diff.c: emit_diff_symbol learns DIFF_SYMBOL_HEADER
[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 {
a29b0a13 564 DIFF_SYMBOL_HEADER,
3ee8b7bf
SB
565 DIFF_SYMBOL_FILEPAIR_PLUS,
566 DIFF_SYMBOL_FILEPAIR_MINUS,
ff958679
SB
567 DIFF_SYMBOL_WORDS_PORCELAIN,
568 DIFF_SYMBOL_WORDS,
091f8e28 569 DIFF_SYMBOL_CONTEXT,
f2bb1218 570 DIFF_SYMBOL_CONTEXT_INCOMPLETE,
091f8e28
SB
571 DIFF_SYMBOL_PLUS,
572 DIFF_SYMBOL_MINUS,
b9cbfde6 573 DIFF_SYMBOL_NO_LF_EOF,
68abc6f1 574 DIFF_SYMBOL_CONTEXT_FRAGINFO,
c64b420b 575 DIFF_SYMBOL_CONTEXT_MARKER,
36a4cefd
SB
576 DIFF_SYMBOL_SEPARATOR
577};
091f8e28
SB
578/*
579 * Flags for content lines:
580 * 0..12 are whitespace rules
581 * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
582 * 16 is marking if the line is blank at EOF
583 */
584#define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF (1<<16)
585#define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
586
587static void emit_line_ws_markup(struct diff_options *o,
588 const char *set, const char *reset,
589 const char *line, int len, char sign,
590 unsigned ws_rule, int blank_at_eof)
591{
592 const char *ws = NULL;
593
594 if (o->ws_error_highlight & ws_rule) {
595 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
596 if (!*ws)
597 ws = NULL;
598 }
599
600 if (!ws)
601 emit_line_0(o, set, reset, sign, line, len);
602 else if (blank_at_eof)
603 /* Blank line at EOF - paint '+' as well */
604 emit_line_0(o, ws, reset, sign, line, len);
605 else {
606 /* Emit just the prefix, then the rest. */
607 emit_line_0(o, set, reset, sign, "", 0);
608 ws_check_emit(line, len, ws_rule,
609 o->file, set, reset, ws);
610 }
611}
36a4cefd
SB
612
613static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
091f8e28 614 const char *line, int len, unsigned flags)
36a4cefd 615{
b9cbfde6 616 static const char *nneof = " No newline at end of file\n";
3ee8b7bf 617 const char *context, *reset, *set, *meta;
36a4cefd 618 switch (s) {
b9cbfde6
SB
619 case DIFF_SYMBOL_NO_LF_EOF:
620 context = diff_get_color_opt(o, DIFF_CONTEXT);
621 reset = diff_get_color_opt(o, DIFF_RESET);
622 putc('\n', o->file);
623 emit_line_0(o, context, reset, '\\',
624 nneof, strlen(nneof));
625 break;
68abc6f1
SB
626 case DIFF_SYMBOL_CONTEXT_FRAGINFO:
627 emit_line(o, "", "", line, len);
628 break;
f2bb1218 629 case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
c64b420b
SB
630 case DIFF_SYMBOL_CONTEXT_MARKER:
631 context = diff_get_color_opt(o, DIFF_CONTEXT);
632 reset = diff_get_color_opt(o, DIFF_RESET);
633 emit_line(o, context, reset, line, len);
634 break;
36a4cefd
SB
635 case DIFF_SYMBOL_SEPARATOR:
636 fprintf(o->file, "%s%c",
637 diff_line_prefix(o),
638 o->line_termination);
639 break;
091f8e28
SB
640 case DIFF_SYMBOL_CONTEXT:
641 set = diff_get_color_opt(o, DIFF_CONTEXT);
642 reset = diff_get_color_opt(o, DIFF_RESET);
643 emit_line_ws_markup(o, set, reset, line, len, ' ',
644 flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
645 break;
646 case DIFF_SYMBOL_PLUS:
647 set = diff_get_color_opt(o, DIFF_FILE_NEW);
648 reset = diff_get_color_opt(o, DIFF_RESET);
649 emit_line_ws_markup(o, set, reset, line, len, '+',
650 flags & DIFF_SYMBOL_CONTENT_WS_MASK,
651 flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
652 break;
653 case DIFF_SYMBOL_MINUS:
654 set = diff_get_color_opt(o, DIFF_FILE_OLD);
655 reset = diff_get_color_opt(o, DIFF_RESET);
656 emit_line_ws_markup(o, set, reset, line, len, '-',
657 flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
658 break;
ff958679
SB
659 case DIFF_SYMBOL_WORDS_PORCELAIN:
660 context = diff_get_color_opt(o, DIFF_CONTEXT);
661 reset = diff_get_color_opt(o, DIFF_RESET);
662 emit_line(o, context, reset, line, len);
663 fputs("~\n", o->file);
664 break;
665 case DIFF_SYMBOL_WORDS:
666 context = diff_get_color_opt(o, DIFF_CONTEXT);
667 reset = diff_get_color_opt(o, DIFF_RESET);
668 /*
669 * Skip the prefix character, if any. With
670 * diff_suppress_blank_empty, there may be
671 * none.
672 */
673 if (line[0] != '\n') {
674 line++;
675 len--;
676 }
677 emit_line(o, context, reset, line, len);
678 break;
3ee8b7bf
SB
679 case DIFF_SYMBOL_FILEPAIR_PLUS:
680 meta = diff_get_color_opt(o, DIFF_METAINFO);
681 reset = diff_get_color_opt(o, DIFF_RESET);
682 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
683 line, reset,
684 strchr(line, ' ') ? "\t" : "");
685 break;
686 case DIFF_SYMBOL_FILEPAIR_MINUS:
687 meta = diff_get_color_opt(o, DIFF_METAINFO);
688 reset = diff_get_color_opt(o, DIFF_RESET);
689 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
690 line, reset,
691 strchr(line, ' ') ? "\t" : "");
692 break;
a29b0a13
SB
693 case DIFF_SYMBOL_HEADER:
694 fprintf(o->file, "%s", line);
695 break;
36a4cefd
SB
696 default:
697 die("BUG: unknown diff symbol");
698 }
699}
700
6957eb9a
JH
701static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
702{
703 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
704 ecbdata->blank_at_eof_in_preimage &&
705 ecbdata->blank_at_eof_in_postimage &&
706 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
707 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
708 return 0;
018cff70 709 return ws_blank_line(line, len, ecbdata->ws_rule);
6957eb9a
JH
710}
711
b8767f79 712static void emit_add_line(const char *reset,
0e383e18
JH
713 struct emit_callback *ecbdata,
714 const char *line, int len)
715{
091f8e28
SB
716 unsigned flags = WSEH_NEW | ecbdata->ws_rule;
717 if (new_blank_line_at_eof(ecbdata, line, len))
718 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
719
720 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
b8767f79 721}
0e383e18 722
b8767f79
JH
723static void emit_del_line(const char *reset,
724 struct emit_callback *ecbdata,
725 const char *line, int len)
726{
091f8e28
SB
727 unsigned flags = WSEH_OLD | ecbdata->ws_rule;
728 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
0e383e18
JH
729}
730
731static void emit_context_line(const char *reset,
732 struct emit_callback *ecbdata,
733 const char *line, int len)
734{
091f8e28
SB
735 unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
736 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
0e383e18
JH
737}
738
89cb73a1
BW
739static void emit_hunk_header(struct emit_callback *ecbdata,
740 const char *line, int len)
741{
8dbf3eb6 742 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
89cb73a1
BW
743 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
744 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
745 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
746 static const char atat[2] = { '@', '@' };
747 const char *cp, *ep;
2efcc977
BY
748 struct strbuf msgbuf = STRBUF_INIT;
749 int org_len = len;
750 int i = 1;
89cb73a1
BW
751
752 /*
753 * As a hunk header must begin with "@@ -<old>, +<new> @@",
754 * it always is at least 10 bytes long.
755 */
756 if (len < 10 ||
757 memcmp(line, atat, 2) ||
758 !(ep = memmem(line + 2, len - 2, atat, 2))) {
c64b420b 759 emit_diff_symbol(ecbdata->opt,
091f8e28 760 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
89cb73a1
BW
761 return;
762 }
763 ep += 2; /* skip over @@ */
764
765 /* The hunk header in fraginfo color */
cedc61a9 766 strbuf_addstr(&msgbuf, frag);
2efcc977 767 strbuf_add(&msgbuf, line, ep - line);
cedc61a9 768 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
769
770 /*
771 * trailing "\r\n"
772 */
773 for ( ; i < 3; i++)
774 if (line[len - i] == '\r' || line[len - i] == '\n')
775 len--;
89cb73a1
BW
776
777 /* blank before the func header */
778 for (cp = ep; ep - line < len; ep++)
779 if (*ep != ' ' && *ep != '\t')
780 break;
2efcc977 781 if (ep != cp) {
8dbf3eb6 782 strbuf_addstr(&msgbuf, context);
2efcc977 783 strbuf_add(&msgbuf, cp, ep - cp);
cedc61a9 784 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
785 }
786
787 if (ep < line + len) {
cedc61a9 788 strbuf_addstr(&msgbuf, func);
2efcc977 789 strbuf_add(&msgbuf, ep, line + len - ep);
cedc61a9 790 strbuf_addstr(&msgbuf, reset);
2efcc977 791 }
89cb73a1 792
2efcc977 793 strbuf_add(&msgbuf, line + len, org_len - len);
dfb7728f 794 strbuf_complete_line(&msgbuf);
68abc6f1 795 emit_diff_symbol(ecbdata->opt,
091f8e28 796 DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
2efcc977 797 strbuf_release(&msgbuf);
89cb73a1
BW
798}
799
479b0ae8
JK
800static struct diff_tempfile *claim_diff_tempfile(void) {
801 int i;
802 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
803 if (!diff_temp[i].name)
804 return diff_temp + i;
805 die("BUG: diff is failing to clean up its tempfiles");
806}
807
479b0ae8
JK
808static void remove_tempfile(void)
809{
810 int i;
a8344abe 811 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
284098f1
MH
812 if (is_tempfile_active(&diff_temp[i].tempfile))
813 delete_tempfile(&diff_temp[i].tempfile);
a8344abe
NR
814 diff_temp[i].name = NULL;
815 }
479b0ae8
JK
816}
817
c0c77734 818static void print_line_count(FILE *file, int count)
6973dcae
JH
819{
820 switch (count) {
821 case 0:
c0c77734 822 fprintf(file, "0,0");
6973dcae
JH
823 break;
824 case 1:
c0c77734 825 fprintf(file, "1");
6973dcae
JH
826 break;
827 default:
c0c77734 828 fprintf(file, "1,%d", count);
6973dcae
JH
829 break;
830 }
831}
832
7f7ee2ff
JH
833static void emit_rewrite_lines(struct emit_callback *ecb,
834 int prefix, const char *data, int size)
6973dcae 835{
7f7ee2ff 836 const char *endp = NULL;
7f7ee2ff
JH
837 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
838
839 while (0 < size) {
840 int len;
841
842 endp = memchr(data, '\n', size);
843 len = endp ? (endp - data + 1) : size;
844 if (prefix != '+') {
845 ecb->lno_in_preimage++;
0e383e18 846 emit_del_line(reset, ecb, data, len);
7f7ee2ff
JH
847 } else {
848 ecb->lno_in_postimage++;
849 emit_add_line(reset, ecb, data, len);
13e36ec5 850 }
7f7ee2ff
JH
851 size -= len;
852 data += len;
853 }
b9cbfde6 854 if (!endp)
091f8e28 855 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
6973dcae
JH
856}
857
858static void emit_rewrite_diff(const char *name_a,
859 const char *name_b,
860 struct diff_filespec *one,
13e36ec5 861 struct diff_filespec *two,
d9bae1a1
JK
862 struct userdiff_driver *textconv_one,
863 struct userdiff_driver *textconv_two,
eab9a40b 864 struct diff_options *o)
6973dcae
JH
865{
866 int lc_a, lc_b;
f1c96261
JK
867 const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
868 const char *reset = diff_get_color(o->use_color, DIFF_RESET);
d5625091 869 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
a5a818ee 870 const char *a_prefix, *b_prefix;
840383b2 871 char *data_one, *data_two;
3aa1f7ca 872 size_t size_one, size_two;
7f7ee2ff 873 struct emit_callback ecbdata;
30997bb8 874 const char *line_prefix = diff_line_prefix(o);
a5a818ee
JH
875
876 if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
877 a_prefix = o->b_prefix;
878 b_prefix = o->a_prefix;
879 } else {
880 a_prefix = o->a_prefix;
881 b_prefix = o->b_prefix;
882 }
1a9eb3b9 883
8a13becc
JH
884 name_a += (*name_a == '/');
885 name_b += (*name_b == '/');
1a9eb3b9 886
d5625091
JH
887 strbuf_reset(&a_name);
888 strbuf_reset(&b_name);
a5a818ee
JH
889 quote_two_c_style(&a_name, a_prefix, name_a, 0);
890 quote_two_c_style(&b_name, b_prefix, name_b, 0);
d5625091 891
840383b2
JK
892 size_one = fill_textconv(textconv_one, one, &data_one);
893 size_two = fill_textconv(textconv_two, two, &data_two);
3aa1f7ca 894
d91ba8fa 895 memset(&ecbdata, 0, sizeof(ecbdata));
daa0c3d9 896 ecbdata.color_diff = want_color(o->use_color);
c189c4f2 897 ecbdata.ws_rule = whitespace_rule(name_b);
a3c158d4 898 ecbdata.opt = o;
d91ba8fa
JH
899 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
900 mmfile_t mf1, mf2;
901 mf1.ptr = (char *)data_one;
902 mf2.ptr = (char *)data_two;
903 mf1.size = size_one;
904 mf2.size = size_two;
905 check_blank_at_eof(&mf1, &mf2, &ecbdata);
906 }
907 ecbdata.lno_in_preimage = 1;
908 ecbdata.lno_in_postimage = 1;
909
3aa1f7ca
JK
910 lc_a = count_lines(data_one, size_one);
911 lc_b = count_lines(data_two, size_two);
3ee8b7bf
SB
912
913 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
914 a_name.buf, a_name.len, 0);
915 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
916 b_name.buf, b_name.len, 0);
917
918 fprintf(o->file, "%s%s@@ -", line_prefix, fraginfo);
467ddc14
JH
919 if (!o->irreversible_delete)
920 print_line_count(o->file, lc_a);
921 else
922 fprintf(o->file, "?,?");
c0c77734
DB
923 fprintf(o->file, " +");
924 print_line_count(o->file, lc_b);
925 fprintf(o->file, " @@%s\n", reset);
467ddc14 926 if (lc_a && !o->irreversible_delete)
d91ba8fa 927 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
6973dcae 928 if (lc_b)
d91ba8fa 929 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
b76c056b 930 if (textconv_one)
aed6ca52 931 free((char *)data_one);
b76c056b 932 if (textconv_two)
aed6ca52 933 free((char *)data_two);
6973dcae
JH
934}
935
f59a59e2
JS
936struct diff_words_buffer {
937 mmfile_t text;
938 long alloc;
2e5d2003
JS
939 struct diff_words_orig {
940 const char *begin, *end;
941 } *orig;
942 int orig_nr, orig_alloc;
f59a59e2
JS
943};
944
945static void diff_words_append(char *line, unsigned long len,
946 struct diff_words_buffer *buffer)
947{
23c1575f 948 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
f59a59e2
JS
949 line++;
950 len--;
951 memcpy(buffer->text.ptr + buffer->text.size, line, len);
952 buffer->text.size += len;
2b6a5417 953 buffer->text.ptr[buffer->text.size] = '\0';
f59a59e2
JS
954}
955
9cba13ca 956struct diff_words_style_elem {
882749a0
TR
957 const char *prefix;
958 const char *suffix;
959 const char *color; /* NULL; filled in by the setup code if
960 * color is enabled */
961};
962
9cba13ca 963struct diff_words_style {
882749a0
TR
964 enum diff_words_type type;
965 struct diff_words_style_elem new, old, ctx;
966 const char *newline;
967};
968
c2e86add 969static struct diff_words_style diff_words_styles[] = {
882749a0
TR
970 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
971 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
972 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
973};
974
f59a59e2 975struct diff_words_data {
f59a59e2 976 struct diff_words_buffer minus, plus;
2e5d2003 977 const char *current_plus;
4297c0ae
BY
978 int last_minus;
979 struct diff_options *opt;
2b6a5417 980 regex_t *word_regex;
882749a0
TR
981 enum diff_words_type type;
982 struct diff_words_style *style;
f59a59e2
JS
983};
984
882749a0
TR
985static int fn_out_diff_words_write_helper(FILE *fp,
986 struct diff_words_style_elem *st_el,
987 const char *newline,
4297c0ae
BY
988 size_t count, const char *buf,
989 const char *line_prefix)
882749a0 990{
4297c0ae
BY
991 int print = 0;
992
882749a0
TR
993 while (count) {
994 char *p = memchr(buf, '\n', count);
4297c0ae
BY
995 if (print)
996 fputs(line_prefix, fp);
882749a0
TR
997 if (p != buf) {
998 if (st_el->color && fputs(st_el->color, fp) < 0)
999 return -1;
1000 if (fputs(st_el->prefix, fp) < 0 ||
1001 fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
1002 fputs(st_el->suffix, fp) < 0)
1003 return -1;
1004 if (st_el->color && *st_el->color
1005 && fputs(GIT_COLOR_RESET, fp) < 0)
1006 return -1;
1007 }
1008 if (!p)
1009 return 0;
1010 if (fputs(newline, fp) < 0)
1011 return -1;
1012 count -= p + 1 - buf;
1013 buf = p + 1;
4297c0ae 1014 print = 1;
882749a0
TR
1015 }
1016 return 0;
1017}
1018
4297c0ae
BY
1019/*
1020 * '--color-words' algorithm can be described as:
1021 *
5621760f 1022 * 1. collect the minus/plus lines of a diff hunk, divided into
4297c0ae
BY
1023 * minus-lines and plus-lines;
1024 *
1025 * 2. break both minus-lines and plus-lines into words and
1026 * place them into two mmfile_t with one word for each line;
1027 *
1028 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1029 *
1030 * And for the common parts of the both file, we output the plus side text.
1031 * diff_words->current_plus is used to trace the current position of the plus file
1032 * which printed. diff_words->last_minus is used to trace the last minus word
1033 * printed.
1034 *
1035 * For '--graph' to work with '--color-words', we need to output the graph prefix
1036 * on each line of color words output. Generally, there are two conditions on
1037 * which we should output the prefix.
1038 *
1039 * 1. diff_words->last_minus == 0 &&
1040 * diff_words->current_plus == diff_words->plus.text.ptr
1041 *
1042 * that is: the plus text must start as a new line, and if there is no minus
1043 * word printed, a graph prefix must be printed.
1044 *
1045 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
1046 * *(diff_words->current_plus - 1) == '\n'
1047 *
1048 * that is: a graph prefix must be printed following a '\n'
1049 */
1050static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1051{
1052 if ((diff_words->last_minus == 0 &&
1053 diff_words->current_plus == diff_words->plus.text.ptr) ||
1054 (diff_words->current_plus > diff_words->plus.text.ptr &&
1055 *(diff_words->current_plus - 1) == '\n')) {
1056 return 1;
1057 } else {
1058 return 0;
1059 }
1060}
1061
f59a59e2 1062static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
f59a59e2 1063{
f59a59e2 1064 struct diff_words_data *diff_words = priv;
882749a0 1065 struct diff_words_style *style = diff_words->style;
2e5d2003
JS
1066 int minus_first, minus_len, plus_first, plus_len;
1067 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
4297c0ae 1068 struct diff_options *opt = diff_words->opt;
30997bb8 1069 const char *line_prefix;
f59a59e2 1070
2e5d2003
JS
1071 if (line[0] != '@' || parse_hunk_header(line, len,
1072 &minus_first, &minus_len, &plus_first, &plus_len))
f59a59e2
JS
1073 return;
1074
4297c0ae 1075 assert(opt);
30997bb8 1076 line_prefix = diff_line_prefix(opt);
4297c0ae 1077
2e5d2003
JS
1078 /* POSIX requires that first be decremented by one if len == 0... */
1079 if (minus_len) {
1080 minus_begin = diff_words->minus.orig[minus_first].begin;
1081 minus_end =
1082 diff_words->minus.orig[minus_first + minus_len - 1].end;
1083 } else
1084 minus_begin = minus_end =
1085 diff_words->minus.orig[minus_first].end;
1086
1087 if (plus_len) {
1088 plus_begin = diff_words->plus.orig[plus_first].begin;
1089 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1090 } else
1091 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1092
4297c0ae
BY
1093 if (color_words_output_graph_prefix(diff_words)) {
1094 fputs(line_prefix, diff_words->opt->file);
1095 }
1096 if (diff_words->current_plus != plus_begin) {
1097 fn_out_diff_words_write_helper(diff_words->opt->file,
882749a0
TR
1098 &style->ctx, style->newline,
1099 plus_begin - diff_words->current_plus,
4297c0ae
BY
1100 diff_words->current_plus, line_prefix);
1101 if (*(plus_begin - 1) == '\n')
1102 fputs(line_prefix, diff_words->opt->file);
1103 }
1104 if (minus_begin != minus_end) {
1105 fn_out_diff_words_write_helper(diff_words->opt->file,
882749a0 1106 &style->old, style->newline,
4297c0ae
BY
1107 minus_end - minus_begin, minus_begin,
1108 line_prefix);
1109 }
1110 if (plus_begin != plus_end) {
1111 fn_out_diff_words_write_helper(diff_words->opt->file,
882749a0 1112 &style->new, style->newline,
4297c0ae
BY
1113 plus_end - plus_begin, plus_begin,
1114 line_prefix);
1115 }
2e5d2003
JS
1116
1117 diff_words->current_plus = plus_end;
4297c0ae 1118 diff_words->last_minus = minus_first;
f59a59e2
JS
1119}
1120
2b6a5417
JS
1121/* This function starts looking at *begin, and returns 0 iff a word was found. */
1122static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1123 int *begin, int *end)
1124{
1125 if (word_regex && *begin < buffer->size) {
1126 regmatch_t match[1];
b7d36ffc
JS
1127 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1128 buffer->size - *begin, 1, match, 0)) {
2b6a5417
JS
1129 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1130 '\n', match[0].rm_eo - match[0].rm_so);
1131 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1132 *begin += match[0].rm_so;
1133 return *begin >= *end;
1134 }
1135 return -1;
f59a59e2
JS
1136 }
1137
2b6a5417
JS
1138 /* find the next word */
1139 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1140 (*begin)++;
1141 if (*begin >= buffer->size)
1142 return -1;
f59a59e2 1143
2b6a5417
JS
1144 /* find the end of the word */
1145 *end = *begin + 1;
1146 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1147 (*end)++;
1148
1149 return 0;
f59a59e2
JS
1150}
1151
23c1575f 1152/*
2e5d2003
JS
1153 * This function splits the words in buffer->text, stores the list with
1154 * newline separator into out, and saves the offsets of the original words
1155 * in buffer->orig.
23c1575f 1156 */
2b6a5417
JS
1157static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1158 regex_t *word_regex)
f59a59e2 1159{
2e5d2003 1160 int i, j;
2b6a5417 1161 long alloc = 0;
f59a59e2 1162
2e5d2003 1163 out->size = 0;
2b6a5417 1164 out->ptr = NULL;
f59a59e2 1165
2e5d2003
JS
1166 /* fake an empty "0th" word */
1167 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1168 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1169 buffer->orig_nr = 1;
1170
1171 for (i = 0; i < buffer->text.size; i++) {
2b6a5417
JS
1172 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1173 return;
2e5d2003
JS
1174
1175 /* store original boundaries */
1176 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1177 buffer->orig_alloc);
1178 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1179 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1180 buffer->orig_nr++;
1181
1182 /* store one word */
2b6a5417 1183 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2e5d2003
JS
1184 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1185 out->ptr[out->size + j - i] = '\n';
1186 out->size += j - i + 1;
1187
1188 i = j - 1;
f59a59e2
JS
1189 }
1190}
1191
1192/* this executes the word diff on the accumulated buffers */
1193static void diff_words_show(struct diff_words_data *diff_words)
1194{
1195 xpparam_t xpp;
1196 xdemitconf_t xecfg;
f59a59e2 1197 mmfile_t minus, plus;
882749a0 1198 struct diff_words_style *style = diff_words->style;
f59a59e2 1199
4297c0ae 1200 struct diff_options *opt = diff_words->opt;
30997bb8 1201 const char *line_prefix;
4297c0ae
BY
1202
1203 assert(opt);
30997bb8 1204 line_prefix = diff_line_prefix(opt);
4297c0ae 1205
2e5d2003
JS
1206 /* special case: only removal */
1207 if (!diff_words->plus.text.size) {
4297c0ae
BY
1208 fputs(line_prefix, diff_words->opt->file);
1209 fn_out_diff_words_write_helper(diff_words->opt->file,
882749a0 1210 &style->old, style->newline,
4297c0ae
BY
1211 diff_words->minus.text.size,
1212 diff_words->minus.text.ptr, line_prefix);
2e5d2003
JS
1213 diff_words->minus.text.size = 0;
1214 return;
1215 }
1216
1217 diff_words->current_plus = diff_words->plus.text.ptr;
4297c0ae 1218 diff_words->last_minus = 0;
f59a59e2 1219
9ccd0a88 1220 memset(&xpp, 0, sizeof(xpp));
30b25010 1221 memset(&xecfg, 0, sizeof(xecfg));
2b6a5417
JS
1222 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1223 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
582aa00b 1224 xpp.flags = 0;
2b6a5417 1225 /* as only the hunk header will be parsed, we need a 0-context */
2e5d2003 1226 xecfg.ctxlen = 0;
3efb9880
JK
1227 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1228 &xpp, &xecfg))
1229 die("unable to generate word diff");
f59a59e2
JS
1230 free(minus.ptr);
1231 free(plus.ptr);
2e5d2003 1232 if (diff_words->current_plus != diff_words->plus.text.ptr +
4297c0ae
BY
1233 diff_words->plus.text.size) {
1234 if (color_words_output_graph_prefix(diff_words))
1235 fputs(line_prefix, diff_words->opt->file);
1236 fn_out_diff_words_write_helper(diff_words->opt->file,
882749a0 1237 &style->ctx, style->newline,
2e5d2003 1238 diff_words->plus.text.ptr + diff_words->plus.text.size
4297c0ae
BY
1239 - diff_words->current_plus, diff_words->current_plus,
1240 line_prefix);
1241 }
f59a59e2 1242 diff_words->minus.text.size = diff_words->plus.text.size = 0;
f59a59e2
JS
1243}
1244
76fd2828
JH
1245/* In "color-words" mode, show word-diff of words accumulated in the buffer */
1246static void diff_words_flush(struct emit_callback *ecbdata)
1247{
1248 if (ecbdata->diff_words->minus.text.size ||
1249 ecbdata->diff_words->plus.text.size)
1250 diff_words_show(ecbdata->diff_words);
1251}
1252
77d1a520
TR
1253static void diff_filespec_load_driver(struct diff_filespec *one)
1254{
1255 /* Use already-loaded driver */
1256 if (one->driver)
1257 return;
1258
1259 if (S_ISREG(one->mode))
1260 one->driver = userdiff_find_by_path(one->path);
1261
1262 /* Fallback to default settings */
1263 if (!one->driver)
1264 one->driver = userdiff_find_by_name("default");
1265}
1266
1267static const char *userdiff_word_regex(struct diff_filespec *one)
1268{
1269 diff_filespec_load_driver(one);
1270 return one->driver->word_regex;
1271}
1272
1273static void init_diff_words_data(struct emit_callback *ecbdata,
6440d341 1274 struct diff_options *orig_opts,
77d1a520
TR
1275 struct diff_filespec *one,
1276 struct diff_filespec *two)
1277{
1278 int i;
6440d341
TR
1279 struct diff_options *o = xmalloc(sizeof(struct diff_options));
1280 memcpy(o, orig_opts, sizeof(struct diff_options));
77d1a520
TR
1281
1282 ecbdata->diff_words =
1283 xcalloc(1, sizeof(struct diff_words_data));
1284 ecbdata->diff_words->type = o->word_diff;
1285 ecbdata->diff_words->opt = o;
1286 if (!o->word_regex)
1287 o->word_regex = userdiff_word_regex(one);
1288 if (!o->word_regex)
1289 o->word_regex = userdiff_word_regex(two);
1290 if (!o->word_regex)
1291 o->word_regex = diff_word_regex_cfg;
1292 if (o->word_regex) {
1293 ecbdata->diff_words->word_regex = (regex_t *)
1294 xmalloc(sizeof(regex_t));
1295 if (regcomp(ecbdata->diff_words->word_regex,
1296 o->word_regex,
1297 REG_EXTENDED | REG_NEWLINE))
1298 die ("Invalid regular expression: %s",
1299 o->word_regex);
1300 }
1301 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1302 if (o->word_diff == diff_words_styles[i].type) {
1303 ecbdata->diff_words->style =
1304 &diff_words_styles[i];
1305 break;
1306 }
1307 }
1308 if (want_color(o->use_color)) {
1309 struct diff_words_style *st = ecbdata->diff_words->style;
1310 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1311 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
8dbf3eb6 1312 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
77d1a520
TR
1313 }
1314}
1315
f59a59e2
JS
1316static void free_diff_words_data(struct emit_callback *ecbdata)
1317{
1318 if (ecbdata->diff_words) {
76fd2828 1319 diff_words_flush(ecbdata);
6440d341 1320 free (ecbdata->diff_words->opt);
8e0f7003 1321 free (ecbdata->diff_words->minus.text.ptr);
2e5d2003 1322 free (ecbdata->diff_words->minus.orig);
8e0f7003 1323 free (ecbdata->diff_words->plus.text.ptr);
2e5d2003 1324 free (ecbdata->diff_words->plus.orig);
ef5644ea
BC
1325 if (ecbdata->diff_words->word_regex) {
1326 regfree(ecbdata->diff_words->word_regex);
1327 free(ecbdata->diff_words->word_regex);
1328 }
6a83d902 1329 FREE_AND_NULL(ecbdata->diff_words);
f59a59e2
JS
1330 }
1331}
1332
ce436973 1333const char *diff_get_color(int diff_use_color, enum color_diff ix)
cd112cef 1334{
daa0c3d9 1335 if (want_color(diff_use_color))
50f575fc
LT
1336 return diff_colors[ix];
1337 return "";
cd112cef
JS
1338}
1339
f1922234
JK
1340const char *diff_line_prefix(struct diff_options *opt)
1341{
1342 struct strbuf *msgbuf;
1343 if (!opt->output_prefix)
1344 return "";
1345
1346 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1347 return msgbuf->buf;
1348}
1349
23707811
JH
1350static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1351{
1352 const char *cp;
1353 unsigned long allot;
1354 size_t l = len;
1355
1356 if (ecb->truncate)
1357 return ecb->truncate(line, len);
1358 cp = line;
1359 allot = l;
1360 while (0 < l) {
1361 (void) utf8_width(&cp, &l);
1362 if (!cp)
1363 break; /* truncated in the middle? */
1364 }
1365 return allot - l;
1366}
1367
d68fe26f 1368static void find_lno(const char *line, struct emit_callback *ecbdata)
690ed843 1369{
d68fe26f
JH
1370 const char *p;
1371 ecbdata->lno_in_preimage = 0;
1372 ecbdata->lno_in_postimage = 0;
1373 p = strchr(line, '-');
690ed843 1374 if (!p)
d68fe26f
JH
1375 return; /* cannot happen */
1376 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1377 p = strchr(p, '+');
1378 if (!p)
1379 return; /* cannot happen */
1380 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
690ed843
JH
1381}
1382
cd112cef 1383static void fn_out_consume(void *priv, char *line, unsigned long len)
6973dcae 1384{
6973dcae 1385 struct emit_callback *ecbdata = priv;
ce436973 1386 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
7be57610 1387 struct diff_options *o = ecbdata->opt;
6973dcae 1388
ba16233c
SB
1389 o->found_changes = 1;
1390
3e97c7c6 1391 if (ecbdata->header) {
a29b0a13
SB
1392 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
1393 ecbdata->header->buf, ecbdata->header->len, 0);
3e97c7c6
GB
1394 strbuf_reset(ecbdata->header);
1395 ecbdata->header = NULL;
1396 }
34a5e1a2 1397
6973dcae 1398 if (ecbdata->label_path[0]) {
3ee8b7bf
SB
1399 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1400 ecbdata->label_path[0],
1401 strlen(ecbdata->label_path[0]), 0);
1402 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1403 ecbdata->label_path[1],
1404 strlen(ecbdata->label_path[1]), 0);
6973dcae
JH
1405 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1406 }
cd112cef 1407
a624eaa7
JM
1408 if (diff_suppress_blank_empty
1409 && len == 2 && line[0] == ' ' && line[1] == '\n') {
1410 line[0] = '\n';
1411 len = 1;
1412 }
1413
b8d9c1a6 1414 if (line[0] == '@') {
76fd2828
JH
1415 if (ecbdata->diff_words)
1416 diff_words_flush(ecbdata);
23707811 1417 len = sane_truncate_line(ecbdata, line, len);
d68fe26f 1418 find_lno(line, ecbdata);
89cb73a1 1419 emit_hunk_header(ecbdata, line, len);
448c3ef1 1420 return;
cd112cef 1421 }
448c3ef1 1422
448c3ef1 1423 if (ecbdata->diff_words) {
ff958679
SB
1424 enum diff_symbol s =
1425 ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
1426 DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
448c3ef1
JH
1427 if (line[0] == '-') {
1428 diff_words_append(line, len,
1429 &ecbdata->diff_words->minus);
1430 return;
1431 } else if (line[0] == '+') {
1432 diff_words_append(line, len,
1433 &ecbdata->diff_words->plus);
1434 return;
59556548 1435 } else if (starts_with(line, "\\ ")) {
c7c2bc0a
TR
1436 /*
1437 * Eat the "no newline at eof" marker as if we
1438 * saw a "+" or "-" line with nothing on it,
1439 * and return without diff_words_flush() to
1440 * defer processing. If this is the end of
1441 * preimage, more "+" lines may come after it.
1442 */
1443 return;
448c3ef1 1444 }
76fd2828 1445 diff_words_flush(ecbdata);
ff958679 1446 emit_diff_symbol(o, s, line, len, 0);
448c3ef1
JH
1447 return;
1448 }
448c3ef1 1449
0e383e18
JH
1450 switch (line[0]) {
1451 case '+':
d68fe26f 1452 ecbdata->lno_in_postimage++;
018cff70 1453 emit_add_line(reset, ecbdata, line + 1, len - 1);
0e383e18
JH
1454 break;
1455 case '-':
1456 ecbdata->lno_in_preimage++;
1457 emit_del_line(reset, ecbdata, line + 1, len - 1);
1458 break;
1459 case ' ':
1460 ecbdata->lno_in_postimage++;
1461 ecbdata->lno_in_preimage++;
1462 emit_context_line(reset, ecbdata, line + 1, len - 1);
1463 break;
1464 default:
1465 /* incomplete line at the end */
1466 ecbdata->lno_in_preimage++;
f2bb1218
SB
1467 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
1468 line, len, 0);
0e383e18 1469 break;
448c3ef1 1470 }
6973dcae
JH
1471}
1472
1473static char *pprint_rename(const char *a, const char *b)
1474{
1475 const char *old = a;
1476 const char *new = b;
f285a2d7 1477 struct strbuf name = STRBUF_INIT;
6973dcae 1478 int pfx_length, sfx_length;
dd281f09 1479 int pfx_adjust_for_slash;
6973dcae
JH
1480 int len_a = strlen(a);
1481 int len_b = strlen(b);
663af342 1482 int a_midlen, b_midlen;
e5bfbf9b
AJ
1483 int qlen_a = quote_c_style(a, NULL, NULL, 0);
1484 int qlen_b = quote_c_style(b, NULL, NULL, 0);
1485
1486 if (qlen_a || qlen_b) {
663af342
PH
1487 quote_c_style(a, &name, NULL, 0);
1488 strbuf_addstr(&name, " => ");
1489 quote_c_style(b, &name, NULL, 0);
b315c5c0 1490 return strbuf_detach(&name, NULL);
e5bfbf9b 1491 }
6973dcae
JH
1492
1493 /* Find common prefix */
1494 pfx_length = 0;
1495 while (*old && *new && *old == *new) {
1496 if (*old == '/')
1497 pfx_length = old - a + 1;
1498 old++;
1499 new++;
1500 }
1501
1502 /* Find common suffix */
1503 old = a + len_a;
1504 new = b + len_b;
1505 sfx_length = 0;
d020e27f 1506 /*
dd281f09
TR
1507 * If there is a common prefix, it must end in a slash. In
1508 * that case we let this loop run 1 into the prefix to see the
1509 * same slash.
1510 *
1511 * If there is no common prefix, we cannot do this as it would
1512 * underrun the input strings.
d020e27f 1513 */
dd281f09
TR
1514 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1515 while (a + pfx_length - pfx_adjust_for_slash <= old &&
1516 b + pfx_length - pfx_adjust_for_slash <= new &&
d020e27f 1517 *old == *new) {
6973dcae
JH
1518 if (*old == '/')
1519 sfx_length = len_a - (old - a);
1520 old--;
1521 new--;
1522 }
1523
1524 /*
1525 * pfx{mid-a => mid-b}sfx
1526 * {pfx-a => pfx-b}sfx
1527 * pfx{sfx-a => sfx-b}
1528 * name-a => name-b
1529 */
663af342
PH
1530 a_midlen = len_a - pfx_length - sfx_length;
1531 b_midlen = len_b - pfx_length - sfx_length;
1532 if (a_midlen < 0)
1533 a_midlen = 0;
1534 if (b_midlen < 0)
1535 b_midlen = 0;
1536
1537 strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
6973dcae 1538 if (pfx_length + sfx_length) {
663af342
PH
1539 strbuf_add(&name, a, pfx_length);
1540 strbuf_addch(&name, '{');
6973dcae 1541 }
663af342
PH
1542 strbuf_add(&name, a + pfx_length, a_midlen);
1543 strbuf_addstr(&name, " => ");
1544 strbuf_add(&name, b + pfx_length, b_midlen);
1545 if (pfx_length + sfx_length) {
1546 strbuf_addch(&name, '}');
1547 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
6973dcae 1548 }
b315c5c0 1549 return strbuf_detach(&name, NULL);
6973dcae
JH
1550}
1551
1552struct diffstat_t {
6973dcae
JH
1553 int nr;
1554 int alloc;
1555 struct diffstat_file {
f604652e 1556 char *from_name;
6973dcae 1557 char *name;
f604652e 1558 char *print_name;
6973dcae
JH
1559 unsigned is_unmerged:1;
1560 unsigned is_binary:1;
1561 unsigned is_renamed:1;
74faaa16 1562 unsigned is_interesting:1;
0974c117 1563 uintmax_t added, deleted;
6973dcae
JH
1564 } **files;
1565};
1566
1567static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1568 const char *name_a,
1569 const char *name_b)
1570{
1571 struct diffstat_file *x;
1a4927c5 1572 x = xcalloc(1, sizeof(*x));
4c960a43 1573 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
6973dcae
JH
1574 diffstat->files[diffstat->nr++] = x;
1575 if (name_b) {
f604652e
JH
1576 x->from_name = xstrdup(name_a);
1577 x->name = xstrdup(name_b);
6973dcae
JH
1578 x->is_renamed = 1;
1579 }
f604652e
JH
1580 else {
1581 x->from_name = NULL;
9befac47 1582 x->name = xstrdup(name_a);
f604652e 1583 }
6973dcae
JH
1584 return x;
1585}
1586
1587static void diffstat_consume(void *priv, char *line, unsigned long len)
1588{
1589 struct diffstat_t *diffstat = priv;
1590 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1591
1592 if (line[0] == '+')
1593 x->added++;
1594 else if (line[0] == '-')
1595 x->deleted++;
1596}
1597
698ce6f8 1598const char mime_boundary_leader[] = "------------";
6973dcae 1599
a2540023
JH
1600static int scale_linear(int it, int width, int max_change)
1601{
2eeeef24
JH
1602 if (!it)
1603 return 0;
a2540023 1604 /*
2eeeef24
JH
1605 * make sure that at least one '-' or '+' is printed if
1606 * there is any change to this path. The easiest way is to
1607 * scale linearly as if the alloted width is one column shorter
1608 * than it is, and then add 1 to the result.
a2540023 1609 */
2eeeef24 1610 return 1 + (it * (width - 1) / max_change);
a2540023
JH
1611}
1612
c0c77734 1613static void show_name(FILE *file,
a408e0e6 1614 const char *prefix, const char *name, int len)
a2540023 1615{
a408e0e6 1616 fprintf(file, " %s%-*s |", prefix, len, name);
a2540023
JH
1617}
1618
c0c77734 1619static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
a2540023
JH
1620{
1621 if (cnt <= 0)
1622 return;
c0c77734 1623 fprintf(file, "%s", set);
a2540023 1624 while (cnt--)
c0c77734
DB
1625 putc(ch, file);
1626 fprintf(file, "%s", reset);
a2540023
JH
1627}
1628
f604652e
JH
1629static void fill_print_name(struct diffstat_file *file)
1630{
1631 char *pname;
1632
1633 if (file->print_name)
1634 return;
1635
1636 if (!file->is_renamed) {
f285a2d7 1637 struct strbuf buf = STRBUF_INIT;
f604652e
JH
1638 if (quote_c_style(file->name, &buf, NULL, 0)) {
1639 pname = strbuf_detach(&buf, NULL);
1640 } else {
1641 pname = file->name;
1642 strbuf_release(&buf);
1643 }
1644 } else {
1645 pname = pprint_rename(file->from_name, file->name);
1646 }
1647 file->print_name = pname;
1648}
1649
7f814632
NTND
1650int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1651{
1652 struct strbuf sb = STRBUF_INIT;
1653 int ret;
1654
1655 if (!files) {
1656 assert(insertions == 0 && deletions == 0);
218adaaa 1657 return fprintf(fp, "%s\n", " 0 files changed");
7f814632
NTND
1658 }
1659
1660 strbuf_addf(&sb,
218adaaa 1661 (files == 1) ? " %d file changed" : " %d files changed",
7f814632
NTND
1662 files);
1663
1664 /*
1665 * For binary diff, the caller may want to print "x files
1666 * changed" with insertions == 0 && deletions == 0.
1667 *
1668 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1669 * is probably less confusing (i.e skip over "2 files changed
1670 * but nothing about added/removed lines? Is this a bug in Git?").
1671 */
1672 if (insertions || deletions == 0) {
7f814632 1673 strbuf_addf(&sb,
218adaaa 1674 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
7f814632
NTND
1675 insertions);
1676 }
1677
1678 if (deletions || insertions == 0) {
7f814632 1679 strbuf_addf(&sb,
218adaaa 1680 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
7f814632
NTND
1681 deletions);
1682 }
1683 strbuf_addch(&sb, '\n');
1684 ret = fputs(sb.buf, fp);
1685 strbuf_release(&sb);
1686 return ret;
1687}
1688
4b25d091 1689static void show_stats(struct diffstat_t *data, struct diff_options *options)
6973dcae 1690{
eb3a9dd3 1691 int i, len, add, del, adds = 0, dels = 0;
0974c117 1692 uintmax_t max_change = 0, max_len = 0;
dc801e71
ZJS
1693 int total_files = data->nr, count;
1694 int width, name_width, graph_width, number_width = 0, bin_width = 0;
c0aa335c 1695 const char *reset, *add_c, *del_c;
7be57610 1696 const char *line_prefix = "";
e5f85df8 1697 int extra_shown = 0;
6973dcae
JH
1698
1699 if (data->nr == 0)
1700 return;
1701
30997bb8 1702 line_prefix = diff_line_prefix(options);
808e1db2 1703 count = options->stat_count ? options->stat_count : data->nr;
a2540023 1704
8f67f8ae 1705 reset = diff_get_color_opt(options, DIFF_RESET);
8f67f8ae
PH
1706 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1707 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
785f7432 1708
1b058bc3
ZJS
1709 /*
1710 * Find the longest filename and max number of changes
1711 */
808e1db2 1712 for (i = 0; (i < count) && (i < data->nr); i++) {
6973dcae 1713 struct diffstat_file *file = data->files[i];
0974c117 1714 uintmax_t change = file->added + file->deleted;
af0ed819
JH
1715
1716 if (!file->is_interesting && (change == 0)) {
808e1db2 1717 count++; /* not shown == room for one more */
358e460e
MG
1718 continue;
1719 }
f604652e
JH
1720 fill_print_name(file);
1721 len = strlen(file->print_name);
6973dcae
JH
1722 if (max_len < len)
1723 max_len = len;
1724
dc801e71
ZJS
1725 if (file->is_unmerged) {
1726 /* "Unmerged" is 8 characters */
1727 bin_width = bin_width < 8 ? 8 : bin_width;
6973dcae 1728 continue;
dc801e71
ZJS
1729 }
1730 if (file->is_binary) {
1731 /* "Bin XXX -> YYY bytes" */
1732 int w = 14 + decimal_width(file->added)
1733 + decimal_width(file->deleted);
1734 bin_width = bin_width < w ? w : bin_width;
1735 /* Display change counts aligned with "Bin" */
1736 number_width = 3;
1737 continue;
1738 }
1739
a2540023
JH
1740 if (max_change < change)
1741 max_change = change;
6973dcae 1742 }
a20d3c0d 1743 count = i; /* where we can stop scanning in data->files[] */
6973dcae 1744
1b058bc3
ZJS
1745 /*
1746 * We have width = stat_width or term_columns() columns total.
1747 * We want a maximum of min(max_len, stat_name_width) for the name part.
969fe57b 1748 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1b058bc3
ZJS
1749 * We also need 1 for " " and 4 + decimal_width(max_change)
1750 * for " | NNNN " and one the empty column at the end, altogether
1751 * 6 + decimal_width(max_change).
1752 *
1753 * If there's not enough space, we will use the smaller of
1754 * stat_name_width (if set) and 5/8*width for the filename,
969fe57b
ZJS
1755 * and the rest for constant elements + graph part, but no more
1756 * than stat_graph_width for the graph part.
1b058bc3
ZJS
1757 * (5/8 gives 50 for filename and 30 for the constant parts + graph
1758 * for the standard terminal size).
a2540023 1759 *
1b058bc3
ZJS
1760 * In other words: stat_width limits the maximum width, and
1761 * stat_name_width fixes the maximum width of the filename,
1762 * and is also used to divide available columns if there
1763 * aren't enough.
dc801e71
ZJS
1764 *
1765 * Binary files are displayed with "Bin XXX -> YYY bytes"
1766 * instead of the change count and graph. This part is treated
1767 * similarly to the graph part, except that it is not
41ccfdd9 1768 * "scaled". If total width is too small to accommodate the
dc801e71
ZJS
1769 * guaranteed minimum width of the filename part and the
1770 * separators and this message, this message will "overflow"
1771 * making the line longer than the maximum width.
a2540023 1772 */
1b058bc3
ZJS
1773
1774 if (options->stat_width == -1)
cd48dadb 1775 width = term_columns() - strlen(line_prefix);
a2540023 1776 else
1b058bc3 1777 width = options->stat_width ? options->stat_width : 80;
dc801e71
ZJS
1778 number_width = decimal_width(max_change) > number_width ?
1779 decimal_width(max_change) : number_width;
a2540023 1780
df44483a
ZJS
1781 if (options->stat_graph_width == -1)
1782 options->stat_graph_width = diff_stat_graph_width;
a2540023 1783
1b058bc3
ZJS
1784 /*
1785 * Guarantee 3/8*16==6 for the graph part
1786 * and 5/8*16==10 for the filename part
1787 */
1788 if (width < 16 + 6 + number_width)
1789 width = 16 + 6 + number_width;
1790
1791 /*
1792 * First assign sizes that are wanted, ignoring available width.
dc801e71
ZJS
1793 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1794 * starting from "XXX" should fit in graph_width.
1b058bc3 1795 */
dc801e71
ZJS
1796 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1797 if (options->stat_graph_width &&
1798 options->stat_graph_width < graph_width)
1799 graph_width = options->stat_graph_width;
1800
1b058bc3
ZJS
1801 name_width = (options->stat_name_width > 0 &&
1802 options->stat_name_width < max_len) ?
1803 options->stat_name_width : max_len;
1804
1805 /*
1806 * Adjust adjustable widths not to exceed maximum width
1807 */
1808 if (name_width + number_width + 6 + graph_width > width) {
678c5741 1809 if (graph_width > width * 3/8 - number_width - 6) {
1b058bc3 1810 graph_width = width * 3/8 - number_width - 6;
678c5741
LP
1811 if (graph_width < 6)
1812 graph_width = 6;
1813 }
1814
969fe57b
ZJS
1815 if (options->stat_graph_width &&
1816 graph_width > options->stat_graph_width)
1817 graph_width = options->stat_graph_width;
1b058bc3
ZJS
1818 if (name_width > width - number_width - 6 - graph_width)
1819 name_width = width - number_width - 6 - graph_width;
1820 else
1821 graph_width = width - number_width - 6 - name_width;
1822 }
1823
1824 /*
1825 * From here name_width is the width of the name area,
1826 * and graph_width is the width of the graph area.
1827 * max_change is used to scale graph properly.
1828 */
808e1db2 1829 for (i = 0; i < count; i++) {
d2543b8e 1830 const char *prefix = "";
af0ed819
JH
1831 struct diffstat_file *file = data->files[i];
1832 char *name = file->print_name;
1833 uintmax_t added = file->added;
1834 uintmax_t deleted = file->deleted;
a2540023 1835 int name_len;
6973dcae 1836
a20d3c0d 1837 if (!file->is_interesting && (added + deleted == 0))
358e460e 1838 continue;
a20d3c0d 1839
6973dcae
JH
1840 /*
1841 * "scale" the filename
1842 */
a2540023
JH
1843 len = name_width;
1844 name_len = strlen(name);
1845 if (name_width < name_len) {
6973dcae
JH
1846 char *slash;
1847 prefix = "...";
a2540023
JH
1848 len -= 3;
1849 name += name_len - len;
6973dcae
JH
1850 slash = strchr(name, '/');
1851 if (slash)
1852 name = slash;
1853 }
6973dcae 1854
af0ed819 1855 if (file->is_binary) {
7be57610 1856 fprintf(options->file, "%s", line_prefix);
a408e0e6 1857 show_name(options->file, prefix, name, len);
fc1320bf 1858 fprintf(options->file, " %*s", number_width, "Bin");
e18872b2
ZJS
1859 if (!added && !deleted) {
1860 putc('\n', options->file);
1861 continue;
1862 }
1863 fprintf(options->file, " %s%"PRIuMAX"%s",
0974c117 1864 del_c, deleted, reset);
c0c77734 1865 fprintf(options->file, " -> ");
0974c117
JK
1866 fprintf(options->file, "%s%"PRIuMAX"%s",
1867 add_c, added, reset);
c0c77734
DB
1868 fprintf(options->file, " bytes");
1869 fprintf(options->file, "\n");
f604652e 1870 continue;
6973dcae 1871 }
af0ed819 1872 else if (file->is_unmerged) {
7be57610 1873 fprintf(options->file, "%s", line_prefix);
a408e0e6 1874 show_name(options->file, prefix, name, len);
dc801e71 1875 fprintf(options->file, " Unmerged\n");
f604652e 1876 continue;
6973dcae 1877 }
6973dcae 1878
a2540023
JH
1879 /*
1880 * scale the add/delete
1881 */
6973dcae
JH
1882 add = added;
1883 del = deleted;
6973dcae 1884
1b058bc3 1885 if (graph_width <= max_change) {
d3c9cf32 1886 int total = scale_linear(add + del, graph_width, max_change);
2eeeef24
JH
1887 if (total < 2 && add && del)
1888 /* width >= 2 due to the sanity check */
1889 total = 2;
1890 if (add < del) {
1b058bc3 1891 add = scale_linear(add, graph_width, max_change);
2eeeef24
JH
1892 del = total - add;
1893 } else {
1b058bc3 1894 del = scale_linear(del, graph_width, max_change);
2eeeef24
JH
1895 add = total - del;
1896 }
6973dcae 1897 }
7be57610 1898 fprintf(options->file, "%s", line_prefix);
a408e0e6 1899 show_name(options->file, prefix, name, len);
dc801e71
ZJS
1900 fprintf(options->file, " %*"PRIuMAX"%s",
1901 number_width, added + deleted,
1902 added + deleted ? " " : "");
c0c77734
DB
1903 show_graph(options->file, '+', add, add_c, reset);
1904 show_graph(options->file, '-', del, del_c, reset);
1905 fprintf(options->file, "\n");
1906 }
a20d3c0d
JH
1907
1908 for (i = 0; i < data->nr; i++) {
af0ed819
JH
1909 struct diffstat_file *file = data->files[i];
1910 uintmax_t added = file->added;
1911 uintmax_t deleted = file->deleted;
82dfc2c4
JH
1912
1913 if (file->is_unmerged ||
1914 (!file->is_interesting && (added + deleted == 0))) {
808e1db2
MG
1915 total_files--;
1916 continue;
1917 }
a20d3c0d 1918
82dfc2c4 1919 if (!file->is_binary) {
a20d3c0d
JH
1920 adds += added;
1921 dels += deleted;
1922 }
1923 if (i < count)
1924 continue;
e5f85df8
JH
1925 if (!extra_shown)
1926 fprintf(options->file, "%s ...\n", line_prefix);
1927 extra_shown = 1;
808e1db2 1928 }
7be57610 1929 fprintf(options->file, "%s", line_prefix);
7f814632 1930 print_stat_summary(options->file, total_files, adds, dels);
6973dcae
JH
1931}
1932
2775d92c 1933static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
ebd124c6
NP
1934{
1935 int i, adds = 0, dels = 0, total_files = data->nr;
1936
1937 if (data->nr == 0)
1938 return;
1939
1940 for (i = 0; i < data->nr; i++) {
e18872b2
ZJS
1941 int added = data->files[i]->added;
1942 int deleted= data->files[i]->deleted;
1943
20c8cde4
JH
1944 if (data->files[i]->is_unmerged ||
1945 (!data->files[i]->is_interesting && (added + deleted == 0))) {
e18872b2 1946 total_files--;
de9658b5 1947 } else if (!data->files[i]->is_binary) { /* don't count bytes */
e18872b2
ZJS
1948 adds += added;
1949 dels += deleted;
ebd124c6 1950 }
ebd124c6 1951 }
30997bb8 1952 fprintf(options->file, "%s", diff_line_prefix(options));
7f814632 1953 print_stat_summary(options->file, total_files, adds, dels);
ebd124c6
NP
1954}
1955
4b25d091 1956static void show_numstat(struct diffstat_t *data, struct diff_options *options)
74e2abe5
JH
1957{
1958 int i;
1959
f604652e
JH
1960 if (data->nr == 0)
1961 return;
1962
74e2abe5
JH
1963 for (i = 0; i < data->nr; i++) {
1964 struct diffstat_file *file = data->files[i];
1965
30997bb8 1966 fprintf(options->file, "%s", diff_line_prefix(options));
7be57610 1967
bfddbc5e 1968 if (file->is_binary)
c0c77734 1969 fprintf(options->file, "-\t-\t");
bfddbc5e 1970 else
c0c77734 1971 fprintf(options->file,
0974c117
JK
1972 "%"PRIuMAX"\t%"PRIuMAX"\t",
1973 file->added, file->deleted);
f604652e
JH
1974 if (options->line_termination) {
1975 fill_print_name(file);
1976 if (!file->is_renamed)
c0c77734 1977 write_name_quoted(file->name, options->file,
f604652e
JH
1978 options->line_termination);
1979 else {
c0c77734
DB
1980 fputs(file->print_name, options->file);
1981 putc(options->line_termination, options->file);
f604652e 1982 }
663af342 1983 } else {
f604652e 1984 if (file->is_renamed) {
c0c77734
DB
1985 putc('\0', options->file);
1986 write_name_quoted(file->from_name, options->file, '\0');
f604652e 1987 }
c0c77734 1988 write_name_quoted(file->name, options->file, '\0');
663af342 1989 }
74e2abe5
JH
1990 }
1991}
1992
c04a7155
JH
1993struct dirstat_file {
1994 const char *name;
1995 unsigned long changed;
7df7c019
LT
1996};
1997
c04a7155
JH
1998struct dirstat_dir {
1999 struct dirstat_file *files;
712d2c7d 2000 int alloc, nr, permille, cumulative;
c04a7155
JH
2001};
2002
7be57610
BY
2003static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2004 unsigned long changed, const char *base, int baselen)
7df7c019
LT
2005{
2006 unsigned long this_dir = 0;
2007 unsigned int sources = 0;
30997bb8 2008 const char *line_prefix = diff_line_prefix(opt);
7df7c019
LT
2009
2010 while (dir->nr) {
c04a7155 2011 struct dirstat_file *f = dir->files;
7df7c019
LT
2012 int namelen = strlen(f->name);
2013 unsigned long this;
2014 char *slash;
2015
2016 if (namelen < baselen)
2017 break;
2018 if (memcmp(f->name, base, baselen))
2019 break;
2020 slash = strchr(f->name + baselen, '/');
2021 if (slash) {
2022 int newbaselen = slash + 1 - f->name;
7be57610 2023 this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
7df7c019
LT
2024 sources++;
2025 } else {
c04a7155 2026 this = f->changed;
7df7c019
LT
2027 dir->files++;
2028 dir->nr--;
2029 sources += 2;
2030 }
2031 this_dir += this;
2032 }
2033
2034 /*
2035 * We don't report dirstat's for
2036 * - the top level
2037 * - or cases where everything came from a single directory
2038 * under this directory (sources == 1).
2039 */
2040 if (baselen && sources != 1) {
58a8756a
JH
2041 if (this_dir) {
2042 int permille = this_dir * 1000 / changed;
712d2c7d 2043 if (permille >= dir->permille) {
7be57610 2044 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
712d2c7d 2045 permille / 10, permille % 10, baselen, base);
7df7c019
LT
2046 if (!dir->cumulative)
2047 return 0;
2048 }
2049 }
2050 }
2051 return this_dir;
2052}
2053
441bca0b
LT
2054static int dirstat_compare(const void *_a, const void *_b)
2055{
2056 const struct dirstat_file *a = _a;
2057 const struct dirstat_file *b = _b;
2058 return strcmp(a->name, b->name);
2059}
2060
c04a7155 2061static void show_dirstat(struct diff_options *options)
7df7c019
LT
2062{
2063 int i;
2064 unsigned long changed;
c04a7155
JH
2065 struct dirstat_dir dir;
2066 struct diff_queue_struct *q = &diff_queued_diff;
2067
2068 dir.files = NULL;
2069 dir.alloc = 0;
2070 dir.nr = 0;
712d2c7d 2071 dir.permille = options->dirstat_permille;
f88d225f 2072 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
7df7c019 2073
7df7c019 2074 changed = 0;
c04a7155
JH
2075 for (i = 0; i < q->nr; i++) {
2076 struct diff_filepair *p = q->queue[i];
2077 const char *name;
2078 unsigned long copied, added, damage;
0133dab7 2079 int content_changed;
c04a7155 2080
2ca86714 2081 name = p->two->path ? p->two->path : p->one->path;
c04a7155 2082
41c9560e 2083 if (p->one->oid_valid && p->two->oid_valid)
a0d12c44 2084 content_changed = oidcmp(&p->one->oid, &p->two->oid);
0133dab7
JH
2085 else
2086 content_changed = 1;
2087
2ff3a803
JH
2088 if (!content_changed) {
2089 /*
2090 * The SHA1 has not changed, so pre-/post-content is
2091 * identical. We can therefore skip looking at the
2092 * file contents altogether.
2093 */
2094 damage = 0;
2095 goto found_damage;
2096 }
2097
0133dab7
JH
2098 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
2099 /*
2100 * In --dirstat-by-file mode, we don't really need to
2101 * look at the actual file contents at all.
2102 * The fact that the SHA1 changed is enough for us to
2103 * add this file to the list of results
2104 * (with each file contributing equal damage).
2105 */
2ff3a803 2106 damage = 1;
0133dab7
JH
2107 goto found_damage;
2108 }
c04a7155
JH
2109
2110 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2111 diff_populate_filespec(p->one, 0);
2112 diff_populate_filespec(p->two, 0);
974e0044 2113 diffcore_count_changes(p->one, p->two, NULL, NULL,
c04a7155
JH
2114 &copied, &added);
2115 diff_free_filespec_data(p->one);
2116 diff_free_filespec_data(p->two);
2117 } else if (DIFF_FILE_VALID(p->one)) {
8e5dd3d6 2118 diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
c04a7155
JH
2119 copied = added = 0;
2120 diff_free_filespec_data(p->one);
2121 } else if (DIFF_FILE_VALID(p->two)) {
8e5dd3d6 2122 diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
c04a7155
JH
2123 copied = 0;
2124 added = p->two->size;
2125 diff_free_filespec_data(p->two);
2126 } else
2b0b551d 2127 continue;
c04a7155
JH
2128
2129 /*
2130 * Original minus copied is the removed material,
2131 * added is the new material. They are both damages
0133dab7 2132 * made to the preimage.
2ff3a803
JH
2133 * If the resulting damage is zero, we know that
2134 * diffcore_count_changes() considers the two entries to
2135 * be identical, but since content_changed is true, we
2136 * know that there must have been _some_ kind of change,
2137 * so we force all entries to have damage > 0.
c04a7155
JH
2138 */
2139 damage = (p->one->size - copied) + added;
2ff3a803 2140 if (!damage)
fd33777b 2141 damage = 1;
c04a7155 2142
0133dab7 2143found_damage:
c04a7155
JH
2144 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2145 dir.files[dir.nr].name = name;
2146 dir.files[dir.nr].changed = damage;
2147 changed += damage;
2148 dir.nr++;
7df7c019
LT
2149 }
2150
2151 /* This can happen even with many files, if everything was renames */
2152 if (!changed)
2153 return;
2154
2155 /* Show all directories with more than x% of the changes */
9ed0d8d6 2156 QSORT(dir.files, dir.nr, dirstat_compare);
7be57610 2157 gather_dirstat(options, &dir, changed, "", 0);
7df7c019
LT
2158}
2159
1c57a627
JH
2160static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2161{
2162 int i;
2163 unsigned long changed;
2164 struct dirstat_dir dir;
2165
2166 if (data->nr == 0)
2167 return;
2168
2169 dir.files = NULL;
2170 dir.alloc = 0;
2171 dir.nr = 0;
2172 dir.permille = options->dirstat_permille;
2173 dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2174
2175 changed = 0;
2176 for (i = 0; i < data->nr; i++) {
2177 struct diffstat_file *file = data->files[i];
2178 unsigned long damage = file->added + file->deleted;
2179 if (file->is_binary)
2180 /*
2181 * binary files counts bytes, not lines. Must find some
2182 * way to normalize binary bytes vs. textual lines.
2183 * The following heuristic assumes that there are 64
2184 * bytes per "line".
2185 * This is stupid and ugly, but very cheap...
2186 */
2187 damage = (damage + 63) / 64;
2188 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2189 dir.files[dir.nr].name = file->name;
2190 dir.files[dir.nr].changed = damage;
2191 changed += damage;
2192 dir.nr++;
2193 }
2194
2195 /* This can happen even with many files, if everything was renames */
2196 if (!changed)
2197 return;
2198
2199 /* Show all directories with more than x% of the changes */
9ed0d8d6 2200 QSORT(dir.files, dir.nr, dirstat_compare);
1c57a627
JH
2201 gather_dirstat(options, &dir, changed, "", 0);
2202}
2203
f604652e
JH
2204static void free_diffstat_info(struct diffstat_t *diffstat)
2205{
2206 int i;
2207 for (i = 0; i < diffstat->nr; i++) {
2208 struct diffstat_file *f = diffstat->files[i];
2209 if (f->name != f->print_name)
2210 free(f->print_name);
2211 free(f->name);
2212 free(f->from_name);
2213 free(f);
2214 }
2215 free(diffstat->files);
2216}
2217
88246898 2218struct checkdiff_t {
88246898 2219 const char *filename;
1ba111d1 2220 int lineno;
a757c646 2221 int conflict_marker_size;
1ba111d1 2222 struct diff_options *o;
cf1b7869 2223 unsigned ws_rule;
62c64895 2224 unsigned status;
88246898
JS
2225};
2226
a757c646 2227static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
04954043
JH
2228{
2229 char firstchar;
2230 int cnt;
2231
a757c646 2232 if (len < marker_size + 1)
04954043
JH
2233 return 0;
2234 firstchar = line[0];
2235 switch (firstchar) {
a757c646 2236 case '=': case '>': case '<': case '|':
04954043
JH
2237 break;
2238 default:
2239 return 0;
2240 }
a757c646 2241 for (cnt = 1; cnt < marker_size; cnt++)
04954043
JH
2242 if (line[cnt] != firstchar)
2243 return 0;
a757c646
JH
2244 /* line[1] thru line[marker_size-1] are same as firstchar */
2245 if (len < marker_size + 1 || !isspace(line[marker_size]))
04954043 2246 return 0;
04954043
JH
2247 return 1;
2248}
2249
88246898
JS
2250static void checkdiff_consume(void *priv, char *line, unsigned long len)
2251{
2252 struct checkdiff_t *data = priv;
a757c646 2253 int marker_size = data->conflict_marker_size;
f1c96261
JK
2254 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2255 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2256 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
c1795bb0 2257 char *err;
30997bb8 2258 const char *line_prefix;
7be57610
BY
2259
2260 assert(data->o);
30997bb8 2261 line_prefix = diff_line_prefix(data->o);
88246898
JS
2262
2263 if (line[0] == '+') {
18374e58 2264 unsigned bad;
0ef617f4 2265 data->lineno++;
a757c646 2266 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
04954043
JH
2267 data->status |= 1;
2268 fprintf(data->o->file,
7be57610
BY
2269 "%s%s:%d: leftover conflict marker\n",
2270 line_prefix, data->filename, data->lineno);
04954043 2271 }
8f8841e9 2272 bad = ws_check(line + 1, len - 1, data->ws_rule);
18374e58 2273 if (!bad)
c1795bb0 2274 return;
18374e58
JH
2275 data->status |= bad;
2276 err = whitespace_error_string(bad);
7be57610
BY
2277 fprintf(data->o->file, "%s%s:%d: %s.\n",
2278 line_prefix, data->filename, data->lineno, err);
c1795bb0 2279 free(err);
a3c158d4 2280 emit_line(data->o, set, reset, line, 1);
8f8841e9 2281 ws_check_emit(line + 1, len - 1, data->ws_rule,
1ba111d1 2282 data->o->file, set, reset, ws);
877f23cc 2283 } else if (line[0] == ' ') {
88246898 2284 data->lineno++;
877f23cc 2285 } else if (line[0] == '@') {
88246898
JS
2286 char *plus = strchr(line, '+');
2287 if (plus)
0ef617f4 2288 data->lineno = strtol(plus, NULL, 10) - 1;
88246898
JS
2289 else
2290 die("invalid diff");
2291 }
2292}
2293
0660626c
JH
2294static unsigned char *deflate_it(char *data,
2295 unsigned long size,
2296 unsigned long *result_size)
051308f6 2297{
0660626c
JH
2298 int bound;
2299 unsigned char *deflated;
ef49a7a0 2300 git_zstream stream;
0660626c 2301
55bb5c91 2302 git_deflate_init(&stream, zlib_compression_level);
225a6f10 2303 bound = git_deflate_bound(&stream, size);
0660626c
JH
2304 deflated = xmalloc(bound);
2305 stream.next_out = deflated;
2306 stream.avail_out = bound;
2307
2308 stream.next_in = (unsigned char *)data;
2309 stream.avail_in = size;
55bb5c91 2310 while (git_deflate(&stream, Z_FINISH) == Z_OK)
0660626c 2311 ; /* nothing */
55bb5c91 2312 git_deflate_end(&stream);
0660626c
JH
2313 *result_size = stream.total_out;
2314 return deflated;
051308f6
JH
2315}
2316
32b367e4
JK
2317static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2318 const char *prefix)
051308f6 2319{
0660626c
JH
2320 void *cp;
2321 void *delta;
2322 void *deflated;
2323 void *data;
2324 unsigned long orig_size;
2325 unsigned long delta_size;
2326 unsigned long deflate_size;
2327 unsigned long data_size;
051308f6 2328
0660626c
JH
2329 /* We could do deflated delta, or we could do just deflated two,
2330 * whichever is smaller.
051308f6 2331 */
0660626c
JH
2332 delta = NULL;
2333 deflated = deflate_it(two->ptr, two->size, &deflate_size);
2334 if (one->size && two->size) {
2335 delta = diff_delta(one->ptr, one->size,
2336 two->ptr, two->size,
2337 &delta_size, deflate_size);
2338 if (delta) {
2339 void *to_free = delta;
2340 orig_size = delta_size;
2341 delta = deflate_it(delta, delta_size, &delta_size);
2342 free(to_free);
051308f6
JH
2343 }
2344 }
051308f6 2345
0660626c 2346 if (delta && delta_size < deflate_size) {
7be57610 2347 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
0660626c
JH
2348 free(deflated);
2349 data = delta;
2350 data_size = delta_size;
2351 }
2352 else {
7be57610 2353 fprintf(file, "%sliteral %lu\n", prefix, two->size);
0660626c
JH
2354 free(delta);
2355 data = deflated;
2356 data_size = deflate_size;
2357 }
051308f6 2358
0660626c
JH
2359 /* emit data encoded in base85 */
2360 cp = data;
2361 while (data_size) {
2362 int bytes = (52 < data_size) ? 52 : data_size;
051308f6 2363 char line[70];
0660626c 2364 data_size -= bytes;
051308f6
JH
2365 if (bytes <= 26)
2366 line[0] = bytes + 'A' - 1;
2367 else
2368 line[0] = bytes - 26 + 'a' - 1;
2369 encode_85(line + 1, cp, bytes);
1d7f171c 2370 cp = (char *) cp + bytes;
7be57610 2371 fprintf(file, "%s", prefix);
c0c77734
DB
2372 fputs(line, file);
2373 fputc('\n', file);
051308f6 2374 }
7be57610 2375 fprintf(file, "%s\n", prefix);
0660626c 2376 free(data);
051308f6
JH
2377}
2378
32b367e4
JK
2379static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2380 const char *prefix)
d4c452f0 2381{
7be57610
BY
2382 fprintf(file, "%sGIT binary patch\n", prefix);
2383 emit_binary_diff_body(file, one, two, prefix);
2384 emit_binary_diff_body(file, two, one, prefix);
d4c452f0
JH
2385}
2386
29a3eefd
JH
2387int diff_filespec_is_binary(struct diff_filespec *one)
2388{
122aa6f9
JK
2389 if (one->is_binary == -1) {
2390 diff_filespec_load_driver(one);
2391 if (one->driver->binary != -1)
2392 one->is_binary = one->driver->binary;
2393 else {
2394 if (!one->data && DIFF_FILE_VALID(one))
6bf3b813
NTND
2395 diff_populate_filespec(one, CHECK_BINARY);
2396 if (one->is_binary == -1 && one->data)
122aa6f9
JK
2397 one->is_binary = buffer_is_binary(one->data,
2398 one->size);
2399 if (one->is_binary == -1)
2400 one->is_binary = 0;
2401 }
2402 }
29a3eefd 2403 return one->is_binary;
6973dcae
JH
2404}
2405
be58e70d 2406static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
f258475a 2407{
122aa6f9
JK
2408 diff_filespec_load_driver(one);
2409 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
f258475a
JH
2410}
2411
a5a818ee
JH
2412void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2413{
2414 if (!options->a_prefix)
2415 options->a_prefix = a;
2416 if (!options->b_prefix)
2417 options->b_prefix = b;
2418}
2419
a788d7d5 2420struct userdiff_driver *get_textconv(struct diff_filespec *one)
04427ac8
JK
2421{
2422 if (!DIFF_FILE_VALID(one))
2423 return NULL;
d391c0ff 2424
04427ac8 2425 diff_filespec_load_driver(one);
3813e690 2426 return userdiff_get_textconv(one->driver);
04427ac8
JK
2427}
2428
6973dcae
JH
2429static void builtin_diff(const char *name_a,
2430 const char *name_b,
2431 struct diff_filespec *one,
2432 struct diff_filespec *two,
2433 const char *xfrm_msg,
296c6bb2 2434 int must_show_header,
051308f6 2435 struct diff_options *o,
6973dcae
JH
2436 int complete_rewrite)
2437{
2438 mmfile_t mf1, mf2;
2439 const char *lbl[2];
2440 char *a_one, *b_two;
d9c552f1 2441 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
8f67f8ae 2442 const char *reset = diff_get_color_opt(o, DIFF_RESET);
a5a818ee 2443 const char *a_prefix, *b_prefix;
d9bae1a1
JK
2444 struct userdiff_driver *textconv_one = NULL;
2445 struct userdiff_driver *textconv_two = NULL;
3e97c7c6 2446 struct strbuf header = STRBUF_INIT;
30997bb8 2447 const char *line_prefix = diff_line_prefix(o);
a5a818ee 2448
fd47ae6a
JK
2449 diff_set_mnemonic_prefix(o, "a/", "b/");
2450 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2451 a_prefix = o->b_prefix;
2452 b_prefix = o->a_prefix;
2453 } else {
2454 a_prefix = o->a_prefix;
2455 b_prefix = o->b_prefix;
2456 }
2457
61cfbc05
JK
2458 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
2459 (!one->mode || S_ISGITLINK(one->mode)) &&
2460 (!two->mode || S_ISGITLINK(two->mode))) {
752c0c24
JS
2461 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2462 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
3b0c18af 2463 show_submodule_summary(o->file, one->path ? one->path : two->path,
0f33a067 2464 line_prefix,
602a283a 2465 &one->oid, &two->oid,
a0d12c44 2466 two->dirty_submodule,
4e215131 2467 meta, del, add, reset);
752c0c24 2468 return;
fd47ae6a
JK
2469 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
2470 (!one->mode || S_ISGITLINK(one->mode)) &&
2471 (!two->mode || S_ISGITLINK(two->mode))) {
2472 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2473 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2474 show_submodule_inline_diff(o->file, one->path ? one->path : two->path,
2475 line_prefix,
2476 &one->oid, &two->oid,
2477 two->dirty_submodule,
2478 meta, del, add, reset, o);
2479 return;
752c0c24
JS
2480 }
2481
3aa1f7ca
JK
2482 if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2483 textconv_one = get_textconv(one);
2484 textconv_two = get_textconv(two);
2485 }
2486
71b989e7
LT
2487 /* Never use a non-valid filename anywhere if at all possible */
2488 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2489 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2490
a5a818ee
JH
2491 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2492 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
6973dcae
JH
2493 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2494 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
d9c552f1 2495 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
6973dcae
JH
2496 if (lbl[0][0] == '/') {
2497 /* /dev/null */
d9c552f1 2498 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
37466447
BW
2499 if (xfrm_msg)
2500 strbuf_addstr(&header, xfrm_msg);
296c6bb2 2501 must_show_header = 1;
6973dcae
JH
2502 }
2503 else if (lbl[1][0] == '/') {
d9c552f1 2504 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
37466447
BW
2505 if (xfrm_msg)
2506 strbuf_addstr(&header, xfrm_msg);
296c6bb2 2507 must_show_header = 1;
6973dcae
JH
2508 }
2509 else {
2510 if (one->mode != two->mode) {
d9c552f1
JK
2511 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2512 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
296c6bb2 2513 must_show_header = 1;
cd112cef 2514 }
37466447
BW
2515 if (xfrm_msg)
2516 strbuf_addstr(&header, xfrm_msg);
3e97c7c6 2517
6973dcae
JH
2518 /*
2519 * we do not run diff between different kind
2520 * of objects.
2521 */
2522 if ((one->mode ^ two->mode) & S_IFMT)
2523 goto free_ab_and_return;
0c01857d 2524 if (complete_rewrite &&
3aa1f7ca
JK
2525 (textconv_one || !diff_filespec_is_binary(one)) &&
2526 (textconv_two || !diff_filespec_is_binary(two))) {
a29b0a13
SB
2527 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2528 header.buf, header.len, 0);
3e97c7c6 2529 strbuf_reset(&header);
3aa1f7ca
JK
2530 emit_rewrite_diff(name_a, name_b, one, two,
2531 textconv_one, textconv_two, o);
34a5e1a2 2532 o->found_changes = 1;
6973dcae
JH
2533 goto free_ab_and_return;
2534 }
2535 }
2536
467ddc14 2537 if (o->irreversible_delete && lbl[1][0] == '/') {
a29b0a13
SB
2538 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
2539 header.len, 0);
467ddc14
JH
2540 strbuf_reset(&header);
2541 goto free_ab_and_return;
2542 } else if (!DIFF_OPT_TST(o, TEXT) &&
b3373982
JK
2543 ( (!textconv_one && diff_filespec_is_binary(one)) ||
2544 (!textconv_two && diff_filespec_is_binary(two)) )) {
1aaf69e6
NTND
2545 if (!one->data && !two->data &&
2546 S_ISREG(one->mode) && S_ISREG(two->mode) &&
2547 !DIFF_OPT_TST(o, BINARY)) {
a0d12c44 2548 if (!oidcmp(&one->oid, &two->oid)) {
1aaf69e6 2549 if (must_show_header)
a29b0a13
SB
2550 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2551 header.buf, header.len,
2552 0);
1aaf69e6
NTND
2553 goto free_ab_and_return;
2554 }
a29b0a13
SB
2555 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2556 header.buf, header.len, 0);
1aaf69e6
NTND
2557 fprintf(o->file, "%sBinary files %s and %s differ\n",
2558 line_prefix, lbl[0], lbl[1]);
2559 goto free_ab_and_return;
2560 }
b3373982
JK
2561 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2562 die("unable to read files to diff");
0660626c
JH
2563 /* Quite common confusing case */
2564 if (mf1.size == mf2.size &&
296c6bb2
CC
2565 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2566 if (must_show_header)
a29b0a13
SB
2567 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2568 header.buf, header.len, 0);
0660626c 2569 goto free_ab_and_return;
296c6bb2 2570 }
a29b0a13 2571 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3e97c7c6 2572 strbuf_reset(&header);
8f67f8ae 2573 if (DIFF_OPT_TST(o, BINARY))
7be57610 2574 emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
051308f6 2575 else
7be57610
BY
2576 fprintf(o->file, "%sBinary files %s and %s differ\n",
2577 line_prefix, lbl[0], lbl[1]);
34a5e1a2 2578 o->found_changes = 1;
467ddc14 2579 } else {
6973dcae
JH
2580 /* Crazy xdl interfaces.. */
2581 const char *diffopts = getenv("GIT_DIFF_OPTS");
ae021d87 2582 const char *v;
6973dcae
JH
2583 xpparam_t xpp;
2584 xdemitconf_t xecfg;
6973dcae 2585 struct emit_callback ecbdata;
be58e70d 2586 const struct userdiff_funcname *pe;
f258475a 2587
b3f01ff2 2588 if (must_show_header) {
a29b0a13
SB
2589 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2590 header.buf, header.len, 0);
3e97c7c6
GB
2591 strbuf_reset(&header);
2592 }
2593
840383b2
JK
2594 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2595 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
04427ac8 2596
45e7ca0f
BC
2597 pe = diff_funcname_pattern(one);
2598 if (!pe)
2599 pe = diff_funcname_pattern(two);
6973dcae 2600
9ccd0a88 2601 memset(&xpp, 0, sizeof(xpp));
30b25010 2602 memset(&xecfg, 0, sizeof(xecfg));
cd112cef 2603 memset(&ecbdata, 0, sizeof(ecbdata));
6973dcae 2604 ecbdata.label_path = lbl;
daa0c3d9 2605 ecbdata.color_diff = want_color(o->use_color);
c189c4f2 2606 ecbdata.ws_rule = whitespace_rule(name_b);
690ed843 2607 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
d68fe26f 2608 check_blank_at_eof(&mf1, &mf2, &ecbdata);
a3c158d4 2609 ecbdata.opt = o;
3e97c7c6 2610 ecbdata.header = header.len ? &header : NULL;
582aa00b 2611 xpp.flags = o->xdl_opts;
ee1e5412 2612 xecfg.ctxlen = o->context;
6d0e674a 2613 xecfg.interhunkctxlen = o->interhunkcontext;
6973dcae 2614 xecfg.flags = XDL_EMIT_FUNCNAMES;
14937c2c
RS
2615 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2616 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
45e7ca0f 2617 if (pe)
a013585b 2618 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
6973dcae
JH
2619 if (!diffopts)
2620 ;
ae021d87
JK
2621 else if (skip_prefix(diffopts, "--unified=", &v))
2622 xecfg.ctxlen = strtoul(v, NULL, 10);
2623 else if (skip_prefix(diffopts, "-u", &v))
2624 xecfg.ctxlen = strtoul(v, NULL, 10);
77d1a520
TR
2625 if (o->word_diff)
2626 init_diff_words_data(&ecbdata, o, one, two);
3efb9880
JK
2627 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2628 &xpp, &xecfg))
2629 die("unable to generate diff for %s", one->path);
882749a0 2630 if (o->word_diff)
f59a59e2 2631 free_diff_words_data(&ecbdata);
04427ac8
JK
2632 if (textconv_one)
2633 free(mf1.ptr);
2634 if (textconv_two)
2635 free(mf2.ptr);
8cfe5f1c 2636 xdiff_clear_find_func(&xecfg);
6973dcae
JH
2637 }
2638
2639 free_ab_and_return:
3e97c7c6 2640 strbuf_release(&header);
fc3abdf5
JH
2641 diff_free_filespec_data(one);
2642 diff_free_filespec_data(two);
6973dcae
JH
2643 free(a_one);
2644 free(b_two);
2645 return;
2646}
2647
2648static void builtin_diffstat(const char *name_a, const char *name_b,
2649 struct diff_filespec *one,
2650 struct diff_filespec *two,
710158e3 2651 struct diffstat_t *diffstat,
0d21efa5 2652 struct diff_options *o,
74faaa16 2653 struct diff_filepair *p)
6973dcae
JH
2654{
2655 mmfile_t mf1, mf2;
2656 struct diffstat_file *data;
352ca4e1 2657 int same_contents;
74faaa16
LT
2658 int complete_rewrite = 0;
2659
2660 if (!DIFF_PAIR_UNMERGED(p)) {
2661 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2662 complete_rewrite = 1;
2663 }
6973dcae
JH
2664
2665 data = diffstat_add(diffstat, name_a, name_b);
99bfd407 2666 data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
6973dcae
JH
2667
2668 if (!one || !two) {
2669 data->is_unmerged = 1;
2670 return;
2671 }
ded0abc7 2672
a0d12c44 2673 same_contents = !oidcmp(&one->oid, &two->oid);
352ca4e1 2674
ded0abc7 2675 if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
ded0abc7 2676 data->is_binary = 1;
352ca4e1 2677 if (same_contents) {
e18872b2
ZJS
2678 data->added = 0;
2679 data->deleted = 0;
2680 } else {
2681 data->added = diff_filespec_size(two);
2682 data->deleted = diff_filespec_size(one);
2683 }
ded0abc7
JK
2684 }
2685
2686 else if (complete_rewrite) {
710158e3
JH
2687 diff_populate_filespec(one, 0);
2688 diff_populate_filespec(two, 0);
2689 data->deleted = count_lines(one->data, one->size);
2690 data->added = count_lines(two->data, two->size);
710158e3 2691 }
6973dcae 2692
352ca4e1 2693 else if (!same_contents) {
6973dcae
JH
2694 /* Crazy xdl interfaces.. */
2695 xpparam_t xpp;
2696 xdemitconf_t xecfg;
6973dcae 2697
ded0abc7
JK
2698 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2699 die("unable to read files to diff");
2700
9ccd0a88 2701 memset(&xpp, 0, sizeof(xpp));
30b25010 2702 memset(&xecfg, 0, sizeof(xecfg));
582aa00b 2703 xpp.flags = o->xdl_opts;
f01cae91
JH
2704 xecfg.ctxlen = o->context;
2705 xecfg.interhunkctxlen = o->interhunkcontext;
3efb9880
JK
2706 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2707 &xpp, &xecfg))
2708 die("unable to generate diffstat for %s", one->path);
6973dcae 2709 }
fc3abdf5 2710
fc3abdf5
JH
2711 diff_free_filespec_data(one);
2712 diff_free_filespec_data(two);
6973dcae
JH
2713}
2714
88246898 2715static void builtin_checkdiff(const char *name_a, const char *name_b,
cd676a51 2716 const char *attr_path,
5ff10dd6
JH
2717 struct diff_filespec *one,
2718 struct diff_filespec *two,
2719 struct diff_options *o)
88246898
JS
2720{
2721 mmfile_t mf1, mf2;
2722 struct checkdiff_t data;
2723
2724 if (!two)
2725 return;
2726
2727 memset(&data, 0, sizeof(data));
88246898
JS
2728 data.filename = name_b ? name_b : name_a;
2729 data.lineno = 0;
1ba111d1 2730 data.o = o;
cd676a51 2731 data.ws_rule = whitespace_rule(attr_path);
a757c646 2732 data.conflict_marker_size = ll_merge_marker_size(attr_path);
88246898
JS
2733
2734 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2735 die("unable to read files to diff");
2736
5ff10dd6
JH
2737 /*
2738 * All the other codepaths check both sides, but not checking
2739 * the "old" side here is deliberate. We are checking the newly
2740 * introduced changes, and as long as the "new" side is text, we
2741 * can and should check what it introduces.
2742 */
29a3eefd 2743 if (diff_filespec_is_binary(two))
fc3abdf5 2744 goto free_and_return;
88246898
JS
2745 else {
2746 /* Crazy xdl interfaces.. */
2747 xpparam_t xpp;
2748 xdemitconf_t xecfg;
88246898 2749
9ccd0a88 2750 memset(&xpp, 0, sizeof(xpp));
30b25010 2751 memset(&xecfg, 0, sizeof(xecfg));
c35539eb 2752 xecfg.ctxlen = 1; /* at least one context line */
582aa00b 2753 xpp.flags = 0;
3efb9880
JK
2754 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2755 &xpp, &xecfg))
2756 die("unable to generate checkdiff for %s", one->path);
877f23cc 2757
467babf8 2758 if (data.ws_rule & WS_BLANK_AT_EOF) {
d68fe26f
JH
2759 struct emit_callback ecbdata;
2760 int blank_at_eof;
2761
2762 ecbdata.ws_rule = data.ws_rule;
2763 check_blank_at_eof(&mf1, &mf2, &ecbdata);
8837d335 2764 blank_at_eof = ecbdata.blank_at_eof_in_postimage;
d68fe26f 2765
467babf8
JH
2766 if (blank_at_eof) {
2767 static char *err;
2768 if (!err)
2769 err = whitespace_error_string(WS_BLANK_AT_EOF);
2770 fprintf(o->file, "%s:%d: %s.\n",
2771 data.filename, blank_at_eof, err);
2772 data.status = 1; /* report errors */
2773 }
877f23cc 2774 }
88246898 2775 }
fc3abdf5
JH
2776 free_and_return:
2777 diff_free_filespec_data(one);
2778 diff_free_filespec_data(two);
62c64895
WC
2779 if (data.status)
2780 DIFF_OPT_SET(o, CHECK_FAILED);
88246898
JS
2781}
2782
6973dcae
JH
2783struct diff_filespec *alloc_filespec(const char *path)
2784{
96ffc06f 2785 struct diff_filespec *spec;
6973dcae 2786
96ffc06f 2787 FLEXPTR_ALLOC_STR(spec, path, path);
9fb88419 2788 spec->count = 1;
122aa6f9 2789 spec->is_binary = -1;
6973dcae
JH
2790 return spec;
2791}
2792
9fb88419
LT
2793void free_filespec(struct diff_filespec *spec)
2794{
2795 if (!--spec->count) {
2796 diff_free_filespec_data(spec);
2797 free(spec);
2798 }
2799}
2800
f9704c2d
BW
2801void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
2802 int oid_valid, unsigned short mode)
6973dcae
JH
2803{
2804 if (mode) {
2805 spec->mode = canon_mode(mode);
f9704c2d
BW
2806 oidcpy(&spec->oid, oid);
2807 spec->oid_valid = oid_valid;
6973dcae
JH
2808 }
2809}
2810
2811/*
5adf317b 2812 * Given a name and sha1 pair, if the index tells us the file in
6973dcae
JH
2813 * the work tree has that object contents, return true, so that
2814 * prepare_temp_file() does not have to inflate and extract.
2815 */
fb4a1c0d 2816static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
6973dcae 2817{
9c5e6c80 2818 const struct cache_entry *ce;
6973dcae
JH
2819 struct stat st;
2820 int pos, len;
2821
150115ad
JH
2822 /*
2823 * We do not read the cache ourselves here, because the
6973dcae
JH
2824 * benchmark with my previous version that always reads cache
2825 * shows that it makes things worse for diff-tree comparing
2826 * two linux-2.6 kernel trees in an already checked out work
2827 * tree. This is because most diff-tree comparisons deal with
2828 * only a small number of files, while reading the cache is
2829 * expensive for a large project, and its cost outweighs the
2830 * savings we get by not inflating the object to a temporary
2831 * file. Practically, this code only helps when we are used
2832 * by diff-cache --cached, which does read the cache before
2833 * calling us.
2834 */
2835 if (!active_cache)
2836 return 0;
2837
1510fea7
SP
2838 /* We want to avoid the working directory if our caller
2839 * doesn't need the data in a normal file, this system
2840 * is rather slow with its stat/open/mmap/close syscalls,
2841 * and the object is contained in a pack file. The pack
2842 * is probably already open and will be faster to obtain
2843 * the data through than the working directory. Loose
2844 * objects however would tend to be slower as they need
2845 * to be individually opened and inflated.
2846 */
fb4a1c0d 2847 if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
1510fea7
SP
2848 return 0;
2849
06dec439
JK
2850 /*
2851 * Similarly, if we'd have to convert the file contents anyway, that
2852 * makes the optimization not worthwhile.
2853 */
82b474e0 2854 if (!want_file && would_convert_to_git(&the_index, name))
06dec439
JK
2855 return 0;
2856
6973dcae
JH
2857 len = strlen(name);
2858 pos = cache_name_pos(name, len);
2859 if (pos < 0)
2860 return 0;
2861 ce = active_cache[pos];
eadb5831
JH
2862
2863 /*
2864 * This is not the sha1 we are looking for, or
2865 * unreusable because it is not a regular file.
2866 */
fb4a1c0d 2867 if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
6973dcae 2868 return 0;
eadb5831 2869
150115ad
JH
2870 /*
2871 * If ce is marked as "assume unchanged", there is no
2872 * guarantee that work tree matches what we are looking for.
2873 */
b4d1690d 2874 if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
150115ad
JH
2875 return 0;
2876
eadb5831
JH
2877 /*
2878 * If ce matches the file in the work tree, we can reuse it.
6973dcae 2879 */
eadb5831
JH
2880 if (ce_uptodate(ce) ||
2881 (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2882 return 1;
2883
2884 return 0;
6973dcae
JH
2885}
2886
04786756
LT
2887static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2888{
b1ddfb91
JK
2889 struct strbuf buf = STRBUF_INIT;
2890 char *dirty = "";
8e08b419
JH
2891
2892 /* Are we looking at the work tree? */
85adbf2f 2893 if (s->dirty_submodule)
8e08b419
JH
2894 dirty = "-dirty";
2895
a0d12c44 2896 strbuf_addf(&buf, "Subproject commit %s%s\n",
2897 oid_to_hex(&s->oid), dirty);
b1ddfb91 2898 s->size = buf.len;
04786756
LT
2899 if (size_only) {
2900 s->data = NULL;
b1ddfb91
JK
2901 strbuf_release(&buf);
2902 } else {
2903 s->data = strbuf_detach(&buf, NULL);
2904 s->should_free = 1;
04786756
LT
2905 }
2906 return 0;
2907}
2908
6973dcae
JH
2909/*
2910 * While doing rename detection and pickaxe operation, we may need to
2911 * grab the data for the blob (or file) for our own in-core comparison.
2912 * diff_filespec has data and size fields for this purpose.
2913 */
8e5dd3d6 2914int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
6973dcae 2915{
8e5dd3d6 2916 int size_only = flags & CHECK_SIZE_ONLY;
6973dcae 2917 int err = 0;
5430bb28
JH
2918 /*
2919 * demote FAIL to WARN to allow inspecting the situation
2920 * instead of refusing.
2921 */
2922 enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2923 ? SAFE_CRLF_WARN
2924 : safe_crlf);
2925
6973dcae
JH
2926 if (!DIFF_FILE_VALID(s))
2927 die("internal error: asking to populate invalid file.");
2928 if (S_ISDIR(s->mode))
2929 return -1;
2930
6973dcae 2931 if (s->data)
fc3abdf5 2932 return 0;
04786756 2933
6e0b8ed6
JH
2934 if (size_only && 0 < s->size)
2935 return 0;
2936
302b9282 2937 if (S_ISGITLINK(s->mode))
04786756
LT
2938 return diff_populate_gitlink(s, size_only);
2939
41c9560e 2940 if (!s->oid_valid ||
fb4a1c0d 2941 reuse_worktree_file(s->path, &s->oid, 0)) {
f285a2d7 2942 struct strbuf buf = STRBUF_INIT;
6973dcae
JH
2943 struct stat st;
2944 int fd;
6c510bee 2945
6973dcae
JH
2946 if (lstat(s->path, &st) < 0) {
2947 if (errno == ENOENT) {
2948 err_empty:
2949 err = -1;
2950 empty:
d2543b8e 2951 s->data = (char *)"";
6973dcae
JH
2952 s->size = 0;
2953 return err;
2954 }
2955 }
dc49cd76 2956 s->size = xsize_t(st.st_size);
6973dcae
JH
2957 if (!s->size)
2958 goto empty;
6973dcae 2959 if (S_ISLNK(st.st_mode)) {
cf219d8c
LT
2960 struct strbuf sb = STRBUF_INIT;
2961
2962 if (strbuf_readlink(&sb, s->path, s->size))
6973dcae 2963 goto err_empty;
0956a6db
RS
2964 s->size = sb.len;
2965 s->data = strbuf_detach(&sb, NULL);
cf219d8c 2966 s->should_free = 1;
6973dcae
JH
2967 return 0;
2968 }
12426e11
JH
2969
2970 /*
2971 * Even if the caller would be happy with getting
2972 * only the size, we cannot return early at this
2973 * point if the path requires us to run the content
2974 * conversion.
2975 */
82b474e0 2976 if (size_only && !would_convert_to_git(&the_index, s->path))
cf219d8c 2977 return 0;
12426e11
JH
2978
2979 /*
2980 * Note: this check uses xsize_t(st.st_size) that may
2981 * not be the true size of the blob after it goes
2982 * through convert_to_git(). This may not strictly be
2983 * correct, but the whole point of big_file_threshold
2984 * and is_binary check being that we want to avoid
2985 * opening the file and inspecting the contents, this
2986 * is probably fine.
2987 */
6bf3b813
NTND
2988 if ((flags & CHECK_BINARY) &&
2989 s->size > big_file_threshold && s->is_binary == -1) {
2990 s->is_binary = 1;
2991 return 0;
2992 }
6973dcae
JH
2993 fd = open(s->path, O_RDONLY);
2994 if (fd < 0)
2995 goto err_empty;
c4712e45 2996 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
6973dcae 2997 close(fd);
6973dcae 2998 s->should_munmap = 1;
6c510bee
LT
2999
3000 /*
3001 * Convert from working tree format to canonical git format
3002 */
82b474e0 3003 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, crlf_warn)) {
c32f749f 3004 size_t size = 0;
6c510bee
LT
3005 munmap(s->data, s->size);
3006 s->should_munmap = 0;
c32f749f
RS
3007 s->data = strbuf_detach(&buf, &size);
3008 s->size = size;
6c510bee
LT
3009 s->should_free = 1;
3010 }
6973dcae
JH
3011 }
3012 else {
21666f1a 3013 enum object_type type;
6bf3b813 3014 if (size_only || (flags & CHECK_BINARY)) {
a0d12c44 3015 type = sha1_object_info(s->oid.hash, &s->size);
c50c4316 3016 if (type < 0)
a0d12c44 3017 die("unable to read %s",
3018 oid_to_hex(&s->oid));
6bf3b813
NTND
3019 if (size_only)
3020 return 0;
3021 if (s->size > big_file_threshold && s->is_binary == -1) {
3022 s->is_binary = 1;
3023 return 0;
3024 }
6973dcae 3025 }
a0d12c44 3026 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
6bf3b813 3027 if (!s->data)
a0d12c44 3028 die("unable to read %s", oid_to_hex(&s->oid));
6bf3b813 3029 s->should_free = 1;
6973dcae
JH
3030 }
3031 return 0;
3032}
3033
8ae92e63 3034void diff_free_filespec_blob(struct diff_filespec *s)
6973dcae
JH
3035{
3036 if (s->should_free)
3037 free(s->data);
3038 else if (s->should_munmap)
3039 munmap(s->data, s->size);
fc3abdf5
JH
3040
3041 if (s->should_free || s->should_munmap) {
3042 s->should_free = s->should_munmap = 0;
3043 s->data = NULL;
3044 }
eede7b7d
JK
3045}
3046
3047void diff_free_filespec_data(struct diff_filespec *s)
3048{
8ae92e63 3049 diff_free_filespec_blob(s);
6a83d902 3050 FREE_AND_NULL(s->cnt_data);
6973dcae
JH
3051}
3052
4e218f54 3053static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
6973dcae
JH
3054 void *blob,
3055 unsigned long size,
09bdff29 3056 const struct object_id *oid,
6973dcae
JH
3057 int mode)
3058{
3059 int fd;
4e218f54 3060 struct strbuf buf = STRBUF_INIT;
003b33a8
DA
3061 struct strbuf template = STRBUF_INIT;
3062 char *path_dup = xstrdup(path);
3063 const char *base = basename(path_dup);
6973dcae 3064
003b33a8
DA
3065 /* Generate "XXXXXX_basename.ext" */
3066 strbuf_addstr(&template, "XXXXXX_");
3067 strbuf_addstr(&template, base);
3068
284098f1 3069 fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
6973dcae 3070 if (fd < 0)
d824cbba 3071 die_errno("unable to create temp-file");
4e218f54
JS
3072 if (convert_to_working_tree(path,
3073 (const char *)blob, (size_t)size, &buf)) {
3074 blob = buf.buf;
3075 size = buf.len;
3076 }
93822c22 3077 if (write_in_full(fd, blob, size) != size)
0721c314 3078 die_errno("unable to write temp-file");
284098f1
MH
3079 close_tempfile(&temp->tempfile);
3080 temp->name = get_tempfile_path(&temp->tempfile);
09bdff29 3081 oid_to_hex_r(temp->hex, oid);
5096d490 3082 xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4e218f54 3083 strbuf_release(&buf);
003b33a8
DA
3084 strbuf_release(&template);
3085 free(path_dup);
6973dcae
JH
3086}
3087
479b0ae8
JK
3088static struct diff_tempfile *prepare_temp_file(const char *name,
3089 struct diff_filespec *one)
6973dcae 3090{
479b0ae8
JK
3091 struct diff_tempfile *temp = claim_diff_tempfile();
3092
6973dcae
JH
3093 if (!DIFF_FILE_VALID(one)) {
3094 not_a_valid_file:
3095 /* A '-' entry produces this for file-2, and
3096 * a '+' entry produces this for file-1.
3097 */
3098 temp->name = "/dev/null";
5096d490
JK
3099 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3100 xsnprintf(temp->mode, sizeof(temp->mode), ".");
479b0ae8
JK
3101 return temp;
3102 }
3103
aba47272 3104 if (!S_ISGITLINK(one->mode) &&
41c9560e 3105 (!one->oid_valid ||
fb4a1c0d 3106 reuse_worktree_file(name, &one->oid, 1))) {
6973dcae
JH
3107 struct stat st;
3108 if (lstat(name, &st) < 0) {
3109 if (errno == ENOENT)
3110 goto not_a_valid_file;
d824cbba 3111 die_errno("stat(%s)", name);
6973dcae
JH
3112 }
3113 if (S_ISLNK(st.st_mode)) {
3cd7388d
JK
3114 struct strbuf sb = STRBUF_INIT;
3115 if (strbuf_readlink(&sb, name, st.st_size) < 0)
0721c314 3116 die_errno("readlink(%s)", name);
3cd7388d 3117 prep_temp_blob(name, temp, sb.buf, sb.len,
41c9560e 3118 (one->oid_valid ?
09bdff29 3119 &one->oid : &null_oid),
41c9560e 3120 (one->oid_valid ?
6973dcae 3121 one->mode : S_IFLNK));
3cd7388d 3122 strbuf_release(&sb);
6973dcae
JH
3123 }
3124 else {
3125 /* we can borrow from the file in the work tree */
3126 temp->name = name;
41c9560e 3127 if (!one->oid_valid)
74014152 3128 oid_to_hex_r(temp->hex, &null_oid);
6973dcae 3129 else
2490574d 3130 oid_to_hex_r(temp->hex, &one->oid);
6973dcae
JH
3131 /* Even though we may sometimes borrow the
3132 * contents from the work tree, we always want
3133 * one->mode. mode is trustworthy even when
74014152 3134 * !(one->oid_valid), as long as
6973dcae
JH
3135 * DIFF_FILE_VALID(one).
3136 */
5096d490 3137 xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
6973dcae 3138 }
479b0ae8 3139 return temp;
6973dcae
JH
3140 }
3141 else {
3142 if (diff_populate_filespec(one, 0))
3143 die("cannot read data blob for %s", one->path);
4e218f54 3144 prep_temp_blob(name, temp, one->data, one->size,
09bdff29 3145 &one->oid, one->mode);
6973dcae 3146 }
479b0ae8 3147 return temp;
6973dcae
JH
3148}
3149
f3efe787
JK
3150static void add_external_diff_name(struct argv_array *argv,
3151 const char *name,
3152 struct diff_filespec *df)
3153{
3154 struct diff_tempfile *temp = prepare_temp_file(name, df);
3155 argv_array_push(argv, temp->name);
3156 argv_array_push(argv, temp->hex);
3157 argv_array_push(argv, temp->mode);
3158}
3159
6973dcae
JH
3160/* An external diff command takes:
3161 *
3162 * diff-cmd name infile1 infile1-sha1 infile1-mode \
3163 * infile2 infile2-sha1 infile2-mode [ rename-to ]
3164 *
3165 */
3166static void run_external_diff(const char *pgm,
3167 const char *name,
3168 const char *other,
3169 struct diff_filespec *one,
3170 struct diff_filespec *two,
3171 const char *xfrm_msg,
ee7fb0b1
ZK
3172 int complete_rewrite,
3173 struct diff_options *o)
6973dcae 3174{
82fbf269 3175 struct argv_array argv = ARGV_ARRAY_INIT;
ae049c95 3176 struct argv_array env = ARGV_ARRAY_INIT;
ee7fb0b1 3177 struct diff_queue_struct *q = &diff_queued_diff;
6973dcae 3178
0d4217d9
JK
3179 argv_array_push(&argv, pgm);
3180 argv_array_push(&argv, name);
6973dcae 3181
6973dcae 3182 if (one && two) {
f3efe787
JK
3183 add_external_diff_name(&argv, name, one);
3184 if (!other)
3185 add_external_diff_name(&argv, name, two);
3186 else {
3187 add_external_diff_name(&argv, other, two);
82fbf269
JK
3188 argv_array_push(&argv, other);
3189 argv_array_push(&argv, xfrm_msg);
6973dcae 3190 }
6973dcae 3191 }
ee7fb0b1 3192
ae049c95
JK
3193 argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3194 argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
ee7fb0b1 3195
89294d14
JK
3196 if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3197 die(_("external diff died, stopping at %s"), name);
ee7fb0b1 3198
6973dcae 3199 remove_tempfile();
82fbf269 3200 argv_array_clear(&argv);
ae049c95 3201 argv_array_clear(&env);
6973dcae
JH
3202}
3203
b67b9612
JH
3204static int similarity_index(struct diff_filepair *p)
3205{
3206 return p->score * 100 / MAX_SCORE;
3207}
3208
4f03666a
JK
3209static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3210{
3211 if (startup_info->have_repository)
3212 return find_unique_abbrev(oid->hash, abbrev);
3213 else {
3214 char *hex = oid_to_hex(oid);
0d9c527d
JH
3215 if (abbrev < 0)
3216 abbrev = FALLBACK_DEFAULT_ABBREV;
3217 if (abbrev > GIT_SHA1_HEXSZ)
4f03666a 3218 die("BUG: oid abbreviation out of range: %d", abbrev);
43d1948b
JB
3219 if (abbrev)
3220 hex[abbrev] = '\0';
4f03666a
JK
3221 return hex;
3222 }
3223}
3224
b67b9612
JH
3225static void fill_metainfo(struct strbuf *msg,
3226 const char *name,
3227 const char *other,
3228 struct diff_filespec *one,
3229 struct diff_filespec *two,
3230 struct diff_options *o,
37466447 3231 struct diff_filepair *p,
5977744d 3232 int *must_show_header,
37466447 3233 int use_color)
b67b9612 3234{
37466447
BW
3235 const char *set = diff_get_color(use_color, DIFF_METAINFO);
3236 const char *reset = diff_get_color(use_color, DIFF_RESET);
30997bb8 3237 const char *line_prefix = diff_line_prefix(o);
7be57610 3238
296c6bb2 3239 *must_show_header = 1;
b67b9612
JH
3240 strbuf_init(msg, PATH_MAX * 2 + 300);
3241 switch (p->status) {
3242 case DIFF_STATUS_COPIED:
98ad90fb
JH
3243 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3244 line_prefix, set, similarity_index(p));
3245 strbuf_addf(msg, "%s\n%s%scopy from ",
3246 reset, line_prefix, set);
b67b9612 3247 quote_c_style(name, msg, NULL, 0);
98ad90fb 3248 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
b67b9612 3249 quote_c_style(other, msg, NULL, 0);
37466447 3250 strbuf_addf(msg, "%s\n", reset);
b67b9612
JH
3251 break;
3252 case DIFF_STATUS_RENAMED:
98ad90fb
JH
3253 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3254 line_prefix, set, similarity_index(p));
3255 strbuf_addf(msg, "%s\n%s%srename from ",
3256 reset, line_prefix, set);
b67b9612 3257 quote_c_style(name, msg, NULL, 0);
98ad90fb
JH
3258 strbuf_addf(msg, "%s\n%s%srename to ",
3259 reset, line_prefix, set);
b67b9612 3260 quote_c_style(other, msg, NULL, 0);
37466447 3261 strbuf_addf(msg, "%s\n", reset);
b67b9612
JH
3262 break;
3263 case DIFF_STATUS_MODIFIED:
3264 if (p->score) {
98ad90fb
JH
3265 strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3266 line_prefix,
37466447 3267 set, similarity_index(p), reset);
b67b9612
JH
3268 break;
3269 }
3270 /* fallthru */
3271 default:
296c6bb2 3272 *must_show_header = 0;
b67b9612 3273 }
a0d12c44 3274 if (one && two && oidcmp(&one->oid, &two->oid)) {
b67b9612
JH
3275 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3276
3277 if (DIFF_OPT_TST(o, BINARY)) {
3278 mmfile_t mf;
3279 if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3280 (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3281 abbrev = 40;
3282 }
4f03666a
JK
3283 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3284 diff_abbrev_oid(&one->oid, abbrev),
3285 diff_abbrev_oid(&two->oid, abbrev));
b67b9612
JH
3286 if (one->mode == two->mode)
3287 strbuf_addf(msg, " %06o", one->mode);
37466447 3288 strbuf_addf(msg, "%s\n", reset);
b67b9612 3289 }
b67b9612
JH
3290}
3291
6973dcae
JH
3292static void run_diff_cmd(const char *pgm,
3293 const char *name,
3294 const char *other,
cd676a51 3295 const char *attr_path,
6973dcae
JH
3296 struct diff_filespec *one,
3297 struct diff_filespec *two,
b67b9612 3298 struct strbuf *msg,
051308f6 3299 struct diff_options *o,
b67b9612 3300 struct diff_filepair *p)
6973dcae 3301{
b67b9612
JH
3302 const char *xfrm_msg = NULL;
3303 int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
296c6bb2 3304 int must_show_header = 0;
b67b9612 3305
bd8c1a9b
JH
3306
3307 if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
be58e70d
JK
3308 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3309 if (drv && drv->external)
3310 pgm = drv->external;
f1af60bd
JH
3311 }
3312
37466447
BW
3313 if (msg) {
3314 /*
3315 * don't use colors when the header is intended for an
3316 * external diff driver
3317 */
3318 fill_metainfo(msg, name, other, one, two, o, p,
5977744d 3319 &must_show_header,
daa0c3d9 3320 want_color(o->use_color) && !pgm);
37466447
BW
3321 xfrm_msg = msg->len ? msg->buf : NULL;
3322 }
3323
6973dcae
JH
3324 if (pgm) {
3325 run_external_diff(pgm, name, other, one, two, xfrm_msg,
ee7fb0b1 3326 complete_rewrite, o);
6973dcae
JH
3327 return;
3328 }
3329 if (one && two)
3330 builtin_diff(name, other ? other : name,
296c6bb2
CC
3331 one, two, xfrm_msg, must_show_header,
3332 o, complete_rewrite);
6973dcae 3333 else
c0c77734 3334 fprintf(o->file, "* Unmerged path %s\n", name);
6973dcae
JH
3335}
3336
94e327e9 3337static void diff_fill_oid_info(struct diff_filespec *one)
6973dcae
JH
3338{
3339 if (DIFF_FILE_VALID(one)) {
41c9560e 3340 if (!one->oid_valid) {
6973dcae 3341 struct stat st;
4682d852 3342 if (one->is_stdin) {
a0d12c44 3343 oidclr(&one->oid);
5332b2af
JS
3344 return;
3345 }
6973dcae 3346 if (lstat(one->path, &st) < 0)
0721c314 3347 die_errno("stat '%s'", one->path);
a0d12c44 3348 if (index_path(one->oid.hash, one->path, &st, 0))
d7530708 3349 die("cannot hash %s", one->path);
6973dcae
JH
3350 }
3351 }
3352 else
a0d12c44 3353 oidclr(&one->oid);
6973dcae
JH
3354}
3355
cd676a51
JH
3356static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3357{
3358 /* Strip the prefix but do not molest /dev/null and absolute paths */
d8faea9d 3359 if (*namep && **namep != '/') {
cd676a51 3360 *namep += prefix_length;
d8faea9d
JN
3361 if (**namep == '/')
3362 ++*namep;
3363 }
3364 if (*otherp && **otherp != '/') {
cd676a51 3365 *otherp += prefix_length;
d8faea9d
JN
3366 if (**otherp == '/')
3367 ++*otherp;
3368 }
cd676a51
JH
3369}
3370
6973dcae
JH
3371static void run_diff(struct diff_filepair *p, struct diff_options *o)
3372{
3373 const char *pgm = external_diff();
663af342 3374 struct strbuf msg;
663af342
PH
3375 struct diff_filespec *one = p->one;
3376 struct diff_filespec *two = p->two;
6973dcae
JH
3377 const char *name;
3378 const char *other;
cd676a51 3379 const char *attr_path;
663af342 3380
f2d2a5de
SB
3381 name = one->path;
3382 other = (strcmp(name, two->path) ? two->path : NULL);
cd676a51
JH
3383 attr_path = name;
3384 if (o->prefix_length)
3385 strip_prefix(o->prefix_length, &name, &other);
6973dcae 3386
bd8c1a9b
JH
3387 if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3388 pgm = NULL;
3389
6973dcae 3390 if (DIFF_PAIR_UNMERGED(p)) {
cd676a51 3391 run_diff_cmd(pgm, name, NULL, attr_path,
b67b9612 3392 NULL, NULL, NULL, o, p);
6973dcae
JH
3393 return;
3394 }
3395
94e327e9
BW
3396 diff_fill_oid_info(one);
3397 diff_fill_oid_info(two);
6973dcae 3398
6973dcae
JH
3399 if (!pgm &&
3400 DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3401 (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
b67b9612
JH
3402 /*
3403 * a filepair that changes between file and symlink
6973dcae
JH
3404 * needs to be split into deletion and creation.
3405 */
3406 struct diff_filespec *null = alloc_filespec(two->path);
cd676a51 3407 run_diff_cmd(NULL, name, other, attr_path,
b67b9612 3408 one, null, &msg, o, p);
6973dcae 3409 free(null);
b67b9612
JH
3410 strbuf_release(&msg);
3411
6973dcae 3412 null = alloc_filespec(one->path);
cd676a51 3413 run_diff_cmd(NULL, name, other, attr_path,
b67b9612 3414 null, two, &msg, o, p);
6973dcae
JH
3415 free(null);
3416 }
3417 else
cd676a51 3418 run_diff_cmd(pgm, name, other, attr_path,
b67b9612 3419 one, two, &msg, o, p);
6973dcae 3420
663af342 3421 strbuf_release(&msg);
6973dcae
JH
3422}
3423
3424static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3425 struct diffstat_t *diffstat)
3426{
3427 const char *name;
3428 const char *other;
3429
3430 if (DIFF_PAIR_UNMERGED(p)) {
3431 /* unmerged */
74faaa16 3432 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
6973dcae
JH
3433 return;
3434 }
3435
3436 name = p->one->path;
3437 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3438
cd676a51
JH
3439 if (o->prefix_length)
3440 strip_prefix(o->prefix_length, &name, &other);
3441
94e327e9
BW
3442 diff_fill_oid_info(p->one);
3443 diff_fill_oid_info(p->two);
6973dcae 3444
74faaa16 3445 builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
6973dcae
JH
3446}
3447
88246898
JS
3448static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3449{
3450 const char *name;
3451 const char *other;
cd676a51 3452 const char *attr_path;
88246898
JS
3453
3454 if (DIFF_PAIR_UNMERGED(p)) {
3455 /* unmerged */
3456 return;
3457 }
3458
3459 name = p->one->path;
3460 other = (strcmp(name, p->two->path) ? p->two->path : NULL);
cd676a51
JH
3461 attr_path = other ? other : name;
3462
3463 if (o->prefix_length)
3464 strip_prefix(o->prefix_length, &name, &other);
88246898 3465
94e327e9
BW
3466 diff_fill_oid_info(p->one);
3467 diff_fill_oid_info(p->two);
88246898 3468
cd676a51 3469 builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
88246898
JS
3470}
3471
6973dcae
JH
3472void diff_setup(struct diff_options *options)
3473{
be4f2b40 3474 memcpy(options, &default_diff_options, sizeof(*options));
c0c77734
DB
3475
3476 options->file = stdout;
3477
43d1948b 3478 options->abbrev = DEFAULT_ABBREV;
6973dcae
JH
3479 options->line_termination = '\n';
3480 options->break_opt = -1;
3481 options->rename_limit = -1;
712d2c7d 3482 options->dirstat_permille = diff_dirstat_permille_default;
6468a4e5 3483 options->context = diff_context_default;
c4888677 3484 options->interhunkcontext = diff_interhunk_context_default;
a17505f2 3485 options->ws_error_highlight = ws_error_highlight_default;
90d43b07 3486 DIFF_OPT_SET(options, RENAME_EMPTY);
6973dcae 3487
72441af7 3488 /* pathchange left =NULL by default */
6973dcae
JH
3489 options->change = diff_change;
3490 options->add_remove = diff_addremove;
f1c96261 3491 options->use_color = diff_use_color_default;
b68ea12e 3492 options->detect_rename = diff_detect_rename_default;
07ab4dec 3493 options->xdl_opts |= diff_algorithm;
433860f3
MH
3494 if (diff_indent_heuristic)
3495 DIFF_XDL_SET(options, INDENT_HEURISTIC);
eab9a40b 3496
6d8940b5
SB
3497 options->orderfile = diff_order_file_cfg;
3498