diff --color-moved-ws: demonstrate false positives
[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"
cbd53a21 16#include "object-store.h"
be58e70d 17#include "userdiff.h"
851e18c3 18#include "submodule-config.h"
752c0c24 19#include "submodule.h"
2e2d5ac1 20#include "hashmap.h"
a757c646 21#include "ll-merge.h"
02e8ca0e 22#include "string-list.h"
82fbf269 23#include "argv-array.h"
660e113c 24#include "graph.h"
150e3001 25#include "packfile.h"
3ac68a93 26#include "help.h"
6973dcae 27
1510fea7
SP
28#ifdef NO_FAST_WORKING_DIRECTORY
29#define FAST_WORKING_DIRECTORY 0
30#else
31#define FAST_WORKING_DIRECTORY 1
32#endif
33
96f1e58f 34static int diff_detect_rename_default;
33de7163 35static int diff_indent_heuristic = 1;
92c57e5c 36static int diff_rename_limit_default = 400;
a624eaa7 37static int diff_suppress_blank_empty;
d2aea137 38static int diff_use_color_default = -1;
2e2d5ac1 39static int diff_color_moved_default;
626c0b5d 40static int diff_color_moved_ws_default;
6468a4e5 41static int diff_context_default = 3;
c4888677 42static int diff_interhunk_context_default;
98a4d87b 43static const char *diff_word_regex_cfg;
cbe02100 44static const char *external_diff_cmd_cfg;
6d8940b5 45static const char *diff_order_file_cfg;
aecbf914 46int diff_auto_refresh_index = 1;
a5a818ee 47static int diff_mnemonic_prefix;
f89504dd 48static int diff_no_prefix;
df44483a 49static int diff_stat_graph_width;
712d2c7d 50static int diff_dirstat_permille_default = 30;
be4f2b40 51static struct diff_options default_diff_options;
07ab4dec 52static long diff_algorithm;
a17505f2 53static unsigned ws_error_highlight_default = WSEH_NEW;
6973dcae 54
7c92fe0e 55static char diff_colors[][COLOR_MAXLEN] = {
dc6ebd4c 56 GIT_COLOR_RESET,
8dbf3eb6 57 GIT_COLOR_NORMAL, /* CONTEXT */
dc6ebd4c
AL
58 GIT_COLOR_BOLD, /* METAINFO */
59 GIT_COLOR_CYAN, /* FRAGINFO */
60 GIT_COLOR_RED, /* OLD */
61 GIT_COLOR_GREEN, /* NEW */
62 GIT_COLOR_YELLOW, /* COMMIT */
63 GIT_COLOR_BG_RED, /* WHITESPACE */
89cb73a1 64 GIT_COLOR_NORMAL, /* FUNCINFO */
86b452e2
SB
65 GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
66 GIT_COLOR_BOLD_BLUE, /* OLD_MOVED ALTERNATIVE */
67 GIT_COLOR_FAINT, /* OLD_MOVED_DIM */
68 GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
69 GIT_COLOR_BOLD_CYAN, /* NEW_MOVED */
70 GIT_COLOR_BOLD_YELLOW, /* NEW_MOVED ALTERNATIVE */
71 GIT_COLOR_FAINT, /* NEW_MOVED_DIM */
72 GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
a7be92ac
JS
73 GIT_COLOR_FAINT, /* CONTEXT_DIM */
74 GIT_COLOR_FAINT_RED, /* OLD_DIM */
75 GIT_COLOR_FAINT_GREEN, /* NEW_DIM */
76 GIT_COLOR_BOLD, /* CONTEXT_BOLD */
77 GIT_COLOR_BOLD_RED, /* OLD_BOLD */
78 GIT_COLOR_BOLD_GREEN, /* NEW_BOLD */
cd112cef
JS
79};
80
a73b3680
NTND
81static const char *color_diff_slots[] = {
82 [DIFF_CONTEXT] = "context",
83 [DIFF_METAINFO] = "meta",
84 [DIFF_FRAGINFO] = "frag",
85 [DIFF_FILE_OLD] = "old",
86 [DIFF_FILE_NEW] = "new",
87 [DIFF_COMMIT] = "commit",
88 [DIFF_WHITESPACE] = "whitespace",
89 [DIFF_FUNCINFO] = "func",
90 [DIFF_FILE_OLD_MOVED] = "oldMoved",
91 [DIFF_FILE_OLD_MOVED_ALT] = "oldMovedAlternative",
92 [DIFF_FILE_OLD_MOVED_DIM] = "oldMovedDimmed",
93 [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
94 [DIFF_FILE_NEW_MOVED] = "newMoved",
95 [DIFF_FILE_NEW_MOVED_ALT] = "newMovedAlternative",
96 [DIFF_FILE_NEW_MOVED_DIM] = "newMovedDimmed",
97 [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
a7be92ac
JS
98 [DIFF_CONTEXT_DIM] = "contextDimmed",
99 [DIFF_FILE_OLD_DIM] = "oldDimmed",
100 [DIFF_FILE_NEW_DIM] = "newDimmed",
101 [DIFF_CONTEXT_BOLD] = "contextBold",
102 [DIFF_FILE_OLD_BOLD] = "oldBold",
103 [DIFF_FILE_NEW_BOLD] = "newBold",
a73b3680
NTND
104};
105
a2f05c94
JNA
106static NORETURN void die_want_option(const char *option_name)
107{
108 die(_("option '%s' requires a value"), option_name);
109}
110
3ac68a93
NTND
111define_list_config_array_extra(color_diff_slots, {"plain"});
112
9e1a5ebe 113static int parse_diff_color_slot(const char *var)
801235c5 114{
a73b3680 115 if (!strcasecmp(var, "plain"))
8dbf3eb6 116 return DIFF_CONTEXT;
a73b3680 117 return LOOKUP_CONFIG(color_diff_slots, var);
801235c5
JH
118}
119
02e8ca0e 120static int parse_dirstat_params(struct diff_options *options, const char *params_string,
51670fc8 121 struct strbuf *errmsg)
333f3fb0 122{
02e8ca0e
MH
123 char *params_copy = xstrdup(params_string);
124 struct string_list params = STRING_LIST_INIT_NODUP;
125 int ret = 0;
126 int i;
51670fc8 127
02e8ca0e
MH
128 if (*params_copy)
129 string_list_split_in_place(&params, params_copy, ',', -1);
130 for (i = 0; i < params.nr; i++) {
131 const char *p = params.items[i].string;
132 if (!strcmp(p, "changes")) {
0d1e0e78
BW
133 options->flags.dirstat_by_line = 0;
134 options->flags.dirstat_by_file = 0;
02e8ca0e 135 } else if (!strcmp(p, "lines")) {
0d1e0e78
BW
136 options->flags.dirstat_by_line = 1;
137 options->flags.dirstat_by_file = 0;
02e8ca0e 138 } else if (!strcmp(p, "files")) {
0d1e0e78
BW
139 options->flags.dirstat_by_line = 0;
140 options->flags.dirstat_by_file = 1;
02e8ca0e 141 } else if (!strcmp(p, "noncumulative")) {
0d1e0e78 142 options->flags.dirstat_cumulative = 0;
02e8ca0e 143 } else if (!strcmp(p, "cumulative")) {
0d1e0e78 144 options->flags.dirstat_cumulative = 1;
333f3fb0
JH
145 } else if (isdigit(*p)) {
146 char *end;
51670fc8
JH
147 int permille = strtoul(p, &end, 10) * 10;
148 if (*end == '.' && isdigit(*++end)) {
712d2c7d 149 /* only use first digit */
51670fc8 150 permille += *end - '0';
712d2c7d 151 /* .. and ignore any further digits */
51670fc8 152 while (isdigit(*++end))
712d2c7d
JH
153 ; /* nothing */
154 }
02e8ca0e 155 if (!*end)
51670fc8
JH
156 options->dirstat_permille = permille;
157 else {
02e8ca0e
MH
158 strbuf_addf(errmsg, _(" Failed to parse dirstat cut-off percentage '%s'\n"),
159 p);
51670fc8
JH
160 ret++;
161 }
162 } else {
02e8ca0e 163 strbuf_addf(errmsg, _(" Unknown dirstat parameter '%s'\n"), p);
51670fc8 164 ret++;
333f3fb0 165 }
51670fc8 166
333f3fb0 167 }
02e8ca0e
MH
168 string_list_clear(&params, 0);
169 free(params_copy);
51670fc8 170 return ret;
333f3fb0
JH
171}
172
c47ef57c
RR
173static int parse_submodule_params(struct diff_options *options, const char *value)
174{
175 if (!strcmp(value, "log"))
61cfbc05 176 options->submodule_format = DIFF_SUBMODULE_LOG;
c47ef57c 177 else if (!strcmp(value, "short"))
61cfbc05 178 options->submodule_format = DIFF_SUBMODULE_SHORT;
fd47ae6a
JK
179 else if (!strcmp(value, "diff"))
180 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
c47ef57c
RR
181 else
182 return -1;
183 return 0;
184}
185
85b46030 186int git_config_rename(const char *var, const char *value)
cced5fbc
LT
187{
188 if (!value)
189 return DIFF_DETECT_RENAME;
190 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
191 return DIFF_DETECT_COPY;
192 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
193}
194
07924d4d 195long parse_algorithm_value(const char *value)
07ab4dec
MP
196{
197 if (!value)
198 return -1;
199 else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
200 return 0;
201 else if (!strcasecmp(value, "minimal"))
202 return XDF_NEED_MINIMAL;
203 else if (!strcasecmp(value, "patience"))
204 return XDF_PATIENCE_DIFF;
205 else if (!strcasecmp(value, "histogram"))
206 return XDF_HISTOGRAM_DIFF;
207 return -1;
208}
209
0b4b42e7
JH
210static int parse_one_token(const char **arg, const char *token)
211{
212 const char *rest;
213 if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
214 *arg = rest;
215 return 1;
216 }
217 return 0;
218}
219
220static int parse_ws_error_highlight(const char *arg)
221{
222 const char *orig_arg = arg;
223 unsigned val = 0;
224
225 while (*arg) {
226 if (parse_one_token(&arg, "none"))
227 val = 0;
228 else if (parse_one_token(&arg, "default"))
229 val = WSEH_NEW;
230 else if (parse_one_token(&arg, "all"))
231 val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
232 else if (parse_one_token(&arg, "new"))
233 val |= WSEH_NEW;
234 else if (parse_one_token(&arg, "old"))
235 val |= WSEH_OLD;
236 else if (parse_one_token(&arg, "context"))
237 val |= WSEH_CONTEXT;
238 else {
239 return -1 - (int)(arg - orig_arg);
240 }
241 if (*arg)
242 arg++;
243 }
244 return val;
245}
246
83ad63cf
JH
247/*
248 * These are to give UI layer defaults.
249 * The core-level commands such as git-diff-files should
250 * never be affected by the setting of diff.renames
251 * the user happens to have in the configuration file.
252 */
5404c116
MM
253void init_diff_ui_defaults(void)
254{
06dba2b0 255 diff_detect_rename_default = DIFF_DETECT_RENAME;
5404c116
MM
256}
257
5b162879
MH
258int git_diff_heuristic_config(const char *var, const char *value, void *cb)
259{
3cde4e02 260 if (!strcmp(var, "diff.indentheuristic"))
5b162879 261 diff_indent_heuristic = git_config_bool(var, value);
5b162879
MH
262 return 0;
263}
264
2e2d5ac1
SB
265static int parse_color_moved(const char *arg)
266{
267 switch (git_parse_maybe_bool(arg)) {
268 case 0:
269 return COLOR_MOVED_NO;
270 case 1:
271 return COLOR_MOVED_DEFAULT;
272 default:
273 break;
274 }
275
276 if (!strcmp(arg, "no"))
277 return COLOR_MOVED_NO;
176841f0
SB
278 else if (!strcmp(arg, "plain"))
279 return COLOR_MOVED_PLAIN;
51da15eb
SB
280 else if (!strcmp(arg, "blocks"))
281 return COLOR_MOVED_BLOCKS;
2e2d5ac1
SB
282 else if (!strcmp(arg, "zebra"))
283 return COLOR_MOVED_ZEBRA;
284 else if (!strcmp(arg, "default"))
285 return COLOR_MOVED_DEFAULT;
e3f2f5f9
ES
286 else if (!strcmp(arg, "dimmed-zebra"))
287 return COLOR_MOVED_ZEBRA_DIM;
86b452e2
SB
288 else if (!strcmp(arg, "dimmed_zebra"))
289 return COLOR_MOVED_ZEBRA_DIM;
2e2d5ac1 290 else
706b0b5e 291 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
2e2d5ac1
SB
292}
293
b3095712
SB
294static int parse_color_moved_ws(const char *arg)
295{
296 int ret = 0;
297 struct string_list l = STRING_LIST_INIT_DUP;
298 struct string_list_item *i;
299
300 string_list_split(&l, arg, ',', -1);
301
302 for_each_string_list_item(i, &l) {
303 struct strbuf sb = STRBUF_INIT;
304 strbuf_addstr(&sb, i->string);
305 strbuf_trim(&sb);
306
b73bcbac
PW
307 if (!strcmp(sb.buf, "no"))
308 ret = 0;
309 else if (!strcmp(sb.buf, "ignore-space-change"))
b3095712
SB
310 ret |= XDF_IGNORE_WHITESPACE_CHANGE;
311 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
312 ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
313 else if (!strcmp(sb.buf, "ignore-all-space"))
314 ret |= XDF_IGNORE_WHITESPACE;
ca1f4ae4
SB
315 else if (!strcmp(sb.buf, "allow-indentation-change"))
316 ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
b3095712
SB
317 else
318 error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
319
320 strbuf_release(&sb);
321 }
322
ca1f4ae4
SB
323 if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
324 (ret & XDF_WHITESPACE_FLAGS))
748aa1aa 325 die(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
ca1f4ae4 326
b3095712
SB
327 string_list_clear(&l, 0);
328
329 return ret;
2e2d5ac1
SB
330}
331
ef90d6d4 332int git_diff_ui_config(const char *var, const char *value, void *cb)
801235c5 333{
a159ca0c 334 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
e269eb79 335 diff_use_color_default = git_config_colorbool(var, value);
801235c5
JH
336 return 0;
337 }
2e2d5ac1
SB
338 if (!strcmp(var, "diff.colormoved")) {
339 int cm = parse_color_moved(value);
340 if (cm < 0)
341 return -1;
342 diff_color_moved_default = cm;
343 return 0;
344 }
626c0b5d
SB
345 if (!strcmp(var, "diff.colormovedws")) {
346 int cm = parse_color_moved_ws(value);
347 if (cm < 0)
348 return -1;
349 diff_color_moved_ws_default = cm;
350 return 0;
351 }
6468a4e5
JM
352 if (!strcmp(var, "diff.context")) {
353 diff_context_default = git_config_int(var, value);
354 if (diff_context_default < 0)
355 return -1;
356 return 0;
357 }
c4888677
VN
358 if (!strcmp(var, "diff.interhunkcontext")) {
359 diff_interhunk_context_default = git_config_int(var, value);
360 if (diff_interhunk_context_default < 0)
361 return -1;
362 return 0;
363 }
b68ea12e 364 if (!strcmp(var, "diff.renames")) {
cced5fbc 365 diff_detect_rename_default = git_config_rename(var, value);
b68ea12e
EW
366 return 0;
367 }
aecbf914
JH
368 if (!strcmp(var, "diff.autorefreshindex")) {
369 diff_auto_refresh_index = git_config_bool(var, value);
370 return 0;
371 }
a5a818ee
JH
372 if (!strcmp(var, "diff.mnemonicprefix")) {
373 diff_mnemonic_prefix = git_config_bool(var, value);
374 return 0;
375 }
f89504dd
EC
376 if (!strcmp(var, "diff.noprefix")) {
377 diff_no_prefix = git_config_bool(var, value);
378 return 0;
379 }
df44483a
ZJS
380 if (!strcmp(var, "diff.statgraphwidth")) {
381 diff_stat_graph_width = git_config_int(var, value);
382 return 0;
383 }
daec808c
BH
384 if (!strcmp(var, "diff.external"))
385 return git_config_string(&external_diff_cmd_cfg, var, value);
98a4d87b
BSSJ
386 if (!strcmp(var, "diff.wordregex"))
387 return git_config_string(&diff_word_regex_cfg, var, value);
6d8940b5
SB
388 if (!strcmp(var, "diff.orderfile"))
389 return git_config_pathname(&diff_order_file_cfg, var, value);
f1af60bd 390
be4f2b40
JS
391 if (!strcmp(var, "diff.ignoresubmodules"))
392 handle_ignore_submodules_arg(&default_diff_options, value);
393
c47ef57c
RR
394 if (!strcmp(var, "diff.submodule")) {
395 if (parse_submodule_params(&default_diff_options, value))
396 warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
397 value);
398 return 0;
399 }
400
07ab4dec
MP
401 if (!strcmp(var, "diff.algorithm")) {
402 diff_algorithm = parse_algorithm_value(value);
403 if (diff_algorithm < 0)
404 return -1;
405 return 0;
406 }
407
a17505f2
JH
408 if (!strcmp(var, "diff.wserrorhighlight")) {
409 int val = parse_ws_error_highlight(value);
410 if (val < 0)
411 return -1;
412 ws_error_highlight_default = val;
413 return 0;
414 }
415
33c643bb
JK
416 if (git_color_config(var, value, cb) < 0)
417 return -1;
418
ef90d6d4 419 return git_diff_basic_config(var, value, cb);
9a1805a8
JK
420}
421
ef90d6d4 422int git_diff_basic_config(const char *var, const char *value, void *cb)
9a1805a8 423{
ae021d87
JK
424 const char *name;
425
2b6ca6df
LT
426 if (!strcmp(var, "diff.renamelimit")) {
427 diff_rename_limit_default = git_config_int(var, value);
428 return 0;
429 }
430
6680a087
JK
431 if (userdiff_config(var, value) < 0)
432 return -1;
c7534ef4 433
ae021d87
JK
434 if (skip_prefix(var, "diff.color.", &name) ||
435 skip_prefix(var, "color.diff.", &name)) {
436 int slot = parse_diff_color_slot(name);
8b8e8624
JK
437 if (slot < 0)
438 return 0;
64f30e94
JH
439 if (!value)
440 return config_error_nonbool(var);
f6c5a296 441 return color_parse(value, diff_colors[slot]);
801235c5 442 }
f1af60bd 443
a624eaa7 444 /* like GNU diff's --suppress-blank-empty option */
950db879
JS
445 if (!strcmp(var, "diff.suppressblankempty") ||
446 /* for backwards compatibility */
447 !strcmp(var, "diff.suppress-blank-empty")) {
a624eaa7
JM
448 diff_suppress_blank_empty = git_config_bool(var, value);
449 return 0;
450 }
451
2d174951 452 if (!strcmp(var, "diff.dirstat")) {
51670fc8 453 struct strbuf errmsg = STRBUF_INIT;
712d2c7d 454 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
51670fc8 455 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
7478ac57 456 warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
51670fc8
JH
457 errmsg.buf);
458 strbuf_release(&errmsg);
712d2c7d 459 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
2d174951
JH
460 return 0;
461 }
462
cf5e7722
MB
463 if (git_diff_heuristic_config(var, value, cb) < 0)
464 return -1;
465
3e1dd17a 466 return git_default_config(var, value, cb);
801235c5
JH
467}
468
6973dcae
JH
469static char *quote_two(const char *one, const char *two)
470{
471 int need_one = quote_c_style(one, NULL, NULL, 1);
472 int need_two = quote_c_style(two, NULL, NULL, 1);
f285a2d7 473 struct strbuf res = STRBUF_INIT;
6973dcae
JH
474
475 if (need_one + need_two) {
663af342
PH
476 strbuf_addch(&res, '"');
477 quote_c_style(one, &res, NULL, 1);
478 quote_c_style(two, &res, NULL, 1);
479 strbuf_addch(&res, '"');
480 } else {
481 strbuf_addstr(&res, one);
482 strbuf_addstr(&res, two);
6973dcae 483 }
b315c5c0 484 return strbuf_detach(&res, NULL);
6973dcae
JH
485}
486
487static const char *external_diff(void)
488{
489 static const char *external_diff_cmd = NULL;
490 static int done_preparing = 0;
491
492 if (done_preparing)
493 return external_diff_cmd;
494 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
cbe02100
JS
495 if (!external_diff_cmd)
496 external_diff_cmd = external_diff_cmd_cfg;
6973dcae
JH
497 done_preparing = 1;
498 return external_diff_cmd;
499}
500
284098f1
MH
501/*
502 * Keep track of files used for diffing. Sometimes such an entry
503 * refers to a temporary file, sometimes to an existing file, and
504 * sometimes to "/dev/null".
505 */
6973dcae 506static struct diff_tempfile {
284098f1
MH
507 /*
508 * filename external diff should read from, or NULL if this
509 * entry is currently not in use:
510 */
511 const char *name;
512
dc01505f 513 char hex[GIT_MAX_HEXSZ + 1];
6973dcae 514 char mode[10];
284098f1
MH
515
516 /*
517 * If this diff_tempfile instance refers to a temporary file,
518 * this tempfile object is used to manage its lifetime.
519 */
076aa2cb 520 struct tempfile *tempfile;
6973dcae
JH
521} diff_temp[2];
522
6957eb9a 523struct emit_callback {
6957eb9a
JH
524 int color_diff;
525 unsigned ws_rule;
526 int blank_at_eof_in_preimage;
527 int blank_at_eof_in_postimage;
528 int lno_in_preimage;
529 int lno_in_postimage;
6957eb9a
JH
530 const char **label_path;
531 struct diff_words_data *diff_words;
a3c158d4 532 struct diff_options *opt;
3e97c7c6 533 struct strbuf *header;
6957eb9a
JH
534};
535
6973dcae
JH
536static int count_lines(const char *data, int size)
537{
538 int count, ch, completely_empty = 1, nl_just_seen = 0;
539 count = 0;
540 while (0 < size--) {
541 ch = *data++;
542 if (ch == '\n') {
543 count++;
544 nl_just_seen = 1;
545 completely_empty = 0;
546 }
547 else {
548 nl_just_seen = 0;
549 completely_empty = 0;
550 }
551 }
552 if (completely_empty)
553 return 0;
554 if (!nl_just_seen)
555 count++; /* no trailing newline */
556 return count;
557}
558
b78ea5fc
NTND
559static int fill_mmfile(struct repository *r, mmfile_t *mf,
560 struct diff_filespec *one)
6957eb9a
JH
561{
562 if (!DIFF_FILE_VALID(one)) {
563 mf->ptr = (char *)""; /* does not matter */
564 mf->size = 0;
565 return 0;
566 }
b78ea5fc 567 else if (diff_populate_filespec(r, one, 0))
6957eb9a 568 return -1;
bb35fefb 569
6957eb9a
JH
570 mf->ptr = one->data;
571 mf->size = one->size;
572 return 0;
573}
574
abb371a1 575/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
b78ea5fc
NTND
576static unsigned long diff_filespec_size(struct repository *r,
577 struct diff_filespec *one)
abb371a1
JK
578{
579 if (!DIFF_FILE_VALID(one))
580 return 0;
b78ea5fc 581 diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
abb371a1
JK
582 return one->size;
583}
584
6957eb9a
JH
585static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
586{
587 char *ptr = mf->ptr;
588 long size = mf->size;
589 int cnt = 0;
590
591 if (!size)
592 return cnt;
593 ptr += size - 1; /* pointing at the very end */
594 if (*ptr != '\n')
595 ; /* incomplete line */
596 else
597 ptr--; /* skip the last LF */
598 while (mf->ptr < ptr) {
599 char *prev_eol;
600 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
601 if (*prev_eol == '\n')
602 break;
603 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
604 break;
605 cnt++;
606 ptr = prev_eol - 1;
607 }
608 return cnt;
609}
610
611static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
612 struct emit_callback *ecbdata)
613{
614 int l1, l2, at;
615 unsigned ws_rule = ecbdata->ws_rule;
616 l1 = count_trailing_blank(mf1, ws_rule);
617 l2 = count_trailing_blank(mf2, ws_rule);
618 if (l2 <= l1) {
619 ecbdata->blank_at_eof_in_preimage = 0;
620 ecbdata->blank_at_eof_in_postimage = 0;
621 return;
622 }
623 at = count_lines(mf1->ptr, mf1->size);
624 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
625
626 at = count_lines(mf2->ptr, mf2->size);
627 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
628}
629
f7c3b4e2 630static void emit_line_0(struct diff_options *o,
017ac45e 631 const char *set_sign, const char *set, unsigned reverse, const char *reset,
250f7993 632 int first, const char *line, int len)
6957eb9a
JH
633{
634 int has_trailing_newline, has_trailing_carriage_return;
44410679 635 int needs_reset = 0; /* at the end of the line */
a3c158d4
BY
636 FILE *file = o->file;
637
f103a6fa 638 fputs(diff_line_prefix(o), file);
6957eb9a 639
44410679
SB
640 has_trailing_newline = (len > 0 && line[len-1] == '\n');
641 if (has_trailing_newline)
642 len--;
643
644 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
645 if (has_trailing_carriage_return)
646 len--;
647
648 if (!len && !first)
649 goto end_of_line;
650
651 if (reverse && want_color(o->use_color)) {
652 fputs(GIT_COLOR_REVERSE, file);
653 needs_reset = 1;
250f7993 654 }
6957eb9a 655
44410679
SB
656 if (set_sign) {
657 fputs(set_sign, file);
658 needs_reset = 1;
06a47552 659 }
44410679
SB
660
661 if (first)
662 fputc(first, file);
663
664 if (!len)
665 goto end_of_line;
666
667 if (set) {
668 if (set_sign && set != set_sign)
669 fputs(reset, file);
06a47552 670 fputs(set, file);
44410679 671 needs_reset = 1;
06a47552 672 }
44410679
SB
673 fwrite(line, len, 1, file);
674 needs_reset = 1; /* 'line' may contain color codes. */
675
676end_of_line:
677 if (needs_reset)
678 fputs(reset, file);
6957eb9a
JH
679 if (has_trailing_carriage_return)
680 fputc('\r', file);
681 if (has_trailing_newline)
682 fputc('\n', file);
683}
684
a3c158d4 685static void emit_line(struct diff_options *o, const char *set, const char *reset,
250f7993
JH
686 const char *line, int len)
687{
44410679 688 emit_line_0(o, set, NULL, 0, reset, 0, line, len);
250f7993
JH
689}
690
36a4cefd 691enum diff_symbol {
4eed0ebd
SB
692 DIFF_SYMBOL_BINARY_DIFF_HEADER,
693 DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
694 DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
695 DIFF_SYMBOL_BINARY_DIFF_BODY,
696 DIFF_SYMBOL_BINARY_DIFF_FOOTER,
0911c475
SB
697 DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
698 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
699 DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
700 DIFF_SYMBOL_STATS_LINE,
bd033291 701 DIFF_SYMBOL_WORD_DIFF,
30b7e1e7 702 DIFF_SYMBOL_STAT_SEP,
146fdb0d 703 DIFF_SYMBOL_SUMMARY,
f3597138
SB
704 DIFF_SYMBOL_SUBMODULE_ADD,
705 DIFF_SYMBOL_SUBMODULE_DEL,
706 DIFF_SYMBOL_SUBMODULE_UNTRACKED,
707 DIFF_SYMBOL_SUBMODULE_MODIFIED,
708 DIFF_SYMBOL_SUBMODULE_HEADER,
709 DIFF_SYMBOL_SUBMODULE_ERROR,
710 DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
5af6ea95 711 DIFF_SYMBOL_REWRITE_DIFF,
4acaaa7a 712 DIFF_SYMBOL_BINARY_FILES,
a29b0a13 713 DIFF_SYMBOL_HEADER,
3ee8b7bf
SB
714 DIFF_SYMBOL_FILEPAIR_PLUS,
715 DIFF_SYMBOL_FILEPAIR_MINUS,
ff958679
SB
716 DIFF_SYMBOL_WORDS_PORCELAIN,
717 DIFF_SYMBOL_WORDS,
091f8e28 718 DIFF_SYMBOL_CONTEXT,
f2bb1218 719 DIFF_SYMBOL_CONTEXT_INCOMPLETE,
091f8e28
SB
720 DIFF_SYMBOL_PLUS,
721 DIFF_SYMBOL_MINUS,
b9cbfde6 722 DIFF_SYMBOL_NO_LF_EOF,
68abc6f1 723 DIFF_SYMBOL_CONTEXT_FRAGINFO,
c64b420b 724 DIFF_SYMBOL_CONTEXT_MARKER,
36a4cefd
SB
725 DIFF_SYMBOL_SEPARATOR
726};
091f8e28
SB
727/*
728 * Flags for content lines:
729 * 0..12 are whitespace rules
730 * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
731 * 16 is marking if the line is blank at EOF
732 */
2e2d5ac1
SB
733#define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF (1<<16)
734#define DIFF_SYMBOL_MOVED_LINE (1<<17)
735#define DIFF_SYMBOL_MOVED_LINE_ALT (1<<18)
86b452e2 736#define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING (1<<19)
091f8e28
SB
737#define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
738
e6e045f8
SB
739/*
740 * This struct is used when we need to buffer the output of the diff output.
741 *
742 * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
743 * into the pre/post image file. This pointer could be a union with the
744 * line pointer. By storing an offset into the file instead of the literal line,
745 * we can decrease the memory footprint for the buffered output. At first we
746 * may want to only have indirection for the content lines, but we could also
747 * enhance the state for emitting prefabricated lines, e.g. the similarity
748 * score line or hunk/file headers would only need to store a number or path
749 * and then the output can be constructed later on depending on state.
750 */
751struct emitted_diff_symbol {
752 const char *line;
753 int len;
754 int flags;
755 enum diff_symbol s;
756};
757#define EMITTED_DIFF_SYMBOL_INIT {NULL}
758
759struct emitted_diff_symbols {
760 struct emitted_diff_symbol *buf;
761 int nr, alloc;
762};
763#define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
764
765static void append_emitted_diff_symbol(struct diff_options *o,
766 struct emitted_diff_symbol *e)
6957eb9a 767{
e6e045f8
SB
768 struct emitted_diff_symbol *f;
769
770 ALLOC_GROW(o->emitted_symbols->buf,
771 o->emitted_symbols->nr + 1,
772 o->emitted_symbols->alloc);
773 f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
774
775 memcpy(f, e, sizeof(struct emitted_diff_symbol));
776 f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
6957eb9a
JH
777}
778
2e2d5ac1
SB
779struct moved_entry {
780 struct hashmap_entry ent;
781 const struct emitted_diff_symbol *es;
782 struct moved_entry *next_line;
783};
784
ca1f4ae4
SB
785/**
786 * The struct ws_delta holds white space differences between moved lines, i.e.
787 * between '+' and '-' lines that have been detected to be a move.
788 * The string contains the difference in leading white spaces, before the
789 * rest of the line is compared using the white space config for move
790 * coloring. The current_longer indicates if the first string in the
791 * comparision is longer than the second.
792 */
793struct ws_delta {
794 char *string;
795 unsigned int current_longer : 1;
2e2d5ac1 796};
ca1f4ae4
SB
797#define WS_DELTA_INIT { NULL, 0 }
798
74d156f4
PW
799struct moved_block {
800 struct moved_entry *match;
801 struct ws_delta wsd;
802};
803
804static void moved_block_clear(struct moved_block *b)
805{
806 FREE_AND_NULL(b->wsd.string);
807 b->match = NULL;
808}
809
ca1f4ae4
SB
810static int compute_ws_delta(const struct emitted_diff_symbol *a,
811 const struct emitted_diff_symbol *b,
812 struct ws_delta *out)
813{
814 const struct emitted_diff_symbol *longer = a->len > b->len ? a : b;
815 const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
816 int d = longer->len - shorter->len;
817
fe4516d1
PW
818 if (strncmp(longer->line + d, shorter->line, shorter->len))
819 return 0;
820
ca1f4ae4
SB
821 out->string = xmemdupz(longer->line, d);
822 out->current_longer = (a == longer);
823
fe4516d1 824 return 1;
ca1f4ae4
SB
825}
826
827static int cmp_in_block_with_wsd(const struct diff_options *o,
828 const struct moved_entry *cur,
829 const struct moved_entry *match,
74d156f4 830 struct moved_block *pmb,
ca1f4ae4
SB
831 int n)
832{
833 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
834 int al = cur->es->len, cl = l->len;
835 const char *a = cur->es->line,
836 *b = match->es->line,
837 *c = l->line;
838
839 int wslen;
840
841 /*
842 * We need to check if 'cur' is equal to 'match'.
843 * As those are from the same (+/-) side, we do not need to adjust for
844 * indent changes. However these were found using fuzzy matching
845 * so we do have to check if they are equal.
846 */
847 if (strcmp(a, b))
848 return 1;
849
74d156f4 850 if (!pmb->wsd.string)
ca1f4ae4 851 /*
74d156f4
PW
852 * The white space delta is not active? This can happen
853 * when we exit early in this function.
ca1f4ae4
SB
854 */
855 return 1;
856
857 /*
74d156f4 858 * The indent changes of the block are known and stored in
ca1f4ae4
SB
859 * pmb->wsd; however we need to check if the indent changes of the
860 * current line are still the same as before.
861 *
862 * To do so we need to compare 'l' to 'cur', adjusting the
863 * one of them for the white spaces, depending which was longer.
864 */
865
74d156f4
PW
866 wslen = strlen(pmb->wsd.string);
867 if (pmb->wsd.current_longer) {
ca1f4ae4
SB
868 c += wslen;
869 cl -= wslen;
870 } else {
871 a += wslen;
872 al -= wslen;
873 }
874
cf074a9b 875 if (al != cl || memcmp(a, c, al))
ca1f4ae4
SB
876 return 1;
877
878 return 0;
879}
2e2d5ac1 880
ee1df66f
SB
881static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
882 const void *entry,
883 const void *entry_or_key,
2e2d5ac1
SB
884 const void *keydata)
885{
ee1df66f
SB
886 const struct diff_options *diffopt = hashmap_cmp_fn_data;
887 const struct moved_entry *a = entry;
888 const struct moved_entry *b = entry_or_key;
b3095712
SB
889 unsigned flags = diffopt->color_moved_ws_handling
890 & XDF_WHITESPACE_FLAGS;
ee1df66f 891
ca1f4ae4
SB
892 if (diffopt->color_moved_ws_handling &
893 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
894 /*
895 * As there is not specific white space config given,
896 * we'd need to check for a new block, so ignore all
897 * white space. The setup of the white space
898 * configuration for the next block is done else where
899 */
900 flags |= XDF_IGNORE_WHITESPACE;
901
01be97c2
SB
902 return !xdiff_compare_lines(a->es->line, a->es->len,
903 b->es->line, b->es->len,
b3095712 904 flags);
2e2d5ac1
SB
905}
906
907static struct moved_entry *prepare_entry(struct diff_options *o,
908 int line_no)
909{
910 struct moved_entry *ret = xmalloc(sizeof(*ret));
911 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
b3095712 912 unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
2e2d5ac1 913
b3095712 914 ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
2e2d5ac1
SB
915 ret->es = l;
916 ret->next_line = NULL;
917
918 return ret;
919}
920
921static void add_lines_to_move_detection(struct diff_options *o,
922 struct hashmap *add_lines,
923 struct hashmap *del_lines)
924{
925 struct moved_entry *prev_line = NULL;
926
927 int n;
928 for (n = 0; n < o->emitted_symbols->nr; n++) {
929 struct hashmap *hm;
930 struct moved_entry *key;
931
932 switch (o->emitted_symbols->buf[n].s) {
933 case DIFF_SYMBOL_PLUS:
934 hm = add_lines;
935 break;
936 case DIFF_SYMBOL_MINUS:
937 hm = del_lines;
938 break;
939 default:
940 prev_line = NULL;
941 continue;
942 }
943
944 key = prepare_entry(o, n);
945 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
946 prev_line->next_line = key;
947
948 hashmap_add(hm, key);
949 prev_line = key;
950 }
951}
952
e2fe6abc
SB
953static void pmb_advance_or_null(struct diff_options *o,
954 struct moved_entry *match,
955 struct hashmap *hm,
74d156f4 956 struct moved_block *pmb,
e2fe6abc
SB
957 int pmb_nr)
958{
959 int i;
960 for (i = 0; i < pmb_nr; i++) {
74d156f4 961 struct moved_entry *prev = pmb[i].match;
e2fe6abc
SB
962 struct moved_entry *cur = (prev && prev->next_line) ?
963 prev->next_line : NULL;
964 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
74d156f4 965 pmb[i].match = cur;
e2fe6abc 966 } else {
74d156f4 967 pmb[i].match = NULL;
e2fe6abc
SB
968 }
969 }
970}
971
ca1f4ae4
SB
972static void pmb_advance_or_null_multi_match(struct diff_options *o,
973 struct moved_entry *match,
974 struct hashmap *hm,
74d156f4 975 struct moved_block *pmb,
ca1f4ae4
SB
976 int pmb_nr, int n)
977{
978 int i;
979 char *got_match = xcalloc(1, pmb_nr);
980
981 for (; match; match = hashmap_get_next(hm, match)) {
982 for (i = 0; i < pmb_nr; i++) {
74d156f4 983 struct moved_entry *prev = pmb[i].match;
ca1f4ae4
SB
984 struct moved_entry *cur = (prev && prev->next_line) ?
985 prev->next_line : NULL;
986 if (!cur)
987 continue;
74d156f4 988 if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
ca1f4ae4
SB
989 got_match[i] |= 1;
990 }
991 }
992
993 for (i = 0; i < pmb_nr; i++) {
994 if (got_match[i]) {
74d156f4
PW
995 /* Advance to the next line */
996 pmb[i].match = pmb[i].match->next_line;
fab01ec5 997 } else {
74d156f4 998 moved_block_clear(&pmb[i]);
fab01ec5 999 }
ca1f4ae4 1000 }
9c1a6c2b
PW
1001
1002 free(got_match);
ca1f4ae4
SB
1003}
1004
74d156f4 1005static int shrink_potential_moved_blocks(struct moved_block *pmb,
2e2d5ac1
SB
1006 int pmb_nr)
1007{
1008 int lp, rp;
1009
1010 /* Shrink the set of potential block to the remaining running */
1011 for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
74d156f4 1012 while (lp < pmb_nr && pmb[lp].match)
2e2d5ac1
SB
1013 lp++;
1014 /* lp points at the first NULL now */
1015
74d156f4 1016 while (rp > -1 && !pmb[rp].match)
2e2d5ac1
SB
1017 rp--;
1018 /* rp points at the last non-NULL */
1019
1020 if (lp < pmb_nr && rp > -1 && lp < rp) {
1021 pmb[lp] = pmb[rp];
74d156f4
PW
1022 pmb[rp].match = NULL;
1023 pmb[rp].wsd.string = NULL;
2e2d5ac1
SB
1024 rp--;
1025 lp++;
1026 }
1027 }
1028
1029 /* Remember the number of running sets */
1030 return rp + 1;
1031}
1032
09153277
JT
1033/*
1034 * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1035 *
f0b8fb6e
JT
1036 * Otherwise, if the last block has fewer alphanumeric characters than
1037 * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
09153277
JT
1038 * that block.
1039 *
1040 * The last block consists of the (n - block_length)'th line up to but not
1041 * including the nth line.
f0b8fb6e
JT
1042 *
1043 * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1044 * Think of a way to unify them.
09153277
JT
1045 */
1046static void adjust_last_block(struct diff_options *o, int n, int block_length)
1047{
f0b8fb6e
JT
1048 int i, alnum_count = 0;
1049 if (o->color_moved == COLOR_MOVED_PLAIN)
09153277 1050 return;
f0b8fb6e
JT
1051 for (i = 1; i < block_length + 1; i++) {
1052 const char *c = o->emitted_symbols->buf[n - i].line;
1053 for (; *c; c++) {
1054 if (!isalnum(*c))
1055 continue;
1056 alnum_count++;
1057 if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1058 return;
1059 }
1060 }
09153277
JT
1061 for (i = 1; i < block_length + 1; i++)
1062 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1063}
1064
2e2d5ac1
SB
1065/* Find blocks of moved code, delegate actual coloring decision to helper */
1066static void mark_color_as_moved(struct diff_options *o,
1067 struct hashmap *add_lines,
1068 struct hashmap *del_lines)
1069{
74d156f4 1070 struct moved_block *pmb = NULL; /* potentially moved blocks */
2e2d5ac1
SB
1071 int pmb_nr = 0, pmb_alloc = 0;
1072 int n, flipped_block = 1, block_length = 0;
1073
1074
1075 for (n = 0; n < o->emitted_symbols->nr; n++) {
1076 struct hashmap *hm = NULL;
1077 struct moved_entry *key;
1078 struct moved_entry *match = NULL;
1079 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
2e2d5ac1
SB
1080
1081 switch (l->s) {
1082 case DIFF_SYMBOL_PLUS:
1083 hm = del_lines;
1084 key = prepare_entry(o, n);
3783aad4 1085 match = hashmap_get(hm, key, NULL);
2e2d5ac1
SB
1086 free(key);
1087 break;
1088 case DIFF_SYMBOL_MINUS:
1089 hm = add_lines;
1090 key = prepare_entry(o, n);
3783aad4 1091 match = hashmap_get(hm, key, NULL);
2e2d5ac1
SB
1092 free(key);
1093 break;
1094 default:
1095 flipped_block = 1;
1096 }
1097
1098 if (!match) {
74d156f4
PW
1099 int i;
1100
09153277 1101 adjust_last_block(o, n, block_length);
74d156f4
PW
1102 for(i = 0; i < pmb_nr; i++)
1103 moved_block_clear(&pmb[i]);
2e2d5ac1
SB
1104 pmb_nr = 0;
1105 block_length = 0;
1106 continue;
1107 }
1108
1109 l->flags |= DIFF_SYMBOL_MOVED_LINE;
2e2d5ac1 1110
176841f0
SB
1111 if (o->color_moved == COLOR_MOVED_PLAIN)
1112 continue;
1113
ca1f4ae4
SB
1114 if (o->color_moved_ws_handling &
1115 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1116 pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1117 else
1118 pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
2e2d5ac1
SB
1119
1120 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1121
1122 if (pmb_nr == 0) {
1123 /*
1124 * The current line is the start of a new block.
1125 * Setup the set of potential blocks.
1126 */
1127 for (; match; match = hashmap_get_next(hm, match)) {
1128 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
ca1f4ae4
SB
1129 if (o->color_moved_ws_handling &
1130 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
74d156f4
PW
1131 if (compute_ws_delta(l, match->es,
1132 &pmb[pmb_nr].wsd))
1133 pmb[pmb_nr++].match = match;
ca1f4ae4 1134 } else {
74d156f4
PW
1135 pmb[pmb_nr].wsd.string = NULL;
1136 pmb[pmb_nr++].match = match;
ca1f4ae4 1137 }
2e2d5ac1
SB
1138 }
1139
1140 flipped_block = (flipped_block + 1) % 2;
f0b8fb6e
JT
1141
1142 adjust_last_block(o, n, block_length);
1143 block_length = 0;
2e2d5ac1
SB
1144 }
1145
f0b8fb6e
JT
1146 block_length++;
1147
51da15eb 1148 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
2e2d5ac1
SB
1149 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1150 }
09153277 1151 adjust_last_block(o, n, block_length);
2e2d5ac1 1152
74d156f4
PW
1153 for(n = 0; n < pmb_nr; n++)
1154 moved_block_clear(&pmb[n]);
2e2d5ac1
SB
1155 free(pmb);
1156}
e6e045f8 1157
86b452e2
SB
1158#define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1159 (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1160static void dim_moved_lines(struct diff_options *o)
1161{
1162 int n;
1163 for (n = 0; n < o->emitted_symbols->nr; n++) {
1164 struct emitted_diff_symbol *prev = (n != 0) ?
1165 &o->emitted_symbols->buf[n - 1] : NULL;
1166 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1167 struct emitted_diff_symbol *next =
1168 (n < o->emitted_symbols->nr - 1) ?
1169 &o->emitted_symbols->buf[n + 1] : NULL;
1170
1171 /* Not a plus or minus line? */
1172 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1173 continue;
1174
1175 /* Not a moved line? */
1176 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1177 continue;
1178
1179 /*
1180 * If prev or next are not a plus or minus line,
1181 * pretend they don't exist
1182 */
1183 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1184 prev->s != DIFF_SYMBOL_MINUS)
1185 prev = NULL;
1186 if (next && next->s != DIFF_SYMBOL_PLUS &&
1187 next->s != DIFF_SYMBOL_MINUS)
1188 next = NULL;
1189
1190 /* Inside a block? */
1191 if ((prev &&
1192 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1193 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1194 (next &&
1195 (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1196 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1197 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1198 continue;
1199 }
1200
1201 /* Check if we are at an interesting bound: */
1202 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1203 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1204 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1205 continue;
1206 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1207 (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1208 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1209 continue;
1210
1211 /*
1212 * The boundary to prev and next are not interesting,
1213 * so this line is not interesting as a whole
1214 */
1215 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1216 }
1217}
1218
091f8e28 1219static void emit_line_ws_markup(struct diff_options *o,
9d1e16bf
SB
1220 const char *set_sign, const char *set,
1221 const char *reset,
bc9feb05 1222 int sign_index, const char *line, int len,
091f8e28 1223 unsigned ws_rule, int blank_at_eof)
6957eb9a 1224{
b8767f79 1225 const char *ws = NULL;
bc9feb05 1226 int sign = o->output_indicators[sign_index];
6957eb9a 1227
091f8e28
SB
1228 if (o->ws_error_highlight & ws_rule) {
1229 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
b8767f79
JH
1230 if (!*ws)
1231 ws = NULL;
1232 }
1233
f7c3b4e2 1234 if (!ws && !set_sign)
017ac45e 1235 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
f7c3b4e2 1236 else if (!ws) {
29ef759d 1237 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
f7c3b4e2 1238 } else if (blank_at_eof)
6957eb9a 1239 /* Blank line at EOF - paint '+' as well */
017ac45e 1240 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
6957eb9a
JH
1241 else {
1242 /* Emit just the prefix, then the rest. */
017ac45e 1243 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
f7c3b4e2 1244 sign, "", 0);
091f8e28
SB
1245 ws_check_emit(line, len, ws_rule,
1246 o->file, set, reset, ws);
6957eb9a
JH
1247 }
1248}
1249
e6e045f8
SB
1250static void emit_diff_symbol_from_struct(struct diff_options *o,
1251 struct emitted_diff_symbol *eds)
36a4cefd 1252{
b9cbfde6 1253 static const char *nneof = " No newline at end of file\n";
f7c3b4e2 1254 const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
0911c475 1255 struct strbuf sb = STRBUF_INIT;
e6e045f8
SB
1256
1257 enum diff_symbol s = eds->s;
1258 const char *line = eds->line;
1259 int len = eds->len;
1260 unsigned flags = eds->flags;
1261
36a4cefd 1262 switch (s) {
b9cbfde6
SB
1263 case DIFF_SYMBOL_NO_LF_EOF:
1264 context = diff_get_color_opt(o, DIFF_CONTEXT);
1265 reset = diff_get_color_opt(o, DIFF_RESET);
1266 putc('\n', o->file);
017ac45e 1267 emit_line_0(o, context, NULL, 0, reset, '\\',
b9cbfde6
SB
1268 nneof, strlen(nneof));
1269 break;
f3597138
SB
1270 case DIFF_SYMBOL_SUBMODULE_HEADER:
1271 case DIFF_SYMBOL_SUBMODULE_ERROR:
1272 case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
0911c475 1273 case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
146fdb0d 1274 case DIFF_SYMBOL_SUMMARY:
0911c475 1275 case DIFF_SYMBOL_STATS_LINE:
4eed0ebd 1276 case DIFF_SYMBOL_BINARY_DIFF_BODY:
68abc6f1
SB
1277 case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1278 emit_line(o, "", "", line, len);
1279 break;
f2bb1218 1280 case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
c64b420b
SB
1281 case DIFF_SYMBOL_CONTEXT_MARKER:
1282 context = diff_get_color_opt(o, DIFF_CONTEXT);
1283 reset = diff_get_color_opt(o, DIFF_RESET);
1284 emit_line(o, context, reset, line, len);
1285 break;
36a4cefd
SB
1286 case DIFF_SYMBOL_SEPARATOR:
1287 fprintf(o->file, "%s%c",
1288 diff_line_prefix(o),
1289 o->line_termination);
1290 break;
091f8e28
SB
1291 case DIFF_SYMBOL_CONTEXT:
1292 set = diff_get_color_opt(o, DIFF_CONTEXT);
1293 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1294 set_sign = NULL;
1295 if (o->flags.dual_color_diffed_diffs) {
1296 char c = !len ? 0 : line[0];
1297
1298 if (c == '+')
1299 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1300 else if (c == '@')
1301 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1302 else if (c == '-')
1303 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1304 }
7648b79e 1305 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1306 OUTPUT_INDICATOR_CONTEXT, line, len,
091f8e28
SB
1307 flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1308 break;
1309 case DIFF_SYMBOL_PLUS:
86b452e2
SB
1310 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1311 DIFF_SYMBOL_MOVED_LINE_ALT |
1312 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1313 case DIFF_SYMBOL_MOVED_LINE |
1314 DIFF_SYMBOL_MOVED_LINE_ALT |
1315 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1316 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1317 break;
1318 case DIFF_SYMBOL_MOVED_LINE |
1319 DIFF_SYMBOL_MOVED_LINE_ALT:
2e2d5ac1 1320 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
86b452e2
SB
1321 break;
1322 case DIFF_SYMBOL_MOVED_LINE |
1323 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1324 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1325 break;
1326 case DIFF_SYMBOL_MOVED_LINE:
2e2d5ac1 1327 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
86b452e2
SB
1328 break;
1329 default:
2e2d5ac1 1330 set = diff_get_color_opt(o, DIFF_FILE_NEW);
86b452e2 1331 }
091f8e28 1332 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1333 if (!o->flags.dual_color_diffed_diffs)
1334 set_sign = NULL;
1335 else {
1336 char c = !len ? 0 : line[0];
1337
1338 set_sign = set;
1339 if (c == '-')
a7be92ac 1340 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
f7c3b4e2
JS
1341 else if (c == '@')
1342 set = diff_get_color_opt(o, DIFF_FRAGINFO);
a7be92ac
JS
1343 else if (c == '+')
1344 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1345 else
1346 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
0b91faa0 1347 flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
f7c3b4e2 1348 }
7648b79e 1349 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1350 OUTPUT_INDICATOR_NEW, line, len,
091f8e28
SB
1351 flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1352 flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1353 break;
1354 case DIFF_SYMBOL_MINUS:
86b452e2
SB
1355 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1356 DIFF_SYMBOL_MOVED_LINE_ALT |
1357 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1358 case DIFF_SYMBOL_MOVED_LINE |
1359 DIFF_SYMBOL_MOVED_LINE_ALT |
1360 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1361 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1362 break;
1363 case DIFF_SYMBOL_MOVED_LINE |
1364 DIFF_SYMBOL_MOVED_LINE_ALT:
2e2d5ac1 1365 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
86b452e2
SB
1366 break;
1367 case DIFF_SYMBOL_MOVED_LINE |
1368 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1369 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1370 break;
1371 case DIFF_SYMBOL_MOVED_LINE:
2e2d5ac1 1372 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
86b452e2
SB
1373 break;
1374 default:
2e2d5ac1 1375 set = diff_get_color_opt(o, DIFF_FILE_OLD);
86b452e2 1376 }
091f8e28 1377 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1378 if (!o->flags.dual_color_diffed_diffs)
1379 set_sign = NULL;
1380 else {
1381 char c = !len ? 0 : line[0];
1382
1383 set_sign = set;
1384 if (c == '+')
a7be92ac 1385 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
f7c3b4e2
JS
1386 else if (c == '@')
1387 set = diff_get_color_opt(o, DIFF_FRAGINFO);
a7be92ac
JS
1388 else if (c == '-')
1389 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1390 else
1391 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
f7c3b4e2 1392 }
7648b79e 1393 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1394 OUTPUT_INDICATOR_OLD, line, len,
091f8e28
SB
1395 flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1396 break;
ff958679
SB
1397 case DIFF_SYMBOL_WORDS_PORCELAIN:
1398 context = diff_get_color_opt(o, DIFF_CONTEXT);
1399 reset = diff_get_color_opt(o, DIFF_RESET);
1400 emit_line(o, context, reset, line, len);
1401 fputs("~\n", o->file);
1402 break;
1403 case DIFF_SYMBOL_WORDS:
1404 context = diff_get_color_opt(o, DIFF_CONTEXT);
1405 reset = diff_get_color_opt(o, DIFF_RESET);
1406 /*
1407 * Skip the prefix character, if any. With
1408 * diff_suppress_blank_empty, there may be
1409 * none.
1410 */
1411 if (line[0] != '\n') {
1412 line++;
1413 len--;
1414 }
1415 emit_line(o, context, reset, line, len);
1416 break;
3ee8b7bf
SB
1417 case DIFF_SYMBOL_FILEPAIR_PLUS:
1418 meta = diff_get_color_opt(o, DIFF_METAINFO);
1419 reset = diff_get_color_opt(o, DIFF_RESET);
1420 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1421 line, reset,
1422 strchr(line, ' ') ? "\t" : "");
1423 break;
1424 case DIFF_SYMBOL_FILEPAIR_MINUS:
1425 meta = diff_get_color_opt(o, DIFF_METAINFO);
1426 reset = diff_get_color_opt(o, DIFF_RESET);
1427 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1428 line, reset,
1429 strchr(line, ' ') ? "\t" : "");
1430 break;
4acaaa7a 1431 case DIFF_SYMBOL_BINARY_FILES:
a29b0a13
SB
1432 case DIFF_SYMBOL_HEADER:
1433 fprintf(o->file, "%s", line);
1434 break;
4eed0ebd
SB
1435 case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1436 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1437 break;
1438 case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1439 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1440 break;
1441 case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1442 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1443 break;
1444 case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1445 fputs(diff_line_prefix(o), o->file);
1446 fputc('\n', o->file);
1447 break;
5af6ea95
SB
1448 case DIFF_SYMBOL_REWRITE_DIFF:
1449 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1450 reset = diff_get_color_opt(o, DIFF_RESET);
1451 emit_line(o, fraginfo, reset, line, len);
1452 break;
f3597138
SB
1453 case DIFF_SYMBOL_SUBMODULE_ADD:
1454 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1455 reset = diff_get_color_opt(o, DIFF_RESET);
1456 emit_line(o, set, reset, line, len);
1457 break;
1458 case DIFF_SYMBOL_SUBMODULE_DEL:
1459 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1460 reset = diff_get_color_opt(o, DIFF_RESET);
1461 emit_line(o, set, reset, line, len);
1462 break;
1463 case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1464 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1465 diff_line_prefix(o), line);
1466 break;
1467 case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1468 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1469 diff_line_prefix(o), line);
1470 break;
0911c475
SB
1471 case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1472 emit_line(o, "", "", " 0 files changed\n",
1473 strlen(" 0 files changed\n"));
1474 break;
1475 case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1476 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1477 break;
bd033291
SB
1478 case DIFF_SYMBOL_WORD_DIFF:
1479 fprintf(o->file, "%.*s", len, line);
1480 break;
30b7e1e7
SB
1481 case DIFF_SYMBOL_STAT_SEP:
1482 fputs(o->stat_sep, o->file);
1483 break;
36a4cefd 1484 default:
033abf97 1485 BUG("unknown diff symbol");
36a4cefd 1486 }
0911c475 1487 strbuf_release(&sb);
36a4cefd
SB
1488}
1489
e6e045f8
SB
1490static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1491 const char *line, int len, unsigned flags)
1492{
1493 struct emitted_diff_symbol e = {line, len, flags, s};
1494
1495 if (o->emitted_symbols)
1496 append_emitted_diff_symbol(o, &e);
1497 else
1498 emit_diff_symbol_from_struct(o, &e);
1499}
1500
f3597138
SB
1501void diff_emit_submodule_del(struct diff_options *o, const char *line)
1502{
1503 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1504}
1505
1506void diff_emit_submodule_add(struct diff_options *o, const char *line)
1507{
1508 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1509}
1510
1511void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1512{
1513 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1514 path, strlen(path), 0);
1515}
1516
1517void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1518{
1519 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1520 path, strlen(path), 0);
1521}
1522
1523void diff_emit_submodule_header(struct diff_options *o, const char *header)
1524{
1525 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1526 header, strlen(header), 0);
1527}
1528
1529void diff_emit_submodule_error(struct diff_options *o, const char *err)
1530{
1531 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1532}
1533
1534void diff_emit_submodule_pipethrough(struct diff_options *o,
1535 const char *line, int len)
1536{
1537 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1538}
1539
6957eb9a
JH
1540static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1541{
1542 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1543 ecbdata->blank_at_eof_in_preimage &&
1544 ecbdata->blank_at_eof_in_postimage &&
1545 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1546 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1547 return 0;
018cff70 1548 return ws_blank_line(line, len, ecbdata->ws_rule);
6957eb9a
JH
1549}
1550
b8767f79 1551static void emit_add_line(const char *reset,
0e383e18
JH
1552 struct emit_callback *ecbdata,
1553 const char *line, int len)
1554{
091f8e28
SB
1555 unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1556 if (new_blank_line_at_eof(ecbdata, line, len))
1557 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1558
1559 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
b8767f79 1560}
0e383e18 1561
b8767f79
JH
1562static void emit_del_line(const char *reset,
1563 struct emit_callback *ecbdata,
1564 const char *line, int len)
1565{
091f8e28
SB
1566 unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1567 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
0e383e18
JH
1568}
1569
1570static void emit_context_line(const char *reset,
1571 struct emit_callback *ecbdata,
1572 const char *line, int len)
1573{
091f8e28
SB
1574 unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1575 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
0e383e18
JH
1576}
1577
89cb73a1
BW
1578static void emit_hunk_header(struct emit_callback *ecbdata,
1579 const char *line, int len)
1580{
8dbf3eb6 1581 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
89cb73a1
BW
1582 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1583 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1584 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
f7c3b4e2 1585 const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
89cb73a1
BW
1586 static const char atat[2] = { '@', '@' };
1587 const char *cp, *ep;
2efcc977
BY
1588 struct strbuf msgbuf = STRBUF_INIT;
1589 int org_len = len;
1590 int i = 1;
89cb73a1
BW
1591
1592 /*
1593 * As a hunk header must begin with "@@ -<old>, +<new> @@",
1594 * it always is at least 10 bytes long.
1595 */
1596 if (len < 10 ||
1597 memcmp(line, atat, 2) ||
1598 !(ep = memmem(line + 2, len - 2, atat, 2))) {
c64b420b 1599 emit_diff_symbol(ecbdata->opt,
091f8e28 1600 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
89cb73a1
BW
1601 return;
1602 }
1603 ep += 2; /* skip over @@ */
1604
1605 /* The hunk header in fraginfo color */
f7c3b4e2
JS
1606 if (ecbdata->opt->flags.dual_color_diffed_diffs)
1607 strbuf_addstr(&msgbuf, reverse);
cedc61a9 1608 strbuf_addstr(&msgbuf, frag);
2efcc977 1609 strbuf_add(&msgbuf, line, ep - line);
cedc61a9 1610 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
1611
1612 /*
1613 * trailing "\r\n"
1614 */
1615 for ( ; i < 3; i++)
1616 if (line[len - i] == '\r' || line[len - i] == '\n')
1617 len--;
89cb73a1
BW
1618
1619 /* blank before the func header */
1620 for (cp = ep; ep - line < len; ep++)
1621 if (*ep != ' ' && *ep != '\t')
1622 break;
2efcc977 1623 if (ep != cp) {
8dbf3eb6 1624 strbuf_addstr(&msgbuf, context);
2efcc977 1625 strbuf_add(&msgbuf, cp, ep - cp);
cedc61a9 1626 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
1627 }
1628
1629 if (ep < line + len) {
cedc61a9 1630 strbuf_addstr(&msgbuf, func);
2efcc977 1631 strbuf_add(&msgbuf, ep, line + len - ep);
cedc61a9 1632 strbuf_addstr(&msgbuf, reset);
2efcc977 1633 }
89cb73a1 1634
2efcc977 1635 strbuf_add(&msgbuf, line + len, org_len - len);
dfb7728f 1636 strbuf_complete_line(&msgbuf);
68abc6f1 1637 emit_diff_symbol(ecbdata->opt,
091f8e28 1638 DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
2efcc977 1639 strbuf_release(&msgbuf);
89cb73a1
BW
1640}
1641
479b0ae8
JK
1642static struct diff_tempfile *claim_diff_tempfile(void) {
1643 int i;
1644 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1645 if (!diff_temp[i].name)
1646 return diff_temp + i;
033abf97 1647 BUG("diff is failing to clean up its tempfiles");
479b0ae8
JK
1648}
1649
479b0ae8
JK
1650static void remove_tempfile(void)
1651{
1652 int i;
a8344abe 1653 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
076aa2cb 1654 if (is_tempfile_active(diff_temp[i].tempfile))
284098f1 1655 delete_tempfile(&diff_temp[i].tempfile);
a8344abe
NR
1656 diff_temp[i].name = NULL;
1657 }
479b0ae8
JK
1658}
1659
5af6ea95 1660static void add_line_count(struct strbuf *out, int count)
6973dcae
JH
1661{
1662 switch (count) {
1663 case 0:
5af6ea95 1664 strbuf_addstr(out, "0,0");
6973dcae
JH
1665 break;
1666 case 1:
5af6ea95 1667 strbuf_addstr(out, "1");
6973dcae
JH
1668 break;
1669 default:
5af6ea95 1670 strbuf_addf(out, "1,%d", count);
6973dcae
JH
1671 break;
1672 }
1673}
1674
7f7ee2ff
JH
1675static void emit_rewrite_lines(struct emit_callback *ecb,
1676 int prefix, const char *data, int size)
6973dcae 1677{
7f7ee2ff 1678 const char *endp = NULL;
7f7ee2ff
JH
1679 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1680
1681 while (0 < size) {
1682 int len;
1683
1684 endp = memchr(data, '\n', size);
1685 len = endp ? (endp - data + 1) : size;
1686 if (prefix != '+') {
1687 ecb->lno_in_preimage++;
0e383e18 1688 emit_del_line(reset, ecb, data, len);
7f7ee2ff
JH
1689 } else {
1690 ecb->lno_in_postimage++;
1691 emit_add_line(reset, ecb, data, len);
13e36ec5 1692 }
7f7ee2ff
JH
1693 size -= len;
1694 data += len;
1695 }
b9cbfde6 1696 if (!endp)
091f8e28 1697 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
6973dcae
JH
1698}
1699
1700static void emit_rewrite_diff(const char *name_a,
1701 const char *name_b,
1702 struct diff_filespec *one,
13e36ec5 1703 struct diff_filespec *two,
d9bae1a1
JK
1704 struct userdiff_driver *textconv_one,
1705 struct userdiff_driver *textconv_two,
eab9a40b 1706 struct diff_options *o)
6973dcae
JH
1707{
1708 int lc_a, lc_b;
d5625091 1709 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
a5a818ee 1710 const char *a_prefix, *b_prefix;
840383b2 1711 char *data_one, *data_two;
3aa1f7ca 1712 size_t size_one, size_two;
7f7ee2ff 1713 struct emit_callback ecbdata;
5af6ea95 1714 struct strbuf out = STRBUF_INIT;
a5a818ee 1715
0d1e0e78 1716 if (diff_mnemonic_prefix && o->flags.reverse_diff) {
a5a818ee
JH
1717 a_prefix = o->b_prefix;
1718 b_prefix = o->a_prefix;
1719 } else {
1720 a_prefix = o->a_prefix;
1721 b_prefix = o->b_prefix;
1722 }
1a9eb3b9 1723
8a13becc
JH
1724 name_a += (*name_a == '/');
1725 name_b += (*name_b == '/');
1a9eb3b9 1726
d5625091
JH
1727 strbuf_reset(&a_name);
1728 strbuf_reset(&b_name);
a5a818ee
JH
1729 quote_two_c_style(&a_name, a_prefix, name_a, 0);
1730 quote_two_c_style(&b_name, b_prefix, name_b, 0);
d5625091 1731
6afaf807
NTND
1732 size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1733 size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
3aa1f7ca 1734
d91ba8fa 1735 memset(&ecbdata, 0, sizeof(ecbdata));
daa0c3d9 1736 ecbdata.color_diff = want_color(o->use_color);
26d024ec 1737 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
a3c158d4 1738 ecbdata.opt = o;
d91ba8fa
JH
1739 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1740 mmfile_t mf1, mf2;
1741 mf1.ptr = (char *)data_one;
1742 mf2.ptr = (char *)data_two;
1743 mf1.size = size_one;
1744 mf2.size = size_two;
1745 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1746 }
1747 ecbdata.lno_in_preimage = 1;
1748 ecbdata.lno_in_postimage = 1;
1749
3aa1f7ca
JK
1750 lc_a = count_lines(data_one, size_one);
1751 lc_b = count_lines(data_two, size_two);
3ee8b7bf
SB
1752
1753 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1754 a_name.buf, a_name.len, 0);
1755 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1756 b_name.buf, b_name.len, 0);
1757
5af6ea95 1758 strbuf_addstr(&out, "@@ -");
467ddc14 1759 if (!o->irreversible_delete)
5af6ea95 1760 add_line_count(&out, lc_a);
467ddc14 1761 else
5af6ea95
SB
1762 strbuf_addstr(&out, "?,?");
1763 strbuf_addstr(&out, " +");
1764 add_line_count(&out, lc_b);
1765 strbuf_addstr(&out, " @@\n");
1766 emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1767 strbuf_release(&out);
1768
467ddc14 1769 if (lc_a && !o->irreversible_delete)
d91ba8fa 1770 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
6973dcae 1771 if (lc_b)
d91ba8fa 1772 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
b76c056b 1773 if (textconv_one)
aed6ca52 1774 free((char *)data_one);
b76c056b 1775 if (textconv_two)
aed6ca52 1776 free((char *)data_two);
6973dcae
JH
1777}
1778
f59a59e2
JS
1779struct diff_words_buffer {
1780 mmfile_t text;
071bcaab 1781 unsigned long alloc;
2e5d2003
JS
1782 struct diff_words_orig {
1783 const char *begin, *end;
1784 } *orig;
1785 int orig_nr, orig_alloc;
f59a59e2
JS
1786};
1787
1788static void diff_words_append(char *line, unsigned long len,
1789 struct diff_words_buffer *buffer)
1790{
23c1575f 1791 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
f59a59e2
JS
1792 line++;
1793 len--;
1794 memcpy(buffer->text.ptr + buffer->text.size, line, len);
1795 buffer->text.size += len;
2b6a5417 1796 buffer->text.ptr[buffer->text.size] = '\0';
f59a59e2
JS
1797}
1798
9cba13ca 1799struct diff_words_style_elem {
882749a0
TR
1800 const char *prefix;
1801 const char *suffix;
1802 const char *color; /* NULL; filled in by the setup code if
1803 * color is enabled */
1804};
1805
9cba13ca 1806struct diff_words_style {
882749a0 1807 enum diff_words_type type;
63a01c3f 1808 struct diff_words_style_elem new_word, old_word, ctx;
882749a0
TR
1809 const char *newline;
1810};
1811
c2e86add 1812static struct diff_words_style diff_words_styles[] = {
882749a0
TR
1813 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1814 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1815 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1816};
1817
f59a59e2 1818struct diff_words_data {
f59a59e2 1819 struct diff_words_buffer minus, plus;
2e5d2003 1820 const char *current_plus;
4297c0ae
BY
1821 int last_minus;
1822 struct diff_options *opt;
2b6a5417 1823 regex_t *word_regex;
882749a0
TR
1824 enum diff_words_type type;
1825 struct diff_words_style *style;
f59a59e2
JS
1826};
1827
bd033291 1828static int fn_out_diff_words_write_helper(struct diff_options *o,
882749a0
TR
1829 struct diff_words_style_elem *st_el,
1830 const char *newline,
bd033291 1831 size_t count, const char *buf)
882749a0 1832{
4297c0ae 1833 int print = 0;
bd033291 1834 struct strbuf sb = STRBUF_INIT;
4297c0ae 1835
882749a0
TR
1836 while (count) {
1837 char *p = memchr(buf, '\n', count);
4297c0ae 1838 if (print)
bd033291
SB
1839 strbuf_addstr(&sb, diff_line_prefix(o));
1840
882749a0 1841 if (p != buf) {
bd033291
SB
1842 const char *reset = st_el->color && *st_el->color ?
1843 GIT_COLOR_RESET : NULL;
1844 if (st_el->color && *st_el->color)
1845 strbuf_addstr(&sb, st_el->color);
1846 strbuf_addstr(&sb, st_el->prefix);
1847 strbuf_add(&sb, buf, p ? p - buf : count);
1848 strbuf_addstr(&sb, st_el->suffix);
1849 if (reset)
1850 strbuf_addstr(&sb, reset);
882749a0
TR
1851 }
1852 if (!p)
bd033291
SB
1853 goto out;
1854
1855 strbuf_addstr(&sb, newline);
882749a0
TR
1856 count -= p + 1 - buf;
1857 buf = p + 1;
4297c0ae 1858 print = 1;
bd033291
SB
1859 if (count) {
1860 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1861 sb.buf, sb.len, 0);
1862 strbuf_reset(&sb);
1863 }
882749a0 1864 }
bd033291
SB
1865
1866out:
1867 if (sb.len)
1868 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1869 sb.buf, sb.len, 0);
1870 strbuf_release(&sb);
882749a0
TR
1871 return 0;
1872}
1873
4297c0ae
BY
1874/*
1875 * '--color-words' algorithm can be described as:
1876 *
5621760f 1877 * 1. collect the minus/plus lines of a diff hunk, divided into
4297c0ae
BY
1878 * minus-lines and plus-lines;
1879 *
1880 * 2. break both minus-lines and plus-lines into words and
1881 * place them into two mmfile_t with one word for each line;
1882 *
1883 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1884 *
1885 * And for the common parts of the both file, we output the plus side text.
1886 * diff_words->current_plus is used to trace the current position of the plus file
1887 * which printed. diff_words->last_minus is used to trace the last minus word
1888 * printed.
1889 *
1890 * For '--graph' to work with '--color-words', we need to output the graph prefix
1891 * on each line of color words output. Generally, there are two conditions on
1892 * which we should output the prefix.
1893 *
1894 * 1. diff_words->last_minus == 0 &&
1895 * diff_words->current_plus == diff_words->plus.text.ptr
1896 *
1897 * that is: the plus text must start as a new line, and if there is no minus
1898 * word printed, a graph prefix must be printed.
1899 *
1900 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
1901 * *(diff_words->current_plus - 1) == '\n'
1902 *
1903 * that is: a graph prefix must be printed following a '\n'
1904 */
1905static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1906{
1907 if ((diff_words->last_minus == 0 &&
1908 diff_words->current_plus == diff_words->plus.text.ptr) ||
1909 (diff_words->current_plus > diff_words->plus.text.ptr &&
1910 *(diff_words->current_plus - 1) == '\n')) {
1911 return 1;
1912 } else {
1913 return 0;
1914 }
1915}
1916
7c61e25f
JK
1917static void fn_out_diff_words_aux(void *priv,
1918 long minus_first, long minus_len,
1919 long plus_first, long plus_len,
1920 const char *func, long funclen)
f59a59e2 1921{
f59a59e2 1922 struct diff_words_data *diff_words = priv;
882749a0 1923 struct diff_words_style *style = diff_words->style;
2e5d2003 1924 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
4297c0ae 1925 struct diff_options *opt = diff_words->opt;
30997bb8 1926 const char *line_prefix;
f59a59e2 1927
4297c0ae 1928 assert(opt);
30997bb8 1929 line_prefix = diff_line_prefix(opt);
4297c0ae 1930
2e5d2003
JS
1931 /* POSIX requires that first be decremented by one if len == 0... */
1932 if (minus_len) {
1933 minus_begin = diff_words->minus.orig[minus_first].begin;
1934 minus_end =
1935 diff_words->minus.orig[minus_first + minus_len - 1].end;
1936 } else
1937 minus_begin = minus_end =
1938 diff_words->minus.orig[minus_first].end;
1939
1940 if (plus_len) {
1941 plus_begin = diff_words->plus.orig[plus_first].begin;
1942 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1943 } else
1944 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1945
4297c0ae
BY
1946 if (color_words_output_graph_prefix(diff_words)) {
1947 fputs(line_prefix, diff_words->opt->file);
1948 }
1949 if (diff_words->current_plus != plus_begin) {
bd033291 1950 fn_out_diff_words_write_helper(diff_words->opt,
882749a0
TR
1951 &style->ctx, style->newline,
1952 plus_begin - diff_words->current_plus,
bd033291 1953 diff_words->current_plus);
4297c0ae
BY
1954 }
1955 if (minus_begin != minus_end) {
bd033291 1956 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 1957 &style->old_word, style->newline,
bd033291 1958 minus_end - minus_begin, minus_begin);
4297c0ae
BY
1959 }
1960 if (plus_begin != plus_end) {
bd033291 1961 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 1962 &style->new_word, style->newline,
bd033291 1963 plus_end - plus_begin, plus_begin);
4297c0ae 1964 }
2e5d2003
JS
1965
1966 diff_words->current_plus = plus_end;
4297c0ae 1967 diff_words->last_minus = minus_first;
f59a59e2
JS
1968}
1969
2b6a5417
JS
1970/* This function starts looking at *begin, and returns 0 iff a word was found. */
1971static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1972 int *begin, int *end)
1973{
1974 if (word_regex && *begin < buffer->size) {
1975 regmatch_t match[1];
b7d36ffc
JS
1976 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1977 buffer->size - *begin, 1, match, 0)) {
2b6a5417
JS
1978 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1979 '\n', match[0].rm_eo - match[0].rm_so);
1980 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1981 *begin += match[0].rm_so;
1982 return *begin >= *end;
1983 }
1984 return -1;
f59a59e2
JS
1985 }
1986
2b6a5417
JS
1987 /* find the next word */
1988 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1989 (*begin)++;
1990 if (*begin >= buffer->size)
1991 return -1;
f59a59e2 1992
2b6a5417
JS
1993 /* find the end of the word */
1994 *end = *begin + 1;
1995 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1996 (*end)++;
1997
1998 return 0;
f59a59e2
JS
1999}
2000
23c1575f 2001/*
2e5d2003
JS
2002 * This function splits the words in buffer->text, stores the list with
2003 * newline separator into out, and saves the offsets of the original words
2004 * in buffer->orig.
23c1575f 2005 */
2b6a5417
JS
2006static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2007 regex_t *word_regex)
f59a59e2 2008{
2e5d2003 2009 int i, j;
2b6a5417 2010 long alloc = 0;
f59a59e2 2011
2e5d2003 2012 out->size = 0;
2b6a5417 2013 out->ptr = NULL;
f59a59e2 2014
2e5d2003
JS
2015 /* fake an empty "0th" word */
2016 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2017 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2018 buffer->orig_nr = 1;
2019
2020 for (i = 0; i < buffer->text.size; i++) {
2b6a5417
JS
2021 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2022 return;
2e5d2003
JS
2023
2024 /* store original boundaries */
2025 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2026 buffer->orig_alloc);
2027 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2028 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2029 buffer->orig_nr++;
2030
2031 /* store one word */
2b6a5417 2032 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2e5d2003
JS
2033 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2034 out->ptr[out->size + j - i] = '\n';
2035 out->size += j - i + 1;
2036
2037 i = j - 1;
f59a59e2
JS
2038 }
2039}
2040
2041/* this executes the word diff on the accumulated buffers */
2042static void diff_words_show(struct diff_words_data *diff_words)
2043{
2044 xpparam_t xpp;
2045 xdemitconf_t xecfg;
f59a59e2 2046 mmfile_t minus, plus;
882749a0 2047 struct diff_words_style *style = diff_words->style;
f59a59e2 2048
4297c0ae 2049 struct diff_options *opt = diff_words->opt;
30997bb8 2050 const char *line_prefix;
4297c0ae
BY
2051
2052 assert(opt);
30997bb8 2053 line_prefix = diff_line_prefix(opt);
4297c0ae 2054
2e5d2003
JS
2055 /* special case: only removal */
2056 if (!diff_words->plus.text.size) {
bd033291
SB
2057 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2058 line_prefix, strlen(line_prefix), 0);
2059 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 2060 &style->old_word, style->newline,
4297c0ae 2061 diff_words->minus.text.size,
bd033291 2062 diff_words->minus.text.ptr);
2e5d2003
JS
2063 diff_words->minus.text.size = 0;
2064 return;
2065 }
2066
2067 diff_words->current_plus = diff_words->plus.text.ptr;
4297c0ae 2068 diff_words->last_minus = 0;
f59a59e2 2069
9ccd0a88 2070 memset(&xpp, 0, sizeof(xpp));
30b25010 2071 memset(&xecfg, 0, sizeof(xecfg));
2b6a5417
JS
2072 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2073 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
582aa00b 2074 xpp.flags = 0;
2b6a5417 2075 /* as only the hunk header will be parsed, we need a 0-context */
2e5d2003 2076 xecfg.ctxlen = 0;
7c61e25f 2077 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
9346d6d1 2078 diff_words, &xpp, &xecfg))
3efb9880 2079 die("unable to generate word diff");
f59a59e2
JS
2080 free(minus.ptr);
2081 free(plus.ptr);
2e5d2003 2082 if (diff_words->current_plus != diff_words->plus.text.ptr +
4297c0ae
BY
2083 diff_words->plus.text.size) {
2084 if (color_words_output_graph_prefix(diff_words))
bd033291
SB
2085 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2086 line_prefix, strlen(line_prefix), 0);
2087 fn_out_diff_words_write_helper(diff_words->opt,
882749a0 2088 &style->ctx, style->newline,
2e5d2003 2089 diff_words->plus.text.ptr + diff_words->plus.text.size
bd033291 2090 - diff_words->current_plus, diff_words->current_plus);
4297c0ae 2091 }
f59a59e2 2092 diff_words->minus.text.size = diff_words->plus.text.size = 0;
f59a59e2
JS
2093}
2094
76fd2828
JH
2095/* In "color-words" mode, show word-diff of words accumulated in the buffer */
2096static void diff_words_flush(struct emit_callback *ecbdata)
2097{
e6e045f8
SB
2098 struct diff_options *wo = ecbdata->diff_words->opt;
2099
76fd2828
JH
2100 if (ecbdata->diff_words->minus.text.size ||
2101 ecbdata->diff_words->plus.text.size)
2102 diff_words_show(ecbdata->diff_words);
e6e045f8
SB
2103
2104 if (wo->emitted_symbols) {
2105 struct diff_options *o = ecbdata->opt;
2106 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2107 int i;
2108
2109 /*
2110 * NEEDSWORK:
2111 * Instead of appending each, concat all words to a line?
2112 */
2113 for (i = 0; i < wol->nr; i++)
2114 append_emitted_diff_symbol(o, &wol->buf[i]);
2115
2116 for (i = 0; i < wol->nr; i++)
2117 free((void *)wol->buf[i].line);
2118
2119 wol->nr = 0;
2120 }
76fd2828
JH
2121}
2122
acd00ea0
NTND
2123static void diff_filespec_load_driver(struct diff_filespec *one,
2124 struct index_state *istate)
77d1a520
TR
2125{
2126 /* Use already-loaded driver */
2127 if (one->driver)
2128 return;
2129
2130 if (S_ISREG(one->mode))
acd00ea0 2131 one->driver = userdiff_find_by_path(istate, one->path);
77d1a520
TR
2132
2133 /* Fallback to default settings */
2134 if (!one->driver)
2135 one->driver = userdiff_find_by_name("default");
2136}
2137
acd00ea0
NTND
2138static const char *userdiff_word_regex(struct diff_filespec *one,
2139 struct index_state *istate)
77d1a520 2140{
acd00ea0 2141 diff_filespec_load_driver(one, istate);
77d1a520
TR
2142 return one->driver->word_regex;
2143}
2144
2145static void init_diff_words_data(struct emit_callback *ecbdata,
6440d341 2146 struct diff_options *orig_opts,
77d1a520
TR
2147 struct diff_filespec *one,
2148 struct diff_filespec *two)
2149{
2150 int i;
6440d341
TR
2151 struct diff_options *o = xmalloc(sizeof(struct diff_options));
2152 memcpy(o, orig_opts, sizeof(struct diff_options));
77d1a520
TR
2153
2154 ecbdata->diff_words =
2155 xcalloc(1, sizeof(struct diff_words_data));
2156 ecbdata->diff_words->type = o->word_diff;
2157 ecbdata->diff_words->opt = o;
e6e045f8
SB
2158
2159 if (orig_opts->emitted_symbols)
2160 o->emitted_symbols =
2161 xcalloc(1, sizeof(struct emitted_diff_symbols));
2162
77d1a520 2163 if (!o->word_regex)
acd00ea0 2164 o->word_regex = userdiff_word_regex(one, o->repo->index);
77d1a520 2165 if (!o->word_regex)
acd00ea0 2166 o->word_regex = userdiff_word_regex(two, o->repo->index);
77d1a520
TR
2167 if (!o->word_regex)
2168 o->word_regex = diff_word_regex_cfg;
2169 if (o->word_regex) {
2170 ecbdata->diff_words->word_regex = (regex_t *)
2171 xmalloc(sizeof(regex_t));
2172 if (regcomp(ecbdata->diff_words->word_regex,
2173 o->word_regex,
2174 REG_EXTENDED | REG_NEWLINE))
1a07e59c
NTND
2175 die("invalid regular expression: %s",
2176 o->word_regex);
77d1a520
TR
2177 }
2178 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2179 if (o->word_diff == diff_words_styles[i].type) {
2180 ecbdata->diff_words->style =
2181 &diff_words_styles[i];
2182 break;
2183 }
2184 }
2185 if (want_color(o->use_color)) {
2186 struct diff_words_style *st = ecbdata->diff_words->style;
63a01c3f
BW
2187 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2188 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
8dbf3eb6 2189 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
77d1a520
TR
2190 }
2191}
2192
f59a59e2
JS
2193static void free_diff_words_data(struct emit_callback *ecbdata)
2194{
2195 if (ecbdata->diff_words) {
76fd2828 2196 diff_words_flush(ecbdata);
e6e045f8 2197 free (ecbdata->diff_words->opt->emitted_symbols);
6440d341 2198 free (ecbdata->diff_words->opt);
8e0f7003 2199 free (ecbdata->diff_words->minus.text.ptr);
2e5d2003 2200 free (ecbdata->diff_words->minus.orig);
8e0f7003 2201 free (ecbdata->diff_words->plus.text.ptr);
2e5d2003 2202 free (ecbdata->diff_words->plus.orig);
ef5644ea
BC
2203 if (ecbdata->diff_words->word_regex) {
2204 regfree(ecbdata->diff_words->word_regex);
2205 free(ecbdata->diff_words->word_regex);
2206 }
6a83d902 2207 FREE_AND_NULL(ecbdata->diff_words);
f59a59e2
JS
2208 }
2209}
2210
ce436973 2211const char *diff_get_color(int diff_use_color, enum color_diff ix)
cd112cef 2212{
daa0c3d9 2213 if (want_color(diff_use_color))
50f575fc
LT
2214 return diff_colors[ix];
2215 return "";
cd112cef
JS
2216}
2217
f1922234
JK
2218const char *diff_line_prefix(struct diff_options *opt)
2219{
2220 struct strbuf *msgbuf;
2221 if (!opt->output_prefix)
2222 return "";
2223
2224 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2225 return msgbuf->buf;
2226}
2227
23707811
JH
2228static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2229{
2230 const char *cp;
2231 unsigned long allot;
2232 size_t l = len;
2233
23707811
JH
2234 cp = line;
2235 allot = l;
2236 while (0 < l) {
2237 (void) utf8_width(&cp, &l);
2238 if (!cp)
2239 break; /* truncated in the middle? */
2240 }
2241 return allot - l;
2242}
2243
d68fe26f 2244static void find_lno(const char *line, struct emit_callback *ecbdata)
690ed843 2245{
d68fe26f
JH
2246 const char *p;
2247 ecbdata->lno_in_preimage = 0;
2248 ecbdata->lno_in_postimage = 0;
2249 p = strchr(line, '-');
690ed843 2250 if (!p)
d68fe26f
JH
2251 return; /* cannot happen */
2252 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2253 p = strchr(p, '+');
2254 if (!p)
2255 return; /* cannot happen */
2256 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
690ed843
JH
2257}
2258
cd112cef 2259static void fn_out_consume(void *priv, char *line, unsigned long len)
6973dcae 2260{
6973dcae 2261 struct emit_callback *ecbdata = priv;
ce436973 2262 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
7be57610 2263 struct diff_options *o = ecbdata->opt;
6973dcae 2264
ba16233c
SB
2265 o->found_changes = 1;
2266
3e97c7c6 2267 if (ecbdata->header) {
a29b0a13
SB
2268 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2269 ecbdata->header->buf, ecbdata->header->len, 0);
3e97c7c6
GB
2270 strbuf_reset(ecbdata->header);
2271 ecbdata->header = NULL;
2272 }
34a5e1a2 2273
6973dcae 2274 if (ecbdata->label_path[0]) {
3ee8b7bf
SB
2275 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2276 ecbdata->label_path[0],
2277 strlen(ecbdata->label_path[0]), 0);
2278 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2279 ecbdata->label_path[1],
2280 strlen(ecbdata->label_path[1]), 0);
6973dcae
JH
2281 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2282 }
cd112cef 2283
a624eaa7
JM
2284 if (diff_suppress_blank_empty
2285 && len == 2 && line[0] == ' ' && line[1] == '\n') {
2286 line[0] = '\n';
2287 len = 1;
2288 }
2289
b8d9c1a6 2290 if (line[0] == '@') {
76fd2828
JH
2291 if (ecbdata->diff_words)
2292 diff_words_flush(ecbdata);
23707811 2293 len = sane_truncate_line(ecbdata, line, len);
d68fe26f 2294 find_lno(line, ecbdata);
89cb73a1 2295 emit_hunk_header(ecbdata, line, len);
448c3ef1 2296 return;
cd112cef 2297 }
448c3ef1 2298
448c3ef1 2299 if (ecbdata->diff_words) {
ff958679
SB
2300 enum diff_symbol s =
2301 ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2302 DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
448c3ef1
JH
2303 if (line[0] == '-') {
2304 diff_words_append(line, len,
2305 &ecbdata->diff_words->minus);
2306 return;
2307 } else if (line[0] == '+') {
2308 diff_words_append(line, len,
2309 &ecbdata->diff_words->plus);
2310 return;
59556548 2311 } else if (starts_with(line, "\\ ")) {
c7c2bc0a
TR
2312 /*
2313 * Eat the "no newline at eof" marker as if we
2314 * saw a "+" or "-" line with nothing on it,
2315 * and return without diff_words_flush() to
2316 * defer processing. If this is the end of
2317 * preimage, more "+" lines may come after it.
2318 */
2319 return;
448c3ef1 2320 }
76fd2828 2321 diff_words_flush(ecbdata);
ff958679 2322 emit_diff_symbol(o, s, line, len, 0);
448c3ef1
JH
2323 return;
2324 }
448c3ef1 2325
0e383e18
JH
2326 switch (line[0]) {
2327 case '+':
d68fe26f 2328 ecbdata->lno_in_postimage++;
018cff70 2329 emit_add_line(reset, ecbdata, line + 1, len - 1);
0e383e18
JH
2330 break;
2331 case '-':
2332 ecbdata->lno_in_preimage++;
2333 emit_del_line(reset, ecbdata, line + 1, len - 1);
2334 break;
2335 case ' ':
2336 ecbdata->lno_in_postimage++;
2337 ecbdata->lno_in_preimage++;
2338 emit_context_line(reset, ecbdata, line + 1, len - 1);
2339 break;
2340 default:
2341 /* incomplete line at the end */
2342 ecbdata->lno_in_preimage++;
f2bb1218
SB
2343 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2344 line, len, 0);
0e383e18 2345 break;
448c3ef1 2346 }
6973dcae
JH
2347}
2348
c905cbc4 2349static void pprint_rename(struct strbuf *name, const char *a, const char *b)
6973dcae 2350{
63a01c3f
BW
2351 const char *old_name = a;
2352 const char *new_name = b;
6973dcae 2353 int pfx_length, sfx_length;
dd281f09 2354 int pfx_adjust_for_slash;
6973dcae
JH
2355 int len_a = strlen(a);
2356 int len_b = strlen(b);
663af342 2357 int a_midlen, b_midlen;
e5bfbf9b
AJ
2358 int qlen_a = quote_c_style(a, NULL, NULL, 0);
2359 int qlen_b = quote_c_style(b, NULL, NULL, 0);
2360
2361 if (qlen_a || qlen_b) {
c905cbc4
NTND
2362 quote_c_style(a, name, NULL, 0);
2363 strbuf_addstr(name, " => ");
2364 quote_c_style(b, name, NULL, 0);
2365 return;
e5bfbf9b 2366 }
6973dcae
JH
2367
2368 /* Find common prefix */
2369 pfx_length = 0;
63a01c3f
BW
2370 while (*old_name && *new_name && *old_name == *new_name) {
2371 if (*old_name == '/')
2372 pfx_length = old_name - a + 1;
2373 old_name++;
2374 new_name++;
6973dcae
JH
2375 }
2376
2377 /* Find common suffix */
63a01c3f
BW
2378 old_name = a + len_a;
2379 new_name = b + len_b;
6973dcae 2380 sfx_length = 0;
d020e27f 2381 /*
dd281f09
TR
2382 * If there is a common prefix, it must end in a slash. In
2383 * that case we let this loop run 1 into the prefix to see the
2384 * same slash.
2385 *
2386 * If there is no common prefix, we cannot do this as it would
2387 * underrun the input strings.
d020e27f 2388 */
dd281f09 2389 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
63a01c3f
BW
2390 while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2391 b + pfx_length - pfx_adjust_for_slash <= new_name &&
2392 *old_name == *new_name) {
2393 if (*old_name == '/')
2394 sfx_length = len_a - (old_name - a);
2395 old_name--;
2396 new_name--;
6973dcae
JH
2397 }
2398
2399 /*
2400 * pfx{mid-a => mid-b}sfx
2401 * {pfx-a => pfx-b}sfx
2402 * pfx{sfx-a => sfx-b}
2403 * name-a => name-b
2404 */
663af342
PH
2405 a_midlen = len_a - pfx_length - sfx_length;
2406 b_midlen = len_b - pfx_length - sfx_length;
2407 if (a_midlen < 0)
2408 a_midlen = 0;
2409 if (b_midlen < 0)
2410 b_midlen = 0;
2411
c905cbc4 2412 strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
6973dcae 2413 if (pfx_length + sfx_length) {
c905cbc4
NTND
2414 strbuf_add(name, a, pfx_length);
2415 strbuf_addch(name, '{');
6973dcae 2416 }
c905cbc4
NTND
2417 strbuf_add(name, a + pfx_length, a_midlen);
2418 strbuf_addstr(name, " => ");
2419 strbuf_add(name, b + pfx_length, b_midlen);
663af342 2420 if (pfx_length + sfx_length) {
c905cbc4
NTND
2421 strbuf_addch(name, '}');
2422 strbuf_add(name, a + len_a - sfx_length, sfx_length);
6973dcae 2423 }
6973dcae
JH
2424}
2425
2426struct diffstat_t {
6973dcae
JH
2427 int nr;
2428 int alloc;
2429 struct diffstat_file {
f604652e 2430 char *from_name;
6973dcae 2431 char *name;
f604652e 2432 char *print_name;
ddf88fa6 2433 const char *comments;
6973dcae
JH
2434 unsigned is_unmerged:1;
2435 unsigned is_binary:1;
2436 unsigned is_renamed:1;
74faaa16 2437 unsigned is_interesting:1;
0974c117 2438 uintmax_t added, deleted;
6973dcae
JH
2439 } **files;
2440};
2441
2442static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2443 const char *name_a,
2444 const char *name_b)
2445{
2446 struct diffstat_file *x;
1a4927c5 2447 x = xcalloc(1, sizeof(*x));
4c960a43 2448 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
6973dcae
JH
2449 diffstat->files[diffstat->nr++] = x;
2450 if (name_b) {
f604652e
JH
2451 x->from_name = xstrdup(name_a);
2452 x->name = xstrdup(name_b);
6973dcae
JH
2453 x->is_renamed = 1;
2454 }
f604652e
JH
2455 else {
2456 x->from_name = NULL;
9befac47 2457 x->name = xstrdup(name_a);
f604652e 2458 }
6973dcae
JH
2459 return x;
2460}
2461
2462static void diffstat_consume(void *priv, char *line, unsigned long len)
2463{
2464 struct diffstat_t *diffstat = priv;
2465 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2466
2467 if (line[0] == '+')
2468 x->added++;
2469 else if (line[0] == '-')
2470 x->deleted++;
2471}
2472
698ce6f8 2473const char mime_boundary_leader[] = "------------";
6973dcae 2474
a2540023
JH
2475static int scale_linear(int it, int width, int max_change)
2476{
2eeeef24
JH
2477 if (!it)
2478 return 0;
a2540023 2479 /*
2eeeef24
JH
2480 * make sure that at least one '-' or '+' is printed if
2481 * there is any change to this path. The easiest way is to
2482 * scale linearly as if the alloted width is one column shorter
2483 * than it is, and then add 1 to the result.
a2540023 2484 */
2eeeef24 2485 return 1 + (it * (width - 1) / max_change);
a2540023
JH
2486}
2487
0911c475
SB
2488static void show_graph(struct strbuf *out, char ch, int cnt,
2489 const char *set, const char *reset)
a2540023
JH
2490{
2491 if (cnt <= 0)
2492 return;
0911c475
SB
2493 strbuf_addstr(out, set);
2494 strbuf_addchars(out, ch, cnt);
2495 strbuf_addstr(out, reset);
a2540023
JH
2496}
2497
f604652e
JH
2498static void fill_print_name(struct diffstat_file *file)
2499{
c905cbc4 2500 struct strbuf pname = STRBUF_INIT;
f604652e
JH
2501
2502 if (file->print_name)
2503 return;
2504
c905cbc4
NTND
2505 if (file->is_renamed)
2506 pprint_rename(&pname, file->from_name, file->name);
2507 else
2508 quote_c_style(file->name, &pname, NULL, 0);
2509
ddf88fa6
NTND
2510 if (file->comments)
2511 strbuf_addf(&pname, " (%s)", file->comments);
2512
c905cbc4 2513 file->print_name = strbuf_detach(&pname, NULL);
f604652e
JH
2514}
2515
0911c475
SB
2516static void print_stat_summary_inserts_deletes(struct diff_options *options,
2517 int files, int insertions, int deletions)
7f814632
NTND
2518{
2519 struct strbuf sb = STRBUF_INIT;
7f814632
NTND
2520
2521 if (!files) {
2522 assert(insertions == 0 && deletions == 0);
0911c475
SB
2523 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2524 NULL, 0, 0);
2525 return;
7f814632
NTND
2526 }
2527
2528 strbuf_addf(&sb,
218adaaa 2529 (files == 1) ? " %d file changed" : " %d files changed",
7f814632
NTND
2530 files);
2531
2532 /*
2533 * For binary diff, the caller may want to print "x files
2534 * changed" with insertions == 0 && deletions == 0.
2535 *
2536 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2537 * is probably less confusing (i.e skip over "2 files changed
2538 * but nothing about added/removed lines? Is this a bug in Git?").
2539 */
2540 if (insertions || deletions == 0) {
7f814632 2541 strbuf_addf(&sb,
218adaaa 2542 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
7f814632
NTND
2543 insertions);
2544 }
2545
2546 if (deletions || insertions == 0) {
7f814632 2547 strbuf_addf(&sb,
218adaaa 2548 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
7f814632
NTND
2549 deletions);
2550 }
2551 strbuf_addch(&sb, '\n');
0911c475
SB
2552 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2553 sb.buf, sb.len, 0);
7f814632 2554 strbuf_release(&sb);
0911c475
SB
2555}
2556
2557void print_stat_summary(FILE *fp, int files,
2558 int insertions, int deletions)
2559{
2560 struct diff_options o;
2561 memset(&o, 0, sizeof(o));
2562 o.file = fp;
2563
2564 print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
7f814632
NTND
2565}
2566
4b25d091 2567static void show_stats(struct diffstat_t *data, struct diff_options *options)
6973dcae 2568{
eb3a9dd3 2569 int i, len, add, del, adds = 0, dels = 0;
0974c117 2570 uintmax_t max_change = 0, max_len = 0;
dc801e71
ZJS
2571 int total_files = data->nr, count;
2572 int width, name_width, graph_width, number_width = 0, bin_width = 0;
c0aa335c 2573 const char *reset, *add_c, *del_c;
e5f85df8 2574 int extra_shown = 0;
0911c475
SB
2575 const char *line_prefix = diff_line_prefix(options);
2576 struct strbuf out = STRBUF_INIT;
6973dcae
JH
2577
2578 if (data->nr == 0)
2579 return;
2580
808e1db2 2581 count = options->stat_count ? options->stat_count : data->nr;
a2540023 2582
8f67f8ae 2583 reset = diff_get_color_opt(options, DIFF_RESET);
8f67f8ae
PH
2584 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2585 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
785f7432 2586
1b058bc3
ZJS
2587 /*
2588 * Find the longest filename and max number of changes
2589 */
808e1db2 2590 for (i = 0; (i < count) && (i < data->nr); i++) {
6973dcae 2591 struct diffstat_file *file = data->files[i];
0974c117 2592 uintmax_t change = file->added + file->deleted;
af0ed819
JH
2593
2594 if (!file->is_interesting && (change == 0)) {
808e1db2 2595 count++; /* not shown == room for one more */
358e460e
MG
2596 continue;
2597 }
f604652e
JH
2598 fill_print_name(file);
2599 len = strlen(file->print_name);
6973dcae
JH
2600 if (max_len < len)
2601 max_len = len;
2602
dc801e71
ZJS
2603 if (file->is_unmerged) {
2604 /* "Unmerged" is 8 characters */
2605 bin_width = bin_width < 8 ? 8 : bin_width;
6973dcae 2606 continue;
dc801e71
ZJS
2607 }
2608 if (file->is_binary) {
2609 /* "Bin XXX -> YYY bytes" */
2610 int w = 14 + decimal_width(file->added)
2611 + decimal_width(file->deleted);
2612 bin_width = bin_width < w ? w : bin_width;
2613 /* Display change counts aligned with "Bin" */
2614 number_width = 3;
2615 continue;
2616 }
2617
a2540023
JH
2618 if (max_change < change)
2619 max_change = change;
6973dcae 2620 }
a20d3c0d 2621 count = i; /* where we can stop scanning in data->files[] */
6973dcae 2622
1b058bc3
ZJS
2623 /*
2624 * We have width = stat_width or term_columns() columns total.
2625 * We want a maximum of min(max_len, stat_name_width) for the name part.
969fe57b 2626 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1b058bc3
ZJS
2627 * We also need 1 for " " and 4 + decimal_width(max_change)
2628 * for " | NNNN " and one the empty column at the end, altogether
2629 * 6 + decimal_width(max_change).
2630 *
2631 * If there's not enough space, we will use the smaller of
2632 * stat_name_width (if set) and 5/8*width for the filename,
969fe57b
ZJS
2633 * and the rest for constant elements + graph part, but no more
2634 * than stat_graph_width for the graph part.
1b058bc3
ZJS
2635 * (5/8 gives 50 for filename and 30 for the constant parts + graph
2636 * for the standard terminal size).
a2540023 2637 *
1b058bc3
ZJS
2638 * In other words: stat_width limits the maximum width, and
2639 * stat_name_width fixes the maximum width of the filename,
2640 * and is also used to divide available columns if there
2641 * aren't enough.
dc801e71
ZJS
2642 *
2643 * Binary files are displayed with "Bin XXX -> YYY bytes"
2644 * instead of the change count and graph. This part is treated
2645 * similarly to the graph part, except that it is not
41ccfdd9 2646 * "scaled". If total width is too small to accommodate the
dc801e71
ZJS
2647 * guaranteed minimum width of the filename part and the
2648 * separators and this message, this message will "overflow"
2649 * making the line longer than the maximum width.
a2540023 2650 */
1b058bc3
ZJS
2651
2652 if (options->stat_width == -1)
cd48dadb 2653 width = term_columns() - strlen(line_prefix);
a2540023 2654 else
1b058bc3 2655 width = options->stat_width ? options->stat_width : 80;
dc801e71
ZJS
2656 number_width = decimal_width(max_change) > number_width ?
2657 decimal_width(max_change) : number_width;
a2540023 2658
df44483a
ZJS
2659 if (options->stat_graph_width == -1)
2660 options->stat_graph_width = diff_stat_graph_width;
a2540023 2661
1b058bc3
ZJS
2662 /*
2663 * Guarantee 3/8*16==6 for the graph part
2664 * and 5/8*16==10 for the filename part
2665 */
2666 if (width < 16 + 6 + number_width)
2667 width = 16 + 6 + number_width;
2668
2669 /*
2670 * First assign sizes that are wanted, ignoring available width.
dc801e71
ZJS
2671 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2672 * starting from "XXX" should fit in graph_width.
1b058bc3 2673 */
dc801e71
ZJS
2674 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2675 if (options->stat_graph_width &&
2676 options->stat_graph_width < graph_width)
2677 graph_width = options->stat_graph_width;
2678
1b058bc3
ZJS
2679 name_width = (options->stat_name_width > 0 &&
2680 options->stat_name_width < max_len) ?
2681 options->stat_name_width : max_len;
2682
2683 /*
2684 * Adjust adjustable widths not to exceed maximum width
2685 */
2686 if (name_width + number_width + 6 + graph_width > width) {
678c5741 2687 if (graph_width > width * 3/8 - number_width - 6) {
1b058bc3 2688 graph_width = width * 3/8 - number_width - 6;
678c5741
LP
2689 if (graph_width < 6)
2690 graph_width = 6;
2691 }
2692
969fe57b
ZJS
2693 if (options->stat_graph_width &&
2694 graph_width > options->stat_graph_width)
2695 graph_width = options->stat_graph_width;
1b058bc3
ZJS
2696 if (name_width > width - number_width - 6 - graph_width)
2697 name_width = width - number_width - 6 - graph_width;
2698 else
2699 graph_width = width - number_width - 6 - name_width;
2700 }
2701
2702 /*
2703 * From here name_width is the width of the name area,
2704 * and graph_width is the width of the graph area.
2705 * max_change is used to scale graph properly.
2706 */
808e1db2 2707 for (i = 0; i < count; i++) {
d2543b8e 2708 const char *prefix = "";
af0ed819
JH
2709 struct diffstat_file *file = data->files[i];
2710 char *name = file->print_name;
2711 uintmax_t added = file->added;
2712 uintmax_t deleted = file->deleted;
a2540023 2713 int name_len;
6973dcae 2714
a20d3c0d 2715 if (!file->is_interesting && (added + deleted == 0))
358e460e 2716 continue;
a20d3c0d 2717
6973dcae
JH
2718 /*
2719 * "scale" the filename
2720 */
a2540023
JH
2721 len = name_width;
2722 name_len = strlen(name);
2723 if (name_width < name_len) {
6973dcae
JH
2724 char *slash;
2725 prefix = "...";
a2540023
JH
2726 len -= 3;
2727 name += name_len - len;
6973dcae
JH
2728 slash = strchr(name, '/');
2729 if (slash)
2730 name = slash;
2731 }
6973dcae 2732
af0ed819 2733 if (file->is_binary) {
0911c475
SB
2734 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2735 strbuf_addf(&out, " %*s", number_width, "Bin");
e18872b2 2736 if (!added && !deleted) {
0911c475
SB
2737 strbuf_addch(&out, '\n');
2738 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2739 out.buf, out.len, 0);
2740 strbuf_reset(&out);
e18872b2
ZJS
2741 continue;
2742 }
0911c475 2743 strbuf_addf(&out, " %s%"PRIuMAX"%s",
0974c117 2744 del_c, deleted, reset);
0911c475
SB
2745 strbuf_addstr(&out, " -> ");
2746 strbuf_addf(&out, "%s%"PRIuMAX"%s",
0974c117 2747 add_c, added, reset);
0911c475
SB
2748 strbuf_addstr(&out, " bytes\n");
2749 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2750 out.buf, out.len, 0);
2751 strbuf_reset(&out);
f604652e 2752 continue;
6973dcae 2753 }
af0ed819 2754 else if (file->is_unmerged) {
0911c475
SB
2755 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2756 strbuf_addstr(&out, " Unmerged\n");
2757 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2758 out.buf, out.len, 0);
2759 strbuf_reset(&out);
f604652e 2760 continue;
6973dcae 2761 }
6973dcae 2762
a2540023
JH
2763 /*
2764 * scale the add/delete
2765 */
6973dcae
JH
2766 add = added;
2767 del = deleted;
6973dcae 2768
1b058bc3 2769 if (graph_width <= max_change) {
d3c9cf32 2770 int total = scale_linear(add + del, graph_width, max_change);
2eeeef24
JH
2771 if (total < 2 && add && del)
2772 /* width >= 2 due to the sanity check */
2773 total = 2;
2774 if (add < del) {
1b058bc3 2775 add = scale_linear(add, graph_width, max_change);
2eeeef24
JH
2776 del = total - add;
2777 } else {
1b058bc3 2778 del = scale_linear(del, graph_width, max_change);
2eeeef24
JH
2779 add = total - del;
2780 }
6973dcae 2781 }
0911c475
SB
2782 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2783 strbuf_addf(&out, " %*"PRIuMAX"%s",
dc801e71
ZJS
2784 number_width, added + deleted,
2785 added + deleted ? " " : "");
0911c475
SB
2786 show_graph(&out, '+', add, add_c, reset);
2787 show_graph(&out, '-', del, del_c, reset);
2788 strbuf_addch(&out, '\n');
2789 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2790 out.buf, out.len, 0);
2791 strbuf_reset(&out);
c0c77734 2792 }
a20d3c0d
JH
2793
2794 for (i = 0; i < data->nr; i++) {
af0ed819
JH
2795 struct diffstat_file *file = data->files[i];
2796 uintmax_t added = file->added;
2797 uintmax_t deleted = file->deleted;
82dfc2c4
JH
2798
2799 if (file->is_unmerged ||
2800 (!file->is_interesting && (added + deleted == 0))) {
808e1db2
MG
2801 total_files--;
2802 continue;
2803 }
a20d3c0d 2804
82dfc2c4 2805 if (!file->is_binary) {
a20d3c0d
JH
2806 adds += added;
2807 dels += deleted;
2808 }
2809 if (i < count)
2810 continue;
e5f85df8 2811 if (!extra_shown)
0911c475
SB
2812 emit_diff_symbol(options,
2813 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2814 NULL, 0, 0);
e5f85df8 2815 extra_shown = 1;
808e1db2 2816 }
0911c475
SB
2817
2818 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
5a612017 2819 strbuf_release(&out);
6973dcae
JH
2820}
2821
2775d92c 2822static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
ebd124c6
NP
2823{
2824 int i, adds = 0, dels = 0, total_files = data->nr;
2825
2826 if (data->nr == 0)
2827 return;
2828
2829 for (i = 0; i < data->nr; i++) {
e18872b2 2830 int added = data->files[i]->added;
0911c475 2831 int deleted = data->files[i]->deleted;
e18872b2 2832
20c8cde4
JH
2833 if (data->files[i]->is_unmerged ||
2834 (!data->files[i]->is_interesting && (added + deleted == 0))) {
e18872b2 2835 total_files--;
de9658b5 2836 } else if (!data->files[i]->is_binary) { /* don't count bytes */
e18872b2
ZJS
2837 adds += added;
2838 dels += deleted;
ebd124c6 2839 }
ebd124c6 2840 }
0911c475 2841 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
ebd124c6
NP
2842}
2843
4b25d091 2844static void show_numstat(struct diffstat_t *data, struct diff_options *options)
74e2abe5
JH
2845{
2846 int i;
2847
f604652e
JH
2848 if (data->nr == 0)
2849 return;
2850
74e2abe5
JH
2851 for (i = 0; i < data->nr; i++) {
2852 struct diffstat_file *file = data->files[i];
2853
30997bb8 2854 fprintf(options->file, "%s", diff_line_prefix(options));
7be57610 2855
bfddbc5e 2856 if (file->is_binary)
c0c77734 2857 fprintf(options->file, "-\t-\t");
bfddbc5e 2858 else
c0c77734 2859 fprintf(options->file,
0974c117
JK
2860 "%"PRIuMAX"\t%"PRIuMAX"\t",
2861 file->added, file->deleted);
f604652e
JH
2862 if (options->line_termination) {
2863 fill_print_name(file);
2864 if (!file->is_renamed)
c0c77734 2865 write_name_quoted(file->name, options->file,
f604652e
JH
2866 options->line_termination);
2867 else {
c0c77734
DB
2868 fputs(file->print_name, options->file);
2869 putc(options->line_termination, options->file);
f604652e 2870 }
663af342 2871 } else {
f604652e 2872 if (file->is_renamed) {
c0c77734
DB
2873 putc('\0', options->file);
2874 write_name_quoted(file->from_name, options->file, '\0');
f604652e 2875 }
c0c77734 2876 write_name_quoted(file->name, options->file, '\0');
663af342 2877 }
74e2abe5
JH
2878 }
2879}
2880
c04a7155
JH
2881struct dirstat_file {
2882 const char *name;
2883 unsigned long changed;
7df7c019
LT
2884};
2885
c04a7155
JH
2886struct dirstat_dir {
2887 struct dirstat_file *files;
712d2c7d 2888 int alloc, nr, permille, cumulative;
c04a7155
JH
2889};
2890
7be57610
BY
2891static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2892 unsigned long changed, const char *base, int baselen)
7df7c019 2893{
585c0e2e 2894 unsigned long sum_changes = 0;
7df7c019 2895 unsigned int sources = 0;
30997bb8 2896 const char *line_prefix = diff_line_prefix(opt);
7df7c019
LT
2897
2898 while (dir->nr) {
c04a7155 2899 struct dirstat_file *f = dir->files;
7df7c019 2900 int namelen = strlen(f->name);
585c0e2e 2901 unsigned long changes;
7df7c019
LT
2902 char *slash;
2903
2904 if (namelen < baselen)
2905 break;
2906 if (memcmp(f->name, base, baselen))
2907 break;
2908 slash = strchr(f->name + baselen, '/');
2909 if (slash) {
2910 int newbaselen = slash + 1 - f->name;
585c0e2e 2911 changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
7df7c019
LT
2912 sources++;
2913 } else {
585c0e2e 2914 changes = f->changed;
7df7c019
LT
2915 dir->files++;
2916 dir->nr--;
2917 sources += 2;
2918 }
585c0e2e 2919 sum_changes += changes;
7df7c019
LT
2920 }
2921
2922 /*
2923 * We don't report dirstat's for
2924 * - the top level
2925 * - or cases where everything came from a single directory
2926 * under this directory (sources == 1).
2927 */
2928 if (baselen && sources != 1) {
585c0e2e
BW
2929 if (sum_changes) {
2930 int permille = sum_changes * 1000 / changed;
712d2c7d 2931 if (permille >= dir->permille) {
7be57610 2932 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
712d2c7d 2933 permille / 10, permille % 10, baselen, base);
7df7c019
LT
2934 if (!dir->cumulative)
2935 return 0;
2936 }
2937 }
2938 }
585c0e2e 2939 return sum_changes;
7df7c019
LT
2940}
2941
441bca0b
LT
2942static int dirstat_compare(const void *_a, const void *_b)
2943{
2944 const struct dirstat_file *a = _a;
2945 const struct dirstat_file *b = _b;
2946 return strcmp(a->name, b->name);
2947}
2948
c04a7155 2949static void show_dirstat(struct diff_options *options)
7df7c019
LT
2950{
2951 int i;
2952 unsigned long changed;
c04a7155
JH
2953 struct dirstat_dir dir;
2954 struct diff_queue_struct *q = &diff_queued_diff;
2955
2956 dir.files = NULL;
2957 dir.alloc = 0;
2958 dir.nr = 0;
712d2c7d 2959 dir.permille = options->dirstat_permille;
0d1e0e78 2960 dir.cumulative = options->flags.dirstat_cumulative;
7df7c019 2961
7df7c019 2962 changed = 0;
c04a7155
JH
2963 for (i = 0; i < q->nr; i++) {
2964 struct diff_filepair *p = q->queue[i];
2965 const char *name;
2966 unsigned long copied, added, damage;
2967
2ca86714 2968 name = p->two->path ? p->two->path : p->one->path;
c04a7155 2969
d9f62dfa
JK
2970 if (p->one->oid_valid && p->two->oid_valid &&
2971 oideq(&p->one->oid, &p->two->oid)) {
2ff3a803
JH
2972 /*
2973 * The SHA1 has not changed, so pre-/post-content is
2974 * identical. We can therefore skip looking at the
2975 * file contents altogether.
2976 */
2977 damage = 0;
2978 goto found_damage;
2979 }
2980
0d1e0e78 2981 if (options->flags.dirstat_by_file) {
0133dab7
JH
2982 /*
2983 * In --dirstat-by-file mode, we don't really need to
2984 * look at the actual file contents at all.
2985 * The fact that the SHA1 changed is enough for us to
2986 * add this file to the list of results
2987 * (with each file contributing equal damage).
2988 */
2ff3a803 2989 damage = 1;
0133dab7
JH
2990 goto found_damage;
2991 }
c04a7155
JH
2992
2993 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
b78ea5fc
NTND
2994 diff_populate_filespec(options->repo, p->one, 0);
2995 diff_populate_filespec(options->repo, p->two, 0);
2996 diffcore_count_changes(options->repo,
2997 p->one, p->two, NULL, NULL,
c04a7155
JH
2998 &copied, &added);
2999 diff_free_filespec_data(p->one);
3000 diff_free_filespec_data(p->two);
3001 } else if (DIFF_FILE_VALID(p->one)) {
b78ea5fc 3002 diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
c04a7155
JH
3003 copied = added = 0;
3004 diff_free_filespec_data(p->one);
3005 } else if (DIFF_FILE_VALID(p->two)) {
b78ea5fc 3006 diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
c04a7155
JH
3007 copied = 0;
3008 added = p->two->size;
3009 diff_free_filespec_data(p->two);
3010 } else
2b0b551d 3011 continue;
c04a7155
JH
3012
3013 /*
3014 * Original minus copied is the removed material,
3015 * added is the new material. They are both damages
0133dab7 3016 * made to the preimage.
2ff3a803
JH
3017 * If the resulting damage is zero, we know that
3018 * diffcore_count_changes() considers the two entries to
d9f62dfa 3019 * be identical, but since the oid changed, we
2ff3a803
JH
3020 * know that there must have been _some_ kind of change,
3021 * so we force all entries to have damage > 0.
c04a7155
JH
3022 */
3023 damage = (p->one->size - copied) + added;
2ff3a803 3024 if (!damage)
fd33777b 3025 damage = 1;
c04a7155 3026
0133dab7 3027found_damage:
c04a7155
JH
3028 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3029 dir.files[dir.nr].name = name;
3030 dir.files[dir.nr].changed = damage;
3031 changed += damage;
3032 dir.nr++;
7df7c019
LT
3033 }
3034
3035 /* This can happen even with many files, if everything was renames */
3036 if (!changed)
3037 return;
3038
3039 /* Show all directories with more than x% of the changes */
9ed0d8d6 3040 QSORT(dir.files, dir.nr, dirstat_compare);
7be57610 3041 gather_dirstat(options, &dir, changed, "", 0);
7df7c019
LT
3042}
3043
1c57a627
JH
3044static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3045{
3046 int i;
3047 unsigned long changed;
3048 struct dirstat_dir dir;
3049
3050 if (data->nr == 0)
3051 return;
3052
3053 dir.files = NULL;
3054 dir.alloc = 0;
3055 dir.nr = 0;
3056 dir.permille = options->dirstat_permille;
0d1e0e78 3057 dir.cumulative = options->flags.dirstat_cumulative;
1c57a627
JH
3058
3059 changed = 0;
3060 for (i = 0; i < data->nr; i++) {
3061 struct diffstat_file *file = data->files[i];
3062 unsigned long damage = file->added + file->deleted;
3063 if (file->is_binary)
3064 /*
3065 * binary files counts bytes, not lines. Must find some
3066 * way to normalize binary bytes vs. textual lines.
3067 * The following heuristic assumes that there are 64
3068 * bytes per "line".
3069 * This is stupid and ugly, but very cheap...
3070 */
42c78a21 3071 damage = DIV_ROUND_UP(damage, 64);
1c57a627
JH
3072 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3073 dir.files[dir.nr].name = file->name;
3074 dir.files[dir.nr].changed = damage;
3075 changed += damage;
3076 dir.nr++;
3077 }
3078
3079 /* This can happen even with many files, if everything was renames */
3080 if (!changed)
3081 return;
3082
3083 /* Show all directories with more than x% of the changes */
9ed0d8d6 3084 QSORT(dir.files, dir.nr, dirstat_compare);
1c57a627
JH
3085 gather_dirstat(options, &dir, changed, "", 0);
3086}
3087
f604652e
JH
3088static void free_diffstat_info(struct diffstat_t *diffstat)
3089{
3090 int i;
3091 for (i = 0; i < diffstat->nr; i++) {
3092 struct diffstat_file *f = diffstat->files[i];
c905cbc4 3093 free(f->print_name);
f604652e
JH
3094 free(f->name);
3095 free(f->from_name);
3096 free(f);
3097 }
3098 free(diffstat->files);
3099}
3100
88246898 3101struct checkdiff_t {
88246898 3102 const char *filename;
1ba111d1 3103 int lineno;
a757c646 3104 int conflict_marker_size;
1ba111d1 3105 struct diff_options *o;
cf1b7869 3106 unsigned ws_rule;
62c64895 3107 unsigned status;
88246898
JS
3108};
3109
a757c646 3110static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
04954043
JH
3111{
3112 char firstchar;
3113 int cnt;
3114
a757c646 3115 if (len < marker_size + 1)
04954043
JH
3116 return 0;
3117 firstchar = line[0];
3118 switch (firstchar) {
a757c646 3119 case '=': case '>': case '<': case '|':
04954043
JH
3120 break;
3121 default:
3122 return 0;
3123 }
a757c646 3124 for (cnt = 1; cnt < marker_size; cnt++)
04954043
JH
3125 if (line[cnt] != firstchar)
3126 return 0;
a757c646
JH
3127 /* line[1] thru line[marker_size-1] are same as firstchar */
3128 if (len < marker_size + 1 || !isspace(line[marker_size]))
04954043 3129 return 0;
04954043
JH
3130 return 1;
3131}
3132
75ab7630
JK
3133static void checkdiff_consume_hunk(void *priv,
3134 long ob, long on, long nb, long nn,
3135 const char *func, long funclen)
3136
3137{
3138 struct checkdiff_t *data = priv;
3139 data->lineno = nb - 1;
3140}
3141
88246898
JS
3142static void checkdiff_consume(void *priv, char *line, unsigned long len)
3143{
3144 struct checkdiff_t *data = priv;
a757c646 3145 int marker_size = data->conflict_marker_size;
f1c96261
JK
3146 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3147 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3148 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
c1795bb0 3149 char *err;
30997bb8 3150 const char *line_prefix;
7be57610
BY
3151
3152 assert(data->o);
30997bb8 3153 line_prefix = diff_line_prefix(data->o);
88246898
JS
3154
3155 if (line[0] == '+') {
18374e58 3156 unsigned bad;
0ef617f4 3157 data->lineno++;
a757c646 3158 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
04954043
JH
3159 data->status |= 1;
3160 fprintf(data->o->file,
7be57610
BY
3161 "%s%s:%d: leftover conflict marker\n",
3162 line_prefix, data->filename, data->lineno);
04954043 3163 }
8f8841e9 3164 bad = ws_check(line + 1, len - 1, data->ws_rule);
18374e58 3165 if (!bad)
c1795bb0 3166 return;
18374e58
JH
3167 data->status |= bad;
3168 err = whitespace_error_string(bad);
7be57610
BY
3169 fprintf(data->o->file, "%s%s:%d: %s.\n",
3170 line_prefix, data->filename, data->lineno, err);
c1795bb0 3171 free(err);
a3c158d4 3172 emit_line(data->o, set, reset, line, 1);
8f8841e9 3173 ws_check_emit(line + 1, len - 1, data->ws_rule,
1ba111d1 3174 data->o->file, set, reset, ws);
877f23cc 3175 } else if (line[0] == ' ') {
88246898 3176 data->lineno++;
88246898
JS
3177 }
3178}
3179
0660626c
JH
3180static unsigned char *deflate_it(char *data,
3181 unsigned long size,
3182 unsigned long *result_size)
051308f6 3183{
0660626c
JH
3184 int bound;
3185 unsigned char *deflated;
ef49a7a0 3186 git_zstream stream;
0660626c 3187
55bb5c91 3188 git_deflate_init(&stream, zlib_compression_level);
225a6f10 3189 bound = git_deflate_bound(&stream, size);
0660626c
JH
3190 deflated = xmalloc(bound);
3191 stream.next_out = deflated;
3192 stream.avail_out = bound;
3193
3194 stream.next_in = (unsigned char *)data;
3195 stream.avail_in = size;
55bb5c91 3196 while (git_deflate(&stream, Z_FINISH) == Z_OK)
0660626c 3197 ; /* nothing */
55bb5c91 3198 git_deflate_end(&stream);
0660626c
JH
3199 *result_size = stream.total_out;
3200 return deflated;
051308f6
JH
3201}
3202
4eed0ebd
SB
3203static void emit_binary_diff_body(struct diff_options *o,
3204 mmfile_t *one, mmfile_t *two)
051308f6 3205{
0660626c
JH
3206 void *cp;
3207 void *delta;
3208 void *deflated;
3209 void *data;
3210 unsigned long orig_size;
3211 unsigned long delta_size;
3212 unsigned long deflate_size;
3213 unsigned long data_size;
051308f6 3214
0660626c
JH
3215 /* We could do deflated delta, or we could do just deflated two,
3216 * whichever is smaller.
051308f6 3217 */
0660626c
JH
3218 delta = NULL;
3219 deflated = deflate_it(two->ptr, two->size, &deflate_size);
3220 if (one->size && two->size) {
3221 delta = diff_delta(one->ptr, one->size,
3222 two->ptr, two->size,
3223 &delta_size, deflate_size);
3224 if (delta) {
3225 void *to_free = delta;
3226 orig_size = delta_size;
3227 delta = deflate_it(delta, delta_size, &delta_size);
3228 free(to_free);
051308f6
JH
3229 }
3230 }
051308f6 3231
0660626c 3232 if (delta && delta_size < deflate_size) {
ca473cef 3233 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
4eed0ebd
SB
3234 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3235 s, strlen(s), 0);
3236 free(s);
0660626c
JH
3237 free(deflated);
3238 data = delta;
3239 data_size = delta_size;
4eed0ebd
SB
3240 } else {
3241 char *s = xstrfmt("%lu", two->size);
3242 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3243 s, strlen(s), 0);
3244 free(s);
0660626c
JH
3245 free(delta);
3246 data = deflated;
3247 data_size = deflate_size;
3248 }
051308f6 3249
0660626c
JH
3250 /* emit data encoded in base85 */
3251 cp = data;
3252 while (data_size) {
4eed0ebd 3253 int len;
0660626c 3254 int bytes = (52 < data_size) ? 52 : data_size;
4eed0ebd 3255 char line[71];
0660626c 3256 data_size -= bytes;
051308f6
JH
3257 if (bytes <= 26)
3258 line[0] = bytes + 'A' - 1;
3259 else
3260 line[0] = bytes - 26 + 'a' - 1;
3261 encode_85(line + 1, cp, bytes);
1d7f171c 3262 cp = (char *) cp + bytes;
4eed0ebd
SB
3263
3264 len = strlen(line);
3265 line[len++] = '\n';
3266 line[len] = '\0';
3267
3268 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3269 line, len, 0);
051308f6 3270 }
4eed0ebd 3271 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
0660626c 3272 free(data);
051308f6
JH
3273}
3274
4eed0ebd
SB
3275static void emit_binary_diff(struct diff_options *o,
3276 mmfile_t *one, mmfile_t *two)
d4c452f0 3277{
4eed0ebd
SB
3278 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3279 emit_binary_diff_body(o, one, two);
3280 emit_binary_diff_body(o, two, one);
d4c452f0
JH
3281}
3282
b78ea5fc
NTND
3283int diff_filespec_is_binary(struct repository *r,
3284 struct diff_filespec *one)
29a3eefd 3285{
122aa6f9 3286 if (one->is_binary == -1) {
acd00ea0 3287 diff_filespec_load_driver(one, r->index);
122aa6f9
JK
3288 if (one->driver->binary != -1)
3289 one->is_binary = one->driver->binary;
3290 else {
3291 if (!one->data && DIFF_FILE_VALID(one))
b78ea5fc 3292 diff_populate_filespec(r, one, CHECK_BINARY);
6bf3b813 3293 if (one->is_binary == -1 && one->data)
122aa6f9
JK
3294 one->is_binary = buffer_is_binary(one->data,
3295 one->size);
3296 if (one->is_binary == -1)
3297 one->is_binary = 0;
3298 }
3299 }
29a3eefd 3300 return one->is_binary;
6973dcae
JH
3301}
3302
acd00ea0
NTND
3303static const struct userdiff_funcname *
3304diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
f258475a 3305{
acd00ea0 3306 diff_filespec_load_driver(one, o->repo->index);
122aa6f9 3307 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
f258475a
JH
3308}
3309
a5a818ee
JH
3310void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3311{
3312 if (!options->a_prefix)
3313 options->a_prefix = a;
3314 if (!options->b_prefix)
3315 options->b_prefix = b;
3316}
3317
acd00ea0
NTND
3318struct userdiff_driver *get_textconv(struct index_state *istate,
3319 struct diff_filespec *one)
04427ac8
JK
3320{
3321 if (!DIFF_FILE_VALID(one))
3322 return NULL;
d391c0ff 3323
acd00ea0 3324 diff_filespec_load_driver(one, istate);
3813e690 3325 return userdiff_get_textconv(one->driver);
04427ac8
JK
3326}
3327
6973dcae
JH
3328static void builtin_diff(const char *name_a,
3329 const char *name_b,
3330 struct diff_filespec *one,
3331 struct diff_filespec *two,
3332 const char *xfrm_msg,
296c6bb2 3333 int must_show_header,
051308f6 3334 struct diff_options *o,
6973dcae
JH
3335 int complete_rewrite)
3336{
3337 mmfile_t mf1, mf2;
3338 const char *lbl[2];
3339 char *a_one, *b_two;
d9c552f1 3340 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
8f67f8ae 3341 const char *reset = diff_get_color_opt(o, DIFF_RESET);
a5a818ee 3342 const char *a_prefix, *b_prefix;
d9bae1a1
JK
3343 struct userdiff_driver *textconv_one = NULL;
3344 struct userdiff_driver *textconv_two = NULL;
3e97c7c6 3345 struct strbuf header = STRBUF_INIT;
30997bb8 3346 const char *line_prefix = diff_line_prefix(o);
a5a818ee 3347
fd47ae6a 3348 diff_set_mnemonic_prefix(o, "a/", "b/");
0d1e0e78 3349 if (o->flags.reverse_diff) {
fd47ae6a
JK
3350 a_prefix = o->b_prefix;
3351 b_prefix = o->a_prefix;
3352 } else {
3353 a_prefix = o->a_prefix;
3354 b_prefix = o->b_prefix;
3355 }
3356
61cfbc05
JK
3357 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3358 (!one->mode || S_ISGITLINK(one->mode)) &&
3359 (!two->mode || S_ISGITLINK(two->mode))) {
f3597138 3360 show_submodule_summary(o, one->path ? one->path : two->path,
602a283a 3361 &one->oid, &two->oid,
f3597138 3362 two->dirty_submodule);
752c0c24 3363 return;
fd47ae6a
JK
3364 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3365 (!one->mode || S_ISGITLINK(one->mode)) &&
3366 (!two->mode || S_ISGITLINK(two->mode))) {
f3597138 3367 show_submodule_inline_diff(o, one->path ? one->path : two->path,
fd47ae6a 3368 &one->oid, &two->oid,
f3597138 3369 two->dirty_submodule);
fd47ae6a 3370 return;
752c0c24
JS
3371 }
3372
0d1e0e78 3373 if (o->flags.allow_textconv) {
acd00ea0
NTND
3374 textconv_one = get_textconv(o->repo->index, one);
3375 textconv_two = get_textconv(o->repo->index, two);
3aa1f7ca
JK
3376 }
3377
71b989e7
LT
3378 /* Never use a non-valid filename anywhere if at all possible */
3379 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3380 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3381
a5a818ee
JH
3382 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3383 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
6973dcae
JH
3384 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3385 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
d9c552f1 3386 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
6973dcae
JH
3387 if (lbl[0][0] == '/') {
3388 /* /dev/null */
d9c552f1 3389 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
37466447
BW
3390 if (xfrm_msg)
3391 strbuf_addstr(&header, xfrm_msg);
296c6bb2 3392 must_show_header = 1;
6973dcae
JH
3393 }
3394 else if (lbl[1][0] == '/') {
d9c552f1 3395 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
37466447
BW
3396 if (xfrm_msg)
3397 strbuf_addstr(&header, xfrm_msg);
296c6bb2 3398 must_show_header = 1;
6973dcae
JH
3399 }
3400 else {
3401 if (one->mode != two->mode) {
d9c552f1
JK
3402 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3403 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
296c6bb2 3404 must_show_header = 1;
cd112cef 3405 }
37466447
BW
3406 if (xfrm_msg)
3407 strbuf_addstr(&header, xfrm_msg);
3e97c7c6 3408
6973dcae
JH
3409 /*
3410 * we do not run diff between different kind
3411 * of objects.
3412 */
3413 if ((one->mode ^ two->mode) & S_IFMT)
3414 goto free_ab_and_return;
0c01857d 3415 if (complete_rewrite &&
b78ea5fc
NTND
3416 (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3417 (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
a29b0a13
SB
3418 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3419 header.buf, header.len, 0);
3e97c7c6 3420 strbuf_reset(&header);
3aa1f7ca 3421 emit_rewrite_diff(name_a, name_b, one, two,
b78ea5fc 3422 textconv_one, textconv_two, o);
34a5e1a2 3423 o->found_changes = 1;
6973dcae
JH
3424 goto free_ab_and_return;
3425 }
3426 }
3427
467ddc14 3428 if (o->irreversible_delete && lbl[1][0] == '/') {
a29b0a13
SB
3429 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3430 header.len, 0);
467ddc14
JH
3431 strbuf_reset(&header);
3432 goto free_ab_and_return;
0d1e0e78 3433 } else if (!o->flags.text &&
b78ea5fc
NTND
3434 ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3435 (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
4acaaa7a 3436 struct strbuf sb = STRBUF_INIT;
1aaf69e6
NTND
3437 if (!one->data && !two->data &&
3438 S_ISREG(one->mode) && S_ISREG(two->mode) &&
0d1e0e78 3439 !o->flags.binary) {
4a7e27e9 3440 if (oideq(&one->oid, &two->oid)) {
1aaf69e6 3441 if (must_show_header)
a29b0a13
SB
3442 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3443 header.buf, header.len,
3444 0);
1aaf69e6
NTND
3445 goto free_ab_and_return;
3446 }
a29b0a13
SB
3447 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3448 header.buf, header.len, 0);
4acaaa7a
SB
3449 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3450 diff_line_prefix(o), lbl[0], lbl[1]);
3451 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3452 sb.buf, sb.len, 0);
3453 strbuf_release(&sb);
1aaf69e6
NTND
3454 goto free_ab_and_return;
3455 }
b78ea5fc
NTND
3456 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3457 fill_mmfile(o->repo, &mf2, two) < 0)
b3373982 3458 die("unable to read files to diff");
0660626c
JH
3459 /* Quite common confusing case */
3460 if (mf1.size == mf2.size &&
296c6bb2
CC
3461 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3462 if (must_show_header)
a29b0a13
SB
3463 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3464 header.buf, header.len, 0);
0660626c 3465 goto free_ab_and_return;
296c6bb2 3466 }
a29b0a13 3467 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3e97c7c6 3468 strbuf_reset(&header);
0d1e0e78 3469 if (o->flags.binary)
4eed0ebd 3470 emit_binary_diff(o, &mf1, &mf2);
4acaaa7a
SB
3471 else {
3472 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3473 diff_line_prefix(o), lbl[0], lbl[1]);
3474 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3475 sb.buf, sb.len, 0);
3476 strbuf_release(&sb);
3477 }
34a5e1a2 3478 o->found_changes = 1;
467ddc14 3479 } else {
6973dcae
JH
3480 /* Crazy xdl interfaces.. */
3481 const char *diffopts = getenv("GIT_DIFF_OPTS");
ae021d87 3482 const char *v;
6973dcae
JH
3483 xpparam_t xpp;
3484 xdemitconf_t xecfg;
6973dcae 3485 struct emit_callback ecbdata;
be58e70d 3486 const struct userdiff_funcname *pe;
f258475a 3487
b3f01ff2 3488 if (must_show_header) {
a29b0a13
SB
3489 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3490 header.buf, header.len, 0);
3e97c7c6
GB
3491 strbuf_reset(&header);
3492 }
3493
6afaf807
NTND
3494 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3495 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
04427ac8 3496
acd00ea0 3497 pe = diff_funcname_pattern(o, one);
45e7ca0f 3498 if (!pe)
acd00ea0 3499 pe = diff_funcname_pattern(o, two);
6973dcae 3500
9ccd0a88 3501 memset(&xpp, 0, sizeof(xpp));
30b25010 3502 memset(&xecfg, 0, sizeof(xecfg));
cd112cef 3503 memset(&ecbdata, 0, sizeof(ecbdata));
1cdde296
JS
3504 if (o->flags.suppress_diff_headers)
3505 lbl[0] = NULL;
6973dcae 3506 ecbdata.label_path = lbl;
daa0c3d9 3507 ecbdata.color_diff = want_color(o->use_color);
26d024ec 3508 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
690ed843 3509 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
d68fe26f 3510 check_blank_at_eof(&mf1, &mf2, &ecbdata);
a3c158d4 3511 ecbdata.opt = o;
1cdde296
JS
3512 if (header.len && !o->flags.suppress_diff_headers)
3513 ecbdata.header = &header;
582aa00b 3514 xpp.flags = o->xdl_opts;
2477ab2e
JT
3515 xpp.anchors = o->anchors;
3516 xpp.anchors_nr = o->anchors_nr;
ee1e5412 3517 xecfg.ctxlen = o->context;
6d0e674a 3518 xecfg.interhunkctxlen = o->interhunkcontext;
6973dcae 3519 xecfg.flags = XDL_EMIT_FUNCNAMES;
0d1e0e78 3520 if (o->flags.funccontext)
14937c2c 3521 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
45e7ca0f 3522 if (pe)
a013585b 3523 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
6973dcae
JH
3524 if (!diffopts)
3525 ;
ae021d87
JK
3526 else if (skip_prefix(diffopts, "--unified=", &v))
3527 xecfg.ctxlen = strtoul(v, NULL, 10);
3528 else if (skip_prefix(diffopts, "-u", &v))
3529 xecfg.ctxlen = strtoul(v, NULL, 10);
77d1a520
TR
3530 if (o->word_diff)
3531 init_diff_words_data(&ecbdata, o, one, two);
9346d6d1
JK
3532 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3533 &ecbdata, &xpp, &xecfg))
3efb9880 3534 die("unable to generate diff for %s", one->path);
882749a0 3535 if (o->word_diff)
f59a59e2 3536 free_diff_words_data(&ecbdata);
04427ac8
JK
3537 if (textconv_one)
3538 free(mf1.ptr);
3539 if (textconv_two)
3540 free(mf2.ptr);
8cfe5f1c 3541 xdiff_clear_find_func(&xecfg);
6973dcae
JH
3542 }
3543
3544 free_ab_and_return:
3e97c7c6 3545 strbuf_release(&header);
fc3abdf5
JH
3546 diff_free_filespec_data(one);
3547 diff_free_filespec_data(two);
6973dcae
JH
3548 free(a_one);
3549 free(b_two);
3550 return;
3551}
3552
ddf88fa6
NTND
3553static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3554{
3555 if (!is_renamed) {
3556 if (p->status == DIFF_STATUS_ADDED) {
3557 if (S_ISLNK(p->two->mode))
3558 return "new +l";
3559 else if ((p-&