
04954043 2280 }
8f8841e9 2281 bad = ws_check(line + 1, len - 1, data->ws_rule);
18374e58 2282 if (!bad)
c1795bb0 2283 return;
18374e58
JH
2284 data->status |= bad;
2285 err = whitespace_error_string(bad);
7be57610
BY
2286 fprintf(data->o->file, "%s%s:%d: %s.\n",
2287 line_prefix, data->filename, data->lineno, err);
c1795bb0 2288 free(err);
a3c158d4 2289 emit_line(data->o, set, reset, line, 1);
8f8841e9 2290 ws_check_emit(line + 1, len - 1, data->ws_rule,
1ba111d1 2291 data->o->file, set, reset, ws);
877f23cc 2292 } else if (line[0] == ' ') {
88246898 2293 data->lineno++;
877f23cc 2294 } else if (line[0] == '@') {
88246898
JS
2295 char *plus = strchr(line, '+');
2296 if (plus)
0ef617f4 2297 data->lineno = strtol(plus, NULL, 10) - 1;
88246898
JS
2298 else
2299 die("invalid diff");
2300 }
2301}
2302
0660626c
JH
2303static unsigned char *deflate_it(char *data,
2304 unsigned long size,
2305 unsigned long *result_size)
051308f6 2306{
0660626c
JH
2307 int bound;
2308 unsigned char *deflated;
ef49a7a0 2309 git_zstream stream;
0660626c 2310
55bb5c91 2311 git_deflate_init(&stream, zlib_compression_level);
225a6f10 2312 bound = git_deflate_bound(&stream, size);
0660626c
JH
2313 deflated = xmalloc(bound);
2314 stream.next_out = deflated;
2315 stream.avail_out = bound;
2316
2317 stream.next_in = (unsigned char *)data;
2318 stream.avail_in = size;
55bb5c91 2319 while (git_deflate(&stream, Z_FINISH) == Z_OK)
0660626c 2320 ; /* nothing */
55bb5c91 2321 git_deflate_end(&stream);
0660626c
JH
2322 *result_size = stream.total_out;
2323 return deflated;
051308f6
JH
2324}
2325
32b367e4
JK
2326static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2327 const char *prefix)
051308f6 2328{
0660626c
JH
2329 void *cp;
2330 void *delta;
2331 void *deflated;
2332 void *data;
2333 unsigned long orig_size;
2334 unsigned long delta_size;
2335 unsigned long deflate_size;
2336 unsigned long data_size;
051308f6 2337
0660626c
JH
2338 /* We could do deflated delta, or we could do just deflated two,
2339 * whichever is smaller.
051308f6 2340 */
0660626c
JH
2341 delta = NULL;
2342 deflated = deflate_it(two->ptr, two->size, &deflate_size);
2343 if (one->size && two->size) {
2344 delta = diff_delta(one->ptr, one->size,
2345 two->ptr, two->size,
2346 &delta_size, deflate_size);
2347 if (delta) {
2348 void *to_free = delta;
2349 orig_size = delta_size;
2350 delta = deflate_it(delta, delta_size, &delta_size);
2351 free(to_free);
051308f6
JH
2352 }
2353 }
051308f6 2354
0660626c 2355 if (delta && delta_size < deflate_size) {
7be57610 2356 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
0660626c
JH
2357 free(deflated);
2358 data = delta;
2359 data_size = delta_size;
2360 }
2361 else {
7be57610 2362 fprintf(file, "%sliteral %lu\n", prefix, two->size);
0660626c
JH
2363 free(delta);
2364 data = deflated;
2365 data_size = deflate_size;
2366 }
051308f6 2367
0660626c
JH
2368 /* emit data encoded in base85 */
2369 cp = data;
2370 while (data_size) {
2371 int bytes = (52 < data_size) ? 52 : data_size;
051308f6 2372 char line[70];
0660626c 2373 data_size -= bytes;
051308f6
JH
2374 if (bytes <= 26)
2375 line[0] = bytes + 'A' - 1;
2376 else
2377 line[0] = bytes - 26 + 'a' - 1;
2378 encode_85(line + 1, cp, bytes);
1d7f171c 2379 cp = (char *) cp + bytes;
7be57610 2380 fprintf(file, "%s", prefix);
c0c77734
DB
2381 fputs(line, file);
2382 fputc('\n', file);
051308f6 2383 }
7be57610 2384 fprintf(file, "%s\n", prefix);
0660626c 2385 free(data);
051308f6
JH
2386}
2387
32b367e4
JK
2388static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2389 const char *prefix)
d4c452f0 2390{
7be57610
BY
2391 fprintf(file, "%sGIT binary patch\n", prefix);
2392 emit_binary_diff_body(file, one, two, prefix);
2393 emit_binary_diff_body(file, two, one, prefix);
d4c452f0
JH
2394}
2395
29a3eefd
JH
2396int diff_filespec_is_binary(struct diff_filespec *one)
2397{
122aa6f9
JK
2398 if (one->is_binary == -1) {
2399 diff_filespec_load_driver(one);
2400 if (one->driver->binary != -1)
2401 one->is_binary = one->driver->binary;
2402 else {
2403 if (!one->data && DIFF_FILE_VALID(one))
6bf3b813
NTND
2404 diff_populate_filespec(one, CHECK_BINARY);
2405 if (one->is_binary == -1 && one->data)
122aa6f9
JK
2406 one->is_binary = buffer_is_binary(one->data,
2407 one->size);
2408 if (one->is_binary == -1)
2409 one->is_binary = 0;
2410 }
2411 }
29a3eefd 2412 return one->is_binary;
6973dcae
JH
2413}
2414
be58e70d 2415static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
f258475a 2416{
122aa6f9
JK
2417 diff_filespec_load_driver(one);
2418 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
f258475a
JH
2419}
2420
a5a818ee
JH
2421void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2422{
2423 if (!options->a_prefix)
2424 options->a_prefix = a;
2425 if (!options->b_prefix)
2426 options->b_prefix = b;
2427}
2428
a788d7d5 2429struct userdiff_driver *get_textconv(struct diff_filespec *one)
04427ac8
JK
2430{
2431 if (!DIFF_FILE_VALID(one))
2432 return NULL;
d391c0ff 2433
04427ac8 2434 diff_filespec_load_driver(one);
3813e690 2435 return userdiff_get_textconv(one->driver);
04427ac8
JK
2436}
2437
6973dcae
JH
2438static void builtin_diff(const char *name_a,
2439 const char *name_b,
2440 struct diff_filespec *one,
2441 struct diff_filespec *two,
2442 const char *xfrm_msg,
296c6bb2 2443 int must_show_header,
051308f6 2444 struct diff_options *o,
6973dcae
JH
2445 int complete_rewrite)
2446{
2447 mmfile_t mf1, mf2;
2448 const char *lbl[2];
2449 char *a_one, *b_two;
d9c552f1 2450 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
8f67f8ae 2451 const char *reset = diff_get_color_opt(o, DIFF_RESET);
a5a818ee 2452 const char *a_prefix, *b_prefix;
d9bae1a1
JK
2453 struct userdiff_driver *textconv_one = NULL;
2454 struct userdiff_driver *textconv_two = NULL;
3e97c7c6 2455 struct strbuf header = STRBUF_INIT;
30997bb8 2456 const char *line_prefix = diff_line_prefix(o);
a5a818ee 2457
fd47ae6a
JK
2458 diff_set_mnemonic_prefix(o, "a/", "b/");
2459 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2460 a_prefix = o->b_prefix;
2461 b_prefix = o->a_prefix;
2462 } else {
2463 a_prefix = o->a_prefix;
2464 b_prefix = o->b_prefix;
2465 }
2466
61cfbc05
JK
2467 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
2468 (!one->mode || S_ISGITLINK(one->mode)) &&
2469 (!two->mode || S_ISGITLINK(two->mode))) {
752c0c24
JS
2470 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2471 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
3b0c18af 2472 show_submodule_summary(o->file, one->path ? one->path : two->path,
0f33a067 2473 line_prefix,
602a283a 2474 &one->oid, &two->oid,
a0d12c44 2475 two->dirty_submodule,
4e215131 2476 meta, del, add, reset);
752c0c24 2477 return;
fd47ae6a
JK
2478 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
2479 (!one->mode || S_ISGITLINK(one->mode)) &&
2480 (!two->mode || S_ISGITLINK(two->mode))) {
2481 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2482 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2483 show_submodule_inline_diff(o->file, one->path ? one->path : two->path,
2484 line_prefix,
2485 &one->oid, &two->oid,
2486 two->dirty_submodule,
2487 meta, del, add, reset, o);
2488 return;
752c0c24
JS
2489 }
2490
3aa1f7ca
JK
2491 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2492 textconv_one = get_textconv(one);
2493 textconv_two = get_textconv(two);
2494 }
2495
71b989e7
LT
2496 /* Never use a non-valid filename anywhere if at all possible */
2497 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2498 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2499
a5a818ee
JH
2500 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2501 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
6973dcae
JH
2502 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2503 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
d9c552f1 2504 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
6973dcae
JH
2505 if (lbl[0][0] == '/') {
2506 /* /dev/null */
d9c552f1 2507 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
37466447
BW
2508 if (xfrm_msg)
2509 strbuf_addstr(&header, xfrm_msg);
296c6bb2 2510 must_show_header = 1;
6973dcae
JH
2511 }
2512 else if (lbl[1][0] == '/') {
d9c552f1 2513 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
37466447
BW
2514 if (xfrm_msg)
2515 strbuf_addstr(&header, xfrm_msg);
296c6bb2 2516 must_show_header = 1;
6973dcae
JH
2517 }
2518 else {
2519 if (one->mode != two->mode) {
d9c552f1
JK
2520 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2521 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
296c6bb2 2522 must_show_header = 1;
cd112cef 2523 }
37466447
BW
2524 if (xfrm_msg)
2525 strbuf_addstr(&header, xfrm_msg);
3e97c7c6 2526
6973dcae
JH
2527 /*
2528 * we do not run diff between different kind
2529 * of objects.
2530 */
2531 if ((one->mode ^ two->mode) & S_IFMT)
2532 goto free_ab_and_return;
0c01857d 2533 if (complete_rewrite &&
3aa1f7ca
JK
2534 (textconv_one || !diff_filespec_is_binary(one)) &&
2535 (textconv_two || !diff_filespec_is_binary(two))) {
a29b0a13
SB
2536 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2537 header.buf, header.len, 0);
3e97c7c6 2538 strbuf_reset(&header);
3aa1f7ca
JK
2539 emit_rewrite_diff(name_a, name_b, one, two,
2540 textconv_one, textconv_two, o);
34a5e1a2 2541 o->found_changes = 1;
6973dcae
JH
2542 goto free_ab_and_return;
2543 }
2544 }
2545
467ddc14 2546 if (o->irreversible_delete && lbl[1][0] == '/') {
a29b0a13
SB
2547 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
2548 header.len, 0);
467ddc14
JH
2549 strbuf_reset(&header);
2550 goto free_ab_and_return;
2551 } else if (!DIFF_OPT_TST(o, TEXT) &&
b3373982
JK
2552 ( (!textconv_one && diff_filespec_is_binary(one)) ||
2553 (!textconv_two && diff_filespec_is_binary(two)) )) {
4acaaa7a 2554 struct strbuf sb = STRBUF_INIT;
1aaf69e6
NTND
2555 if (!one->data && !two->data &&
2556 S_ISREG(one->mode) && S_ISREG(two->mode) &&
2557 !DIFF_OPT_TST(o, BINARY)) {
a0d12c44 2558 if (!oidcmp(&one->oid, &two->oid)) {
1aaf69e6 2559 if (must_show_header)
a29b0a13
SB
2560 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2561 header.buf, header.len,
2562 0);
1aaf69e6
NTND
2563 goto free_ab_and_return;
2564 }
a29b0a13
SB
2565 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2566 header.buf, header.len, 0);
4acaaa7a
SB
2567 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
2568 diff_line_prefix(o), lbl[0], lbl[1]);
2569 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
2570 sb.buf, sb.len, 0);
2571 strbuf_release(&sb);
1aaf69e6
NTND
2572 goto free_ab_and_return;
2573 }
b3373982
JK
2574 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2575 die("unable to read files to diff");
0660626c
JH
2576 /* Quite common confusing case */
2577 if (mf1.size == mf2.size &&
296c6bb2
CC
2578 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2579 if (must_show_header)
a29b0a13
SB
2580 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2581 header.buf, header.len, 0);
0660626c 2582 goto free_ab_and_return;
296c6bb2 2583 }
a29b0a13 2584 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3e97c7c6 2585 strbuf_reset(&header);
8f67f8ae 2586 if (DIFF_OPT_TST(o, BINARY))
7be57610 2587 emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
4acaaa7a
SB
2588 else {
2589 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
2590 diff_line_prefix(o), lbl[0], lbl[1]);
2591 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
2592 sb.buf, sb.len, 0);
2593 strbuf_release(&sb);
2594 }
34a5e1a2 2595 o->found_changes = 1;
467ddc14 2596 } else {
6973dcae
JH
2597 /* Crazy xdl interfaces.. */
2598 const char *diffopts = getenv("GIT_DIFF_OPTS");
ae021d87 2599 const char *v;
6973dcae
JH
2600 xpparam_t xpp;
2601 xdemitconf_t xecfg;
6973dcae 2602 struct emit_callback ecbdata;
be58e70d 2603 const struct userdiff_funcname *pe;
f258475a 2604
b3f01ff2 2605 if (must_show_header) {
a29b0a13
SB
2606 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2607 header.buf, header.len, 0);
3e97c7c6
GB
2608 strbuf_reset(&header);
2609 }
2610
840383b2
JK
2611 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2612 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
04427ac8 2613
45e7ca0f
BC
2614 pe = diff_funcname_pattern(one);
2615 if (!pe)
2616 pe = diff_funcname_pattern(two);
6973dcae 2617
9ccd0a88 2618 memset(&xpp, 0, sizeof(xpp));
30b25010 2619 memset(&xecfg, 0, sizeof(xecfg));
cd112cef 2620 memset(&ecbdata, 0, sizeof(ecbdata));
6973dcae 2621 ecbdata.label_path = lbl;
daa0c3d9 2622 ecbdata.color_diff = want_color(o->use_color);
c189c4f2 2623 ecbdata.ws_rule = whitespace_rule(name_b);
690ed843 2624 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
d68fe26f 2625 check_blank_at_eof(&mf1, &mf2, &ecbdata);
a3c158d4 2626 ecbdata.opt = o;
3e97c7c6 2627 ecbdata.header = header.len ? &header : NULL;
582aa00b 2628 xpp.flags = o->xdl_opts;
ee1e5412 2629 xecfg.ctxlen = o->context;
6d0e674a 2630 xecfg.interhunkctxlen = o->interhunkcontext;
6973dcae 2631 xecfg.flags = XDL_EMIT_FUNCNAMES;
14937c2c
RS
2632 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2633 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
45e7ca0f 2634 if (pe)
a013585b 2635 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
6973dcae
JH
2636 if (!diffopts)
2637 ;
ae021d87
JK
2638 else if (skip_prefix(diffopts, "--unified=", &v))
2639 xecfg.ctxlen = strtoul(v, NULL, 10);
2640 else if (skip_prefix(diffopts, "-u", &v))
2641 xecfg.ctxlen = strtoul(v, NULL, 10);
77d1a520
TR
2642 if (o->word_diff)
2643 init_diff_words_data(&ecbdata, o, one, two);
3efb9880
JK
2644 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2645 &xpp, &xecfg))
2646 die("unable to generate diff for %s", one->path);
882749a0 2647 if (o->word_diff)
f59a59e2 2648 free_diff_words_data(&ecbdata);
04427ac8
JK
2649 if (textconv_one)
2650 free(mf1.ptr);
2651 if (textconv_two)
2652 free(mf2.ptr);
8cfe5f1c 2653 xdiff_clear_find_func(&xecfg);
6973dcae
JH
2654 }
2655
2656 free_ab_and_return:
3e97c7c6 2657 strbuf_release(&header);
fc3abdf5
JH
2658 diff_free_filespec_data(one);
2659 diff_free_filespec_data(two);
6973dcae
JH
2660 free(a_one);
2661 free(b_two);
2662 return;
2663}
2664
2665static void builtin_diffstat(const char *name_a, const char *name_b,
2666 struct diff_filespec *one,
2667 struct diff_filespec *two,
710158e3 2668 struct diffstat_t *diffstat,
0d21efa5 2669 struct diff_options *o,
74faaa16 2670 struct diff_filepair *p)
6973dcae
JH
2671{
2672 mmfile_t mf1, mf2;
2673 struct diffstat_file *data;
352ca4e1 2674 int same_contents;
74faaa16
LT
2675 int complete_rewrite = 0;
2676
2677 if (!DIFF_PAIR_UNMERGED(p)) {
2678 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2679 complete_rewrite = 1;
2680 }
6973dcae
JH
2681
2682 data = diffstat_add(diffstat, name_a, name_b);
99bfd407 2683 data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
6973dcae
JH
2684
2685 if (!one || !two) {
2686 data->is_unmerged = 1;
2687 return;
2688 }
ded0abc7 2689
a0d12c44 2690 same_contents = !oidcmp(&one->oid, &two->oid);
352ca4e1 2691
ded0abc7 2692 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
ded0abc7 2693 data->is_binary = 1;
352ca4e1 2694 if (same_contents) {
e18872b2
ZJS
2695 data->added = 0;
2696 data->deleted = 0;
2697 } else {
2698 data->added = diff_filespec_size(two);
2699 data->deleted = diff_filespec_size(one);
2700 }
ded0abc7
JK
2701 }
2702
2703 else if (complete_rewrite) {
710158e3
JH
2704 diff_populate_filespec(one, 0);
2705 diff_populate_filespec(two, 0);
2706 data->deleted = count_lines(one->data, one->size);
2707 data->added = count_lines(two->data, two->size);
710158e3 2708 }
6973dcae 2709
352ca4e1 2710 else if (!same_contents) {
6973dcae
JH
2711 /* Crazy xdl interfaces.. */
2712 xpparam_t xpp;
2713 xdemitconf_t xecfg;
6973dcae 2714
ded0abc7
JK
2715 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2716 die("unable to read files to diff");
2717
9ccd0a88 2718 memset(&xpp, 0, sizeof(xpp));
30b25010 2719 memset(&xecfg, 0, sizeof(xecfg));
582aa00b 2720 xpp.flags = o->xdl_opts;
f01cae91
JH
2721 xecfg.ctxlen = o->context;
2722 xecfg.interhunkctxlen = o->interhunkcontext;
3efb9880
JK
2723 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2724 &xpp, &xecfg))
2725 die("unable to generate diffstat for %s", one->path);
6973dcae 2726 }
fc3abdf5 2727
fc3abdf5
JH
2728 diff_free_filespec_data(one);
2729 diff_free_filespec_data(two);
6973dcae
JH
2730}
2731
88246898 2732static void builtin_checkdiff(const char *name_a, const char *name_b,
cd676a51 2733 const char *attr_path,
5ff10dd6
JH
2734 struct diff_filespec *one,
2735 struct diff_filespec *two,
2736 struct diff_options *o)
88246898
JS
2737{
2738 mmfile_t mf1, mf2;
2739 struct checkdiff_t data;
2740
2741 if (!two)
2742 return;
2743
2744 memset(&data, 0, sizeof(data));
88246898
JS
2745 data.filename = name_b ? name_b : name_a;
2746 data.lineno = 0;
1ba111d1 2747 data.o = o;
cd676a51 2748 data.ws_rule = whitespace_rule(attr_path);
a757c646 2749 data.conflict_marker_size = ll_merge_marker_size(attr_path);
88246898
JS
2750
2751 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2752 die("unable to read files to diff");
2753
5ff10dd6
JH
2754 /*
2755 * All the other codepaths check both sides, but not checking
2756 * the "old" side here is deliberate. We are checking the newly
2757 * introduced changes, and as long as the "new" side is text, we
2758 * can and should check what it introduces.
2759 */
29a3eefd 2760 if (diff_filespec_is_binary(two))
fc3abdf5 2761 goto free_and_return;
88246898
JS
2762 else {
2763 /* Crazy xdl interfaces.. */
2764 xpparam_t xpp;
2765 xdemitconf_t xecfg;
88246898 2766
9ccd0a88 2767 memset(&xpp, 0, sizeof(xpp));
30b25010 2768 memset(&xecfg, 0, sizeof(xecfg));
c35539eb 2769 xecfg.ctxlen = 1; /* at least one context line */
582aa00b 2770 xpp.flags = 0;
3efb9880
JK
2771 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2772 &xpp, &xecfg))
2773 die("unable to generate checkdiff for %s", one->path);
877f23cc 2774
467babf8 2775 if (data.ws_rule & WS_BLANK_AT_EOF) {
d68fe26f
JH
2776 struct emit_callback ecbdata;
2777 int blank_at_eof;
2778
2779 ecbdata.ws_rule = data.ws_rule;
2780 check_blank_at_eof(&mf1, &mf2, &ecbdata);
8837d335 2781 blank_at_eof = ecbdata.blank_at_eof_in_postimage;
d68fe26f 2782
467babf8
JH
2783 if (blank_at_eof) {
2784 static char *err;
2785 if (!err)
2786 err = whitespace_error_string(WS_BLANK_AT_EOF);
2787 fprintf(o->file, "%s:%d: %s.\n",
2788 data.filename, blank_at_eof, err);
2789 data.status = 1; /* report errors */
2790 }
877f23cc 2791 }
88246898 2792 }
fc3abdf5
JH
2793 free_and_return:
2794 diff_free_filespec_data(one);
2795 diff_free_filespec_data(two);
62c64895
WC
2796 if (data.status)
2797 DIFF_OPT_SET(o, CHECK_FAILED);
88246898
JS
2798}
2799
6973dcae
JH
2800struct diff_filespec *alloc_filespec(const char *path)
2801{
96ffc06f 2802 struct diff_filespec *spec;
6973dcae 2803
96ffc06f 2804 FLEXPTR_ALLOC_STR(spec, path, path);
9fb88419 2805 spec->count = 1;
122aa6f9 2806 spec->is_binary = -1;
6973dcae
JH
2807 return spec;
2808}
2809
9fb88419
LT
2810void free_filespec(struct diff_filespec *spec)
2811{
2812 if (!--spec->count) {
2813 diff_free_filespec_data(spec);
2814 free(spec);
2815 }
2816}
2817
f9704c2d
BW
2818void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
2819 int oid_valid, unsigned short mode)
6973dcae
JH
2820{
2821 if (mode) {
2822 spec->mode = canon_mode(mode);
f9704c2d
BW
2823 oidcpy(&spec->oid, oid);
2824 spec->oid_valid = oid_valid;
6973dcae
JH
2825 }
2826}
2827
2828/*
5adf317b 2829 * Given a name and sha1 pair, if the index tells us the file in
6973dcae
JH
2830 * the work tree has that object contents, return true, so that
2831 * prepare_temp_file() does not have to inflate and extract.
2832 */
fb4a1c0d 2833static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
6973dcae 2834{
9c5e6c80 2835 const struct cache_entry *ce;
6973dcae
JH
2836 struct stat st;
2837 int pos, len;
2838
150115ad
JH
2839 /*
2840 * We do not read the cache ourselves here, because the
6973dcae
JH
2841 * benchmark with my previous version that always reads cache
2842 * shows that it makes things worse for diff-tree comparing
2843 * two linux-2.6 kernel trees in an already checked out work
2844 * tree. This is because most diff-tree comparisons deal with
2845 * only a small number of files, while reading the cache is
2846 * expensive for a large project, and its cost outweighs the
2847 * savings we get by not inflating the object to a temporary
2848 * file. Practically, this code only helps when we are used
2849 * by diff-cache --cached, which does read the cache before
2850 * calling us.
2851 */
2852 if (!active_cache)
2853 return 0;
2854
1510fea7
SP
2855 /* We want to avoid the working directory if our caller
2856 * doesn't need the data in a normal file, this system
2857 * is rather slow with its stat/open/mmap/close syscalls,
2858 * and the object is contained in a pack file. The pack
2859 * is probably already open and will be faster to obtain
2860 * the data through than the working directory. Loose
2861 * objects however would tend to be slower as they need
2862 * to be individually opened and inflated.
2863 */
fb4a1c0d 2864 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
1510fea7
SP
2865 return 0;
2866
06dec439
JK
2867 /*
2868 * Similarly, if we'd have to convert the file contents anyway, that
2869 * makes the optimization not worthwhile.
2870 */
82b474e0 2871 if (!want_file && would_convert_to_git(&the_index, name))
06dec439
JK
2872 return 0;
2873
6973dcae
JH
2874 len = strlen(name);
2875 pos = cache_name_pos(name, len);
2876 if (pos < 0)
2877 return 0;
2878 ce = active_cache[pos];
eadb5831
JH
2879
2880 /*
2881 * This is not the sha1 we are looking for, or
2882 * unreusable because it is not a regular file.
2883 */
fb4a1c0d 2884 if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
6973dcae 2885 return 0;
eadb5831 2886
150115ad
JH
2887 /*
2888 * If ce is marked as "assume unchanged", there is no
2889 * guarantee that work tree matches what we are looking for.
2890 */
b4d1690d 2891 if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
150115ad
JH
2892 return 0;
2893
eadb5831
JH
2894 /*
2895 * If ce matches the file in the work tree, we can reuse it.
6973dcae 2896 */
eadb5831
JH
2897 if (ce_uptodate(ce) ||
2898 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2899 return 1;
2900
2901 return 0;
6973dcae
JH
2902}
2903
04786756
LT
2904static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2905{
b1ddfb91
JK
2906 struct strbuf buf = STRBUF_INIT;
2907 char *dirty = "";
8e08b419
JH
2908
2909 /* Are we looking at the work tree? */
85adbf2f 2910 if (s->dirty_submodule)
8e08b419
JH
2911 dirty = "-dirty";
2912
a0d12c44 2913 strbuf_addf(&buf, "Subproject commit %s%s\n",
2914 oid_to_hex(&s->oid), dirty);
b1ddfb91 2915 s->size = buf.len;
04786756
LT
2916 if (size_only) {
2917 s->data = NULL;
b1ddfb91
JK
2918 strbuf_release(&buf);
2919 } else {
2920 s->data = strbuf_detach(&buf, NULL);
2921 s->should_free = 1;
04786756
LT
2922 }
2923 return 0;
2924}
2925
6973dcae
JH
2926/*
2927 * While doing rename detection and pickaxe operation, we may need to
2928 * grab the data for the blob (or file) for our own in-core comparison.
2929 * diff_filespec has data and size fields for this purpose.
2930 */
8e5dd3d6 2931int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
6973dcae 2932{
8e5dd3d6 2933 int size_only = flags & CHECK_SIZE_ONLY;
6973dcae 2934 int err = 0;
5430bb28
JH
2935 /*
2936 * demote FAIL to WARN to allow inspecting the situation
2937 * instead of refusing.
2938 */
2939 enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2940 ? SAFE_CRLF_WARN
2941 : safe_crlf);
2942
6973dcae
JH
2943 if (!DIFF_FILE_VALID(s))
2944 die("internal error: asking to populate invalid file.");
2945 if (S_ISDIR(s->mode))
2946 return -1;
2947
6973dcae 2948 if (s->data)
fc3abdf5 2949 return 0;
04786756 2950
6e0b8ed6
JH
2951 if (size_only && 0 < s->size)
2952 return 0;
2953
302b9282 2954 if (S_ISGITLINK(s->mode))
04786756
LT
2955 return diff_populate_gitlink(s, size_only);
2956
41c9560e 2957 if (!s->oid_valid ||
fb4a1c0d 2958 reuse_worktree_file(s->path, &s->oid, 0)) {
f285a2d7 2959 struct strbuf buf = STRBUF_INIT;
6973dcae
JH
2960 struct stat st;
2961 int fd;
6c510bee 2962
6973dcae
JH
2963 if (lstat(s->path, &st) < 0) {
2964 if (errno == ENOENT) {
2965 err_empty:
2966 err = -1;
2967 empty:
d2543b8e 2968 s->data = (char *)"";
6973dcae
JH
2969 s->size = 0;
2970 return err;
2971 }
2972 }
dc49cd76 2973 s->size = xsize_t(st.st_size);
6973dcae
JH
2974 if (!s->size)
2975 goto empty;
6973dcae 2976 if (S_ISLNK(st.st_mode)) {
cf219d8c
LT
2977 struct strbuf sb = STRBUF_INIT;
2978
2979 if (strbuf_readlink(&sb, s->path, s->size))
6973dcae 2980 goto err_empty;
0956a6db
RS
2981 s->size = sb.len;
2982 s->data = strbuf_detach(&sb, NULL);
cf219d8c 2983 s->should_free = 1;
6973dcae
JH
2984 return 0;
2985 }
12426e11
JH
2986
2987 /*
2988 * Even if the caller would be happy with getting
2989 * only the size, we cannot return early at this
2990 * point if the path requires us to run the content
2991 * conversion.
2992 */
82b474e0 2993 if (size_only && !would_convert_to_git(&the_index, s->path))
cf219d8c 2994 return 0;
12426e11
JH
2995
2996 /*
2997 * Note: this check uses xsize_t(st.st_size) that may
2998 * not be the true size of the blob after it goes
2999 * through convert_to_git(). This may not strictly be
3000 * correct, but the whole point of big_file_threshold
3001 * and is_binary check being that we want to avoid
3002 * opening the file and inspecting the contents, this
3003 * is probably fine.
3004 */
6bf3b813
NTND
3005 if ((flags & CHECK_BINARY) &&
3006 s->size > big_file_threshold && s->is_binary == -1) {
3007 s->is_binary = 1;
3008 return 0;
3009 }
6973dcae
JH
3010 fd = open(s->path, O_RDONLY);
3011 if (fd < 0)
3012 goto err_empty;
c4712e45 3013 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
6973dcae 3014 close(fd);
6973dcae 3015 s->should_munmap = 1;
6c510bee
LT
3016
3017 /*
3018 * Convert from working tree format to canonical git format
3019 */
82b474e0 3020 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, crlf_warn)) {
c32f749f 3021 size_t size = 0;
6c510bee
LT
3022 munmap(s->data, s->size);
3023 s->should_munmap = 0;
c32f749f
RS
3024 s->data = strbuf_detach(&buf, &size);
3025 s->size = size;
6c510bee
LT
3026 s->should_free = 1;
3027 }
6973dcae
JH
3028 }
3029 else {
21666f1a 3030 enum object_type type;
6bf3b813 3031 if (size_only || (flags & CHECK_BINARY)) {
a0d12c44 3032 type = sha1_object_info(s->oid.hash, &s->size);
c50c4316 3033 if (type < 0)
a0d12c44 3034 die("unable to read %s",
3035 oid_to_hex(&s->oid));
6bf3b813
NTND
3036 if (size_only)
3037 return 0;
3038 if (s->size > big_file_threshold && s->is_binary == -1) {
3039 s->is_binary = 1;
3040 return 0;
3041 }
6973dcae 3042 }
a0d12c44 3043 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
6bf3b813 3044 if (!s->data)
a0d12c44 3045 die("unable to read %s", oid_to_hex(&s->oid));
6bf3b813 3046 s->should_free = 1;
6973dcae
JH
3047 }
3048 return 0;
3049}
3050
8ae92e63 3051void diff_free_filespec_blob(struct diff_filespec *s)
6973dcae
JH
3052{
3053 if (s->should_free)
3054 free(s->data);
3055 else if (s->should_munmap)
3056 munmap(s->data, s->size);
fc3abdf5
JH
3057
3058 if (s->should_free || s->should_munmap) {
3059 s->should_free = s->should_munmap = 0;
3060 s->data = NULL;
3061 }
eede7b7d
JK
3062}
3063
3064void diff_free_filespec_data(struct diff_filespec *s)
3065{
8ae92e63 3066 diff_free_filespec_blob(s);
6a83d902 3067 FREE_AND_NULL(s->cnt_data);
6973dcae
JH
3068}
3069
4e218f54 3070static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
6973dcae
JH
3071 void *blob,
3072 unsigned long size,
09bdff29 3073 const struct object_id *oid,
6973dcae
JH
3074 int mode)
3075{
3076 int fd;
4e218f54 3077 struct strbuf buf = STRBUF_INIT;
003b33a8
DA
3078 struct strbuf template = STRBUF_INIT;
3079 char *path_dup = xstrdup(path);
3080 const char *base = basename(path_dup);
6973dcae 3081
003b33a8
DA
3082 /* Generate "XXXXXX_basename.ext" */
3083 strbuf_addstr(&template, "XXXXXX_");
3084 strbuf_addstr(&template, base);
3085
284098f1 3086 fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
6973dcae 3087 if (fd < 0)
d824cbba 3088 die_errno("unable to create temp-file");
4e218f54
JS
3089 if (convert_to_working_tree(path,
3090 (const char *)blob, (size_t)size, &buf)) {
3091 blob = buf.buf;
3092 size = buf.len;
3093 }
93822c22 3094 if (write_in_full(fd, blob, size) != size)
0721c314 3095 die_errno("unable to write temp-file");
284098f1
MH
3096 close_tempfile(&temp->tempfile);
3097 temp->name = get_tempfile_path(&temp->tempfile);
09bdff29 3098 oid_to_hex_r(temp->hex, oid);
5096d490 3099 xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4e218f54 3100 strbuf_release(&buf);
003b33a8
DA
3101 strbuf_release(&template);
3102 free(path_dup);
6973dcae
JH
3103}
3104
479b0ae8
JK
3105static struct diff_tempfile *prepare_temp_file(const char *name,
3106 struct diff_filespec *one)
6973dcae 3107{
479b0ae8
JK
3108 struct diff_tempfile *temp = claim_diff_tempfile();
3109
6973dcae
JH
3110 if (!DIFF_FILE_VALID(one)) {
3111 not_a_valid_file:
3112 /* A '-' entry produces this for file-2, and
3113 * a '+' entry produces this for file-1.
3114 */
3115 temp->name = "/dev/null";
5096d490
JK
3116 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3117 xsnprintf(temp->mode, sizeof(temp->mode), ".");
479b0ae8
JK
3118 return temp;
3119 }
3120
aba47272 3121 if (!S_ISGITLINK(one->mode) &&
41c9560e 3122 (!one->oid_valid ||
fb4a1c0d 3123 reuse_worktree_file(name, &one->oid, 1))) {
6973dcae
JH
3124 struct stat st;
3125 if (lstat(name, &st) < 0) {
3126 if (errno == ENOENT)
3127 goto not_a_valid_file;
d824cbba 3128 die_errno("stat(%s)", name);
6973dcae
JH
3129 }
3130 if (S_ISLNK(st.st_mode)) {
3cd7388d
JK
3131 struct strbuf sb = STRBUF_INIT;
3132 if (strbuf_readlink(&sb, name, st.st_size) < 0)
0721c314 3133 die_errno("readlink(%s)", name);
3cd7388d 3134 prep_temp_blob(name, temp, sb.buf, sb.len,
41c9560e 3135 (one->oid_valid ?
09bdff29 3136 &one->oid : &null_oid),
41c9560e 3137 (one->oid_valid ?
6973dcae 3138 one->mode : S_IFLNK));
3cd7388d 3139 strbuf_release(&sb);
6973dcae
JH
3140 }
3141 else {
3142 /* we can borrow from the file in the work tree */
3143 temp->name = name;
41c9560e 3144 if (!one->oid_valid)
74014152 3145 oid_to_hex_r(temp->hex, &null_oid);
6973dcae 3146 else
2490574d 3147 oid_to_hex_r(temp->hex, &one->oid);
6973dcae
JH
3148 /* Even though we may sometimes borrow the
3149 * contents from the work tree, we always want
3150 * one->mode. mode is trustworthy even when
74014152 3151 * !(one->oid_valid), as long as
6973dcae
JH
3152 * DIFF_FILE_VALID(one).
3153 */
5096d490 3154 xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
6973dcae 3155 }
479b0ae8 3156 return temp;
6973dcae
JH
3157 }
3158 else {
3159 if (diff_populate_filespec(one, 0))
3160 die("cannot read data blob for %s", one->path);
4e218f54 3161 prep_temp_blob(name, temp, one->data, one->size,
09bdff29 3162 &one->oid, one->mode);
6973dcae 3163 }
479b0ae8 3164 return temp;
6973dcae
JH
3165}
3166
f3efe787
JK
3167static void add_external_diff_name(struct argv_array *argv,
3168 const char *name,
3169 struct diff_filespec *df)
3170{
3171 struct diff_tempfile *temp = prepare_temp_file(name, df);
3172 argv_array_push(argv, temp->name);
3173 argv_array_push(argv, temp->hex);
3174 argv_array_push(argv, temp->mode);
3175}
3176
6973dcae
JH
3177/* An external diff command takes:
3178 *
3179 * diff-cmd name infile1 infile1-sha1 infile1-mode \
3180 * infile2 infile2-sha1 infile2-mode [ rename-to ]
3181 *
3182 */
3183static void run_external_diff(const char *pgm,
3184 const char *name,
3185 const char *other,
3186 struct diff_filespec *one,
3187 struct diff_filespec *two,
3188 const char *xfrm_msg,
ee7fb0b1
ZK
3189 int complete_rewrite,
3190 struct diff_options *o)
6973dcae 3191{
82fbf269 3192 struct argv_array argv = ARGV_ARRAY_INIT;
ae049c95 3193 struct argv_array env = ARGV_ARRAY_INIT;
ee7fb0b1 3194 struct diff_queue_struct *q = &diff_queued_diff;
6973dcae 3195
0d4217d9
JK
3196 argv_array_push(&argv, pgm);
3197 argv_array_push(&argv, name);
6973dcae 3198
6973dcae 3199 if (one && two) {
f3efe787
JK
3200 add_external_diff_name(&argv, name, one);
3201 if (!other)
3202 add_external_diff_name(&argv, name, two);
3203 else {
3204 add_external_diff_name(&argv, other, two);
82fbf269
JK
3205 argv_array_push(&argv, other);
3206 argv_array_push(&argv, xfrm_msg);
6973dcae 3207 }
6973dcae 3208 }
ee7fb0b1 3209
ae049c95
JK
3210 argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3211 argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
ee7fb0b1 3212
89294d14
JK
3213 if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3214 die(_("external diff died, stopping at %s"), name);
ee7fb0b1 3215
6973dcae 3216 remove_tempfile();
82fbf269 3217 argv_array_clear(&argv);
ae049c95 3218 argv_array_clear(&env);
6973dcae
JH
3219}
3220
b67b9612
JH
3221static int similarity_index(struct diff_filepair *p)
3222{
3223 return p->score * 100 / MAX_SCORE;
3224}
3225
4f03666a
JK
3226static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3227{
3228 if (startup_info->have_repository)
3229 return find_unique_abbrev(oid->hash, abbrev);
3230 else {
3231 char *hex = oid_to_hex(oid);
0d9c527d
JH
3232 if (abbrev < 0)
3233 abbrev = FALLBACK_DEFAULT_ABBREV;
3234 if (abbrev > GIT_SHA1_HEXSZ)
4f03666a 3235 die("BUG: oid abbreviation out of range: %d", abbrev);
43d1948b
JB
3236 if (abbrev)
3237 hex[abbrev] = '\0';
4f03666a
JK
3238 return hex;
3239 }
3240}
3241
b67b9612
JH
3242static void fill_metainfo(struct strbuf *msg,
3243 const char *name,
3244 const char *other,
3245 struct diff_filespec *one,
3246 struct diff_filespec *two,
3247 struct diff_options *o,
37466447 3248 struct diff_filepair *p,
5977744d 3249 int *must_show_header,
37466447 3250 int use_color)
b67b9612 3251{
37466447
BW
3252 const char *set = diff_get_color(use_color, DIFF_METAINFO);
3253 const char *reset = diff_get_color(use_color, DIFF_RESET);
30997bb8 3254 const char *line_prefix = diff_line_prefix(o);
7be57610 3255
296c6bb2 3256 *must_show_header = 1;
b67b9612
JH
3257 strbuf_init(msg, PATH_MAX * 2 + 300);
3258 switch (p->status) {
3259 case DIFF_STATUS_COPIED:
98ad90fb
JH
3260 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3261 line_prefix, set, similarity_index(p));
3262 strbuf_addf(msg, "%s\n%s%scopy from ",
3263 reset, line_prefix, set);
b67b9612 3264 quote_c_style(name, msg, NULL, 0);
98ad90fb 3265 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
b67b9612 3266 quote_c_style(other, msg, NULL, 0);
37466447 3267 strbuf_addf(msg, "%s\n", reset);
b67b9612
JH
3268 break;
3269 case DIFF_STATUS_RENAMED:
98ad90fb
JH
3270 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3271 line_prefix, set, similarity_index(p));
3272 strbuf_addf(msg, "%s\n%s%srename from ",
3273 reset, line_prefix, set);
b67b9612 3274 quote_c_style(name, msg, NULL, 0);
98ad90fb
JH
3275 strbuf_addf(msg, "%s\n%s%srename to ",
3276 reset, line_prefix, set);
b67b9612 3277 quote_c_style(other, msg, NULL, 0);
37466447 3278 strbuf_addf(msg, "%s\n", reset);
b67b9612
JH
3279 break;
3280 case DIFF_STATUS_MODIFIED:
3281 if (p->score) {
98ad90fb
JH
3282 strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3283 line_prefix,
37466447 3284 set, similarity_index(p), reset);
b67b9612
JH
3285 break;
3286 }
3287 /* fallthru */
3288 default:
296c6bb2 3289 *must_show_header = 0;
b67b9612 3290 }
a0d12c44 3291 if (one && two && oidcmp(&one->oid, &two->oid)) {
b67b9612
JH
3292 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3293
3294 if (DIFF_OPT_TST(o, BINARY)) {
3295 mmfile_t mf;
3296 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3297 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3298 abbrev = 40;
3299 }
4f03666a
JK
3300 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3301 diff_abbrev_oid(&one->oid, abbrev),
3302 diff_abbrev_oid(&two->oid, abbrev));
b67b9612
JH
3303 if (one->mode == two->mode)
3304 strbuf_addf(msg, " %06o", one->mode);
37466447 3305 strbuf_addf(msg, "%s\n", reset);
b67b9612 3306 }
b67b9612
JH
3307}
3308
6973dcae
JH
3309static void run_diff_cmd(const char *pgm,
3310 const char *name,
3311 const char *other,
cd676a51 3312 const char *attr_path,
6973dcae
JH
3313 struct diff_filespec *one,
3314 struct diff_filespec *two,
b67b9612 3315 struct strbuf *msg,
051308f6 3316 struct diff_options *o,
b67b9612 3317 struct diff_filepair *p)
6973dcae 3318{
b67b9612
JH
3319 const char *xfrm_msg = NULL;
3320 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
296c6bb2 3321 int must_show_header = 0;
b67b9612 3322
bd8c1a9b
JH
3323
3324 if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
be58e70d
JK
3325 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3326 if (drv && drv->external)
3327 pgm = drv->external;
f1af60bd
JH
3328 }
3329
37466447
BW
3330 if (msg) {
3331 /*
3332 * don't use colors when the header is intended for an
3333 * external diff driver
3334 */
3335 fill_metainfo(msg, name, other, one, two, o, p,
5977744d 3336 &must_show_header,
daa0c3d9 3337 want_color(o->use_color) && !pgm);
37466447
BW
3338 xfrm_msg = msg->len ? msg->buf : NULL;
3339 }
3340
6973dcae
JH
3341 if (pgm) {
3342 run_external_diff(pgm, name, other, one, two, xfrm_msg,
ee7fb0b1 3343 complete_rewrite, o);
6973dcae
JH
3344 return;
3345 }
3346 if (one && two)
3347 builtin_diff(name, other ? other : name,
296c6bb2
CC
3348 one, two, xfrm_msg, must_show_header,
3349 o, complete_rewrite);
6973dcae 3350 else
c0c77734 3351 fprintf(o->file, "* Unmerged path %s\n", name);
6973dcae
JH
3352}
3353
94e327e9 3354static void diff_fill_oid_info(struct diff_filespec *one)
6973dcae
JH
3355{
3356 if (DIFF_FILE_VALID(one)) {
41c9560e 3357 if (!one->oid_valid) {
6973dcae 3358 struct stat st;
4682d852 3359 if (one->is_stdin) {
a0d12c44 3360 oidclr(&one->oid);
5332b2af
JS
3361 return;
3362 }
6973dcae 3363 if (lstat(one->path, &st) < 0)
0721c314 3364 die_errno("stat '%s'", one->path);
a0d12c44 3365 if (index_path(one->oid.hash, one->path, &st, 0))
d7530708 3366 die("cannot hash %s", one->path);
6973dcae
JH
3367 }
3368 }
3369 else
a0d12c44 3370 oidclr(&one->oid);
6973dcae
JH
3371}
3372
cd676a51
JH
3373static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3374{
3375 /* Strip the prefix but do not molest /dev/null and absolute paths */
d8faea9d 3376 if (*namep && **namep != '/') {
cd676a51 3377 *namep += prefix_length;
d8faea9d
JN
3378 if (**namep == '/')
3379 ++*namep;
3380 }
3381 if (*otherp && **otherp != '/') {
cd676a51 3382 *otherp += prefix_length;
d8faea9d
JN
3383 if (**otherp == '/')
3384 ++*otherp;
3385 }
cd676a51
JH
3386}
3387
6973dcae
JH
3388static void run_diff(struct diff_filepair *p, struct diff_options *o)
3389{
3390 const char *pgm = external_diff();
663af342 3391 struct strbuf msg;
663af342
PH
3392 struct diff_filespec *one = p->one;
3393 struct diff_filespec *two = p->two;
6973dcae
JH
3394 const char *name;
3395 const char *other;
cd676a51 3396 const char *attr_path;
663af342 3397
f2d2a5de
SB
3398 name = one->path;
3399 other = (strcmp(name, two->path) ? two->path : NULL);
cd676a51
JH
3400 attr_path = name;
3401 if (o->prefix_length)
3402 strip_prefix(o->prefix_length, &name, &other);
6973dcae 3403
bd8c1a9b
JH
3404 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3405 pgm = NULL;
3406
6973dcae 3407 if (DIFF_PAIR_UNMERGED(p)) {
cd676a51 3408 run_diff_cmd(pgm, name, NULL, attr_path,
b67b9612 3409 NULL, NULL, NULL, o, p);
6973dcae
JH
3410 return;
3411 }
3412
94e327e9
BW
3413 diff_fill_oid_info(one);
3414 diff_fill_oid_info(two);
6973dcae 3415
6973dcae
JH
3416 if (!pgm &&
3417 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3418 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
b67b9612
JH
3419 /*
3420 * a filepair that changes between file and symlink
6973dcae
JH
3421 * needs to be split into deletion and creation.
3422 */
3423 struct diff_filespec *null = alloc_filespec(two->path);
cd676a51 3424 run_diff_cmd(NULL, name, other, attr_path,
b67b9612 3425 one, null, &msg, o, p);
6973dcae 3426 free(null);
b67b9612
JH
3427 strbuf_release(&msg);
3428
6973dcae 3429 null = alloc_filespec(one->path);
cd676a51 3430 run_diff_cmd(NULL, name, other, attr_path,
b67b9612 3431 null, two, &msg, o, p);
6973dcae
JH
3432 free(null);
3433 }
3434 else
cd676a51 3435 run_diff_cmd(pgm, name, other, attr_path,
b67b9612 3436 one, two, &msg, o, p);
6973dcae 3437
663af342 3438 strbuf_release(&msg);
6973dcae
JH
3439}
3440
3441static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3442 struct diffstat_t *diffstat)
3443{
3444 const char *name;
3445 const char *other;
3446
3447 if (DIFF_PAIR_UNMERGED(p)) {
3448 /* unmerged */
74faaa16 3449 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
6973dcae
JH
3450 return;
3451 }
3452
3453 name = p->one->path;
3454 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3455
cd676a51
JH
3456 if (o->prefix_length)
3457 strip_prefix(o->prefix_length, &name, &other);
3458
94e327e9
BW
3459 diff_fill_oid_info(p->one);
3460 diff_fill_oid_info(p->two);
6973dcae 3461
74faaa16 3462 builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
6973dcae
JH
3463}
3464
88246898
JS
3465static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3466{
3467 const char *name;
3468 const char *other;
cd676a51 3469 const char *attr_path;
88246898
JS
3470
3471 if (DIFF_PAIR_UNMERGED(p)) {
3472 /* unmerged */
3473 return;
3474 }
3475
3476 name = p->one->path;
3477 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
cd676a51
JH
3478 attr_path = other ? other : name;
3479
3480 if (o->prefix_length)
3481 strip_prefix(o->prefix_length, &name, &other);
88246898 3482
94e327e9
BW
3483 diff_fill_oid_info(p->one);
3484 diff_fill_oid_info(p->two);
88246898 3485
cd676a51 3486 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
88246898
JS
3487}
3488
6973dcae
JH
3489void diff_setup(struct diff_options *options)
3490{
be4f2b40 3491 memcpy(options, &default_diff_options, sizeof(*options));
c0c77734
DB
3492
3493 options->file = stdout;
3494
43d1948b 3495 options->abbrev = DEFAULT_ABBREV;
6973dcae
JH
3496 options->line_termination = '\n';
3497 options->break_opt = -1;
3498 options->rename_limit = -1;
712d2c7d 3499 options->dirstat_permille = diff_dirstat_permille_default;
6468a4e5 3500 options->context = diff_context_default;
c4888677 3501 options->interhunkcontext = diff_interhunk_context_default;
a17505f2